Sync with trunk (r49303)
[reactos.git] / dll / win32 / urlmon / uri.c
index d4d6b47..2aee38a 100644 (file)
@@ -1,5 +1,6 @@
 /*
  * Copyright 2010 Jacek Caban for CodeWeavers
+ * Copyright 2010 Thomas Mullaly
  *
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
 #include "urlmon_main.h"
 #include "wine/debug.h"
 
+#define NO_SHLWAPI_REG
+#include "shlwapi.h"
+
+#define UINT_MAX 0xffffffff
+#define USHORT_MAX 0xffff
+
+#define ALLOW_NULL_TERM_SCHEME          0x01
+#define ALLOW_NULL_TERM_USER_NAME       0x02
+#define ALLOW_NULL_TERM_PASSWORD        0x04
+#define ALLOW_BRACKETLESS_IP_LITERAL    0x08
+#define SKIP_IP_FUTURE_CHECK            0x10
+#define IGNORE_PORT_DELIMITER           0x20
+
 WINE_DEFAULT_DEBUG_CHANNEL(urlmon);
 
-typedef struct {
-    const IUriVtbl  *lpIUriVtbl;
-    LONG ref;
-} Uri;
+static const IID IID_IUriObj = {0x4b364760,0x9f51,0x11df,{0x98,0x1c,0x08,0x00,0x20,0x0c,0x9a,0x66}};
+
+typedef struct {
+    const IUriVtbl  *lpIUriVtbl;
+    LONG ref;
+
+    BSTR            raw_uri;
+
+    /* Information about the canonicalized URI's buffer. */
+    WCHAR           *canon_uri;
+    DWORD           canon_size;
+    DWORD           canon_len;
+    BOOL            display_absolute;
+    DWORD           create_flags;
+
+    INT             scheme_start;
+    DWORD           scheme_len;
+    URL_SCHEME      scheme_type;
+
+    INT             userinfo_start;
+    DWORD           userinfo_len;
+    INT             userinfo_split;
+
+    INT             host_start;
+    DWORD           host_len;
+    Uri_HOST_TYPE   host_type;
+
+    USHORT          port;
+    BOOL            has_port;
+
+    INT             authority_start;
+    DWORD           authority_len;
+
+    INT             domain_offset;
+
+    INT             path_start;
+    DWORD           path_len;
+    INT             extension_offset;
+
+    INT             query_start;
+    DWORD           query_len;
+
+    INT             fragment_start;
+    DWORD           fragment_len;
+} Uri;
+
+typedef struct {
+    const IUriBuilderVtbl  *lpIUriBuilderVtbl;
+    LONG ref;
+
+    Uri *uri;
+    DWORD modified_props;
+
+    WCHAR   *fragment;
+    DWORD   fragment_len;
+
+    WCHAR   *host;
+    DWORD   host_len;
+
+    WCHAR   *password;
+    DWORD   password_len;
+
+    WCHAR   *path;
+    DWORD   path_len;
+
+    BOOL    has_port;
+    DWORD   port;
+
+    WCHAR   *query;
+    DWORD   query_len;
+
+    WCHAR   *scheme;
+    DWORD   scheme_len;
+
+    WCHAR   *username;
+    DWORD   username_len;
+} UriBuilder;
+
+typedef struct {
+    const WCHAR *str;
+    DWORD       len;
+} h16;
+
+typedef struct {
+    /* IPv6 addresses can hold up to 8 h16 components. */
+    h16         components[8];
+    DWORD       h16_count;
+
+    /* An IPv6 can have 1 elision ("::"). */
+    const WCHAR *elision;
+
+    /* An IPv6 can contain 1 IPv4 address as the last 32bits of the address. */
+    const WCHAR *ipv4;
+    DWORD       ipv4_len;
+
+    INT         components_size;
+    INT         elision_size;
+} ipv6_address;
+
+typedef struct {
+    BSTR            uri;
+
+    BOOL            is_relative;
+    BOOL            is_opaque;
+    BOOL            has_implicit_scheme;
+    BOOL            has_implicit_ip;
+    UINT            implicit_ipv4;
+
+    const WCHAR     *scheme;
+    DWORD           scheme_len;
+    URL_SCHEME      scheme_type;
+
+    const WCHAR     *username;
+    DWORD           username_len;
+
+    const WCHAR     *password;
+    DWORD           password_len;
+
+    const WCHAR     *host;
+    DWORD           host_len;
+    Uri_HOST_TYPE   host_type;
+
+    BOOL            has_ipv6;
+    ipv6_address    ipv6_address;
+
+    BOOL            has_port;
+    const WCHAR     *port;
+    DWORD           port_len;
+    DWORD           port_value;
+
+    const WCHAR     *path;
+    DWORD           path_len;
+
+    const WCHAR     *query;
+    DWORD           query_len;
+
+    const WCHAR     *fragment;
+    DWORD           fragment_len;
+} parse_data;
+
+static const CHAR hexDigits[] = "0123456789ABCDEF";
+
+/* List of scheme types/scheme names that are recognized by the IUri interface as of IE 7. */
+static const struct {
+    URL_SCHEME  scheme;
+    WCHAR       scheme_name[16];
+} recognized_schemes[] = {
+    {URL_SCHEME_FTP,            {'f','t','p',0}},
+    {URL_SCHEME_HTTP,           {'h','t','t','p',0}},
+    {URL_SCHEME_GOPHER,         {'g','o','p','h','e','r',0}},
+    {URL_SCHEME_MAILTO,         {'m','a','i','l','t','o',0}},
+    {URL_SCHEME_NEWS,           {'n','e','w','s',0}},
+    {URL_SCHEME_NNTP,           {'n','n','t','p',0}},
+    {URL_SCHEME_TELNET,         {'t','e','l','n','e','t',0}},
+    {URL_SCHEME_WAIS,           {'w','a','i','s',0}},
+    {URL_SCHEME_FILE,           {'f','i','l','e',0}},
+    {URL_SCHEME_MK,             {'m','k',0}},
+    {URL_SCHEME_HTTPS,          {'h','t','t','p','s',0}},
+    {URL_SCHEME_SHELL,          {'s','h','e','l','l',0}},
+    {URL_SCHEME_SNEWS,          {'s','n','e','w','s',0}},
+    {URL_SCHEME_LOCAL,          {'l','o','c','a','l',0}},
+    {URL_SCHEME_JAVASCRIPT,     {'j','a','v','a','s','c','r','i','p','t',0}},
+    {URL_SCHEME_VBSCRIPT,       {'v','b','s','c','r','i','p','t',0}},
+    {URL_SCHEME_ABOUT,          {'a','b','o','u','t',0}},
+    {URL_SCHEME_RES,            {'r','e','s',0}},
+    {URL_SCHEME_MSSHELLROOTED,  {'m','s','-','s','h','e','l','l','-','r','o','o','t','e','d',0}},
+    {URL_SCHEME_MSSHELLIDLIST,  {'m','s','-','s','h','e','l','l','-','i','d','l','i','s','t',0}},
+    {URL_SCHEME_MSHELP,         {'h','c','p',0}},
+    {URL_SCHEME_WILDCARD,       {'*',0}}
+};
+
+/* List of default ports Windows recognizes. */
+static const struct {
+    URL_SCHEME  scheme;
+    USHORT      port;
+} default_ports[] = {
+    {URL_SCHEME_FTP,    21},
+    {URL_SCHEME_HTTP,   80},
+    {URL_SCHEME_GOPHER, 70},
+    {URL_SCHEME_NNTP,   119},
+    {URL_SCHEME_TELNET, 23},
+    {URL_SCHEME_WAIS,   210},
+    {URL_SCHEME_HTTPS,  443},
+};
+
+/* List of 3 character top level domain names Windows seems to recognize.
+ * There might be more, but, these are the only ones I've found so far.
+ */
+static const struct {
+    WCHAR tld_name[4];
+} recognized_tlds[] = {
+    {{'c','o','m',0}},
+    {{'e','d','u',0}},
+    {{'g','o','v',0}},
+    {{'i','n','t',0}},
+    {{'m','i','l',0}},
+    {{'n','e','t',0}},
+    {{'o','r','g',0}}
+};
+
+static Uri *get_uri_obj(IUri *uri)
+{
+    Uri *ret;
+    HRESULT hres;
+
+    hres = IUri_QueryInterface(uri, &IID_IUriObj, (void**)&ret);
+    return SUCCEEDED(hres) ? ret : NULL;
+}
+
+static inline BOOL is_alpha(WCHAR val) {
+       return ((val >= 'a' && val <= 'z') || (val >= 'A' && val <= 'Z'));
+}
+
+static inline BOOL is_num(WCHAR val) {
+       return (val >= '0' && val <= '9');
+}
+
+static inline BOOL is_drive_path(const WCHAR *str) {
+    return (is_alpha(str[0]) && (str[1] == ':' || str[1] == '|'));
+}
+
+static inline BOOL is_unc_path(const WCHAR *str) {
+    return (str[0] == '\\' && str[0] == '\\');
+}
+
+static inline BOOL is_forbidden_dos_path_char(WCHAR val) {
+    return (val == '>' || val == '<' || val == '\"');
+}
+
+/* A URI is implicitly a file path if it begins with
+ * a drive letter (eg X:) or starts with "\\" (UNC path).
+ */
+static inline BOOL is_implicit_file_path(const WCHAR *str) {
+    return (is_unc_path(str) || (is_alpha(str[0]) && str[1] == ':'));
+}
+
+/* Checks if the URI is a hierarchical URI. A hierarchical
+ * URI is one that has "//" after the scheme.
+ */
+static BOOL check_hierarchical(const WCHAR **ptr) {
+    const WCHAR *start = *ptr;
+
+    if(**ptr != '/')
+        return FALSE;
+
+    ++(*ptr);
+    if(**ptr != '/') {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    return TRUE;
+}
+
+/* unreserved    = ALPHA / DIGIT / "-" / "." / "_" / "~" */
+static inline BOOL is_unreserved(WCHAR val) {
+    return (is_alpha(val) || is_num(val) || val == '-' || val == '.' ||
+            val == '_' || val == '~');
+}
+
+/* sub-delims    = "!" / "$" / "&" / "'" / "(" / ")"
+ *               / "*" / "+" / "," / ";" / "="
+ */
+static inline BOOL is_subdelim(WCHAR val) {
+    return (val == '!' || val == '$' || val == '&' ||
+            val == '\'' || val == '(' || val == ')' ||
+            val == '*' || val == '+' || val == ',' ||
+            val == ';' || val == '=');
+}
+
+/* gen-delims  = ":" / "/" / "?" / "#" / "[" / "]" / "@" */
+static inline BOOL is_gendelim(WCHAR val) {
+    return (val == ':' || val == '/' || val == '?' ||
+            val == '#' || val == '[' || val == ']' ||
+            val == '@');
+}
+
+/* Characters that delimit the end of the authority
+ * section of a URI. Sometimes a '\\' is considered
+ * an authority delimeter.
+ */
+static inline BOOL is_auth_delim(WCHAR val, BOOL acceptSlash) {
+    return (val == '#' || val == '/' || val == '?' ||
+            val == '\0' || (acceptSlash && val == '\\'));
+}
+
+/* reserved = gen-delims / sub-delims */
+static inline BOOL is_reserved(WCHAR val) {
+    return (is_subdelim(val) || is_gendelim(val));
+}
+
+static inline BOOL is_hexdigit(WCHAR val) {
+    return ((val >= 'a' && val <= 'f') ||
+            (val >= 'A' && val <= 'F') ||
+            (val >= '0' && val <= '9'));
+}
+
+static inline BOOL is_path_delim(WCHAR val) {
+    return (!val || val == '#' || val == '?');
+}
+
+/* List of schemes types Windows seems to expect to be hierarchical. */
+static inline BOOL is_hierarchical_scheme(URL_SCHEME type) {
+    return(type == URL_SCHEME_HTTP || type == URL_SCHEME_FTP ||
+           type == URL_SCHEME_GOPHER || type == URL_SCHEME_NNTP ||
+           type == URL_SCHEME_TELNET || type == URL_SCHEME_WAIS ||
+           type == URL_SCHEME_FILE || type == URL_SCHEME_HTTPS ||
+           type == URL_SCHEME_RES);
+}
+
+/* Checks if 'flags' contains an invalid combination of Uri_CREATE flags. */
+static inline BOOL has_invalid_flag_combination(DWORD flags) {
+    return((flags & Uri_CREATE_DECODE_EXTRA_INFO && flags & Uri_CREATE_NO_DECODE_EXTRA_INFO) ||
+           (flags & Uri_CREATE_CANONICALIZE && flags & Uri_CREATE_NO_CANONICALIZE) ||
+           (flags & Uri_CREATE_CRACK_UNKNOWN_SCHEMES && flags & Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES) ||
+           (flags & Uri_CREATE_PRE_PROCESS_HTML_URI && flags & Uri_CREATE_NO_PRE_PROCESS_HTML_URI) ||
+           (flags & Uri_CREATE_IE_SETTINGS && flags & Uri_CREATE_NO_IE_SETTINGS));
+}
+
+/* Applies each default Uri_CREATE flags to 'flags' if it
+ * doesn't cause a flag conflict.
+ */
+static void apply_default_flags(DWORD *flags) {
+    if(!(*flags & Uri_CREATE_NO_CANONICALIZE))
+        *flags |= Uri_CREATE_CANONICALIZE;
+    if(!(*flags & Uri_CREATE_NO_DECODE_EXTRA_INFO))
+        *flags |= Uri_CREATE_DECODE_EXTRA_INFO;
+    if(!(*flags & Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES))
+        *flags |= Uri_CREATE_CRACK_UNKNOWN_SCHEMES;
+    if(!(*flags & Uri_CREATE_NO_PRE_PROCESS_HTML_URI))
+        *flags |= Uri_CREATE_PRE_PROCESS_HTML_URI;
+    if(!(*flags & Uri_CREATE_IE_SETTINGS))
+        *flags |= Uri_CREATE_NO_IE_SETTINGS;
+}
+
+/* Determines if the URI is hierarchical using the information already parsed into
+ * data and using the current location of parsing in the URI string.
+ *
+ * Windows considers a URI hierarchical if on of the following is true:
+ *  A.) It's a wildcard scheme.
+ *  B.) It's an implicit file scheme.
+ *  C.) It's a known hierarchical scheme and it has two '\\' after the scheme name.
+ *      (the '\\' will be converted into "//" during canonicalization).
+ *  D.) It's not a relative URI and "//" appears after the scheme name.
+ */
+static inline BOOL is_hierarchical_uri(const WCHAR **ptr, const parse_data *data) {
+    const WCHAR *start = *ptr;
+
+    if(data->scheme_type == URL_SCHEME_WILDCARD)
+        return TRUE;
+    else if(data->scheme_type == URL_SCHEME_FILE && data->has_implicit_scheme)
+        return TRUE;
+    else if(is_hierarchical_scheme(data->scheme_type) && (*ptr)[0] == '\\' && (*ptr)[1] == '\\') {
+        *ptr += 2;
+        return TRUE;
+    } else if(!data->is_relative && check_hierarchical(ptr))
+        return TRUE;
+
+    *ptr = start;
+    return FALSE;
+}
+
+/* Checks if the two Uri's are logically equivalent. It's a simple
+ * comparison, since they are both of type Uri, and it can access
+ * the properties of each Uri directly without the need to go
+ * through the "IUri_Get*" interface calls.
+ */
+static BOOL are_equal_simple(const Uri *a, const Uri *b) {
+    if(a->scheme_type == b->scheme_type) {
+        const BOOL known_scheme = a->scheme_type != URL_SCHEME_UNKNOWN;
+        const BOOL are_hierarchical =
+                (a->authority_start > -1 && b->authority_start > -1);
+
+        if(a->scheme_type == URL_SCHEME_FILE) {
+            if(a->canon_len == b->canon_len)
+                return !StrCmpIW(a->canon_uri, b->canon_uri);
+        }
+
+        /* Only compare the scheme names (if any) if their unknown scheme types. */
+        if(!known_scheme) {
+            if((a->scheme_start > -1 && b->scheme_start > -1) &&
+               (a->scheme_len == b->scheme_len)) {
+                /* Make sure the schemes are the same. */
+                if(StrCmpNW(a->canon_uri+a->scheme_start, b->canon_uri+b->scheme_start, a->scheme_len))
+                    return FALSE;
+            } else if(a->scheme_len != b->scheme_len)
+                /* One of the Uri's has a scheme name, while the other doesn't. */
+                return FALSE;
+        }
+
+        /* If they have a userinfo component, perform case sensitive compare. */
+        if((a->userinfo_start > -1 && b->userinfo_start > -1) &&
+           (a->userinfo_len == b->userinfo_len)) {
+            if(StrCmpNW(a->canon_uri+a->userinfo_start, b->canon_uri+b->userinfo_start, a->userinfo_len))
+                return FALSE;
+        } else if(a->userinfo_len != b->userinfo_len)
+            /* One of the Uri's had a userinfo, while the other one doesn't. */
+            return FALSE;
+
+        /* Check if they have a host name. */
+        if((a->host_start > -1 && b->host_start > -1) &&
+           (a->host_len == b->host_len)) {
+            /* Perform a case insensitive compare if they are a known scheme type. */
+            if(known_scheme) {
+                if(StrCmpNIW(a->canon_uri+a->host_start, b->canon_uri+b->host_start, a->host_len))
+                    return FALSE;
+            } else if(StrCmpNW(a->canon_uri+a->host_start, b->canon_uri+b->host_start, a->host_len))
+                return FALSE;
+        } else if(a->host_len != b->host_len)
+            /* One of the Uri's had a host, while the other one didn't. */
+            return FALSE;
+
+        if(a->has_port && b->has_port) {
+            if(a->port != b->port)
+                return FALSE;
+        } else if(a->has_port || b->has_port)
+            /* One had a port, while the other one didn't. */
+            return FALSE;
+
+        /* Windows is weird with how it handles paths. For example
+         * One URI could be "http://google.com" (after canonicalization)
+         * and one could be "http://google.com/" and the IsEqual function
+         * would still evaluate to TRUE, but, only if they are both hierarchical
+         * URIs.
+         */
+        if((a->path_start > -1 && b->path_start > -1) &&
+           (a->path_len == b->path_len)) {
+            if(StrCmpNW(a->canon_uri+a->path_start, b->canon_uri+b->path_start, a->path_len))
+                return FALSE;
+        } else if(are_hierarchical && a->path_len == -1 && b->path_len == 0) {
+            if(*(a->canon_uri+a->path_start) != '/')
+                return FALSE;
+        } else if(are_hierarchical && b->path_len == 1 && a->path_len == 0) {
+            if(*(b->canon_uri+b->path_start) != '/')
+                return FALSE;
+        } else if(a->path_len != b->path_len)
+            return FALSE;
+
+        /* Compare the query strings of the two URIs. */
+        if((a->query_start > -1 && b->query_start > -1) &&
+           (a->query_len == b->query_len)) {
+            if(StrCmpNW(a->canon_uri+a->query_start, b->canon_uri+b->query_start, a->query_len))
+                return FALSE;
+        } else if(a->query_len != b->query_len)
+            return FALSE;
+
+        if((a->fragment_start > -1 && b->fragment_start > -1) &&
+           (a->fragment_len == b->fragment_len)) {
+            if(StrCmpNW(a->canon_uri+a->fragment_start, b->canon_uri+b->fragment_start, a->fragment_len))
+                return FALSE;
+        } else if(a->fragment_len != b->fragment_len)
+            return FALSE;
+
+        /* If we get here, the two URIs are equivalent. */
+        return TRUE;
+    }
+
+    return FALSE;
+}
+
+/* Computes the size of the given IPv6 address.
+ * Each h16 component is 16bits, if there is an IPv4 address, it's
+ * 32bits. If there's an elision it can be 16bits to 128bits, depending
+ * on the number of other components.
+ *
+ * Modeled after google-url's CheckIPv6ComponentsSize function
+ */
+static void compute_ipv6_comps_size(ipv6_address *address) {
+    address->components_size = address->h16_count * 2;
+
+    if(address->ipv4)
+        /* IPv4 address is 4 bytes. */
+        address->components_size += 4;
+
+    if(address->elision) {
+        /* An elision can be anywhere from 2 bytes up to 16 bytes.
+         * It size depends on the size of the h16 and IPv4 components.
+         */
+        address->elision_size = 16 - address->components_size;
+        if(address->elision_size < 2)
+            address->elision_size = 2;
+    } else
+        address->elision_size = 0;
+}
+
+/* Taken from dlls/jscript/lex.c */
+static int hex_to_int(WCHAR val) {
+    if(val >= '0' && val <= '9')
+        return val - '0';
+    else if(val >= 'a' && val <= 'f')
+        return val - 'a' + 10;
+    else if(val >= 'A' && val <= 'F')
+        return val - 'A' + 10;
+
+    return -1;
+}
+
+/* Helper function for converting a percent encoded string
+ * representation of a WCHAR value into its actual WCHAR value. If
+ * the two characters following the '%' aren't valid hex values then
+ * this function returns the NULL character.
+ *
+ * Eg.
+ *  "%2E" will result in '.' being returned by this function.
+ */
+static WCHAR decode_pct_val(const WCHAR *ptr) {
+    WCHAR ret = '\0';
+
+    if(*ptr == '%' && is_hexdigit(*(ptr + 1)) && is_hexdigit(*(ptr + 2))) {
+        INT a = hex_to_int(*(ptr + 1));
+        INT b = hex_to_int(*(ptr + 2));
+
+        ret = a << 4;
+        ret += b;
+    }
+
+    return ret;
+}
+
+/* Helper function for percent encoding a given character
+ * and storing the encoded value into a given buffer (dest).
+ *
+ * It's up to the calling function to ensure that there is
+ * at least enough space in 'dest' for the percent encoded
+ * value to be stored (so dest + 3 spaces available).
+ */
+static inline void pct_encode_val(WCHAR val, WCHAR *dest) {
+    dest[0] = '%';
+    dest[1] = hexDigits[(val >> 4) & 0xf];
+    dest[2] = hexDigits[val & 0xf];
+}
+
+/* Scans the range of characters [str, end] and returns the last occurrence
+ * of 'ch' or returns NULL.
+ */
+static const WCHAR *str_last_of(const WCHAR *str, const WCHAR *end, WCHAR ch) {
+    const WCHAR *ptr = end;
+
+    while(ptr >= str) {
+        if(*ptr == ch)
+            return ptr;
+        --ptr;
+    }
+
+    return NULL;
+}
+
+/* Attempts to parse the domain name from the host.
+ *
+ * This function also includes the Top-level Domain (TLD) name
+ * of the host when it tries to find the domain name. If it finds
+ * a valid domain name it will assign 'domain_start' the offset
+ * into 'host' where the domain name starts.
+ *
+ * It's implied that if a domain name its range is implied to be
+ * [host+domain_start, host+host_len).
+ */
+static void find_domain_name(const WCHAR *host, DWORD host_len,
+                             INT *domain_start) {
+    const WCHAR *last_tld, *sec_last_tld, *end;
+
+    end = host+host_len-1;
+
+    *domain_start = -1;
+
+    /* There has to be at least enough room for a '.' followed by a
+     * 3 character TLD for a domain to even exist in the host name.
+     */
+    if(host_len < 4)
+        return;
+
+    last_tld = str_last_of(host, end, '.');
+    if(!last_tld)
+        /* http://hostname -> has no domain name. */
+        return;
+
+    sec_last_tld = str_last_of(host, last_tld-1, '.');
+    if(!sec_last_tld) {
+        /* If the '.' is at the beginning of the host there
+         * has to be at least 3 characters in the TLD for it
+         * to be valid.
+         *  Ex: .com -> .com as the domain name.
+         *      .co  -> has no domain name.
+         */
+        if(last_tld-host == 0) {
+            if(end-(last_tld-1) < 3)
+                return;
+        } else if(last_tld-host == 3) {
+            DWORD i;
+
+            /* If there's three characters in front of last_tld and
+             * they are on the list of recognized TLDs, then this
+             * host doesn't have a domain (since the host only contains
+             * a TLD name.
+             *  Ex: edu.uk -> has no domain name.
+             *      foo.uk -> foo.uk as the domain name.
+             */
+            for(i = 0; i < sizeof(recognized_tlds)/sizeof(recognized_tlds[0]); ++i) {
+                if(!StrCmpNIW(host, recognized_tlds[i].tld_name, 3))
+                    return;
+            }
+        } else if(last_tld-host < 3)
+            /* Anything less than 3 characters is considered part
+             * of the TLD name.
+             *  Ex: ak.uk -> Has no domain name.
+             */
+            return;
+
+        /* Otherwise the domain name is the whole host name. */
+        *domain_start = 0;
+    } else if(end+1-last_tld > 3) {
+        /* If the last_tld has more than 3 characters, then it's automatically
+         * considered the TLD of the domain name.
+         *  Ex: www.winehq.org.uk.test -> uk.test as the domain name.
+         */
+        *domain_start = (sec_last_tld+1)-host;
+    } else if(last_tld - (sec_last_tld+1) < 4) {
+        DWORD i;
+        /* If the sec_last_tld is 3 characters long it HAS to be on the list of
+         * recognized to still be considered part of the TLD name, otherwise
+         * its considered the domain name.
+         *  Ex: www.google.com.uk -> google.com.uk as the domain name.
+         *      www.google.foo.uk -> foo.uk as the domain name.
+         */
+        if(last_tld - (sec_last_tld+1) == 3) {
+            for(i = 0; i < sizeof(recognized_tlds)/sizeof(recognized_tlds[0]); ++i) {
+                if(!StrCmpNIW(sec_last_tld+1, recognized_tlds[i].tld_name, 3)) {
+                    const WCHAR *domain = str_last_of(host, sec_last_tld-1, '.');
+
+                    if(!domain)
+                        *domain_start = 0;
+                    else
+                        *domain_start = (domain+1) - host;
+                    TRACE("Found domain name %s\n", debugstr_wn(host+*domain_start,
+                                                        (host+host_len)-(host+*domain_start)));
+                    return;
+                }
+            }
+
+            *domain_start = (sec_last_tld+1)-host;
+        } else {
+            /* Since the sec_last_tld is less than 3 characters it's considered
+             * part of the TLD.
+             *  Ex: www.google.fo.uk -> google.fo.uk as the domain name.
+             */
+            const WCHAR *domain = str_last_of(host, sec_last_tld-1, '.');
+
+            if(!domain)
+                *domain_start = 0;
+            else
+                *domain_start = (domain+1) - host;
+        }
+    } else {
+        /* The second to last TLD has more than 3 characters making it
+         * the domain name.
+         *  Ex: www.google.test.us -> test.us as the domain name.
+         */
+        *domain_start = (sec_last_tld+1)-host;
+    }
+
+    TRACE("Found domain name %s\n", debugstr_wn(host+*domain_start,
+                                        (host+host_len)-(host+*domain_start)));
+}
+
+/* Removes the dot segments from a hierarchical URIs path component. This
+ * function performs the removal in place.
+ *
+ * This is a modified version of Qt's QUrl function "removeDotsFromPath".
+ *
+ * This function returns the new length of the path string.
+ */
+static DWORD remove_dot_segments(WCHAR *path, DWORD path_len) {
+    WCHAR *out = path;
+    const WCHAR *in = out;
+    const WCHAR *end = out + path_len;
+    DWORD len;
+
+    while(in < end) {
+        /* A.  if the input buffer begins with a prefix of "/./" or "/.",
+         *     where "." is a complete path segment, then replace that
+         *     prefix with "/" in the input buffer; otherwise,
+         */
+        if(in <= end - 3 && in[0] == '/' && in[1] == '.' && in[2] == '/') {
+            in += 2;
+            continue;
+        } else if(in == end - 2 && in[0] == '/' && in[1] == '.') {
+            *out++ = '/';
+            in += 2;
+            break;
+        }
+
+        /* B.  if the input buffer begins with a prefix of "/../" or "/..",
+         *     where ".." is a complete path segment, then replace that
+         *     prefix with "/" in the input buffer and remove the last
+         *     segment and its preceding "/" (if any) from the output
+         *     buffer; otherwise,
+         */
+        if(in <= end - 4 && in[0] == '/' && in[1] == '.' && in[2] == '.' && in[3] == '/') {
+            while(out > path && *(--out) != '/');
+
+            in += 3;
+            continue;
+        } else if(in == end - 3 && in[0] == '/' && in[1] == '.' && in[2] == '.') {
+            while(out > path && *(--out) != '/');
+
+            if(*out == '/')
+                ++out;
+
+            in += 3;
+            break;
+        }
+
+        /* C.  move the first path segment in the input buffer to the end of
+         *     the output buffer, including the initial "/" character (if
+         *     any) and any subsequent characters up to, but not including,
+         *     the next "/" character or the end of the input buffer.
+         */
+        *out++ = *in++;
+        while(in < end && *in != '/')
+            *out++ = *in++;
+    }
+
+    len = out - path;
+    TRACE("(%p %d): Path after dot segments removed %s len=%d\n", path, path_len,
+        debugstr_wn(path, len), len);
+    return len;
+}
+
+/* Attempts to find the file extension in a given path. */
+static INT find_file_extension(const WCHAR *path, DWORD path_len) {
+    const WCHAR *end;
+
+    for(end = path+path_len-1; end >= path && *end != '/' && *end != '\\'; --end) {
+        if(*end == '.')
+            return end-path;
+    }
+
+    return -1;
+}
+
+/* Computes the location where the elision should occur in the IPv6
+ * address using the numerical values of each component stored in
+ * 'values'. If the address shouldn't contain an elision then 'index'
+ * is assigned -1 as it's value. Otherwise 'index' will contain the
+ * starting index (into values) where the elision should be, and 'count'
+ * will contain the number of cells the elision covers.
+ *
+ * NOTES:
+ *  Windows will expand an elision if the elision only represents 1 h16
+ *  component of the URI.
+ *
+ *  Ex: [1::2:3:4:5:6:7] -> [1:0:2:3:4:5:6:7]
+ *
+ *  If the IPv6 address contains an IPv4 address, the IPv4 address is also
+ *  considered for being included as part of an elision if all it's components
+ *  are zeros.
+ *
+ *  Ex: [1:2:3:4:5:6:0.0.0.0] -> [1:2:3:4:5:6::]
+ */
+static void compute_elision_location(const ipv6_address *address, const USHORT values[8],
+                                     INT *index, DWORD *count) {
+    DWORD i, max_len, cur_len;
+    INT max_index, cur_index;
+
+    max_len = cur_len = 0;
+    max_index = cur_index = -1;
+    for(i = 0; i < 8; ++i) {
+        BOOL check_ipv4 = (address->ipv4 && i == 6);
+        BOOL is_end = (check_ipv4 || i == 7);
+
+        if(check_ipv4) {
+            /* Check if the IPv4 address contains only zeros. */
+            if(values[i] == 0 && values[i+1] == 0) {
+                if(cur_index == -1)
+                    cur_index = i;
+
+                cur_len += 2;
+                ++i;
+            }
+        } else if(values[i] == 0) {
+            if(cur_index == -1)
+                cur_index = i;
+
+            ++cur_len;
+        }
+
+        if(is_end || values[i] != 0) {
+            /* We only consider it for an elision if it's
+             * more than 1 component long.
+             */
+            if(cur_len > 1 && cur_len > max_len) {
+                /* Found the new elision location. */
+                max_len = cur_len;
+                max_index = cur_index;
+            }
+
+            /* Reset the current range for the next range of zeros. */
+            cur_index = -1;
+            cur_len = 0;
+        }
+    }
+
+    *index = max_index;
+    *count = max_len;
+}
+
+/* Removes all the leading and trailing white spaces or
+ * control characters from the URI and removes all control
+ * characters inside of the URI string.
+ */
+static BSTR pre_process_uri(LPCWSTR uri) {
+    BSTR ret;
+    DWORD len;
+    const WCHAR *start, *end;
+    WCHAR *buf, *ptr;
+
+    len = lstrlenW(uri);
+
+    start = uri;
+    /* Skip leading controls and whitespace. */
+    while(iscntrlW(*start) || isspaceW(*start)) ++start;
+
+    end = uri+len-1;
+    if(start == end)
+        /* URI consisted only of control/whitespace. */
+        ret = SysAllocStringLen(NULL, 0);
+    else {
+        while(iscntrlW(*end) || isspaceW(*end)) --end;
+
+        buf = heap_alloc(((end+1)-start)*sizeof(WCHAR));
+        if(!buf)
+            return NULL;
+
+        for(ptr = buf; start < end+1; ++start) {
+            if(!iscntrlW(*start))
+                *ptr++ = *start;
+        }
+
+        ret = SysAllocStringLen(buf, ptr-buf);
+        heap_free(buf);
+    }
+
+    return ret;
+}
+
+/* Converts the specified IPv4 address into an uint value.
+ *
+ * This function assumes that the IPv4 address has already been validated.
+ */
+static UINT ipv4toui(const WCHAR *ip, DWORD len) {
+    UINT ret = 0;
+    DWORD comp_value = 0;
+    const WCHAR *ptr;
+
+    for(ptr = ip; ptr < ip+len; ++ptr) {
+        if(*ptr == '.') {
+            ret <<= 8;
+            ret += comp_value;
+            comp_value = 0;
+        } else
+            comp_value = comp_value*10 + (*ptr-'0');
+    }
+
+    ret <<= 8;
+    ret += comp_value;
+
+    return ret;
+}
+
+/* Converts an IPv4 address in numerical form into it's fully qualified
+ * string form. This function returns the number of characters written
+ * to 'dest'. If 'dest' is NULL this function will return the number of
+ * characters that would have been written.
+ *
+ * It's up to the caller to ensure there's enough space in 'dest' for the
+ * address.
+ */
+static DWORD ui2ipv4(WCHAR *dest, UINT address) {
+    static const WCHAR formatW[] =
+        {'%','u','.','%','u','.','%','u','.','%','u',0};
+    DWORD ret = 0;
+    UCHAR digits[4];
+
+    digits[0] = (address >> 24) & 0xff;
+    digits[1] = (address >> 16) & 0xff;
+    digits[2] = (address >> 8) & 0xff;
+    digits[3] = address & 0xff;
+
+    if(!dest) {
+        WCHAR tmp[16];
+        ret = sprintfW(tmp, formatW, digits[0], digits[1], digits[2], digits[3]);
+    } else
+        ret = sprintfW(dest, formatW, digits[0], digits[1], digits[2], digits[3]);
+
+    return ret;
+}
+
+/* Converts an h16 component (from an IPv6 address) into it's
+ * numerical value.
+ *
+ * This function assumes that the h16 component has already been validated.
+ */
+static USHORT h16tous(h16 component) {
+    DWORD i;
+    USHORT ret = 0;
+
+    for(i = 0; i < component.len; ++i) {
+        ret <<= 4;
+        ret += hex_to_int(component.str[i]);
+    }
+
+    return ret;
+}
+
+/* Converts an IPv6 address into it's 128 bits (16 bytes) numerical value.
+ *
+ * This function assumes that the ipv6_address has already been validated.
+ */
+static BOOL ipv6_to_number(const ipv6_address *address, USHORT number[8]) {
+    DWORD i, cur_component = 0;
+    BOOL already_passed_elision = FALSE;
+
+    for(i = 0; i < address->h16_count; ++i) {
+        if(address->elision) {
+            if(address->components[i].str > address->elision && !already_passed_elision) {
+                /* Means we just passed the elision and need to add it's values to
+                 * 'number' before we do anything else.
+                 */
+                DWORD j = 0;
+                for(j = 0; j < address->elision_size; j+=2)
+                    number[cur_component++] = 0;
+
+                already_passed_elision = TRUE;
+            }
+        }
+
+        number[cur_component++] = h16tous(address->components[i]);
+    }
+
+    /* Case when the elision appears after the h16 components. */
+    if(!already_passed_elision && address->elision) {
+        for(i = 0; i < address->elision_size; i+=2)
+            number[cur_component++] = 0;
+        already_passed_elision = TRUE;
+    }
+
+    if(address->ipv4) {
+        UINT value = ipv4toui(address->ipv4, address->ipv4_len);
+
+        if(cur_component != 6) {
+            ERR("(%p %p): Failed sanity check with %d\n", address, number, cur_component);
+            return FALSE;
+        }
+
+        number[cur_component++] = (value >> 16) & 0xffff;
+        number[cur_component] = value & 0xffff;
+    }
+
+    return TRUE;
+}
+
+/* Checks if the characters pointed to by 'ptr' are
+ * a percent encoded data octet.
+ *
+ * pct-encoded = "%" HEXDIG HEXDIG
+ */
+static BOOL check_pct_encoded(const WCHAR **ptr) {
+    const WCHAR *start = *ptr;
+
+    if(**ptr != '%')
+        return FALSE;
+
+    ++(*ptr);
+    if(!is_hexdigit(**ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    if(!is_hexdigit(**ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    return TRUE;
+}
+
+/* dec-octet   = DIGIT                 ; 0-9
+ *             / %x31-39 DIGIT         ; 10-99
+ *             / "1" 2DIGIT            ; 100-199
+ *             / "2" %x30-34 DIGIT     ; 200-249
+ *             / "25" %x30-35          ; 250-255
+ */
+static BOOL check_dec_octet(const WCHAR **ptr) {
+    const WCHAR *c1, *c2, *c3;
+
+    c1 = *ptr;
+    /* A dec-octet must be at least 1 digit long. */
+    if(*c1 < '0' || *c1 > '9')
+        return FALSE;
+
+    ++(*ptr);
+
+    c2 = *ptr;
+    /* Since the 1 digit requirment was meet, it doesn't
+     * matter if this is a DIGIT value, it's considered a
+     * dec-octet.
+     */
+    if(*c2 < '0' || *c2 > '9')
+        return TRUE;
+
+    ++(*ptr);
+
+    c3 = *ptr;
+    /* Same explanation as above. */
+    if(*c3 < '0' || *c3 > '9')
+        return TRUE;
+
+    /* Anything > 255 isn't a valid IP dec-octet. */
+    if(*c1 >= '2' && *c2 >= '5' && *c3 >= '5') {
+        *ptr = c1;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    return TRUE;
+}
+
+/* Checks if there is an implicit IPv4 address in the host component of the URI.
+ * The max value of an implicit IPv4 address is UINT_MAX.
+ *
+ *  Ex:
+ *      "234567" would be considered an implicit IPv4 address.
+ */
+static BOOL check_implicit_ipv4(const WCHAR **ptr, UINT *val) {
+    const WCHAR *start = *ptr;
+    ULONGLONG ret = 0;
+    *val = 0;
+
+    while(is_num(**ptr)) {
+        ret = ret*10 + (**ptr - '0');
+
+        if(ret > UINT_MAX) {
+            *ptr = start;
+            return FALSE;
+        }
+        ++(*ptr);
+    }
+
+    if(*ptr == start)
+        return FALSE;
+
+    *val = ret;
+    return TRUE;
+}
+
+/* Checks if the string contains an IPv4 address.
+ *
+ * This function has a strict mode or a non-strict mode of operation
+ * When 'strict' is set to FALSE this function will return TRUE if
+ * the string contains at least 'dec-octet "." dec-octet' since partial
+ * IPv4 addresses will be normalized out into full IPv4 addresses. When
+ * 'strict' is set this function expects there to be a full IPv4 address.
+ *
+ * IPv4address = dec-octet "." dec-octet "." dec-octet "." dec-octet
+ */
+static BOOL check_ipv4address(const WCHAR **ptr, BOOL strict) {
+    const WCHAR *start = *ptr;
+
+    if(!check_dec_octet(ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    if(**ptr != '.') {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    if(!check_dec_octet(ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    if(**ptr != '.') {
+        if(strict) {
+            *ptr = start;
+            return FALSE;
+        } else
+            return TRUE;
+    }
+
+    ++(*ptr);
+    if(!check_dec_octet(ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    if(**ptr != '.') {
+        if(strict) {
+            *ptr = start;
+            return FALSE;
+        } else
+            return TRUE;
+    }
+
+    ++(*ptr);
+    if(!check_dec_octet(ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    /* Found a four digit ip address. */
+    return TRUE;
+}
+/* Tries to parse the scheme name of the URI.
+ *
+ * scheme = ALPHA *(ALPHA | NUM | '+' | '-' | '.') as defined by RFC 3896.
+ * NOTE: Windows accepts a number as the first character of a scheme.
+ */
+static BOOL parse_scheme_name(const WCHAR **ptr, parse_data *data, DWORD extras) {
+    const WCHAR *start = *ptr;
+
+    data->scheme = NULL;
+    data->scheme_len = 0;
+
+    while(**ptr) {
+        if(**ptr == '*' && *ptr == start) {
+            /* Might have found a wildcard scheme. If it is the next
+             * char has to be a ':' for it to be a valid URI
+             */
+            ++(*ptr);
+            break;
+        } else if(!is_num(**ptr) && !is_alpha(**ptr) && **ptr != '+' &&
+           **ptr != '-' && **ptr != '.')
+            break;
+
+        (*ptr)++;
+    }
+
+    if(*ptr == start)
+        return FALSE;
+
+    /* Schemes must end with a ':' */
+    if(**ptr != ':' && !((extras & ALLOW_NULL_TERM_SCHEME) && !**ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    data->scheme = start;
+    data->scheme_len = *ptr - start;
+
+    ++(*ptr);
+    return TRUE;
+}
+
+/* Tries to deduce the corresponding URL_SCHEME for the given URI. Stores
+ * the deduced URL_SCHEME in data->scheme_type.
+ */
+static BOOL parse_scheme_type(parse_data *data) {
+    /* If there's scheme data then see if it's a recognized scheme. */
+    if(data->scheme && data->scheme_len) {
+        DWORD i;
+
+        for(i = 0; i < sizeof(recognized_schemes)/sizeof(recognized_schemes[0]); ++i) {
+            if(lstrlenW(recognized_schemes[i].scheme_name) == data->scheme_len) {
+                /* Has to be a case insensitive compare. */
+                if(!StrCmpNIW(recognized_schemes[i].scheme_name, data->scheme, data->scheme_len)) {
+                    data->scheme_type = recognized_schemes[i].scheme;
+                    return TRUE;
+                }
+            }
+        }
+
+        /* If we get here it means it's not a recognized scheme. */
+        data->scheme_type = URL_SCHEME_UNKNOWN;
+        return TRUE;
+    } else if(data->is_relative) {
+        /* Relative URI's have no scheme. */
+        data->scheme_type = URL_SCHEME_UNKNOWN;
+        return TRUE;
+    } else {
+        /* Should never reach here! what happened... */
+        FIXME("(%p): Unable to determine scheme type for URI %s\n", data, debugstr_w(data->uri));
+        return FALSE;
+    }
+}
+
+/* Tries to parse (or deduce) the scheme_name of a URI. If it can't
+ * parse a scheme from the URI it will try to deduce the scheme_name and scheme_type
+ * using the flags specified in 'flags' (if any). Flags that affect how this function
+ * operates are the Uri_CREATE_ALLOW_* flags.
+ *
+ * All parsed/deduced information will be stored in 'data' when the function returns.
+ *
+ * Returns TRUE if it was able to successfully parse the information.
+ */
+static BOOL parse_scheme(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
+    static const WCHAR fileW[] = {'f','i','l','e',0};
+    static const WCHAR wildcardW[] = {'*',0};
+
+    /* First check to see if the uri could implicitly be a file path. */
+    if(is_implicit_file_path(*ptr)) {
+        if(flags & Uri_CREATE_ALLOW_IMPLICIT_FILE_SCHEME) {
+            data->scheme = fileW;
+            data->scheme_len = lstrlenW(fileW);
+            data->has_implicit_scheme = TRUE;
+
+            TRACE("(%p %p %x): URI is an implicit file path.\n", ptr, data, flags);
+        } else {
+            /* Window's does not consider anything that can implicitly be a file
+             * path to be a valid URI if the ALLOW_IMPLICIT_FILE_SCHEME flag is not set...
+             */
+            TRACE("(%p %p %x): URI is implicitly a file path, but, the ALLOW_IMPLICIT_FILE_SCHEME flag wasn't set.\n",
+                    ptr, data, flags);
+            return FALSE;
+        }
+    } else if(!parse_scheme_name(ptr, data, extras)) {
+        /* No Scheme was found, this means it could be:
+         *      a) an implicit Wildcard scheme
+         *      b) a relative URI
+         *      c) a invalid URI.
+         */
+        if(flags & Uri_CREATE_ALLOW_IMPLICIT_WILDCARD_SCHEME) {
+            data->scheme = wildcardW;
+            data->scheme_len = lstrlenW(wildcardW);
+            data->has_implicit_scheme = TRUE;
+
+            TRACE("(%p %p %x): URI is an implicit wildcard scheme.\n", ptr, data, flags);
+        } else if (flags & Uri_CREATE_ALLOW_RELATIVE) {
+            data->is_relative = TRUE;
+            TRACE("(%p %p %x): URI is relative.\n", ptr, data, flags);
+        } else {
+            TRACE("(%p %p %x): Malformed URI found. Unable to deduce scheme name.\n", ptr, data, flags);
+            return FALSE;
+        }
+    }
+
+    if(!data->is_relative)
+        TRACE("(%p %p %x): Found scheme=%s scheme_len=%d\n", ptr, data, flags,
+                debugstr_wn(data->scheme, data->scheme_len), data->scheme_len);
+
+    if(!parse_scheme_type(data))
+        return FALSE;
+
+    TRACE("(%p %p %x): Assigned %d as the URL_SCHEME.\n", ptr, data, flags, data->scheme_type);
+    return TRUE;
+}
+
+static BOOL parse_username(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
+    data->username = *ptr;
+
+    while(**ptr != ':' && **ptr != '@') {
+        if(**ptr == '%') {
+            if(!check_pct_encoded(ptr)) {
+                if(data->scheme_type != URL_SCHEME_UNKNOWN) {
+                    *ptr = data->username;
+                    data->username = NULL;
+                    return FALSE;
+                }
+            } else
+                continue;
+        } else if(extras & ALLOW_NULL_TERM_USER_NAME && !**ptr)
+            break;
+        else if(is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)) {
+            *ptr = data->username;
+            data->username = NULL;
+            return FALSE;
+        }
+
+        ++(*ptr);
+    }
+
+    data->username_len = *ptr - data->username;
+    return TRUE;
+}
+
+static BOOL parse_password(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
+    const WCHAR *start = *ptr;
+
+    if(**ptr != ':')
+        return TRUE;
+
+    ++(*ptr);
+    data->password = *ptr;
+
+    while(**ptr != '@') {
+        if(**ptr == '%') {
+            if(!check_pct_encoded(ptr)) {
+                if(data->scheme_type != URL_SCHEME_UNKNOWN) {
+                    *ptr = start;
+                    data->password = NULL;
+                    return FALSE;
+                }
+            } else
+                continue;
+        } else if(extras & ALLOW_NULL_TERM_PASSWORD && !**ptr)
+            break;
+        else if(is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)) {
+            *ptr = start;
+            data->password = NULL;
+            return FALSE;
+        }
+
+        ++(*ptr);
+    }
+
+    data->password_len = *ptr - data->password;
+    return TRUE;
+}
+
+/* Parses the userinfo part of the URI (if it exists). The userinfo field of
+ * a URI can consist of "username:password@", or just "username@".
+ *
+ * RFC def:
+ * userinfo    = *( unreserved / pct-encoded / sub-delims / ":" )
+ *
+ * NOTES:
+ *  1)  If there is more than one ':' in the userinfo part of the URI Windows
+ *      uses the first occurrence of ':' to delimit the username and password
+ *      components.
+ *
+ *      ex:
+ *          ftp://user:pass:word@winehq.org
+ *
+ *      Would yield, "user" as the username and "pass:word" as the password.
+ *
+ *  2)  Windows allows any character to appear in the "userinfo" part of
+ *      a URI, as long as it's not an authority delimeter character set.
+ */
+static void parse_userinfo(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const WCHAR *start = *ptr;
+
+    if(!parse_username(ptr, data, flags, 0)) {
+        TRACE("(%p %p %x): URI contained no userinfo.\n", ptr, data, flags);
+        return;
+    }
+
+    if(!parse_password(ptr, data, flags, 0)) {
+        *ptr = start;
+        data->username = NULL;
+        data->username_len = 0;
+        TRACE("(%p %p %x): URI contained no userinfo.\n", ptr, data, flags);
+        return;
+    }
+
+    if(**ptr != '@') {
+        *ptr = start;
+        data->username = NULL;
+        data->username_len = 0;
+        data->password = NULL;
+        data->password_len = 0;
+
+        TRACE("(%p %p %x): URI contained no userinfo.\n", ptr, data, flags);
+        return;
+    }
+
+    if(data->username)
+        TRACE("(%p %p %x): Found username %s len=%d.\n", ptr, data, flags,
+            debugstr_wn(data->username, data->username_len), data->username_len);
+
+    if(data->password)
+        TRACE("(%p %p %x): Found password %s len=%d.\n", ptr, data, flags,
+            debugstr_wn(data->password, data->password_len), data->password_len);
+
+    ++(*ptr);
+}
+
+/* Attempts to parse a port from the URI.
+ *
+ * NOTES:
+ *  Windows seems to have a cap on what the maximum value
+ *  for a port can be. The max value is USHORT_MAX.
+ *
+ * port = *DIGIT
+ */
+static BOOL parse_port(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    UINT port = 0;
+    data->port = *ptr;
+
+    while(!is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)) {
+        if(!is_num(**ptr)) {
+            *ptr = data->port;
+            data->port = NULL;
+            return FALSE;
+        }
+
+        port = port*10 + (**ptr-'0');
+
+        if(port > USHORT_MAX) {
+            *ptr = data->port;
+            data->port = NULL;
+            return FALSE;
+        }
+
+        ++(*ptr);
+    }
+
+    data->has_port = TRUE;
+    data->port_value = port;
+    data->port_len = *ptr - data->port;
+
+    TRACE("(%p %p %x): Found port %s len=%d value=%u\n", ptr, data, flags,
+        debugstr_wn(data->port, data->port_len), data->port_len, data->port_value);
+    return TRUE;
+}
+
+/* Attempts to parse a IPv4 address from the URI.
+ *
+ * NOTES:
+ *  Window's normalizes IPv4 addresses, This means there's three
+ *  possibilities for the URI to contain an IPv4 address.
+ *      1)  A well formed address (ex. 192.2.2.2).
+ *      2)  A partially formed address. For example "192.0" would
+ *          normalize to "192.0.0.0" during canonicalization.
+ *      3)  An implicit IPv4 address. For example "256" would
+ *          normalize to "0.0.1.0" during canonicalization. Also
+ *          note that the maximum value for an implicit IP address
+ *          is UINT_MAX, if the value in the URI exceeds this then
+ *          it is not considered an IPv4 address.
+ */
+static BOOL parse_ipv4address(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const BOOL is_unknown = data->scheme_type == URL_SCHEME_UNKNOWN;
+    data->host = *ptr;
+
+    if(!check_ipv4address(ptr, FALSE)) {
+        if(!check_implicit_ipv4(ptr, &data->implicit_ipv4)) {
+            TRACE("(%p %p %x): URI didn't contain anything looking like an IPv4 address.\n",
+                ptr, data, flags);
+            *ptr = data->host;
+            data->host = NULL;
+            return FALSE;
+        } else
+            data->has_implicit_ip = TRUE;
+    }
+
+    /* Check if what we found is the only part of the host name (if it isn't
+     * we don't have an IPv4 address).
+     */
+    if(**ptr == ':') {
+        ++(*ptr);
+        if(!parse_port(ptr, data, flags)) {
+            *ptr = data->host;
+            data->host = NULL;
+            return FALSE;
+        }
+    } else if(!is_auth_delim(**ptr, !is_unknown)) {
+        /* Found more data which belongs the host, so this isn't an IPv4. */
+        *ptr = data->host;
+        data->host = NULL;
+        data->has_implicit_ip = FALSE;
+        return FALSE;
+    }
+
+    data->host_len = *ptr - data->host;
+    data->host_type = Uri_HOST_IPV4;
+
+    TRACE("(%p %p %x): IPv4 address found. host=%s host_len=%d host_type=%d\n",
+        ptr, data, flags, debugstr_wn(data->host, data->host_len),
+        data->host_len, data->host_type);
+    return TRUE;
+}
+
+/* Attempts to parse the reg-name from the URI.
+ *
+ * Because of the way Windows handles ':' this function also
+ * handles parsing the port.
+ *
+ * reg-name = *( unreserved / pct-encoded / sub-delims )
+ *
+ * NOTE:
+ *  Windows allows everything, but, the characters in "auth_delims" and ':'
+ *  to appear in a reg-name, unless it's an unknown scheme type then ':' is
+ *  allowed to appear (even if a valid port isn't after it).
+ *
+ *  Windows doesn't like host names which start with '[' and end with ']'
+ *  and don't contain a valid IP literal address in between them.
+ *
+ *  On Windows if an '[' is encountered in the host name the ':' no longer
+ *  counts as a delimiter until you reach the next ']' or an "authority delimeter".
+ *
+ *  A reg-name CAN be empty.
+ */
+static BOOL parse_reg_name(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
+    const BOOL has_start_bracket = **ptr == '[';
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+    BOOL inside_brackets = has_start_bracket;
+    BOOL ignore_col = extras & IGNORE_PORT_DELIMITER;
+
+    /* We have to be careful with file schemes. */
+    if(data->scheme_type == URL_SCHEME_FILE) {
+        /* This is because an implicit file scheme could be "C:\\test" and it
+         * would trick this function into thinking the host is "C", when after
+         * canonicalization the host would end up being an empty string. A drive
+         * path can also have a '|' instead of a ':' after the drive letter.
+         */
+        if(is_drive_path(*ptr)) {
+            /* Regular old drive paths don't have a host type (or host name). */
+            data->host_type = Uri_HOST_UNKNOWN;
+            data->host = *ptr;
+            data->host_len = 0;
+            return TRUE;
+        } else if(is_unc_path(*ptr))
+            /* Skip past the "\\" of a UNC path. */
+            *ptr += 2;
+    }
+
+    data->host = *ptr;
+
+    while(!is_auth_delim(**ptr, known_scheme)) {
+        if(**ptr == ':' && !ignore_col) {
+            /* We can ignore ':' if were inside brackets.*/
+            if(!inside_brackets) {
+                const WCHAR *tmp = (*ptr)++;
+
+                /* Attempt to parse the port. */
+                if(!parse_port(ptr, data, flags)) {
+                    /* Windows expects there to be a valid port for known scheme types. */
+                    if(data->scheme_type != URL_SCHEME_UNKNOWN) {
+                        *ptr = data->host;
+                        data->host = NULL;
+                        TRACE("(%p %p %x %x): Expected valid port\n", ptr, data, flags, extras);
+                        return FALSE;
+                    } else
+                        /* Windows gives up on trying to parse a port when it
+                         * encounters 1 invalid port.
+                         */
+                        ignore_col = TRUE;
+                } else {
+                    data->host_len = tmp - data->host;
+                    break;
+                }
+            }
+        } else if(**ptr == '%' && known_scheme) {
+            /* Has to be a legit % encoded value. */
+            if(!check_pct_encoded(ptr)) {
+                *ptr = data->host;
+                data->host = NULL;
+                return FALSE;
+            } else
+                continue;
+        } else if(**ptr == ']')
+            inside_brackets = FALSE;
+        else if(**ptr == '[')
+            inside_brackets = TRUE;
+
+        ++(*ptr);
+    }
+
+    if(has_start_bracket) {
+        /* Make sure the last character of the host wasn't a ']'. */
+        if(*(*ptr-1) == ']') {
+            TRACE("(%p %p %x %x): Expected an IP literal inside of the host\n",
+                ptr, data, flags, extras);
+            *ptr = data->host;
+            data->host = NULL;
+            return FALSE;
+        }
+    }
+
+    /* Don't overwrite our length if we found a port earlier. */
+    if(!data->port)
+        data->host_len = *ptr - data->host;
+
+    /* If the host is empty, then it's an unknown host type. */
+    if(data->host_len == 0)
+        data->host_type = Uri_HOST_UNKNOWN;
+    else
+        data->host_type = Uri_HOST_DNS;
+
+    TRACE("(%p %p %x %x): Parsed reg-name. host=%s len=%d\n", ptr, data, flags, extras,
+        debugstr_wn(data->host, data->host_len), data->host_len);
+    return TRUE;
+}
+
+/* Attempts to parse an IPv6 address out of the URI.
+ *
+ * IPv6address =                               6( h16 ":" ) ls32
+ *                /                       "::" 5( h16 ":" ) ls32
+ *                / [               h16 ] "::" 4( h16 ":" ) ls32
+ *                / [ *1( h16 ":" ) h16 ] "::" 3( h16 ":" ) ls32
+ *                / [ *2( h16 ":" ) h16 ] "::" 2( h16 ":" ) ls32
+ *                / [ *3( h16 ":" ) h16 ] "::"    h16 ":"   ls32
+ *                / [ *4( h16 ":" ) h16 ] "::"              ls32
+ *                / [ *5( h16 ":" ) h16 ] "::"              h16
+ *                / [ *6( h16 ":" ) h16 ] "::"
+ *
+ * ls32        = ( h16 ":" h16 ) / IPv4address
+ *             ; least-significant 32 bits of address.
+ *
+ * h16         = 1*4HEXDIG
+ *             ; 16 bits of address represented in hexadecimal.
+ *
+ * Modeled after google-url's 'DoParseIPv6' function.
+ */
+static BOOL parse_ipv6address(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const WCHAR *start, *cur_start;
+    ipv6_address ip;
+
+    start = cur_start = *ptr;
+    memset(&ip, 0, sizeof(ipv6_address));
+
+    for(;; ++(*ptr)) {
+        /* Check if we're on the last character of the host. */
+        BOOL is_end = (is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)
+                        || **ptr == ']');
+
+        BOOL is_split = (**ptr == ':');
+        BOOL is_elision = (is_split && !is_end && *(*ptr+1) == ':');
+
+        /* Check if we're at the end of a component, or
+         * if we're at the end of the IPv6 address.
+         */
+        if(is_split || is_end) {
+            DWORD cur_len = 0;
+
+            cur_len = *ptr - cur_start;
+
+            /* h16 can't have a length > 4. */
+            if(cur_len > 4) {
+                *ptr = start;
+
+                TRACE("(%p %p %x): h16 component to long.\n",
+                    ptr, data, flags);
+                return FALSE;
+            }
+
+            if(cur_len == 0) {
+                /* An h16 component can't have the length of 0 unless
+                 * the elision is at the beginning of the address, or
+                 * at the end of the address.
+                 */
+                if(!((*ptr == start && is_elision) ||
+                    (is_end && (*ptr-2) == ip.elision))) {
+                    *ptr = start;
+                    TRACE("(%p %p %x): IPv6 component cannot have a length of 0.\n",
+                        ptr, data, flags);
+                    return FALSE;
+                }
+            }
+
+            if(cur_len > 0) {
+                /* An IPv6 address can have no more than 8 h16 components. */
+                if(ip.h16_count >= 8) {
+                    *ptr = start;
+                    TRACE("(%p %p %x): Not a IPv6 address, to many h16 components.\n",
+                        ptr, data, flags);
+                    return FALSE;
+                }
+
+                ip.components[ip.h16_count].str = cur_start;
+                ip.components[ip.h16_count].len = cur_len;
+
+                TRACE("(%p %p %x): Found h16 component %s, len=%d, h16_count=%d\n",
+                    ptr, data, flags, debugstr_wn(cur_start, cur_len), cur_len,
+                    ip.h16_count);
+                ++ip.h16_count;
+            }
+        }
+
+        if(is_end)
+            break;
+
+        if(is_elision) {
+            /* A IPv6 address can only have 1 elision ('::'). */
+            if(ip.elision) {
+                *ptr = start;
+
+                TRACE("(%p %p %x): IPv6 address cannot have 2 elisions.\n",
+                    ptr, data, flags);
+                return FALSE;
+            }
+
+            ip.elision = *ptr;
+            ++(*ptr);
+        }
+
+        if(is_split)
+            cur_start = *ptr+1;
+        else {
+            if(!check_ipv4address(ptr, TRUE)) {
+                if(!is_hexdigit(**ptr)) {
+                    /* Not a valid character for an IPv6 address. */
+                    *ptr = start;
+                    return FALSE;
+                }
+            } else {
+                /* Found an IPv4 address. */
+                ip.ipv4 = cur_start;
+                ip.ipv4_len = *ptr - cur_start;
+
+                TRACE("(%p %p %x): Found an attached IPv4 address %s len=%d.\n",
+                    ptr, data, flags, debugstr_wn(ip.ipv4, ip.ipv4_len),
+                    ip.ipv4_len);
+
+                /* IPv4 addresses can only appear at the end of a IPv6. */
+                break;
+            }
+        }
+    }
+
+    compute_ipv6_comps_size(&ip);
+
+    /* Make sure the IPv6 address adds up to 16 bytes. */
+    if(ip.components_size + ip.elision_size != 16) {
+        *ptr = start;
+        TRACE("(%p %p %x): Invalid IPv6 address, did not add up to 16 bytes.\n",
+            ptr, data, flags);
+        return FALSE;
+    }
+
+    if(ip.elision_size == 2) {
+        /* For some reason on Windows if an elision that represents
+         * only 1 h16 component is encountered at the very begin or
+         * end of an IPv6 address, Windows does not consider it a
+         * valid IPv6 address.
+         *
+         *  Ex: [::2:3:4:5:6:7] is not valid, even though the sum
+         *      of all the components == 128bits.
+         */
+         if(ip.elision < ip.components[0].str ||
+            ip.elision > ip.components[ip.h16_count-1].str) {
+            *ptr = start;
+            TRACE("(%p %p %x): Invalid IPv6 address. Detected elision of 2 bytes at the beginning or end of the address.\n",
+                ptr, data, flags);
+            return FALSE;
+        }
+    }
+
+    data->host_type = Uri_HOST_IPV6;
+    data->has_ipv6 = TRUE;
+    data->ipv6_address = ip;
+
+    TRACE("(%p %p %x): Found valid IPv6 literal %s len=%d\n",
+        ptr, data, flags, debugstr_wn(start, *ptr-start),
+        *ptr-start);
+    return TRUE;
+}
+
+/*  IPvFuture  = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" ) */
+static BOOL parse_ipvfuture(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const WCHAR *start = *ptr;
+
+    /* IPvFuture has to start with a 'v' or 'V'. */
+    if(**ptr != 'v' && **ptr != 'V')
+        return FALSE;
+
+    /* Following the v there must be at least 1 hex digit. */
+    ++(*ptr);
+    if(!is_hexdigit(**ptr)) {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    while(is_hexdigit(**ptr))
+        ++(*ptr);
+
+    /* End of the hexdigit sequence must be a '.' */
+    if(**ptr != '.') {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    if(!is_unreserved(**ptr) && !is_subdelim(**ptr) && **ptr != ':') {
+        *ptr = start;
+        return FALSE;
+    }
+
+    ++(*ptr);
+    while(is_unreserved(**ptr) || is_subdelim(**ptr) || **ptr == ':')
+        ++(*ptr);
+
+    data->host_type = Uri_HOST_UNKNOWN;
+
+    TRACE("(%p %p %x): Parsed IPvFuture address %s len=%d\n", ptr, data, flags,
+        debugstr_wn(start, *ptr-start), *ptr-start);
+
+    return TRUE;
+}
+
+/* IP-literal = "[" ( IPv6address / IPvFuture  ) "]" */
+static BOOL parse_ip_literal(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
+    data->host = *ptr;
+
+    if(**ptr != '[' && !(extras & ALLOW_BRACKETLESS_IP_LITERAL)) {
+        data->host = NULL;
+        return FALSE;
+    } else if(**ptr == '[')
+        ++(*ptr);
+
+    if(!parse_ipv6address(ptr, data, flags)) {
+        if(extras & SKIP_IP_FUTURE_CHECK || !parse_ipvfuture(ptr, data, flags)) {
+            *ptr = data->host;
+            data->host = NULL;
+            return FALSE;
+        }
+    }
+
+    if(**ptr != ']' && !(extras & ALLOW_BRACKETLESS_IP_LITERAL)) {
+        *ptr = data->host;
+        data->host = NULL;
+        return FALSE;
+    } else if(!**ptr && extras & ALLOW_BRACKETLESS_IP_LITERAL) {
+        /* The IP literal didn't contain brackets and was followed by
+         * a NULL terminator, so no reason to even check the port.
+         */
+        data->host_len = *ptr - data->host;
+        return TRUE;
+    }
+
+    ++(*ptr);
+    if(**ptr == ':') {
+        ++(*ptr);
+        /* If a valid port is not found, then let it trickle down to
+         * parse_reg_name.
+         */
+        if(!parse_port(ptr, data, flags)) {
+            *ptr = data->host;
+            data->host = NULL;
+            return FALSE;
+        }
+    } else
+        data->host_len = *ptr - data->host;
+
+    return TRUE;
+}
+
+/* Parses the host information from the URI.
+ *
+ * host = IP-literal / IPv4address / reg-name
+ */
+static BOOL parse_host(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
+    if(!parse_ip_literal(ptr, data, flags, extras)) {
+        if(!parse_ipv4address(ptr, data, flags)) {
+            if(!parse_reg_name(ptr, data, flags, extras)) {
+                TRACE("(%p %p %x %x): Malformed URI, Unknown host type.\n",
+                    ptr, data, flags, extras);
+                return FALSE;
+            }
+        }
+    }
+
+    return TRUE;
+}
+
+/* Parses the authority information from the URI.
+ *
+ * authority   = [ userinfo "@" ] host [ ":" port ]
+ */
+static BOOL parse_authority(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    parse_userinfo(ptr, data, flags);
+
+    /* Parsing the port will happen during one of the host parsing
+     * routines (if the URI has a port).
+     */
+    if(!parse_host(ptr, data, flags, 0))
+        return FALSE;
+
+    return TRUE;
+}
+
+/* Attempts to parse the path information of a hierarchical URI. */
+static BOOL parse_path_hierarchical(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const WCHAR *start = *ptr;
+    static const WCHAR slash[] = {'/',0};
+    const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
+
+    if(is_path_delim(**ptr)) {
+        if(data->scheme_type == URL_SCHEME_WILDCARD) {
+            /* Wildcard schemes don't get a '/' attached if their path is
+             * empty.
+             */
+            data->path = NULL;
+            data->path_len = 0;
+        } else if(!(flags & Uri_CREATE_NO_CANONICALIZE)) {
+            /* If the path component is empty, then a '/' is added. */
+            data->path = slash;
+            data->path_len = 1;
+        }
+    } else {
+        while(!is_path_delim(**ptr)) {
+            if(**ptr == '%' && data->scheme_type != URL_SCHEME_UNKNOWN && !is_file) {
+                if(!check_pct_encoded(ptr)) {
+                    *ptr = start;
+                    return FALSE;
+                } else
+                    continue;
+            } else if(is_forbidden_dos_path_char(**ptr) && is_file &&
+                      (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+                /* File schemes with USE_DOS_PATH set aren't allowed to have
+                 * a '<' or '>' or '\"' appear in them.
+                 */
+                *ptr = start;
+                return FALSE;
+            } else if(**ptr == '\\') {
+                /* Not allowed to have a backslash if NO_CANONICALIZE is set
+                 * and the scheme is known type (but not a file scheme).
+                 */
+                if(flags & Uri_CREATE_NO_CANONICALIZE) {
+                    if(data->scheme_type != URL_SCHEME_FILE &&
+                       data->scheme_type != URL_SCHEME_UNKNOWN) {
+                        *ptr = start;
+                        return FALSE;
+                    }
+                }
+            }
+
+            ++(*ptr);
+        }
+
+        /* The only time a URI doesn't have a path is when
+         * the NO_CANONICALIZE flag is set and the raw URI
+         * didn't contain one.
+         */
+        if(*ptr == start) {
+            data->path = NULL;
+            data->path_len = 0;
+        } else {
+            data->path = start;
+            data->path_len = *ptr - start;
+        }
+    }
+
+    if(data->path)
+        TRACE("(%p %p %x): Parsed path %s len=%d\n", ptr, data, flags,
+            debugstr_wn(data->path, data->path_len), data->path_len);
+    else
+        TRACE("(%p %p %x): The URI contained no path\n", ptr, data, flags);
+
+    return TRUE;
+}
+
+/* Parses the path of a opaque URI (much less strict then the parser
+ * for a hierarchical URI).
+ *
+ * NOTE:
+ *  Windows allows invalid % encoded data to appear in opaque URI paths
+ *  for unknown scheme types.
+ *
+ *  File schemes with USE_DOS_PATH set aren't allowed to have '<', '>', or '\"'
+ *  appear in them.
+ */
+static BOOL parse_path_opaque(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+    const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
+
+    data->path = *ptr;
+
+    while(!is_path_delim(**ptr)) {
+        if(**ptr == '%' && known_scheme) {
+            if(!check_pct_encoded(ptr)) {
+                *ptr = data->path;
+                data->path = NULL;
+                return FALSE;
+            } else
+                continue;
+        } else if(is_forbidden_dos_path_char(**ptr) && is_file &&
+                  (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+            *ptr = data->path;
+            data->path = NULL;
+            return FALSE;
+        }
+
+        ++(*ptr);
+    }
+
+    data->path_len = *ptr - data->path;
+    TRACE("(%p %p %x): Parsed opaque URI path %s len=%d\n", ptr, data, flags,
+        debugstr_wn(data->path, data->path_len), data->path_len);
+    return TRUE;
+}
+
+/* Determines how the URI should be parsed after the scheme information.
+ *
+ * If the scheme is followed, by "//" then, it is treated as an hierarchical URI
+ * which then the authority and path information will be parsed out. Otherwise, the
+ * URI will be treated as an opaque URI which the authority information is not parsed
+ * out.
+ *
+ * RFC 3896 definition of hier-part:
+ *
+ * hier-part   = "//" authority path-abempty
+ *                 / path-absolute
+ *                 / path-rootless
+ *                 / path-empty
+ *
+ * MSDN opaque URI definition:
+ *  scheme ":" path [ "#" fragment ]
+ *
+ * NOTES:
+ *  If the URI is of an unknown scheme type and has a "//" following the scheme then it
+ *  is treated as a hierarchical URI, but, if the CREATE_NO_CRACK_UNKNOWN_SCHEMES flag is
+ *  set then it is considered an opaque URI reguardless of what follows the scheme information
+ *  (per MSDN documentation).
+ */
+static BOOL parse_hierpart(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const WCHAR *start = *ptr;
+
+    /* Checks if the authority information needs to be parsed. */
+    if(is_hierarchical_uri(ptr, data)) {
+        /* Only treat it as a hierarchical URI if the scheme_type is known or
+         * the Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES flag is not set.
+         */
+        if(data->scheme_type != URL_SCHEME_UNKNOWN ||
+           !(flags & Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES)) {
+            TRACE("(%p %p %x): Treating URI as an hierarchical URI.\n", ptr, data, flags);
+            data->is_opaque = FALSE;
+
+            /* TODO: Handle hierarchical URI's, parse authority then parse the path. */
+            if(!parse_authority(ptr, data, flags))
+                return FALSE;
+
+            return parse_path_hierarchical(ptr, data, flags);
+        } else
+            /* Reset ptr to it's starting position so opaque path parsing
+             * begins at the correct location.
+             */
+            *ptr = start;
+    }
+
+    /* If it reaches here, then the URI will be treated as an opaque
+     * URI.
+     */
+
+    TRACE("(%p %p %x): Treating URI as an opaque URI.\n", ptr, data, flags);
+
+    data->is_opaque = TRUE;
+    if(!parse_path_opaque(ptr, data, flags))
+        return FALSE;
+
+    return TRUE;
+}
+
+/* Attempts to parse the query string from the URI.
+ *
+ * NOTES:
+ *  If NO_DECODE_EXTRA_INFO flag is set, then invalid percent encoded
+ *  data is allowed appear in the query string. For unknown scheme types
+ *  invalid percent encoded data is allowed to appear reguardless.
+ */
+static BOOL parse_query(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+
+    if(**ptr != '?') {
+        TRACE("(%p %p %x): URI didn't contain a query string.\n", ptr, data, flags);
+        return TRUE;
+    }
+
+    data->query = *ptr;
+
+    ++(*ptr);
+    while(**ptr && **ptr != '#') {
+        if(**ptr == '%' && known_scheme &&
+           !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
+            if(!check_pct_encoded(ptr)) {
+                *ptr = data->query;
+                data->query = NULL;
+                return FALSE;
+            } else
+                continue;
+        }
+
+        ++(*ptr);
+    }
+
+    data->query_len = *ptr - data->query;
+
+    TRACE("(%p %p %x): Parsed query string %s len=%d\n", ptr, data, flags,
+        debugstr_wn(data->query, data->query_len), data->query_len);
+    return TRUE;
+}
+
+/* Attempts to parse the fragment from the URI.
+ *
+ * NOTES:
+ *  If NO_DECODE_EXTRA_INFO flag is set, then invalid percent encoded
+ *  data is allowed appear in the query string. For unknown scheme types
+ *  invalid percent encoded data is allowed to appear reguardless.
+ */
+static BOOL parse_fragment(const WCHAR **ptr, parse_data *data, DWORD flags) {
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+
+    if(**ptr != '#') {
+        TRACE("(%p %p %x): URI didn't contain a fragment.\n", ptr, data, flags);
+        return TRUE;
+    }
+
+    data->fragment = *ptr;
+
+    ++(*ptr);
+    while(**ptr) {
+        if(**ptr == '%' && known_scheme &&
+           !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
+            if(!check_pct_encoded(ptr)) {
+                *ptr = data->fragment;
+                data->fragment = NULL;
+                return FALSE;
+            } else
+                continue;
+        }
+
+        ++(*ptr);
+    }
+
+    data->fragment_len = *ptr - data->fragment;
+
+    TRACE("(%p %p %x): Parsed fragment %s len=%d\n", ptr, data, flags,
+        debugstr_wn(data->fragment, data->fragment_len), data->fragment_len);
+    return TRUE;
+}
+
+/* Parses and validates the components of the specified by data->uri
+ * and stores the information it parses into 'data'.
+ *
+ * Returns TRUE if it successfully parsed the URI. False otherwise.
+ */
+static BOOL parse_uri(parse_data *data, DWORD flags) {
+    const WCHAR *ptr;
+    const WCHAR **pptr;
+
+    ptr = data->uri;
+    pptr = &ptr;
+
+    TRACE("(%p %x): BEGINNING TO PARSE URI %s.\n", data, flags, debugstr_w(data->uri));
+
+    if(!parse_scheme(pptr, data, flags, 0))
+        return FALSE;
+
+    if(!parse_hierpart(pptr, data, flags))
+        return FALSE;
+
+    if(!parse_query(pptr, data, flags))
+        return FALSE;
+
+    if(!parse_fragment(pptr, data, flags))
+        return FALSE;
+
+    TRACE("(%p %x): FINISHED PARSING URI.\n", data, flags);
+    return TRUE;
+}
+
+static BOOL canonicalize_username(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    const WCHAR *ptr;
+
+    if(!data->username) {
+        uri->userinfo_start = -1;
+        return TRUE;
+    }
+
+    uri->userinfo_start = uri->canon_len;
+    for(ptr = data->username; ptr < data->username+data->username_len; ++ptr) {
+        if(*ptr == '%') {
+            /* Only decode % encoded values for known scheme types. */
+            if(data->scheme_type != URL_SCHEME_UNKNOWN) {
+                /* See if the value really needs decoded. */
+                WCHAR val = decode_pct_val(ptr);
+                if(is_unreserved(val)) {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = val;
+
+                    ++uri->canon_len;
+
+                    /* Move pass the hex characters. */
+                    ptr += 2;
+                    continue;
+                }
+            }
+        } else if(!is_reserved(*ptr) && !is_unreserved(*ptr) && *ptr != '\\') {
+            /* Only percent encode forbidden characters if the NO_ENCODE_FORBIDDEN_CHARACTERS flag
+             * is NOT set.
+             */
+            if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS)) {
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri + uri->canon_len);
+
+                uri->canon_len += 3;
+                continue;
+            }
+        }
+
+        if(!computeOnly)
+            /* Nothing special, so just copy the character over. */
+            uri->canon_uri[uri->canon_len] = *ptr;
+        ++uri->canon_len;
+    }
+
+    return TRUE;
+}
+
+static BOOL canonicalize_password(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    const WCHAR *ptr;
+
+    if(!data->password) {
+        uri->userinfo_split = -1;
+        return TRUE;
+    }
+
+    if(uri->userinfo_start == -1)
+        /* Has a password, but, doesn't have a username. */
+        uri->userinfo_start = uri->canon_len;
+
+    uri->userinfo_split = uri->canon_len - uri->userinfo_start;
+
+    /* Add the ':' to the userinfo component. */
+    if(!computeOnly)
+        uri->canon_uri[uri->canon_len] = ':';
+    ++uri->canon_len;
+
+    for(ptr = data->password; ptr < data->password+data->password_len; ++ptr) {
+        if(*ptr == '%') {
+            /* Only decode % encoded values for known scheme types. */
+            if(data->scheme_type != URL_SCHEME_UNKNOWN) {
+                /* See if the value really needs decoded. */
+                WCHAR val = decode_pct_val(ptr);
+                if(is_unreserved(val)) {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = val;
+
+                    ++uri->canon_len;
+
+                    /* Move pass the hex characters. */
+                    ptr += 2;
+                    continue;
+                }
+            }
+        } else if(!is_reserved(*ptr) && !is_unreserved(*ptr) && *ptr != '\\') {
+            /* Only percent encode forbidden characters if the NO_ENCODE_FORBIDDEN_CHARACTERS flag
+             * is NOT set.
+             */
+            if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS)) {
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri + uri->canon_len);
+
+                uri->canon_len += 3;
+                continue;
+            }
+        }
+
+        if(!computeOnly)
+            /* Nothing special, so just copy the character over. */
+            uri->canon_uri[uri->canon_len] = *ptr;
+        ++uri->canon_len;
+    }
+
+    return TRUE;
+}
+
+/* Canonicalizes the userinfo of the URI represented by the parse_data.
+ *
+ * Canonicalization of the userinfo is a simple process. If there are any percent
+ * encoded characters that fall in the "unreserved" character set, they are decoded
+ * to their actual value. If a character is not in the "unreserved" or "reserved" sets
+ * then it is percent encoded. Other than that the characters are copied over without
+ * change.
+ */
+static BOOL canonicalize_userinfo(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    uri->userinfo_start = uri->userinfo_split = -1;
+    uri->userinfo_len = 0;
+
+    if(!data->username && !data->password)
+        /* URI doesn't have userinfo, so nothing to do here. */
+        return TRUE;
+
+    if(!canonicalize_username(data, uri, flags, computeOnly))
+        return FALSE;
+
+    if(!canonicalize_password(data, uri, flags, computeOnly))
+        return FALSE;
+
+    uri->userinfo_len = uri->canon_len - uri->userinfo_start;
+    if(!computeOnly)
+        TRACE("(%p %p %x %d): Canonicalized userinfo, userinfo_start=%d, userinfo=%s, userinfo_split=%d userinfo_len=%d.\n",
+                data, uri, flags, computeOnly, uri->userinfo_start, debugstr_wn(uri->canon_uri + uri->userinfo_start, uri->userinfo_len),
+                uri->userinfo_split, uri->userinfo_len);
+
+    /* Now insert the '@' after the userinfo. */
+    if(!computeOnly)
+        uri->canon_uri[uri->canon_len] = '@';
+    ++uri->canon_len;
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize a reg_name.
+ *
+ * Things that happen:
+ *  1)  If Uri_CREATE_NO_CANONICALIZE flag is not set, then the reg_name is
+ *      lower cased. Unless it's an unknown scheme type, which case it's
+ *      no lower cased reguardless.
+ *
+ *  2)  Unreserved % encoded characters are decoded for known
+ *      scheme types.
+ *
+ *  3)  Forbidden characters are % encoded as long as
+ *      Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS flag is not set and
+ *      it isn't an unknown scheme type.
+ *
+ *  4)  If it's a file scheme and the host is "localhost" it's removed.
+ */
+static BOOL canonicalize_reg_name(const parse_data *data, Uri *uri,
+                                  DWORD flags, BOOL computeOnly) {
+    static const WCHAR localhostW[] =
+            {'l','o','c','a','l','h','o','s','t',0};
+    const WCHAR *ptr;
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+
+    uri->host_start = uri->canon_len;
+
+    if(data->scheme_type == URL_SCHEME_FILE &&
+       data->host_len == lstrlenW(localhostW)) {
+        if(!StrCmpNIW(data->host, localhostW, data->host_len)) {
+            uri->host_start = -1;
+            uri->host_len = 0;
+            uri->host_type = Uri_HOST_UNKNOWN;
+            return TRUE;
+        }
+    }
+
+    for(ptr = data->host; ptr < data->host+data->host_len; ++ptr) {
+        if(*ptr == '%' && known_scheme) {
+            WCHAR val = decode_pct_val(ptr);
+            if(is_unreserved(val)) {
+                /* If NO_CANONICALZE is not set, then windows lower cases the
+                 * decoded value.
+                 */
+                if(!(flags & Uri_CREATE_NO_CANONICALIZE) && isupperW(val)) {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = tolowerW(val);
+                } else {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = val;
+                }
+                ++uri->canon_len;
+
+                /* Skip past the % encoded character. */
+                ptr += 2;
+                continue;
+            } else {
+                /* Just copy the % over. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            }
+        } else if(*ptr == '\\') {
+            /* Only unknown scheme types could have made it here with a '\\' in the host name. */
+            if(!computeOnly)
+                uri->canon_uri[uri->canon_len] = *ptr;
+            ++uri->canon_len;
+        } else if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) &&
+                  !is_unreserved(*ptr) && !is_reserved(*ptr) && known_scheme) {
+            if(!computeOnly) {
+                pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
+
+                /* The percent encoded value gets lower cased also. */
+                if(!(flags & Uri_CREATE_NO_CANONICALIZE)) {
+                    uri->canon_uri[uri->canon_len+1] = tolowerW(uri->canon_uri[uri->canon_len+1]);
+                    uri->canon_uri[uri->canon_len+2] = tolowerW(uri->canon_uri[uri->canon_len+2]);
+                }
+            }
+
+            uri->canon_len += 3;
+        } else {
+            if(!computeOnly) {
+                if(!(flags & Uri_CREATE_NO_CANONICALIZE) && known_scheme)
+                    uri->canon_uri[uri->canon_len] = tolowerW(*ptr);
+                else
+                    uri->canon_uri[uri->canon_len] = *ptr;
+            }
+
+            ++uri->canon_len;
+        }
+    }
+
+    uri->host_len = uri->canon_len - uri->host_start;
+
+    if(!computeOnly)
+        TRACE("(%p %p %x %d): Canonicalize reg_name=%s len=%d\n", data, uri, flags,
+            computeOnly, debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
+            uri->host_len);
+
+    if(!computeOnly)
+        find_domain_name(uri->canon_uri+uri->host_start, uri->host_len,
+            &(uri->domain_offset));
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize an implicit IPv4 address. */
+static BOOL canonicalize_implicit_ipv4address(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    uri->host_start = uri->canon_len;
+
+    TRACE("%u\n", data->implicit_ipv4);
+    /* For unknown scheme types Window's doesn't convert
+     * the value into an IP address, but, it still considers
+     * it an IPv4 address.
+     */
+    if(data->scheme_type == URL_SCHEME_UNKNOWN) {
+        if(!computeOnly)
+            memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
+        uri->canon_len += data->host_len;
+    } else {
+        if(!computeOnly)
+            uri->canon_len += ui2ipv4(uri->canon_uri+uri->canon_len, data->implicit_ipv4);
+        else
+            uri->canon_len += ui2ipv4(NULL, data->implicit_ipv4);
+    }
+
+    uri->host_len = uri->canon_len - uri->host_start;
+    uri->host_type = Uri_HOST_IPV4;
+
+    if(!computeOnly)
+        TRACE("%p %p %x %d): Canonicalized implicit IP address=%s len=%d\n",
+            data, uri, flags, computeOnly,
+            debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
+            uri->host_len);
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize an IPv4 address.
+ *
+ * If the parse_data represents a URI that has an implicit IPv4 address
+ * (ex. http://256/, this function will convert 256 into 0.0.1.0). If
+ * the implicit IP address exceeds the value of UINT_MAX (maximum value
+ * for an IPv4 address) it's canonicalized as if were a reg-name.
+ *
+ * If the parse_data contains a partial or full IPv4 address it normalizes it.
+ * A partial IPv4 address is something like "192.0" and would be normalized to
+ * "192.0.0.0". With a full (or partial) IPv4 address like "192.002.01.003" would
+ * be normalized to "192.2.1.3".
+ *
+ * NOTES:
+ *  Window's ONLY normalizes IPv4 address for known scheme types (one that isn't
+ *  URL_SCHEME_UNKNOWN). For unknown scheme types, it simply copies the data from
+ *  the original URI into the canonicalized URI, but, it still recognizes URI's
+ *  host type as HOST_IPV4.
+ */
+static BOOL canonicalize_ipv4address(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    if(data->has_implicit_ip)
+        return canonicalize_implicit_ipv4address(data, uri, flags, computeOnly);
+    else {
+        uri->host_start = uri->canon_len;
+
+        /* Windows only normalizes for known scheme types. */
+        if(data->scheme_type != URL_SCHEME_UNKNOWN) {
+            /* parse_data contains a partial or full IPv4 address, so normalize it. */
+            DWORD i, octetDigitCount = 0, octetCount = 0;
+            BOOL octetHasDigit = FALSE;
+
+            for(i = 0; i < data->host_len; ++i) {
+                if(data->host[i] == '0' && !octetHasDigit) {
+                    /* Can ignore leading zeros if:
+                     *  1) It isn't the last digit of the octet.
+                     *  2) i+1 != data->host_len
+                     *  3) i+1 != '.'
+                     */
+                    if(octetDigitCount == 2 ||
+                       i+1 == data->host_len ||
+                       data->host[i+1] == '.') {
+                        if(!computeOnly)
+                            uri->canon_uri[uri->canon_len] = data->host[i];
+                        ++uri->canon_len;
+                        TRACE("Adding zero\n");
+                    }
+                } else if(data->host[i] == '.') {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = data->host[i];
+                    ++uri->canon_len;
+
+                    octetDigitCount = 0;
+                    octetHasDigit = FALSE;
+                    ++octetCount;
+                } else {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = data->host[i];
+                    ++uri->canon_len;
+
+                    ++octetDigitCount;
+                    octetHasDigit = TRUE;
+                }
+            }
+
+            /* Make sure the canonicalized IP address has 4 dec-octets.
+             * If doesn't add "0" ones until there is 4;
+             */
+            for( ; octetCount < 3; ++octetCount) {
+                if(!computeOnly) {
+                    uri->canon_uri[uri->canon_len] = '.';
+                    uri->canon_uri[uri->canon_len+1] = '0';
+                }
+
+                uri->canon_len += 2;
+            }
+        } else {
+            /* Windows doesn't normalize addresses in unknown schemes. */
+            if(!computeOnly)
+                memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
+            uri->canon_len += data->host_len;
+        }
+
+        uri->host_len = uri->canon_len - uri->host_start;
+        if(!computeOnly)
+            TRACE("(%p %p %x %d): Canonicalized IPv4 address, ip=%s len=%d\n",
+                data, uri, flags, computeOnly,
+                debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
+                uri->host_len);
+    }
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize the IPv6 address of the URI.
+ *
+ * Multiple things happen during the canonicalization of an IPv6 address:
+ *  1)  Any leading zero's in an h16 component are removed.
+ *      Ex: [0001:0022::] -> [1:22::]
+ *
+ *  2)  The longest sequence of zero h16 components are compressed
+ *      into a "::" (elision). If there's a tie, the first is choosen.
+ *
+ *      Ex: [0:0:0:0:1:6:7:8]   -> [::1:6:7:8]
+ *          [0:0:0:0:1:2::]     -> [::1:2:0:0]
+ *          [0:0:1:2:0:0:7:8]   -> [::1:2:0:0:7:8]
+ *
+ *  3)  If an IPv4 address is attached to the IPv6 address, it's
+ *      also normalized.
+ *      Ex: [::001.002.022.000] -> [::1.2.22.0]
+ *
+ *  4)  If an elision is present, but, only represents 1 h16 component
+ *      it's expanded.
+ *
+ *      Ex: [1::2:3:4:5:6:7] -> [1:0:2:3:4:5:6:7]
+ *
+ *  5)  If the IPv6 address contains an IPv4 address and there exists
+ *      at least 1 non-zero h16 component the IPv4 address is converted
+ *      into two h16 components, otherwise it's normalized and kept as is.
+ *
+ *      Ex: [::192.200.003.4]       -> [::192.200.3.4]
+ *          [ffff::192.200.003.4]   -> [ffff::c0c8:3041]
+ *
+ * NOTE:
+ *  For unknown scheme types Windows simply copies the address over without any
+ *  changes.
+ *
+ *  IPv4 address can be included in an elision if all its components are 0's.
+ */
+static BOOL canonicalize_ipv6address(const parse_data *data, Uri *uri,
+                                     DWORD flags, BOOL computeOnly) {
+    uri->host_start = uri->canon_len;
+
+    if(data->scheme_type == URL_SCHEME_UNKNOWN) {
+        if(!computeOnly)
+            memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
+        uri->canon_len += data->host_len;
+    } else {
+        USHORT values[8];
+        INT elision_start;
+        DWORD i, elision_len;
+
+        if(!ipv6_to_number(&(data->ipv6_address), values)) {
+            TRACE("(%p %p %x %d): Failed to compute numerical value for IPv6 address.\n",
+                data, uri, flags, computeOnly);
+            return FALSE;
+        }
+
+        if(!computeOnly)
+            uri->canon_uri[uri->canon_len] = '[';
+        ++uri->canon_len;
+
+        /* Find where the elision should occur (if any). */
+        compute_elision_location(&(data->ipv6_address), values, &elision_start, &elision_len);
+
+        TRACE("%p %p %x %d): Elision starts at %d, len=%u\n", data, uri, flags,
+            computeOnly, elision_start, elision_len);
+
+        for(i = 0; i < 8; ++i) {
+            BOOL in_elision = (elision_start > -1 && i >= elision_start &&
+                               i < elision_start+elision_len);
+            BOOL do_ipv4 = (i == 6 && data->ipv6_address.ipv4 && !in_elision &&
+                            data->ipv6_address.h16_count == 0);
+
+            if(i == elision_start) {
+                if(!computeOnly) {
+                    uri->canon_uri[uri->canon_len] = ':';
+                    uri->canon_uri[uri->canon_len+1] = ':';
+                }
+                uri->canon_len += 2;
+            }
+
+            /* We can ignore the current component if we're in the elision. */
+            if(in_elision)
+                continue;
+
+            /* We only add a ':' if we're not at i == 0, or when we're at
+             * the very end of elision range since the ':' colon was handled
+             * earlier. Otherwise we would end up with ":::" after elision.
+             */
+            if(i != 0 && !(elision_start > -1 && i == elision_start+elision_len)) {
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = ':';
+                ++uri->canon_len;
+            }
+
+            if(do_ipv4) {
+                UINT val;
+                DWORD len;
+
+                /* Combine the two parts of the IPv4 address values. */
+                val = values[i];
+                val <<= 16;
+                val += values[i+1];
+
+                if(!computeOnly)
+                    len = ui2ipv4(uri->canon_uri+uri->canon_len, val);
+                else
+                    len = ui2ipv4(NULL, val);
+
+                uri->canon_len += len;
+                ++i;
+            } else {
+                /* Write a regular h16 component to the URI. */
+
+                /* Short circuit for the trivial case. */
+                if(values[i] == 0) {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = '0';
+                    ++uri->canon_len;
+                } else {
+                    static const WCHAR formatW[] = {'%','x',0};
+
+                    if(!computeOnly)
+                        uri->canon_len += sprintfW(uri->canon_uri+uri->canon_len,
+                                            formatW, values[i]);
+                    else {
+                        WCHAR tmp[5];
+                        uri->canon_len += sprintfW(tmp, formatW, values[i]);
+                    }
+                }
+            }
+        }
+
+        /* Add the closing ']'. */
+        if(!computeOnly)
+            uri->canon_uri[uri->canon_len] = ']';
+        ++uri->canon_len;
+    }
+
+    uri->host_len = uri->canon_len - uri->host_start;
+
+    if(!computeOnly)
+        TRACE("(%p %p %x %d): Canonicalized IPv6 address %s, len=%d\n", data, uri, flags,
+            computeOnly, debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
+            uri->host_len);
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize the host of the URI (if any). */
+static BOOL canonicalize_host(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    uri->host_start = -1;
+    uri->host_len = 0;
+    uri->domain_offset = -1;
+
+    if(data->host) {
+        switch(data->host_type) {
+        case Uri_HOST_DNS:
+            uri->host_type = Uri_HOST_DNS;
+            if(!canonicalize_reg_name(data, uri, flags, computeOnly))
+                return FALSE;
+
+            break;
+        case Uri_HOST_IPV4:
+            uri->host_type = Uri_HOST_IPV4;
+            if(!canonicalize_ipv4address(data, uri, flags, computeOnly))
+                return FALSE;
+
+            break;
+        case Uri_HOST_IPV6:
+            if(!canonicalize_ipv6address(data, uri, flags, computeOnly))
+                return FALSE;
+
+            uri->host_type = Uri_HOST_IPV6;
+            break;
+        case Uri_HOST_UNKNOWN:
+            if(data->host_len > 0 || data->scheme_type != URL_SCHEME_FILE) {
+                uri->host_start = uri->canon_len;
+
+                /* Nothing happens to unknown host types. */
+                if(!computeOnly)
+                    memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
+                uri->canon_len += data->host_len;
+                uri->host_len = data->host_len;
+            }
+
+            uri->host_type = Uri_HOST_UNKNOWN;
+            break;
+        default:
+            FIXME("(%p %p %x %d): Canonicalization for host type %d not supported.\n", data,
+                    uri, flags, computeOnly, data->host_type);
+            return FALSE;
+       }
+   }
+
+   return TRUE;
+}
+
+static BOOL canonicalize_port(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    BOOL has_default_port = FALSE;
+    USHORT default_port = 0;
+    DWORD i;
+
+    uri->has_port = FALSE;
+
+    /* Check if the scheme has a default port. */
+    for(i = 0; i < sizeof(default_ports)/sizeof(default_ports[0]); ++i) {
+        if(default_ports[i].scheme == data->scheme_type) {
+            has_default_port = TRUE;
+            default_port = default_ports[i].port;
+            break;
+        }
+    }
+
+    if(data->port || has_default_port)
+        uri->has_port = TRUE;
+
+    /* Possible cases:
+     *  1)  Has a port which is the default port.
+     *  2)  Has a port (not the default).
+     *  3)  Doesn't have a port, but, scheme has a default port.
+     *  4)  No port.
+     */
+    if(has_default_port && data->port && data->port_value == default_port) {
+        /* If it's the default port and this flag isn't set, don't do anything. */
+        if(flags & Uri_CREATE_NO_CANONICALIZE) {
+            /* Copy the original port over. */
+            if(!computeOnly) {
+                uri->canon_uri[uri->canon_len] = ':';
+                memcpy(uri->canon_uri+uri->canon_len+1, data->port, data->port_len*sizeof(WCHAR));
+            }
+            uri->canon_len += data->port_len+1;
+        }
+
+        uri->port = default_port;
+    } else if(data->port) {
+        if(!computeOnly)
+            uri->canon_uri[uri->canon_len] = ':';
+        ++uri->canon_len;
+
+        if(flags & Uri_CREATE_NO_CANONICALIZE) {
+            /* Copy the original over without changes. */
+            if(!computeOnly)
+                memcpy(uri->canon_uri+uri->canon_len, data->port, data->port_len*sizeof(WCHAR));
+            uri->canon_len += data->port_len;
+        } else {
+            const WCHAR formatW[] = {'%','u',0};
+            INT len = 0;
+            if(!computeOnly)
+                len = sprintfW(uri->canon_uri+uri->canon_len, formatW, data->port_value);
+            else {
+                WCHAR tmp[6];
+                len = sprintfW(tmp, formatW, data->port_value);
+            }
+            uri->canon_len += len;
+        }
+
+        uri->port = data->port_value;
+    } else if(has_default_port)
+        uri->port = default_port;
+
+    return TRUE;
+}
+
+/* Canonicalizes the authority of the URI represented by the parse_data. */
+static BOOL canonicalize_authority(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    uri->authority_start = uri->canon_len;
+    uri->authority_len = 0;
+
+    if(!canonicalize_userinfo(data, uri, flags, computeOnly))
+        return FALSE;
+
+    if(!canonicalize_host(data, uri, flags, computeOnly))
+        return FALSE;
+
+    if(!canonicalize_port(data, uri, flags, computeOnly))
+        return FALSE;
+
+    if(uri->host_start != -1)
+        uri->authority_len = uri->canon_len - uri->authority_start;
+    else
+        uri->authority_start = -1;
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize the path of a hierarchical URI.
+ *
+ * Things that happen:
+ *  1). Forbidden characters are percent encoded, unless the NO_ENCODE_FORBIDDEN
+ *      flag is set or it's a file URI. Forbidden characters are always encoded
+ *      for file schemes reguardless and forbidden characters are never encoded
+ *      for unknown scheme types.
+ *
+ *  2). For known scheme types '\\' are changed to '/'.
+ *
+ *  3). Percent encoded, unreserved characters are decoded to their actual values.
+ *      Unless the scheme type is unknown. For file schemes any percent encoded
+ *      character in the unreserved or reserved set is decoded.
+ *
+ *  4). For File schemes if the path is starts with a drive letter and doesn't
+ *      start with a '/' then one is appended.
+ *      Ex: file://c:/test.mp3 -> file:///c:/test.mp3
+ *
+ *  5). Dot segments are removed from the path for all scheme types
+ *      unless NO_CANONICALIZE flag is set. Dot segments aren't removed
+ *      for wildcard scheme types.
+ *
+ * NOTES:
+ *      file://c:/test%20test   -> file:///c:/test%2520test
+ *      file://c:/test%3Etest   -> file:///c:/test%253Etest
+ *      file:///c:/test%20test  -> file:///c:/test%20test
+ *      file:///c:/test%test    -> file:///c:/test%25test
+ */
+static BOOL canonicalize_path_hierarchical(const parse_data *data, Uri *uri,
+                                           DWORD flags, BOOL computeOnly) {
+    const WCHAR *ptr;
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+    const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
+
+    BOOL escape_pct = FALSE;
+
+    if(!data->path) {
+        uri->path_start = -1;
+        uri->path_len = 0;
+        return TRUE;
+    }
+
+    uri->path_start = uri->canon_len;
+    ptr = data->path;
+
+    if(is_file && uri->host_start == -1) {
+        /* Check if a '/' needs to be appended for the file scheme. */
+        if(data->path_len > 1 && is_drive_path(ptr) && !(flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+            if(!computeOnly)
+                uri->canon_uri[uri->canon_len] = '/';
+            uri->canon_len++;
+            escape_pct = TRUE;
+        } else if(*ptr == '/') {
+            if(!(flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+                /* Copy the extra '/' over. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = '/';
+                ++uri->canon_len;
+            }
+            ++ptr;
+        }
+
+        if(is_drive_path(ptr)) {
+            if(!computeOnly) {
+                uri->canon_uri[uri->canon_len] = *ptr;
+                /* If theres a '|' after the drive letter, convert it to a ':'. */
+                uri->canon_uri[uri->canon_len+1] = ':';
+            }
+            ptr += 2;
+            uri->canon_len += 2;
+        }
+    }
+
+    for(; ptr < data->path+data->path_len; ++ptr) {
+        if(*ptr == '%') {
+            const WCHAR *tmp = ptr;
+            WCHAR val;
+
+            /* Check if the % represents a valid encoded char, or if it needs encoded. */
+            BOOL force_encode = !check_pct_encoded(&tmp) && is_file;
+            val = decode_pct_val(ptr);
+
+            if(force_encode || escape_pct) {
+                /* Escape the percent sign in the file URI. */
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
+                uri->canon_len += 3;
+            } else if((is_unreserved(val) && known_scheme) ||
+                      (is_file && (is_unreserved(val) || is_reserved(val)))) {
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = val;
+                ++uri->canon_len;
+
+                ptr += 2;
+                continue;
+            } else {
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            }
+        } else if(*ptr == '/' && is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+            /* Convert the '/' back to a '\\'. */
+            if(!computeOnly)
+                uri->canon_uri[uri->canon_len] = '\\';
+            ++uri->canon_len;
+        } else if(*ptr == '\\' && known_scheme) {
+            if(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+                /* Don't convert the '\\' to a '/'. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            } else {
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = '/';
+                ++uri->canon_len;
+            }
+        } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr) &&
+                  (!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) || is_file)) {
+            if(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+                /* Don't escape the character. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            } else {
+                /* Escape the forbidden character. */
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
+                uri->canon_len += 3;
+            }
+        } else {
+            if(!computeOnly)
+                uri->canon_uri[uri->canon_len] = *ptr;
+            ++uri->canon_len;
+        }
+    }
+
+    uri->path_len = uri->canon_len - uri->path_start;
+
+    /* Removing the dot segments only happens when it's not in
+     * computeOnly mode and it's not a wildcard scheme. File schemes
+     * with USE_DOS_PATH set don't get dot segments removed.
+     */
+    if(!(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) &&
+       data->scheme_type != URL_SCHEME_WILDCARD) {
+        if(!(flags & Uri_CREATE_NO_CANONICALIZE) && !computeOnly) {
+            /* Remove the dot segments (if any) and reset everything to the new
+             * correct length.
+             */
+            DWORD new_len = remove_dot_segments(uri->canon_uri+uri->path_start, uri->path_len);
+            uri->canon_len -= uri->path_len-new_len;
+            uri->path_len = new_len;
+        }
+    }
+
+    if(!computeOnly)
+        TRACE("Canonicalized path %s len=%d\n",
+            debugstr_wn(uri->canon_uri+uri->path_start, uri->path_len),
+            uri->path_len);
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize the path for an opaque URI.
+ *
+ * For known scheme types:
+ *  1)  forbidden characters are percent encoded if
+ *      NO_ENCODE_FORBIDDEN_CHARACTERS isn't set.
+ *
+ *  2)  Percent encoded, unreserved characters are decoded
+ *      to their actual values, for known scheme types.
+ *
+ *  3)  '\\' are changed to '/' for known scheme types
+ *      except for mailto schemes.
+ *
+ *  4)  For file schemes, if USE_DOS_PATH is set all '/'
+ *      are converted to backslashes.
+ *
+ *  5)  For file schemes, if USE_DOS_PATH isn't set all '\'
+ *      are converted to forward slashes.
+ */
+static BOOL canonicalize_path_opaque(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    const WCHAR *ptr;
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+    const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
+
+    if(!data->path) {
+        uri->path_start = -1;
+        uri->path_len = 0;
+        return TRUE;
+    }
+
+    uri->path_start = uri->canon_len;
+
+    /* Windows doesn't allow a "//" to appear after the scheme
+     * of a URI, if it's an opaque URI.
+     */
+    if(data->scheme && *(data->path) == '/' && *(data->path+1) == '/') {
+        /* So it inserts a "/." before the "//" if it exists. */
+        if(!computeOnly) {
+            uri->canon_uri[uri->canon_len] = '/';
+            uri->canon_uri[uri->canon_len+1] = '.';
+        }
+
+        uri->canon_len += 2;
+    }
+
+    for(ptr = data->path; ptr < data->path+data->path_len; ++ptr) {
+        if(*ptr == '%' && known_scheme) {
+            WCHAR val = decode_pct_val(ptr);
+
+            if(is_unreserved(val)) {
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = val;
+                ++uri->canon_len;
+
+                ptr += 2;
+                continue;
+            } else {
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            }
+        } else if(*ptr == '/' && is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+            if(!computeOnly)
+                uri->canon_uri[uri->canon_len] = '\\';
+            ++uri->canon_len;
+        } else if(*ptr == '\\' && is_file) {
+            if(!(flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+                /* Convert to a '/'. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = '/';
+                ++uri->canon_len;
+            } else {
+                /* Just copy it over. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            }
+        } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr) &&
+                  !(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS)) {
+            if(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
+                /* Forbidden characters aren't percent encoded for file schemes
+                 * with USE_DOS_PATH set.
+                 */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            } else if(data->scheme_type == URL_SCHEME_MK && *ptr == '\\') {
+                /* MK URIs don't get '\\' percent encoded. */
+                if(!computeOnly)
+                    uri->canon_uri[uri->canon_len] = *ptr;
+                ++uri->canon_len;
+            } else {
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
+                uri->canon_len += 3;
+            }
+        } else {
+            if(!computeOnly)
+                uri->canon_uri[uri->canon_len] = *ptr;
+            ++uri->canon_len;
+        }
+    }
+
+    uri->path_len = uri->canon_len - uri->path_start;
+
+    TRACE("(%p %p %x %d): Canonicalized opaque URI path %s len=%d\n", data, uri, flags, computeOnly,
+        debugstr_wn(uri->canon_uri+uri->path_start, uri->path_len), uri->path_len);
+    return TRUE;
+}
+
+/* Determines how the URI represented by the parse_data should be canonicalized.
+ *
+ * Essentially, if the parse_data represents an hierarchical URI then it calls
+ * canonicalize_authority and the canonicalization functions for the path. If the
+ * URI is opaque it canonicalizes the path of the URI.
+ */
+static BOOL canonicalize_hierpart(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    uri->display_absolute = TRUE;
+
+    if(!data->is_opaque) {
+        /* "//" is only added for non-wildcard scheme types. */
+        if(data->scheme_type != URL_SCHEME_WILDCARD) {
+            if(!computeOnly) {
+                INT pos = uri->canon_len;
+
+                uri->canon_uri[pos] = '/';
+                uri->canon_uri[pos+1] = '/';
+           }
+           uri->canon_len += 2;
+        }
+
+        if(!canonicalize_authority(data, uri, flags, computeOnly))
+            return FALSE;
+
+        /* TODO: Canonicalize the path of the URI. */
+        if(!canonicalize_path_hierarchical(data, uri, flags, computeOnly))
+            return FALSE;
+
+    } else {
+        /* Opaque URI's don't have an authority. */
+        uri->userinfo_start = uri->userinfo_split = -1;
+        uri->userinfo_len = 0;
+        uri->host_start = -1;
+        uri->host_len = 0;
+        uri->host_type = Uri_HOST_UNKNOWN;
+        uri->has_port = FALSE;
+        uri->authority_start = -1;
+        uri->authority_len = 0;
+        uri->domain_offset = -1;
+
+        if(is_hierarchical_scheme(data->scheme_type)) {
+            DWORD i;
+
+            /* Absolute URIs aren't displayed for known scheme types
+             * which should be hierarchical URIs.
+             */
+            uri->display_absolute = FALSE;
+
+            /* Windows also sets the port for these (if they have one). */
+            for(i = 0; i < sizeof(default_ports)/sizeof(default_ports[0]); ++i) {
+                if(data->scheme_type == default_ports[i].scheme) {
+                    uri->has_port = TRUE;
+                    uri->port = default_ports[i].port;
+                    break;
+                }
+            }
+        }
+
+        if(!canonicalize_path_opaque(data, uri, flags, computeOnly))
+            return FALSE;
+    }
+
+    if(uri->path_start > -1 && !computeOnly)
+        /* Finding file extensions happens for both types of URIs. */
+        uri->extension_offset = find_file_extension(uri->canon_uri+uri->path_start, uri->path_len);
+    else
+        uri->extension_offset = -1;
+
+    return TRUE;
+}
+
+/* Attempts to canonicalize the query string of the URI.
+ *
+ * Things that happen:
+ *  1)  For known scheme types forbidden characters
+ *      are percent encoded, unless the NO_DECODE_EXTRA_INFO flag is set
+ *      or NO_ENCODE_FORBIDDEN_CHARACTERS is set.
+ *
+ *  2)  For known scheme types, percent encoded, unreserved characters
+ *      are decoded as long as the NO_DECODE_EXTRA_INFO flag isn't set.
+ */
+static BOOL canonicalize_query(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    const WCHAR *ptr, *end;
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+
+    if(!data->query) {
+        uri->query_start = -1;
+        uri->query_len = 0;
+        return TRUE;
+    }
+
+    uri->query_start = uri->canon_len;
+
+    end = data->query+data->query_len;
+    for(ptr = data->query; ptr < end; ++ptr) {
+        if(*ptr == '%') {
+            if(known_scheme && !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
+                WCHAR val = decode_pct_val(ptr);
+                if(is_unreserved(val)) {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = val;
+                    ++uri->canon_len;
+
+                    ptr += 2;
+                    continue;
+                }
+            }
+        } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr)) {
+            if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) &&
+               !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
+                uri->canon_len += 3;
+                continue;
+            }
+        }
+
+        if(!computeOnly)
+            uri->canon_uri[uri->canon_len] = *ptr;
+        ++uri->canon_len;
+    }
+
+    uri->query_len = uri->canon_len - uri->query_start;
+
+    if(!computeOnly)
+        TRACE("(%p %p %x %d): Canonicalized query string %s len=%d\n", data, uri, flags,
+            computeOnly, debugstr_wn(uri->canon_uri+uri->query_start, uri->query_len),
+            uri->query_len);
+    return TRUE;
+}
+
+static BOOL canonicalize_fragment(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    const WCHAR *ptr, *end;
+    const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
+
+    if(!data->fragment) {
+        uri->fragment_start = -1;
+        uri->fragment_len = 0;
+        return TRUE;
+    }
+
+    uri->fragment_start = uri->canon_len;
+
+    end = data->fragment + data->fragment_len;
+    for(ptr = data->fragment; ptr < end; ++ptr) {
+        if(*ptr == '%') {
+            if(known_scheme && !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
+                WCHAR val = decode_pct_val(ptr);
+                if(is_unreserved(val)) {
+                    if(!computeOnly)
+                        uri->canon_uri[uri->canon_len] = val;
+                    ++uri->canon_len;
+
+                    ptr += 2;
+                    continue;
+                }
+            }
+        } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr)) {
+            if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) &&
+               !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
+                if(!computeOnly)
+                    pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
+                uri->canon_len += 3;
+                continue;
+            }
+        }
+
+        if(!computeOnly)
+            uri->canon_uri[uri->canon_len] = *ptr;
+        ++uri->canon_len;
+    }
+
+    uri->fragment_len = uri->canon_len - uri->fragment_start;
+
+    if(!computeOnly)
+        TRACE("(%p %p %x %d): Canonicalized fragment %s len=%d\n", data, uri, flags,
+            computeOnly, debugstr_wn(uri->canon_uri+uri->fragment_start, uri->fragment_len),
+            uri->fragment_len);
+    return TRUE;
+}
+
+/* Canonicalizes the scheme information specified in the parse_data using the specified flags. */
+static BOOL canonicalize_scheme(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
+    uri->scheme_start = -1;
+    uri->scheme_len = 0;
+
+    if(!data->scheme) {
+        /* The only type of URI that doesn't have to have a scheme is a relative
+         * URI.
+         */
+        if(!data->is_relative) {
+            FIXME("(%p %p %x): Unable to determine the scheme type of %s.\n", data,
+                    uri, flags, debugstr_w(data->uri));
+            return FALSE;
+        }
+    } else {
+        if(!computeOnly) {
+            DWORD i;
+            INT pos = uri->canon_len;
+
+            for(i = 0; i < data->scheme_len; ++i) {
+                /* Scheme name must be lower case after canonicalization. */
+                uri->canon_uri[i + pos] = tolowerW(data->scheme[i]);
+            }
+
+            uri->canon_uri[i + pos] = ':';
+            uri->scheme_start = pos;
+
+            TRACE("(%p %p %x): Canonicalized scheme=%s, len=%d.\n", data, uri, flags,
+                    debugstr_wn(uri->canon_uri,  uri->scheme_len), data->scheme_len);
+        }
+
+        /* This happens in both computation modes. */
+        uri->canon_len += data->scheme_len + 1;
+        uri->scheme_len = data->scheme_len;
+    }
+    return TRUE;
+}
+
+/* Compute's what the length of the URI specified by the parse_data will be
+ * after canonicalization occurs using the specified flags.
+ *
+ * This function will return a non-zero value indicating the length of the canonicalized
+ * URI, or -1 on error.
+ */
+static int compute_canonicalized_length(const parse_data *data, DWORD flags) {
+    Uri uri;
+
+    memset(&uri, 0, sizeof(Uri));
+
+    TRACE("(%p %x): Beginning to compute canonicalized length for URI %s\n", data, flags,
+            debugstr_w(data->uri));
+
+    if(!canonicalize_scheme(data, &uri, flags, TRUE)) {
+        ERR("(%p %x): Failed to compute URI scheme length.\n", data, flags);
+        return -1;
+    }
+
+    if(!canonicalize_hierpart(data, &uri, flags, TRUE)) {
+        ERR("(%p %x): Failed to compute URI hierpart length.\n", data, flags);
+        return -1;
+    }
+
+    if(!canonicalize_query(data, &uri, flags, TRUE)) {
+        ERR("(%p %x): Failed to compute query string length.\n", data, flags);
+        return -1;
+    }
+
+    if(!canonicalize_fragment(data, &uri, flags, TRUE)) {
+        ERR("(%p %x): Failed to compute fragment length.\n", data, flags);
+        return -1;
+    }
+
+    TRACE("(%p %x): Finished computing canonicalized URI length. length=%d\n", data, flags, uri.canon_len);
+
+    return uri.canon_len;
+}
+
+/* Canonicalizes the URI data specified in the parse_data, using the given flags. If the
+ * canonicalization succeededs it will store all the canonicalization information
+ * in the pointer to the Uri.
+ *
+ * To canonicalize a URI this function first computes what the length of the URI
+ * specified by the parse_data will be. Once this is done it will then perfom the actual
+ * canonicalization of the URI.
+ */
+static HRESULT canonicalize_uri(const parse_data *data, Uri *uri, DWORD flags) {
+    INT len;
+
+    uri->canon_uri = NULL;
+    len = uri->canon_size = uri->canon_len = 0;
+
+    TRACE("(%p %p %x): beginning to canonicalize URI %s.\n", data, uri, flags, debugstr_w(data->uri));
+
+    /* First try to compute the length of the URI. */
+    len = compute_canonicalized_length(data, flags);
+    if(len == -1) {
+        ERR("(%p %p %x): Could not compute the canonicalized length of %s.\n", data, uri, flags,
+                debugstr_w(data->uri));
+        return E_INVALIDARG;
+    }
+
+    uri->canon_uri = heap_alloc((len+1)*sizeof(WCHAR));
+    if(!uri->canon_uri)
+        return E_OUTOFMEMORY;
+
+    uri->canon_size = len;
+    if(!canonicalize_scheme(data, uri, flags, FALSE)) {
+        ERR("(%p %p %x): Unable to canonicalize the scheme of the URI.\n", data, uri, flags);
+        heap_free(uri->canon_uri);
+        return E_INVALIDARG;
+    }
+    uri->scheme_type = data->scheme_type;
+
+    if(!canonicalize_hierpart(data, uri, flags, FALSE)) {
+        ERR("(%p %p %x): Unable to canonicalize the heirpart of the URI\n", data, uri, flags);
+        heap_free(uri->canon_uri);
+        return E_INVALIDARG;
+    }
+
+    if(!canonicalize_query(data, uri, flags, FALSE)) {
+        ERR("(%p %p %x): Unable to canonicalize query string of the URI.\n",
+            data, uri, flags);
+        return E_INVALIDARG;
+    }
+
+    if(!canonicalize_fragment(data, uri, flags, FALSE)) {
+        ERR("(%p %p %x): Unable to canonicalize fragment of the URI.\n",
+            data, uri, flags);
+        return E_INVALIDARG;
+    }
+
+    /* There's a possibility we didn't use all the space we allocated
+     * earlier.
+     */
+    if(uri->canon_len < uri->canon_size) {
+        /* This happens if the URI is hierarchical and dot
+         * segments were removed from it's path.
+         */
+        WCHAR *tmp = heap_realloc(uri->canon_uri, (uri->canon_len+1)*sizeof(WCHAR));
+        if(!tmp)
+            return E_OUTOFMEMORY;
+
+        uri->canon_uri = tmp;
+        uri->canon_size = uri->canon_len;
+    }
+
+    uri->canon_uri[uri->canon_len] = '\0';
+    TRACE("(%p %p %x): finished canonicalizing the URI. uri=%s\n", data, uri, flags, debugstr_w(uri->canon_uri));
+
+    return S_OK;
+}
+
+static HRESULT get_builder_component(LPWSTR *component, DWORD *component_len,
+                                     LPCWSTR source, DWORD source_len,
+                                     LPCWSTR *output, DWORD *output_len)
+{
+    if(!output_len) {
+        if(output)
+            *output = NULL;
+        return E_POINTER;
+    }
+
+    if(!output) {
+        *output_len = 0;
+        return E_POINTER;
+    }
+
+    if(!(*component) && source) {
+        /* Allocate 'component', and copy the contents from 'source'
+         * into the new allocation.
+         */
+        *component = heap_alloc((source_len+1)*sizeof(WCHAR));
+        if(!(*component))
+            return E_OUTOFMEMORY;
+
+        memcpy(*component, source, source_len*sizeof(WCHAR));
+        (*component)[source_len] = '\0';
+        *component_len = source_len;
+    }
+
+    *output = *component;
+    *output_len = *component_len;
+    return *output ? S_OK : S_FALSE;
+}
+
+/* Allocates 'component' and copies the string from 'new_value' into 'component'.
+ * If 'prefix' is set and 'new_value' isn't NULL, then it checks if 'new_value'
+ * starts with 'prefix'. If it doesn't then 'prefix' is prepended to 'component'.
+ *
+ * If everything is successful, then will set 'success_flag' in 'flags'.
+ */
+static HRESULT set_builder_component(LPWSTR *component, DWORD *component_len, LPCWSTR new_value,
+                                     WCHAR prefix, DWORD *flags, DWORD success_flag)
+{
+    heap_free(*component);
+
+    if(!new_value) {
+        *component = NULL;
+        *component_len = 0;
+    } else {
+        BOOL add_prefix = FALSE;
+        DWORD len = lstrlenW(new_value);
+        DWORD pos = 0;
+
+        if(prefix && *new_value != prefix) {
+            add_prefix = TRUE;
+            *component = heap_alloc((len+2)*sizeof(WCHAR));
+        } else
+            *component = heap_alloc((len+1)*sizeof(WCHAR));
+
+        if(!(*component))
+            return E_OUTOFMEMORY;
+
+        if(add_prefix)
+            (*component)[pos++] = prefix;
+
+        memcpy(*component+pos, new_value, (len+1)*sizeof(WCHAR));
+        *component_len = len+pos;
+    }
+
+    *flags |= success_flag;
+    return S_OK;
+}
+
+#define URI(x)         ((IUri*)  &(x)->lpIUriVtbl)
+#define URIBUILDER(x)  ((IUriBuilder*)  &(x)->lpIUriBuilderVtbl)
+
+static void reset_builder(UriBuilder *builder) {
+    if(builder->uri)
+        IUri_Release(URI(builder->uri));
+    builder->uri = NULL;
+
+    heap_free(builder->fragment);
+    builder->fragment = NULL;
+    builder->fragment_len = 0;
+
+    heap_free(builder->host);
+    builder->host = NULL;
+    builder->host_len = 0;
+
+    heap_free(builder->password);
+    builder->password = NULL;
+    builder->password_len = 0;
+
+    heap_free(builder->path);
+    builder->path = NULL;
+    builder->path_len = 0;
+
+    heap_free(builder->query);
+    builder->query = NULL;
+    builder->query_len = 0;
+
+    heap_free(builder->scheme);
+    builder->scheme = NULL;
+    builder->scheme_len = 0;
+
+    heap_free(builder->username);
+    builder->username = NULL;
+    builder->username_len = 0;
+
+    builder->has_port = FALSE;
+    builder->port = 0;
+    builder->modified_props = 0;
+}
+
+static HRESULT validate_scheme_name(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *component;
+    const WCHAR *ptr;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->scheme) {
+        ptr = builder->scheme;
+        expected_len = builder->scheme_len;
+    } else if(builder->uri && builder->uri->scheme_start > -1) {
+        ptr = builder->uri->canon_uri+builder->uri->scheme_start;
+        expected_len = builder->uri->scheme_len;
+    } else {
+        static const WCHAR nullW[] = {0};
+        ptr = nullW;
+        expected_len = 0;
+    }
+
+    component = ptr;
+    pptr = &ptr;
+    if(parse_scheme(pptr, data, flags, ALLOW_NULL_TERM_SCHEME) &&
+       data->scheme_len == expected_len) {
+        if(data->scheme)
+            TRACE("(%p %p %x): Found valid scheme component %s len=%d.\n", builder, data, flags,
+               debugstr_wn(data->scheme, data->scheme_len), data->scheme_len);
+    } else {
+        TRACE("(%p %p %x): Invalid scheme component found %s.\n", builder, data, flags,
+            debugstr_wn(component, expected_len));
+        return INET_E_INVALID_URL;
+   }
+
+    return S_OK;
+}
+
+static HRESULT validate_username(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *ptr;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->username) {
+        ptr = builder->username;
+        expected_len = builder->username_len;
+    } else if(!(builder->modified_props & Uri_HAS_USER_NAME) && builder->uri &&
+              builder->uri->userinfo_start > -1 && builder->uri->userinfo_split != 0) {
+        /* Just use the username from the base Uri. */
+        data->username = builder->uri->canon_uri+builder->uri->userinfo_start;
+        data->username_len = (builder->uri->userinfo_split > -1) ?
+                                        builder->uri->userinfo_split : builder->uri->userinfo_len;
+        ptr = NULL;
+    } else {
+        ptr = NULL;
+        expected_len = 0;
+    }
+
+    if(ptr) {
+        const WCHAR *component = ptr;
+        pptr = &ptr;
+        if(parse_username(pptr, data, flags, ALLOW_NULL_TERM_USER_NAME) &&
+           data->username_len == expected_len)
+            TRACE("(%p %p %x): Found valid username component %s len=%d.\n", builder, data, flags,
+                debugstr_wn(data->username, data->username_len), data->username_len);
+        else {
+            TRACE("(%p %p %x): Invalid username component found %s.\n", builder, data, flags,
+                debugstr_wn(component, expected_len));
+            return INET_E_INVALID_URL;
+        }
+    }
+
+    return S_OK;
+}
+
+static HRESULT validate_password(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *ptr;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->password) {
+        ptr = builder->password;
+        expected_len = builder->password_len;
+    } else if(!(builder->modified_props & Uri_HAS_PASSWORD) && builder->uri &&
+              builder->uri->userinfo_split > -1) {
+        data->password = builder->uri->canon_uri+builder->uri->userinfo_start+builder->uri->userinfo_split+1;
+        data->password_len = builder->uri->userinfo_len-builder->uri->userinfo_split-1;
+        ptr = NULL;
+    } else {
+        ptr = NULL;
+        expected_len = 0;
+    }
+
+    if(ptr) {
+        const WCHAR *component = ptr;
+        pptr = &ptr;
+        if(parse_password(pptr, data, flags, ALLOW_NULL_TERM_PASSWORD) &&
+           data->password_len == expected_len)
+            TRACE("(%p %p %x): Found valid password component %s len=%d.\n", builder, data, flags,
+                debugstr_wn(data->password, data->password_len), data->password_len);
+        else {
+            TRACE("(%p %p %x): Invalid password component found %s.\n", builder, data, flags,
+                debugstr_wn(component, expected_len));
+            return INET_E_INVALID_URL;
+        }
+    }
+
+    return S_OK;
+}
+
+static HRESULT validate_userinfo(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    HRESULT hr;
+
+    hr = validate_username(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    hr = validate_password(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    return S_OK;
+}
+
+static HRESULT validate_host(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *ptr;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->host) {
+        ptr = builder->host;
+        expected_len = builder->host_len;
+    } else if(!(builder->modified_props & Uri_HAS_HOST) && builder->uri && builder->uri->host_start > -1) {
+        ptr = builder->uri->canon_uri + builder->uri->host_start;
+        expected_len = builder->uri->host_len;
+    } else
+        ptr = NULL;
+
+    if(ptr) {
+        const WCHAR *component = ptr;
+        DWORD extras = ALLOW_BRACKETLESS_IP_LITERAL|IGNORE_PORT_DELIMITER|SKIP_IP_FUTURE_CHECK;
+        pptr = &ptr;
+
+        if(parse_host(pptr, data, flags, extras) && data->host_len == expected_len)
+            TRACE("(%p %p %x): Found valid host name %s len=%d type=%d.\n", builder, data, flags,
+                debugstr_wn(data->host, data->host_len), data->host_len, data->host_type);
+        else {
+            TRACE("(%p %p %x): Invalid host name found %s.\n", builder, data, flags,
+                debugstr_wn(component, expected_len));
+            return INET_E_INVALID_URL;
+        }
+    }
+
+    return S_OK;
+}
+
+static void setup_port(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    if(builder->modified_props & Uri_HAS_PORT) {
+        if(builder->has_port) {
+            data->has_port = TRUE;
+            data->port_value = builder->port;
+        }
+    } else if(builder->uri && builder->uri->has_port) {
+        data->has_port = TRUE;
+        data->port_value = builder->uri->port;
+    }
+
+    if(data->has_port)
+        TRACE("(%p %p %x): Using %u as port for IUri.\n", builder, data, flags, data->port_value);
+}
+
+static HRESULT validate_path(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *ptr = NULL;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->path) {
+        ptr = builder->path;
+        expected_len = builder->path_len;
+    } else if(!(builder->modified_props & Uri_HAS_PATH) &&
+              builder->uri && builder->uri->path_start > -1) {
+        ptr = builder->uri->canon_uri+builder->uri->path_start;
+        expected_len = builder->uri->path_len;
+    }
+
+    if(ptr) {
+        BOOL valid = FALSE;
+        const WCHAR *component = ptr;
+        pptr = &ptr;
+
+        /* How the path is validated depends on what type of
+         * URI it is.
+         */
+        valid = data->is_opaque ?
+            parse_path_opaque(pptr, data, flags) : parse_path_hierarchical(pptr, data, flags);
+
+        if(!valid || expected_len != data->path_len) {
+            TRACE("(%p %p %x): Invalid path componet %s.\n", builder, data, flags,
+                debugstr_wn(component, expected_len));
+            return INET_E_INVALID_URL;
+        }
+
+        TRACE("(%p %p %x): Valid path component %s len=%d.\n", builder, data, flags,
+            debugstr_wn(data->path, data->path_len), data->path_len);
+    }
+
+    return S_OK;
+}
+
+static HRESULT validate_query(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *ptr = NULL;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->query) {
+        ptr = builder->query;
+        expected_len = builder->query_len;
+    } else if(!(builder->modified_props & Uri_HAS_QUERY) && builder->uri &&
+              builder->uri->query_start > -1) {
+        ptr = builder->uri->canon_uri+builder->uri->query_start;
+        expected_len = builder->uri->query_len;
+    }
+
+    if(ptr) {
+        const WCHAR *component = ptr;
+        pptr = &ptr;
+
+        if(parse_query(pptr, data, flags) && expected_len == data->query_len)
+            TRACE("(%p %p %x): Valid query component %s len=%d.\n", builder, data, flags,
+                debugstr_wn(data->query, data->query_len), data->query_len);
+        else {
+            TRACE("(%p %p %x): Invalid query component %s.\n", builder, data, flags,
+                debugstr_wn(component, expected_len));
+            return INET_E_INVALID_URL;
+        }
+    }
+
+    return S_OK;
+}
+
+static HRESULT validate_fragment(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    const WCHAR *ptr = NULL;
+    const WCHAR **pptr;
+    DWORD expected_len;
+
+    if(builder->fragment) {
+        ptr = builder->fragment;
+        expected_len = builder->fragment_len;
+    } else if(!(builder->modified_props & Uri_HAS_FRAGMENT) && builder->uri &&
+              builder->uri->fragment_start > -1) {
+        ptr = builder->uri->canon_uri+builder->uri->fragment_start;
+        expected_len = builder->uri->fragment_len;
+    }
+
+    if(ptr) {
+        const WCHAR *component = ptr;
+        pptr = &ptr;
+
+        if(parse_query(pptr, data, flags) && expected_len == data->fragment_len)
+            TRACE("(%p %p %x): Valid fragment component %s len=%d.\n", builder, data, flags,
+                debugstr_wn(data->fragment, data->fragment_len), data->fragment_len);
+        else {
+            TRACE("(%p %p %x): Invalid fragment component %s.\n", builder, data, flags,
+                debugstr_wn(component, expected_len));
+            return INET_E_INVALID_URL;
+        }
+    }
+
+    return S_OK;
+}
+
+static HRESULT validate_components(const UriBuilder *builder, parse_data *data, DWORD flags) {
+    HRESULT hr;
+
+    memset(data, 0, sizeof(parse_data));
+
+    TRACE("(%p %p %x): Beginning to validate builder components.\n", builder, data, flags);
+
+    hr = validate_scheme_name(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    /* Extra validation for file schemes. */
+    if(data->scheme_type == URL_SCHEME_FILE) {
+        if((builder->password || (builder->uri && builder->uri->userinfo_split > -1)) ||
+           (builder->username || (builder->uri && builder->uri->userinfo_start > -1))) {
+            TRACE("(%p %p %x): File schemes can't contain a username or password.\n",
+                builder, data, flags);
+            return INET_E_INVALID_URL;
+        }
+    }
+
+    hr = validate_userinfo(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    hr = validate_host(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    /* The URI is opaque if it doesn't have an authority component. */
+    data->is_opaque = !data->username && !data->password && !data->host;
+
+    setup_port(builder, data, flags);
+
+    hr = validate_path(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    hr = validate_query(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    hr = validate_fragment(builder, data, flags);
+    if(FAILED(hr))
+        return hr;
+
+    TRACE("(%p %p %x): Finished validating builder components.\n", builder, data, flags);
+
+    return S_OK;
+}
+
+static HRESULT build_uri(const UriBuilder *builder, IUri **uri, DWORD create_flags,
+                         DWORD use_orig_flags, DWORD encoding_mask)
+{
+    HRESULT hr;
+    parse_data data;
+
+    if(!uri)
+        return E_POINTER;
+
+    if(encoding_mask && (!builder->uri || builder->modified_props)) {
+        *uri = NULL;
+        return E_NOTIMPL;
+    }
+
+    /* Decide what flags should be used when creating the Uri. */
+    if((use_orig_flags & UriBuilder_USE_ORIGINAL_FLAGS) && builder->uri)
+        create_flags = builder->uri->create_flags;
+    else {
+        if(has_invalid_flag_combination(create_flags)) {
+            *uri = NULL;
+            return E_INVALIDARG;
+        }
+
+        /* Set the default flags if they don't cause a conflict. */
+        apply_default_flags(&create_flags);
+    }
+
+    /* Return the base IUri if no changes have been made and the create_flags match. */
+    if(builder->uri && !builder->modified_props && builder->uri->create_flags == create_flags) {
+        *uri = URI(builder->uri);
+        IUri_AddRef(*uri);
+        return S_OK;
+    }
+
+    hr = validate_components(builder, &data, create_flags);
+    if(FAILED(hr)) {
+        *uri = NULL;
+        return hr;
+    }
+
+    return E_NOTIMPL;
+}
+
+#define URI_THIS(iface) DEFINE_THIS(Uri, IUri, iface)
+
+static HRESULT WINAPI Uri_QueryInterface(IUri *iface, REFIID riid, void **ppv)
+{
+    Uri *This = URI_THIS(iface);
+
+    if(IsEqualGUID(&IID_IUnknown, riid)) {
+        TRACE("(%p)->(IID_IUnknown %p)\n", This, ppv);
+        *ppv = URI(This);
+    }else if(IsEqualGUID(&IID_IUri, riid)) {
+        TRACE("(%p)->(IID_IUri %p)\n", This, ppv);
+        *ppv = URI(This);
+    }else if(IsEqualGUID(&IID_IUriObj, riid)) {
+        TRACE("(%p)->(IID_IUriObj %p)\n", This, ppv);
+        *ppv = This;
+        return S_OK;
+    }else {
+        TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppv);
+        *ppv = NULL;
+        return E_NOINTERFACE;
+    }
+
+    IUnknown_AddRef((IUnknown*)*ppv);
+    return S_OK;
+}
+
+static ULONG WINAPI Uri_AddRef(IUri *iface)
+{
+    Uri *This = URI_THIS(iface);
+    LONG ref = InterlockedIncrement(&This->ref);
+
+    TRACE("(%p) ref=%d\n", This, ref);
+
+    return ref;
+}
+
+static ULONG WINAPI Uri_Release(IUri *iface)
+{
+    Uri *This = URI_THIS(iface);
+    LONG ref = InterlockedDecrement(&This->ref);
+
+    TRACE("(%p) ref=%d\n", This, ref);
+
+    if(!ref) {
+        SysFreeString(This->raw_uri);
+        heap_free(This->canon_uri);
+        heap_free(This);
+    }
+
+    return ref;
+}
+
+static HRESULT WINAPI Uri_GetPropertyBSTR(IUri *iface, Uri_PROPERTY uriProp, BSTR *pbstrProperty, DWORD dwFlags)
+{
+    Uri *This = URI_THIS(iface);
+    HRESULT hres;
+    TRACE("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
+
+    if(!pbstrProperty)
+        return E_POINTER;
+
+    if(uriProp > Uri_PROPERTY_STRING_LAST) {
+        /* Windows allocates an empty BSTR for invalid Uri_PROPERTY's. */
+        *pbstrProperty = SysAllocStringLen(NULL, 0);
+        if(!(*pbstrProperty))
+            return E_OUTOFMEMORY;
+
+        /* It only returns S_FALSE for the ZONE property... */
+        if(uriProp == Uri_PROPERTY_ZONE)
+            return S_FALSE;
+        else
+            return S_OK;
+    }
+
+    /* Don't have support for flags yet. */
+    if(dwFlags) {
+        FIXME("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
+        return E_NOTIMPL;
+    }
+
+    switch(uriProp) {
+    case Uri_PROPERTY_ABSOLUTE_URI:
+        if(!This->display_absolute) {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        } else {
+            if(This->scheme_type != URL_SCHEME_UNKNOWN && This->userinfo_start > -1) {
+                if(This->userinfo_len == 0) {
+                    /* Don't include the '@' after the userinfo component. */
+                    *pbstrProperty = SysAllocStringLen(NULL, This->canon_len-1);
+                    hres = S_OK;
+                    if(*pbstrProperty) {
+                        /* Copy everything before it. */
+                        memcpy(*pbstrProperty, This->canon_uri, This->userinfo_start*sizeof(WCHAR));
+
+                        /* And everything after it. */
+                        memcpy(*pbstrProperty+This->userinfo_start, This->canon_uri+This->userinfo_start+1,
+                               (This->canon_len-This->userinfo_start-1)*sizeof(WCHAR));
+                    }
+                } else if(This->userinfo_split == 0 && This->userinfo_len == 1) {
+                    /* Don't include the ":@" */
+                    *pbstrProperty = SysAllocStringLen(NULL, This->canon_len-2);
+                    hres = S_OK;
+                    if(*pbstrProperty) {
+                        memcpy(*pbstrProperty, This->canon_uri, This->userinfo_start*sizeof(WCHAR));
+                        memcpy(*pbstrProperty+This->userinfo_start, This->canon_uri+This->userinfo_start+2,
+                               (This->canon_len-This->userinfo_start-2)*sizeof(WCHAR));
+                    }
+                } else {
+                    *pbstrProperty = SysAllocString(This->canon_uri);
+                    hres = S_OK;
+                }
+            } else {
+                *pbstrProperty = SysAllocString(This->canon_uri);
+                hres = S_OK;
+            }
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_AUTHORITY:
+        if(This->authority_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->authority_start, This->authority_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_DISPLAY_URI:
+        /* The Display URI contains everything except for the userinfo for known
+         * scheme types.
+         */
+        if(This->scheme_type != URL_SCHEME_UNKNOWN && This->userinfo_start > -1) {
+            *pbstrProperty = SysAllocStringLen(NULL, This->canon_len-This->userinfo_len);
+
+            if(*pbstrProperty) {
+                /* Copy everything before the userinfo over. */
+                memcpy(*pbstrProperty, This->canon_uri, This->userinfo_start*sizeof(WCHAR));
+                /* Copy everything after the userinfo over. */
+                memcpy(*pbstrProperty+This->userinfo_start,
+                   This->canon_uri+This->userinfo_start+This->userinfo_len+1,
+                   (This->canon_len-(This->userinfo_start+This->userinfo_len+1))*sizeof(WCHAR));
+            }
+        } else
+            *pbstrProperty = SysAllocString(This->canon_uri);
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+        else
+            hres = S_OK;
+
+        break;
+    case Uri_PROPERTY_DOMAIN:
+        if(This->domain_offset > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->host_start+This->domain_offset,
+                                               This->host_len-This->domain_offset);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_EXTENSION:
+        if(This->extension_offset > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->path_start+This->extension_offset,
+                                               This->path_len-This->extension_offset);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_FRAGMENT:
+        if(This->fragment_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->fragment_start, This->fragment_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_HOST:
+        if(This->host_start > -1) {
+            /* The '[' and ']' aren't included for IPv6 addresses. */
+            if(This->host_type == Uri_HOST_IPV6)
+                *pbstrProperty = SysAllocStringLen(This->canon_uri+This->host_start+1, This->host_len-2);
+            else
+                *pbstrProperty = SysAllocStringLen(This->canon_uri+This->host_start, This->host_len);
+
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_PASSWORD:
+        if(This->userinfo_split > -1) {
+            *pbstrProperty = SysAllocStringLen(
+                This->canon_uri+This->userinfo_start+This->userinfo_split+1,
+                This->userinfo_len-This->userinfo_split-1);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            return E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_PATH:
+        if(This->path_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->path_start, This->path_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_PATH_AND_QUERY:
+        if(This->path_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->path_start, This->path_len+This->query_len);
+            hres = S_OK;
+        } else if(This->query_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->query_start, This->query_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_QUERY:
+        if(This->query_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->query_start, This->query_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_RAW_URI:
+        *pbstrProperty = SysAllocString(This->raw_uri);
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+        else
+            hres = S_OK;
+        break;
+    case Uri_PROPERTY_SCHEME_NAME:
+        if(This->scheme_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri + This->scheme_start, This->scheme_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_USER_INFO:
+        if(This->userinfo_start > -1) {
+            *pbstrProperty = SysAllocStringLen(This->canon_uri+This->userinfo_start, This->userinfo_len);
+            hres = S_OK;
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            hres = E_OUTOFMEMORY;
+
+        break;
+    case Uri_PROPERTY_USER_NAME:
+        if(This->userinfo_start > -1 && This->userinfo_split != 0) {
+            /* If userinfo_split is set, that means a password exists
+             * so the username is only from userinfo_start to userinfo_split.
+             */
+            if(This->userinfo_split > -1) {
+                *pbstrProperty = SysAllocStringLen(This->canon_uri + This->userinfo_start, This->userinfo_split);
+                hres = S_OK;
+            } else {
+                *pbstrProperty = SysAllocStringLen(This->canon_uri + This->userinfo_start, This->userinfo_len);
+                hres = S_OK;
+            }
+        } else {
+            *pbstrProperty = SysAllocStringLen(NULL, 0);
+            hres = S_FALSE;
+        }
+
+        if(!(*pbstrProperty))
+            return E_OUTOFMEMORY;
+
+        break;
+    default:
+        FIXME("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
+        hres = E_NOTIMPL;
+    }
+
+    return hres;
+}
+
+static HRESULT WINAPI Uri_GetPropertyLength(IUri *iface, Uri_PROPERTY uriProp, DWORD *pcchProperty, DWORD dwFlags)
+{
+    Uri *This = URI_THIS(iface);
+    HRESULT hres;
+    TRACE("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
+
+    if(!pcchProperty)
+        return E_INVALIDARG;
+
+    /* Can only return a length for a property if it's a string. */
+    if(uriProp > Uri_PROPERTY_STRING_LAST)
+        return E_INVALIDARG;
+
+    /* Don't have support for flags yet. */
+    if(dwFlags) {
+        FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
+        return E_NOTIMPL;
+    }
+
+    switch(uriProp) {
+    case Uri_PROPERTY_ABSOLUTE_URI:
+        if(!This->display_absolute) {
+            *pcchProperty = 0;
+            hres = S_FALSE;
+        } else {
+            if(This->scheme_type != URL_SCHEME_UNKNOWN) {
+                if(This->userinfo_start > -1 && This->userinfo_len == 0)
+                    /* Don't include the '@' in the length. */
+                    *pcchProperty = This->canon_len-1;
+                else if(This->userinfo_start > -1 && This->userinfo_len == 1 &&
+                        This->userinfo_split == 0)
+                    /* Don't include the ":@" in the length. */
+                    *pcchProperty = This->canon_len-2;
+                else
+                    *pcchProperty = This->canon_len;
+            } else
+                *pcchProperty = This->canon_len;
+
+            hres = S_OK;
+        }
+
+        break;
+    case Uri_PROPERTY_AUTHORITY:
+        *pcchProperty = This->authority_len;
+        hres = (This->authority_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_DISPLAY_URI:
+        if(This->scheme_type != URL_SCHEME_UNKNOWN && This->userinfo_start > -1)
+            *pcchProperty = This->canon_len-This->userinfo_len-1;
+        else
+            *pcchProperty = This->canon_len;
+
+        hres = S_OK;
+        break;
+    case Uri_PROPERTY_DOMAIN:
+        if(This->domain_offset > -1)
+            *pcchProperty = This->host_len - This->domain_offset;
+        else
+            *pcchProperty = 0;
+
+        hres = (This->domain_offset > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_EXTENSION:
+        if(This->extension_offset > -1) {
+            *pcchProperty = This->path_len - This->extension_offset;
+            hres = S_OK;
+        } else {
+            *pcchProperty = 0;
+            hres = S_FALSE;
+        }
+
+        break;
+    case Uri_PROPERTY_FRAGMENT:
+        *pcchProperty = This->fragment_len;
+        hres = (This->fragment_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_HOST:
+        *pcchProperty = This->host_len;
+
+        /* '[' and ']' aren't included in the length. */
+        if(This->host_type == Uri_HOST_IPV6)
+            *pcchProperty -= 2;
+
+        hres = (This->host_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_PASSWORD:
+        *pcchProperty = (This->userinfo_split > -1) ? This->userinfo_len-This->userinfo_split-1 : 0;
+        hres = (This->userinfo_split > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_PATH:
+        *pcchProperty = This->path_len;
+        hres = (This->path_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_PATH_AND_QUERY:
+        *pcchProperty = This->path_len+This->query_len;
+        hres = (This->path_start > -1 || This->query_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_QUERY:
+        *pcchProperty = This->query_len;
+        hres = (This->query_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_RAW_URI:
+        *pcchProperty = SysStringLen(This->raw_uri);
+        hres = S_OK;
+        break;
+    case Uri_PROPERTY_SCHEME_NAME:
+        *pcchProperty = This->scheme_len;
+        hres = (This->scheme_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_USER_INFO:
+        *pcchProperty = This->userinfo_len;
+        hres = (This->userinfo_start > -1) ? S_OK : S_FALSE;
+        break;
+    case Uri_PROPERTY_USER_NAME:
+        *pcchProperty = (This->userinfo_split > -1) ? This->userinfo_split : This->userinfo_len;
+        if(This->userinfo_split == 0)
+            hres = S_FALSE;
+        else
+            hres = (This->userinfo_start > -1) ? S_OK : S_FALSE;
+        break;
+    default:
+        FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
+        hres = E_NOTIMPL;
+    }
+
+    return hres;
+}
+
+static HRESULT WINAPI Uri_GetPropertyDWORD(IUri *iface, Uri_PROPERTY uriProp, DWORD *pcchProperty, DWORD dwFlags)
+{
+    Uri *This = URI_THIS(iface);
+    HRESULT hres;
+
+    TRACE("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
+
+    if(!pcchProperty)
+        return E_INVALIDARG;
+
+    /* Microsoft's implementation for the ZONE property of a URI seems to be lacking...
+     * From what I can tell, instead of checking which URLZONE the URI belongs to it
+     * simply assigns URLZONE_INVALID and returns E_NOTIMPL. This also applies to the GetZone
+     * function.
+     */
+    if(uriProp == Uri_PROPERTY_ZONE) {
+        *pcchProperty = URLZONE_INVALID;
+        return E_NOTIMPL;
+    }
+
+    if(uriProp < Uri_PROPERTY_DWORD_START) {
+        *pcchProperty = 0;
+        return E_INVALIDARG;
+    }
+
+    switch(uriProp) {
+    case Uri_PROPERTY_HOST_TYPE:
+        *pcchProperty = This->host_type;
+        hres = S_OK;
+        break;
+    case Uri_PROPERTY_PORT:
+        if(!This->has_port) {
+            *pcchProperty = 0;
+            hres = S_FALSE;
+        } else {
+            *pcchProperty = This->port;
+            hres = S_OK;
+        }
+
+        break;
+    case Uri_PROPERTY_SCHEME:
+        *pcchProperty = This->scheme_type;
+        hres = S_OK;
+        break;
+    default:
+        FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
+        hres = E_NOTIMPL;
+    }
+
+    return hres;
+}
+
+static HRESULT WINAPI Uri_HasProperty(IUri *iface, Uri_PROPERTY uriProp, BOOL *pfHasProperty)
+{
+    Uri *This = URI_THIS(iface);
+    TRACE("(%p)->(%d %p)\n", This, uriProp, pfHasProperty);
+
+    if(!pfHasProperty)
+        return E_INVALIDARG;
+
+    switch(uriProp) {
+    case Uri_PROPERTY_ABSOLUTE_URI:
+        *pfHasProperty = This->display_absolute;
+        break;
+    case Uri_PROPERTY_AUTHORITY:
+        *pfHasProperty = This->authority_start > -1;
+        break;
+    case Uri_PROPERTY_DISPLAY_URI:
+        *pfHasProperty = TRUE;
+        break;
+    case Uri_PROPERTY_DOMAIN:
+        *pfHasProperty = This->domain_offset > -1;
+        break;
+    case Uri_PROPERTY_EXTENSION:
+        *pfHasProperty = This->extension_offset > -1;
+        break;
+    case Uri_PROPERTY_FRAGMENT:
+        *pfHasProperty = This->fragment_start > -1;
+        break;
+    case Uri_PROPERTY_HOST:
+        *pfHasProperty = This->host_start > -1;
+        break;
+    case Uri_PROPERTY_PASSWORD:
+        *pfHasProperty = This->userinfo_split > -1;
+        break;
+    case Uri_PROPERTY_PATH:
+        *pfHasProperty = This->path_start > -1;
+        break;
+    case Uri_PROPERTY_PATH_AND_QUERY:
+        *pfHasProperty = (This->path_start > -1 || This->query_start > -1);
+        break;
+    case Uri_PROPERTY_QUERY:
+        *pfHasProperty = This->query_start > -1;
+        break;
+    case Uri_PROPERTY_RAW_URI:
+        *pfHasProperty = TRUE;
+        break;
+    case Uri_PROPERTY_SCHEME_NAME:
+        *pfHasProperty = This->scheme_start > -1;
+        break;
+    case Uri_PROPERTY_USER_INFO:
+        *pfHasProperty = This->userinfo_start > -1;
+        break;
+    case Uri_PROPERTY_USER_NAME:
+        if(This->userinfo_split == 0)
+            *pfHasProperty = FALSE;
+        else
+            *pfHasProperty = This->userinfo_start > -1;
+        break;
+    case Uri_PROPERTY_HOST_TYPE:
+        *pfHasProperty = TRUE;
+        break;
+    case Uri_PROPERTY_PORT:
+        *pfHasProperty = This->has_port;
+        break;
+    case Uri_PROPERTY_SCHEME:
+        *pfHasProperty = TRUE;
+        break;
+    case Uri_PROPERTY_ZONE:
+        *pfHasProperty = FALSE;
+        break;
+    default:
+        FIXME("(%p)->(%d %p): Unsupported property type.\n", This, uriProp, pfHasProperty);
+        return E_NOTIMPL;
+    }
+
+    return S_OK;
+}
+
+static HRESULT WINAPI Uri_GetAbsoluteUri(IUri *iface, BSTR *pstrAbsoluteUri)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrAbsoluteUri);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_ABSOLUTE_URI, pstrAbsoluteUri, 0);
+}
+
+static HRESULT WINAPI Uri_GetAuthority(IUri *iface, BSTR *pstrAuthority)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrAuthority);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_AUTHORITY, pstrAuthority, 0);
+}
+
+static HRESULT WINAPI Uri_GetDisplayUri(IUri *iface, BSTR *pstrDisplayUri)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrDisplayUri);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_DISPLAY_URI, pstrDisplayUri, 0);
+}
+
+static HRESULT WINAPI Uri_GetDomain(IUri *iface, BSTR *pstrDomain)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrDomain);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_DOMAIN, pstrDomain, 0);
+}
+
+static HRESULT WINAPI Uri_GetExtension(IUri *iface, BSTR *pstrExtension)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrExtension);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_EXTENSION, pstrExtension, 0);
+}
+
+static HRESULT WINAPI Uri_GetFragment(IUri *iface, BSTR *pstrFragment)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrFragment);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_FRAGMENT, pstrFragment, 0);
+}
+
+static HRESULT WINAPI Uri_GetHost(IUri *iface, BSTR *pstrHost)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrHost);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_HOST, pstrHost, 0);
+}
+
+static HRESULT WINAPI Uri_GetPassword(IUri *iface, BSTR *pstrPassword)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrPassword);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_PASSWORD, pstrPassword, 0);
+}
+
+static HRESULT WINAPI Uri_GetPath(IUri *iface, BSTR *pstrPath)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrPath);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_PATH, pstrPath, 0);
+}
+
+static HRESULT WINAPI Uri_GetPathAndQuery(IUri *iface, BSTR *pstrPathAndQuery)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrPathAndQuery);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_PATH_AND_QUERY, pstrPathAndQuery, 0);
+}
+
+static HRESULT WINAPI Uri_GetQuery(IUri *iface, BSTR *pstrQuery)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrQuery);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_QUERY, pstrQuery, 0);
+}
+
+static HRESULT WINAPI Uri_GetRawUri(IUri *iface, BSTR *pstrRawUri)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrRawUri);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_RAW_URI, pstrRawUri, 0);
+}
+
+static HRESULT WINAPI Uri_GetSchemeName(IUri *iface, BSTR *pstrSchemeName)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrSchemeName);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_SCHEME_NAME, pstrSchemeName, 0);
+}
+
+static HRESULT WINAPI Uri_GetUserInfo(IUri *iface, BSTR *pstrUserInfo)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrUserInfo);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_USER_INFO, pstrUserInfo, 0);
+}
+
+static HRESULT WINAPI Uri_GetUserName(IUri *iface, BSTR *pstrUserName)
+{
+    TRACE("(%p)->(%p)\n", iface, pstrUserName);
+    return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_USER_NAME, pstrUserName, 0);
+}
+
+static HRESULT WINAPI Uri_GetHostType(IUri *iface, DWORD *pdwHostType)
+{
+    TRACE("(%p)->(%p)\n", iface, pdwHostType);
+    return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_HOST_TYPE, pdwHostType, 0);
+}
+
+static HRESULT WINAPI Uri_GetPort(IUri *iface, DWORD *pdwPort)
+{
+    TRACE("(%p)->(%p)\n", iface, pdwPort);
+    return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_PORT, pdwPort, 0);
+}
+
+static HRESULT WINAPI Uri_GetScheme(IUri *iface, DWORD *pdwScheme)
+{
+    Uri *This = URI_THIS(iface);
+    TRACE("(%p)->(%p)\n", This, pdwScheme);
+    return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_SCHEME, pdwScheme, 0);
+}
+
+static HRESULT WINAPI Uri_GetZone(IUri *iface, DWORD *pdwZone)
+{
+    TRACE("(%p)->(%p)\n", iface, pdwZone);
+    return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_ZONE,pdwZone, 0);
+}
+
+static HRESULT WINAPI Uri_GetProperties(IUri *iface, DWORD *pdwProperties)
+{
+    Uri *This = URI_THIS(iface);
+    TRACE("(%p)->(%p)\n", This, pdwProperties);
+
+    if(!pdwProperties)
+        return E_INVALIDARG;
+
+    /* All URIs have these. */
+    *pdwProperties = Uri_HAS_DISPLAY_URI|Uri_HAS_RAW_URI|Uri_HAS_SCHEME|Uri_HAS_HOST_TYPE;
+
+    if(This->display_absolute)
+        *pdwProperties |= Uri_HAS_ABSOLUTE_URI;
+
+    if(This->scheme_start > -1)
+        *pdwProperties |= Uri_HAS_SCHEME_NAME;
+
+    if(This->authority_start > -1) {
+        *pdwProperties |= Uri_HAS_AUTHORITY;
+        if(This->userinfo_start > -1) {
+            *pdwProperties |= Uri_HAS_USER_INFO;
+            if(This->userinfo_split != 0)
+                *pdwProperties |= Uri_HAS_USER_NAME;
+        }
+        if(This->userinfo_split > -1)
+            *pdwProperties |= Uri_HAS_PASSWORD;
+        if(This->host_start > -1)
+            *pdwProperties |= Uri_HAS_HOST;
+        if(This->domain_offset > -1)
+            *pdwProperties |= Uri_HAS_DOMAIN;
+    }
+
+    if(This->has_port)
+        *pdwProperties |= Uri_HAS_PORT;
+    if(This->path_start > -1)
+        *pdwProperties |= Uri_HAS_PATH|Uri_HAS_PATH_AND_QUERY;
+    if(This->query_start > -1)
+        *pdwProperties |= Uri_HAS_QUERY|Uri_HAS_PATH_AND_QUERY;
+
+    if(This->extension_offset > -1)
+        *pdwProperties |= Uri_HAS_EXTENSION;
+
+    if(This->fragment_start > -1)
+        *pdwProperties |= Uri_HAS_FRAGMENT;
+
+    return S_OK;
+}
+
+static HRESULT WINAPI Uri_IsEqual(IUri *iface, IUri *pUri, BOOL *pfEqual)
+{
+    Uri *This = URI_THIS(iface);
+    Uri *other;
+
+    TRACE("(%p)->(%p %p)\n", This, pUri, pfEqual);
+
+    if(!pfEqual)
+        return E_POINTER;
+
+    if(!pUri) {
+        *pfEqual = FALSE;
+
+        /* For some reason Windows returns S_OK here... */
+        return S_OK;
+    }
+
+    /* Try to convert it to a Uri (allows for a more simple comparison). */
+    if((other = get_uri_obj(pUri)))
+        *pfEqual = are_equal_simple(This, other);
+    else {
+        /* Do it the hard way. */
+        FIXME("(%p)->(%p %p) No support for unknown IUri's yet.\n", iface, pUri, pfEqual);
+        return E_NOTIMPL;
+    }
+
+    return S_OK;
+}
+
+#undef URI_THIS
+
+static const IUriVtbl UriVtbl = {
+    Uri_QueryInterface,
+    Uri_AddRef,
+    Uri_Release,
+    Uri_GetPropertyBSTR,
+    Uri_GetPropertyLength,
+    Uri_GetPropertyDWORD,
+    Uri_HasProperty,
+    Uri_GetAbsoluteUri,
+    Uri_GetAuthority,
+    Uri_GetDisplayUri,
+    Uri_GetDomain,
+    Uri_GetExtension,
+    Uri_GetFragment,
+    Uri_GetHost,
+    Uri_GetPassword,
+    Uri_GetPath,
+    Uri_GetPathAndQuery,
+    Uri_GetQuery,
+    Uri_GetRawUri,
+    Uri_GetSchemeName,
+    Uri_GetUserInfo,
+    Uri_GetUserName,
+    Uri_GetHostType,
+    Uri_GetPort,
+    Uri_GetScheme,
+    Uri_GetZone,
+    Uri_GetProperties,
+    Uri_IsEqual
+};
+
+/***********************************************************************
+ *           CreateUri (urlmon.@)
+ *
+ * Creates a new IUri object using the URI represented by pwzURI. This function
+ * parses and validates the components of pwzURI and then canonicalizes the
+ * parsed components.
+ *
+ * PARAMS
+ *  pwzURI      [I] The URI to parse, validate, and canonicalize.
+ *  dwFlags     [I] Flags which can affect how the parsing/canonicalization is performed.
+ *  dwReserved  [I] Reserved (not used).
+ *  ppURI       [O] The resulting IUri after parsing/canonicalization occurs.
+ *
+ * RETURNS
+ *  Success: Returns S_OK. ppURI contains the pointer to the newly allocated IUri.
+ *  Failure: E_INVALIDARG if there's invalid flag combinations in dwFlags, or an
+ *           invalid parameters, or pwzURI doesn't represnt a valid URI.
+ *           E_OUTOFMEMORY if any memory allocation fails.
+ *
+ * NOTES
+ *  Default flags:
+ *      Uri_CREATE_CANONICALIZE, Uri_CREATE_DECODE_EXTRA_INFO, Uri_CREATE_CRACK_UNKNOWN_SCHEMES,
+ *      Uri_CREATE_PRE_PROCESS_HTML_URI, Uri_CREATE_NO_IE_SETTINGS.
+ */
+HRESULT WINAPI CreateUri(LPCWSTR pwzURI, DWORD dwFlags, DWORD_PTR dwReserved, IUri **ppURI)
+{
+    const DWORD supported_flags = Uri_CREATE_ALLOW_RELATIVE|Uri_CREATE_ALLOW_IMPLICIT_WILDCARD_SCHEME|
+        Uri_CREATE_ALLOW_IMPLICIT_FILE_SCHEME|Uri_CREATE_NO_CANONICALIZE|Uri_CREATE_CANONICALIZE|
+        Uri_CREATE_DECODE_EXTRA_INFO|Uri_CREATE_NO_DECODE_EXTRA_INFO|Uri_CREATE_CRACK_UNKNOWN_SCHEMES|
+        Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES|Uri_CREATE_PRE_PROCESS_HTML_URI|Uri_CREATE_NO_PRE_PROCESS_HTML_URI|
+        Uri_CREATE_NO_IE_SETTINGS|Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS|Uri_CREATE_FILE_USE_DOS_PATH;
+    Uri *ret;
+    HRESULT hr;
+    parse_data data;
 
-#define URI(x)  ((IUri*)  &(x)->lpIUriVtbl)
+    TRACE("(%s %x %x %p)\n", debugstr_w(pwzURI), dwFlags, (DWORD)dwReserved, ppURI);
 
-#define URI_THIS(iface) DEFINE_THIS(Uri, IUri, iface)
+    if(!ppURI)
+        return E_INVALIDARG;
 
-static HRESULT WINAPI Uri_QueryInterface(IUri *iface, REFIID riid, void **ppv)
+    if(!pwzURI || !*pwzURI) {
+        *ppURI = NULL;
+        return E_INVALIDARG;
+    }
+
+    /* Check for invalid flags. */
+    if(has_invalid_flag_combination(dwFlags)) {
+        *ppURI = NULL;
+        return E_INVALIDARG;
+    }
+
+    /* Currently unsupported. */
+    if(dwFlags & ~supported_flags)
+        FIXME("Ignoring unsupported flag(s) %x\n", dwFlags & ~supported_flags);
+
+    ret = heap_alloc(sizeof(Uri));
+    if(!ret)
+        return E_OUTOFMEMORY;
+
+    ret->lpIUriVtbl = &UriVtbl;
+    ret->ref = 1;
+
+    /* Explicitly set the default flags if it doesn't cause a flag conflict. */
+    apply_default_flags(&dwFlags);
+
+    /* Pre process the URI, unless told otherwise. */
+    if(!(dwFlags & Uri_CREATE_NO_PRE_PROCESS_HTML_URI))
+        ret->raw_uri = pre_process_uri(pwzURI);
+    else
+        ret->raw_uri = SysAllocString(pwzURI);
+
+    if(!ret->raw_uri) {
+        heap_free(ret);
+        return E_OUTOFMEMORY;
+    }
+
+    memset(&data, 0, sizeof(parse_data));
+    data.uri = ret->raw_uri;
+
+    /* Validate and parse the URI into it's components. */
+    if(!parse_uri(&data, dwFlags)) {
+        /* Encountered an unsupported or invalid URI */
+        SysFreeString(ret->raw_uri);
+        heap_free(ret);
+        *ppURI = NULL;
+        return E_INVALIDARG;
+    }
+
+    /* Canonicalize the URI. */
+    hr = canonicalize_uri(&data, ret, dwFlags);
+    if(FAILED(hr)) {
+        SysFreeString(ret->raw_uri);
+        heap_free(ret);
+        *ppURI = NULL;
+        return hr;
+    }
+
+    ret->create_flags = dwFlags;
+
+    *ppURI = URI(ret);
+    return S_OK;
+}
+
+/***********************************************************************
+ *           CreateUriWithFragment (urlmon.@)
+ *
+ * Creates a new IUri object. This is almost the same as CreateUri, expect that
+ * it allows you to explicitly specify a fragment (pwzFragment) for pwzURI.
+ *
+ * PARAMS
+ *  pwzURI      [I] The URI to parse and perform canonicalization on.
+ *  pwzFragment [I] The explict fragment string which should be added to pwzURI.
+ *  dwFlags     [I] The flags which will be passed to CreateUri.
+ *  dwReserved  [I] Reserved (not used).
+ *  ppURI       [O] The resulting IUri after parsing/canonicalization.
+ *
+ * RETURNS
+ *  Success: S_OK. ppURI contains the pointer to the newly allocated IUri.
+ *  Failure: E_INVALIDARG if pwzURI already contains a fragment and pwzFragment
+ *           isn't NULL. Will also return E_INVALIDARG for the same reasons as
+ *           CreateUri will. E_OUTOFMEMORY if any allocations fail.
+ */
+HRESULT WINAPI CreateUriWithFragment(LPCWSTR pwzURI, LPCWSTR pwzFragment, DWORD dwFlags,
+                                     DWORD_PTR dwReserved, IUri **ppURI)
 {
-    Uri *This = URI_THIS(iface);
+    HRESULT hres;
+    TRACE("(%s %s %x %x %p)\n", debugstr_w(pwzURI), debugstr_w(pwzFragment), dwFlags, (DWORD)dwReserved, ppURI);
+
+    if(!ppURI)
+        return E_INVALIDARG;
+
+    if(!pwzURI) {
+        *ppURI = NULL;
+        return E_INVALIDARG;
+    }
+
+    /* Check if a fragment should be appended to the URI string. */
+    if(pwzFragment) {
+        WCHAR *uriW;
+        DWORD uri_len, frag_len;
+        BOOL add_pound;
+
+        /* Check if the original URI already has a fragment component. */
+        if(StrChrW(pwzURI, '#')) {
+            *ppURI = NULL;
+            return E_INVALIDARG;
+        }
+
+        uri_len = lstrlenW(pwzURI);
+        frag_len = lstrlenW(pwzFragment);
+
+        /* If the fragment doesn't start with a '#', one will be added. */
+        add_pound = *pwzFragment != '#';
+
+        if(add_pound)
+            uriW = heap_alloc((uri_len+frag_len+2)*sizeof(WCHAR));
+        else
+            uriW = heap_alloc((uri_len+frag_len+1)*sizeof(WCHAR));
+
+        if(!uriW)
+            return E_OUTOFMEMORY;
+
+        memcpy(uriW, pwzURI, uri_len*sizeof(WCHAR));
+        if(add_pound)
+            uriW[uri_len++] = '#';
+        memcpy(uriW+uri_len, pwzFragment, (frag_len+1)*sizeof(WCHAR));
+
+        hres = CreateUri(uriW, dwFlags, 0, ppURI);
+
+        heap_free(uriW);
+    } else
+        /* A fragment string wasn't specified, so just forward the call. */
+        hres = CreateUri(pwzURI, dwFlags, 0, ppURI);
+
+    return hres;
+}
+
+#define URIBUILDER_THIS(iface) DEFINE_THIS(UriBuilder, IUriBuilder, iface)
+
+static HRESULT WINAPI UriBuilder_QueryInterface(IUriBuilder *iface, REFIID riid, void **ppv)
+{
+    UriBuilder *This = URIBUILDER_THIS(iface);
 
     if(IsEqualGUID(&IID_IUnknown, riid)) {
         TRACE("(%p)->(IID_IUnknown %p)\n", This, ppv);
-        *ppv = URI(This);
-    }else if(IsEqualGUID(&IID_IUri, riid)) {
+        *ppv = URIBUILDER(This);
+    }else if(IsEqualGUID(&IID_IUriBuilder, riid)) {
         TRACE("(%p)->(IID_IUri %p)\n", This, ppv);
-        *ppv = URI(This);
+        *ppv = URIBUILDER(This);
     }else {
         TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppv);
         *ppv = NULL;
@@ -50,9 +4930,9 @@ static HRESULT WINAPI Uri_QueryInterface(IUri *iface, REFIID riid, void **ppv)
     return S_OK;
 }
 
-static ULONG WINAPI Uri_AddRef(IUri *iface)
+static ULONG WINAPI UriBuilder_AddRef(IUriBuilder *iface)
 {
-    Uri *This = URI_THIS(iface);
+    UriBuilder *This = URIBUILDER_THIS(iface);
     LONG ref = InterlockedIncrement(&This->ref);
 
     TRACE("(%p) ref=%d\n", This, ref);
@@ -60,243 +4940,454 @@ static ULONG WINAPI Uri_AddRef(IUri *iface)
     return ref;
 }
 
-static ULONG WINAPI Uri_Release(IUri *iface)
+static ULONG WINAPI UriBuilder_Release(IUriBuilder *iface)
 {
-    Uri *This = URI_THIS(iface);
+    UriBuilder *This = URIBUILDER_THIS(iface);
     LONG ref = InterlockedDecrement(&This->ref);
 
     TRACE("(%p) ref=%d\n", This, ref);
 
-    if(!ref)
+    if(!ref) {
+        if(This->uri) IUri_Release(URI(This->uri));
+        heap_free(This->fragment);
+        heap_free(This->host);
+        heap_free(This->password);
+        heap_free(This->path);
+        heap_free(This->query);
+        heap_free(This->scheme);
+        heap_free(This->username);
         heap_free(This);
+    }
 
     return ref;
 }
 
-static HRESULT WINAPI Uri_GetPropertyBSTR(IUri *iface, Uri_PROPERTY uriProp, BSTR *pbstrProperty, DWORD dwFlags)
+static HRESULT WINAPI UriBuilder_CreateUriSimple(IUriBuilder *iface,
+                                                 DWORD        dwAllowEncodingPropertyMask,
+                                                 DWORD_PTR    dwReserved,
+                                                 IUri       **ppIUri)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
-    return E_NOTIMPL;
-}
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    HRESULT hr;
+    TRACE("(%p)->(%d %d %p)\n", This, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
 
-static HRESULT WINAPI Uri_GetPropertyLength(IUri *iface, Uri_PROPERTY uriProp, DWORD *pcchProperty, DWORD dwFlags)
-{
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
-    return E_NOTIMPL;
+    hr = build_uri(This, ppIUri, 0, UriBuilder_USE_ORIGINAL_FLAGS, dwAllowEncodingPropertyMask);
+    if(hr == E_NOTIMPL)
+        FIXME("(%p)->(%d %d %p)\n", This, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
+    return hr;
 }
 
-static HRESULT WINAPI Uri_GetPropertyDWORD(IUri *iface, Uri_PROPERTY uriProp, DWORD *pcchProperty, DWORD dwFlags)
+static HRESULT WINAPI UriBuilder_CreateUri(IUriBuilder *iface,
+                                           DWORD        dwCreateFlags,
+                                           DWORD        dwAllowEncodingPropertyMask,
+                                           DWORD_PTR    dwReserved,
+                                           IUri       **ppIUri)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
-    return E_NOTIMPL;
-}
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    HRESULT hr;
+    TRACE("(%p)->(0x%08x %d %d %p)\n", This, dwCreateFlags, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
 
-static HRESULT WINAPI Uri_HasProperty(IUri *iface, Uri_PROPERTY uriProp, BOOL *pfHasProperty)
-{
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->()\n", This);
-    return E_NOTIMPL;
+    if(dwCreateFlags == -1)
+        hr = build_uri(This, ppIUri, 0, UriBuilder_USE_ORIGINAL_FLAGS, dwAllowEncodingPropertyMask);
+    else
+        hr = build_uri(This, ppIUri, dwCreateFlags, 0, dwAllowEncodingPropertyMask);
+
+    if(hr == E_NOTIMPL)
+        FIXME("(%p)->(0x%08x %d %d %p)\n", This, dwCreateFlags, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
+    return hr;
 }
 
-static HRESULT WINAPI Uri_GetAbsoluteUri(IUri *iface, BSTR *pstrAbsoluteUri)
+static HRESULT WINAPI UriBuilder_CreateUriWithFlags(IUriBuilder *iface,
+                                         DWORD        dwCreateFlags,
+                                         DWORD        dwUriBuilderFlags,
+                                         DWORD        dwAllowEncodingPropertyMask,
+                                         DWORD_PTR    dwReserved,
+                                         IUri       **ppIUri)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrAbsoluteUri);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    HRESULT hr;
+    TRACE("(%p)->(0x%08x 0x%08x %d %d %p)\n", This, dwCreateFlags, dwUriBuilderFlags,
+        dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
+
+    hr = build_uri(This, ppIUri, dwCreateFlags, dwUriBuilderFlags, dwAllowEncodingPropertyMask);
+    if(hr == E_NOTIMPL)
+        FIXME("(%p)->(0x%08x 0x%08x %d %d %p)\n", This, dwCreateFlags, dwUriBuilderFlags,
+            dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
+    return hr;
 }
 
-static HRESULT WINAPI Uri_GetAuthority(IUri *iface, BSTR *pstrAuthority)
+static HRESULT WINAPI  UriBuilder_GetIUri(IUriBuilder *iface, IUri **ppIUri)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrAuthority);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p)\n", This, ppIUri);
+
+    if(!ppIUri)
+        return E_POINTER;
+
+    if(This->uri) {
+        IUri *uri = URI(This->uri);
+        IUri_AddRef(uri);
+        *ppIUri = uri;
+    } else
+        *ppIUri = NULL;
+
+    return S_OK;
 }
 
-static HRESULT WINAPI Uri_GetDisplayUri(IUri *iface, BSTR *pstrDisplayUri)
+static HRESULT WINAPI UriBuilder_SetIUri(IUriBuilder *iface, IUri *pIUri)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrDisplayUri);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p)\n", This, pIUri);
+
+    if(pIUri) {
+        Uri *uri;
+
+        if((uri = get_uri_obj(pIUri))) {
+            /* Only reset the builder if it's Uri isn't the same as
+             * the Uri passed to the function.
+             */
+            if(This->uri != uri) {
+                reset_builder(This);
+
+                This->uri = uri;
+                if(uri->has_port)
+                    This->port = uri->port;
+
+                IUri_AddRef(pIUri);
+            }
+        } else {
+            FIXME("(%p)->(%p) Unknown IUri types not supported yet.\n", This, pIUri);
+            return E_NOTIMPL;
+        }
+    } else if(This->uri)
+        /* Only reset the builder if it's Uri isn't NULL. */
+        reset_builder(This);
+
+    return S_OK;
 }
 
-static HRESULT WINAPI Uri_GetDomain(IUri *iface, BSTR *pstrDomain)
+static HRESULT WINAPI UriBuilder_GetFragment(IUriBuilder *iface, DWORD *pcchFragment, LPCWSTR *ppwzFragment)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrDomain);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchFragment, ppwzFragment);
+
+    if(!This->uri || This->uri->fragment_start == -1 || This->modified_props & Uri_HAS_FRAGMENT)
+        return get_builder_component(&This->fragment, &This->fragment_len, NULL, 0, ppwzFragment, pcchFragment);
+    else
+        return get_builder_component(&This->fragment, &This->fragment_len, This->uri->canon_uri+This->uri->fragment_start,
+                                     This->uri->fragment_len, ppwzFragment, pcchFragment);
 }
 
-static HRESULT WINAPI Uri_GetExtension(IUri *iface, BSTR *pstrExtension)
+static HRESULT WINAPI UriBuilder_GetHost(IUriBuilder *iface, DWORD *pcchHost, LPCWSTR *ppwzHost)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrExtension);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchHost, ppwzHost);
+
+    if(!This->uri || This->uri->host_start == -1 || This->modified_props & Uri_HAS_HOST)
+        return get_builder_component(&This->host, &This->host_len, NULL, 0, ppwzHost, pcchHost);
+    else {
+        if(This->uri->host_type == Uri_HOST_IPV6)
+            /* Don't include the '[' and ']' around the address. */
+            return get_builder_component(&This->host, &This->host_len, This->uri->canon_uri+This->uri->host_start+1,
+                                         This->uri->host_len-2, ppwzHost, pcchHost);
+        else
+            return get_builder_component(&This->host, &This->host_len, This->uri->canon_uri+This->uri->host_start,
+                                         This->uri->host_len, ppwzHost, pcchHost);
+    }
 }
 
-static HRESULT WINAPI Uri_GetFragment(IUri *iface, BSTR *pstrFragment)
+static HRESULT WINAPI UriBuilder_GetPassword(IUriBuilder *iface, DWORD *pcchPassword, LPCWSTR *ppwzPassword)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrFragment);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchPassword, ppwzPassword);
+
+    if(!This->uri || This->uri->userinfo_split == -1 || This->modified_props & Uri_HAS_PASSWORD)
+        return get_builder_component(&This->password, &This->password_len, NULL, 0, ppwzPassword, pcchPassword);
+    else {
+        const WCHAR *start = This->uri->canon_uri+This->uri->userinfo_start+This->uri->userinfo_split+1;
+        DWORD len = This->uri->userinfo_len-This->uri->userinfo_split-1;
+        return get_builder_component(&This->password, &This->password_len, start, len, ppwzPassword, pcchPassword);
+    }
 }
 
-static HRESULT WINAPI Uri_GetHost(IUri *iface, BSTR *pstrHost)
+static HRESULT WINAPI UriBuilder_GetPath(IUriBuilder *iface, DWORD *pcchPath, LPCWSTR *ppwzPath)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrHost);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchPath, ppwzPath);
+
+    if(!This->uri || This->uri->path_start == -1 || This->modified_props & Uri_HAS_PATH)
+        return get_builder_component(&This->path, &This->path_len, NULL, 0, ppwzPath, pcchPath);
+    else
+        return get_builder_component(&This->path, &This->path_len, This->uri->canon_uri+This->uri->path_start,
+                                     This->uri->path_len, ppwzPath, pcchPath);
 }
 
-static HRESULT WINAPI Uri_GetPassword(IUri *iface, BSTR *pstrPassword)
+static HRESULT WINAPI UriBuilder_GetPort(IUriBuilder *iface, BOOL *pfHasPort, DWORD *pdwPort)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrPassword);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pfHasPort, pdwPort);
+
+    if(!pfHasPort) {
+        if(pdwPort)
+            *pdwPort = 0;
+        return E_POINTER;
+    }
+
+    if(!pdwPort) {
+        *pfHasPort = FALSE;
+        return E_POINTER;
+    }
+
+    *pfHasPort = This->has_port;
+    *pdwPort = This->port;
+    return S_OK;
 }
 
-static HRESULT WINAPI Uri_GetPath(IUri *iface, BSTR *pstrPath)
+static HRESULT WINAPI UriBuilder_GetQuery(IUriBuilder *iface, DWORD *pcchQuery, LPCWSTR *ppwzQuery)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrPath);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchQuery, ppwzQuery);
+
+    if(!This->uri || This->uri->query_start == -1 || This->modified_props & Uri_HAS_QUERY)
+        return get_builder_component(&This->query, &This->query_len, NULL, 0, ppwzQuery, pcchQuery);
+    else
+        return get_builder_component(&This->query, &This->query_len, This->uri->canon_uri+This->uri->query_start,
+                                     This->uri->query_len, ppwzQuery, pcchQuery);
 }
 
-static HRESULT WINAPI Uri_GetPathAndQuery(IUri *iface, BSTR *pstrPathAndQuery)
+static HRESULT WINAPI UriBuilder_GetSchemeName(IUriBuilder *iface, DWORD *pcchSchemeName, LPCWSTR *ppwzSchemeName)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrPathAndQuery);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchSchemeName, ppwzSchemeName);
+
+    if(!This->uri || This->uri->scheme_start == -1 || This->modified_props & Uri_HAS_SCHEME_NAME)
+        return get_builder_component(&This->scheme, &This->scheme_len, NULL, 0, ppwzSchemeName, pcchSchemeName);
+    else
+        return get_builder_component(&This->scheme, &This->scheme_len, This->uri->canon_uri+This->uri->scheme_start,
+                                     This->uri->scheme_len, ppwzSchemeName, pcchSchemeName);
 }
 
-static HRESULT WINAPI Uri_GetQuery(IUri *iface, BSTR *pstrQuery)
+static HRESULT WINAPI UriBuilder_GetUserName(IUriBuilder *iface, DWORD *pcchUserName, LPCWSTR *ppwzUserName)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrQuery);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p %p)\n", This, pcchUserName, ppwzUserName);
+
+    if(!This->uri || This->uri->userinfo_start == -1 || This->uri->userinfo_split == 0 ||
+       This->modified_props & Uri_HAS_USER_NAME)
+        return get_builder_component(&This->username, &This->username_len, NULL, 0, ppwzUserName, pcchUserName);
+    else {
+        const WCHAR *start = This->uri->canon_uri+This->uri->userinfo_start;
+
+        /* Check if there's a password in the userinfo section. */
+        if(This->uri->userinfo_split > -1)
+            /* Don't include the password. */
+            return get_builder_component(&This->username, &This->username_len, start,
+                                         This->uri->userinfo_split, ppwzUserName, pcchUserName);
+        else
+            return get_builder_component(&This->username, &This->username_len, start,
+                                         This->uri->userinfo_len, ppwzUserName, pcchUserName);
+    }
 }
 
-static HRESULT WINAPI Uri_GetRawUri(IUri *iface, BSTR *pstrRawUri)
+static HRESULT WINAPI UriBuilder_SetFragment(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrRawUri);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+    return set_builder_component(&This->fragment, &This->fragment_len, pwzNewValue, '#',
+                                 &This->modified_props, Uri_HAS_FRAGMENT);
 }
 
-static HRESULT WINAPI Uri_GetSchemeName(IUri *iface, BSTR *pstrSchemeName)
+static HRESULT WINAPI UriBuilder_SetHost(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrSchemeName);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+
+    /* Host name can't be set to NULL. */
+    if(!pwzNewValue)
+        return E_INVALIDARG;
+
+    return set_builder_component(&This->host, &This->host_len, pwzNewValue, 0,
+                                 &This->modified_props, Uri_HAS_HOST);
 }
 
-static HRESULT WINAPI Uri_GetUserInfo(IUri *iface, BSTR *pstrUserInfo)
+static HRESULT WINAPI UriBuilder_SetPassword(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrUserInfo);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+    return set_builder_component(&This->password, &This->password_len, pwzNewValue, 0,
+                                 &This->modified_props, Uri_HAS_PASSWORD);
 }
 
-static HRESULT WINAPI Uri_GetUserName(IUri *iface, BSTR *pstrUserName)
+static HRESULT WINAPI UriBuilder_SetPath(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pstrUserName);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+    return set_builder_component(&This->path, &This->path_len, pwzNewValue, 0,
+                                 &This->modified_props, Uri_HAS_PATH);
 }
 
-static HRESULT WINAPI Uri_GetHostType(IUri *iface, DWORD *pdwHostType)
+static HRESULT WINAPI UriBuilder_SetPort(IUriBuilder *iface, BOOL fHasPort, DWORD dwNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pdwHostType);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%d %d)\n", This, fHasPort, dwNewValue);
+
+    This->has_port = fHasPort;
+    This->port = dwNewValue;
+    This->modified_props |= Uri_HAS_PORT;
+    return S_OK;
 }
 
-static HRESULT WINAPI Uri_GetPort(IUri *iface, DWORD *pdwPort)
+static HRESULT WINAPI UriBuilder_SetQuery(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pdwPort);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+    return set_builder_component(&This->query, &This->query_len, pwzNewValue, '?',
+                                 &This->modified_props, Uri_HAS_QUERY);
 }
 
-static HRESULT WINAPI Uri_GetScheme(IUri *iface, DWORD *pdwScheme)
+static HRESULT WINAPI UriBuilder_SetSchemeName(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pdwScheme);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+
+    /* Only set the scheme name if it's not NULL or empty. */
+    if(!pwzNewValue || !*pwzNewValue)
+        return E_INVALIDARG;
+
+    return set_builder_component(&This->scheme, &This->scheme_len, pwzNewValue, 0,
+                                 &This->modified_props, Uri_HAS_SCHEME_NAME);
 }
 
-static HRESULT WINAPI Uri_GetZone(IUri *iface, DWORD *pdwZone)
+static HRESULT WINAPI UriBuilder_SetUserName(IUriBuilder *iface, LPCWSTR pwzNewValue)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pdwZone);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
+    return set_builder_component(&This->username, &This->username_len, pwzNewValue, 0,
+                                 &This->modified_props, Uri_HAS_USER_NAME);
 }
 
-static HRESULT WINAPI Uri_GetProperties(IUri *iface, DWORD *pdwProperties)
+static HRESULT WINAPI UriBuilder_RemoveProperties(IUriBuilder *iface, DWORD dwPropertyMask)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p)\n", This, pdwProperties);
-    return E_NOTIMPL;
+    const DWORD accepted_flags = Uri_HAS_AUTHORITY|Uri_HAS_DOMAIN|Uri_HAS_EXTENSION|Uri_HAS_FRAGMENT|Uri_HAS_HOST|
+                                 Uri_HAS_PASSWORD|Uri_HAS_PATH|Uri_HAS_PATH_AND_QUERY|Uri_HAS_QUERY|
+                                 Uri_HAS_USER_INFO|Uri_HAS_USER_NAME;
+
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(0x%08x)\n", This, dwPropertyMask);
+
+    if(dwPropertyMask & ~accepted_flags)
+        return E_INVALIDARG;
+
+    if(dwPropertyMask & Uri_HAS_FRAGMENT)
+        UriBuilder_SetFragment(iface, NULL);
+
+    /* Even though you can't set the host name to NULL or an
+     * empty string, you can still remove it... for some reason.
+     */
+    if(dwPropertyMask & Uri_HAS_HOST)
+        set_builder_component(&This->host, &This->host_len, NULL, 0,
+                              &This->modified_props, Uri_HAS_HOST);
+
+    if(dwPropertyMask & Uri_HAS_PASSWORD)
+        UriBuilder_SetPassword(iface, NULL);
+
+    if(dwPropertyMask & Uri_HAS_PATH)
+        UriBuilder_SetPath(iface, NULL);
+
+    if(dwPropertyMask & Uri_HAS_PORT)
+        UriBuilder_SetPort(iface, FALSE, 0);
+
+    if(dwPropertyMask & Uri_HAS_QUERY)
+        UriBuilder_SetQuery(iface, NULL);
+
+    if(dwPropertyMask & Uri_HAS_USER_NAME)
+        UriBuilder_SetUserName(iface, NULL);
+
+    return S_OK;
 }
 
-static HRESULT WINAPI Uri_IsEqual(IUri *iface, IUri *pUri, BOOL *pfEqual)
+static HRESULT WINAPI UriBuilder_HasBeenModified(IUriBuilder *iface, BOOL *pfModified)
 {
-    Uri *This = URI_THIS(iface);
-    FIXME("(%p)->(%p %p)\n", This, pUri, pfEqual);
-    return E_NOTIMPL;
+    UriBuilder *This = URIBUILDER_THIS(iface);
+    TRACE("(%p)->(%p)\n", This, pfModified);
+
+    if(!pfModified)
+        return E_POINTER;
+
+    *pfModified = This->modified_props > 0;
+    return S_OK;
 }
 
-#undef URI_THIS
+#undef URIBUILDER_THIS
 
-static const IUriVtbl UriVtbl = {
-    Uri_QueryInterface,
-    Uri_AddRef,
-    Uri_Release,
-    Uri_GetPropertyBSTR,
-    Uri_GetPropertyLength,
-    Uri_GetPropertyDWORD,
-    Uri_HasProperty,
-    Uri_GetAbsoluteUri,
-    Uri_GetAuthority,
-    Uri_GetDisplayUri,
-    Uri_GetDomain,
-    Uri_GetExtension,
-    Uri_GetFragment,
-    Uri_GetHost,
-    Uri_GetPassword,
-    Uri_GetPath,
-    Uri_GetPathAndQuery,
-    Uri_GetQuery,
-    Uri_GetRawUri,
-    Uri_GetSchemeName,
-    Uri_GetUserInfo,
-    Uri_GetUserName,
-    Uri_GetHostType,
-    Uri_GetPort,
-    Uri_GetScheme,
-    Uri_GetZone,
-    Uri_GetProperties,
-    Uri_IsEqual
+static const IUriBuilderVtbl UriBuilderVtbl = {
+    UriBuilder_QueryInterface,
+    UriBuilder_AddRef,
+    UriBuilder_Release,
+    UriBuilder_CreateUriSimple,
+    UriBuilder_CreateUri,
+    UriBuilder_CreateUriWithFlags,
+    UriBuilder_GetIUri,
+    UriBuilder_SetIUri,
+    UriBuilder_GetFragment,
+    UriBuilder_GetHost,
+    UriBuilder_GetPassword,
+    UriBuilder_GetPath,
+    UriBuilder_GetPort,
+    UriBuilder_GetQuery,
+    UriBuilder_GetSchemeName,
+    UriBuilder_GetUserName,
+    UriBuilder_SetFragment,
+    UriBuilder_SetHost,
+    UriBuilder_SetPassword,
+    UriBuilder_SetPath,
+    UriBuilder_SetPort,
+    UriBuilder_SetQuery,
+    UriBuilder_SetSchemeName,
+    UriBuilder_SetUserName,
+    UriBuilder_RemoveProperties,
+    UriBuilder_HasBeenModified,
 };
 
 /***********************************************************************
- *           CreateUri (urlmon.@)
+ *           CreateIUriBuilder (urlmon.@)
  */
-HRESULT WINAPI CreateUri(LPCWSTR pwzURI, DWORD dwFlags, DWORD_PTR dwReserved, IUri **ppURI)
+HRESULT WINAPI CreateIUriBuilder(IUri *pIUri, DWORD dwFlags, DWORD_PTR dwReserved, IUriBuilder **ppIUriBuilder)
 {
-    Uri *ret;
+    UriBuilder *ret;
 
-    TRACE("(%s %x %x %p)\n", debugstr_w(pwzURI), dwFlags, (DWORD)dwReserved, ppURI);
+    TRACE("(%p %x %x %p)\n", pIUri, dwFlags, (DWORD)dwReserved, ppIUriBuilder);
 
-    ret = heap_alloc(sizeof(Uri));
+    if(!ppIUriBuilder)
+        return E_POINTER;
+
+    ret = heap_alloc_zero(sizeof(UriBuilder));
     if(!ret)
         return E_OUTOFMEMORY;
 
-    ret->lpIUriVtbl = &UriVtbl;
+    ret->lpIUriBuilderVtbl = &UriBuilderVtbl;
     ret->ref = 1;
 
-    *ppURI = URI(ret);
+    if(pIUri) {
+        Uri *uri;
+
+        if((uri = get_uri_obj(pIUri))) {
+            IUri_AddRef(pIUri);
+            ret->uri = uri;
+
+            if(uri->has_port)
+                /* Windows doesn't set 'has_port' to TRUE in this case. */
+                ret->port = uri->port;
+
+        } else {
+            heap_free(ret);
+            *ppIUriBuilder = NULL;
+            FIXME("(%p %x %x %p): Unknown IUri types not supported yet.\n", pIUri, dwFlags,
+                (DWORD)dwReserved, ppIUriBuilder);
+            return E_NOTIMPL;
+        }
+    }
+
+    *ppIUriBuilder = URIBUILDER(ret);
     return S_OK;
 }