2 * Copyright 2010 Jacek Caban for CodeWeavers
3 * Copyright 2010 Thomas Mullaly
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2.1 of the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
20 #include "urlmon_main.h"
21 #include "wine/debug.h"
23 #define NO_SHLWAPI_REG
26 #define UINT_MAX 0xffffffff
27 #define USHORT_MAX 0xffff
29 #define ALLOW_NULL_TERM_SCHEME 0x01
30 #define ALLOW_NULL_TERM_USER_NAME 0x02
31 #define ALLOW_NULL_TERM_PASSWORD 0x04
32 #define ALLOW_BRACKETLESS_IP_LITERAL 0x08
33 #define SKIP_IP_FUTURE_CHECK 0x10
34 #define IGNORE_PORT_DELIMITER 0x20
36 WINE_DEFAULT_DEBUG_CHANNEL(urlmon);
38 static const IID IID_IUriObj = {0x4b364760,0x9f51,0x11df,{0x98,0x1c,0x08,0x00,0x20,0x0c,0x9a,0x66}};
41 const IUriVtbl *lpIUriVtbl;
46 /* Information about the canonicalized URI's buffer. */
50 BOOL display_absolute;
55 URL_SCHEME scheme_type;
63 Uri_HOST_TYPE host_type;
85 const IUriBuilderVtbl *lpIUriBuilderVtbl;
122 /* IPv6 addresses can hold up to 8 h16 components. */
126 /* An IPv6 can have 1 elision ("::"). */
127 const WCHAR *elision;
129 /* An IPv6 can contain 1 IPv4 address as the last 32bits of the address. */
142 BOOL has_implicit_scheme;
143 BOOL has_implicit_ip;
148 URL_SCHEME scheme_type;
150 const WCHAR *username;
153 const WCHAR *password;
158 Uri_HOST_TYPE host_type;
161 ipv6_address ipv6_address;
174 const WCHAR *fragment;
178 static const CHAR hexDigits[] = "0123456789ABCDEF";
180 /* List of scheme types/scheme names that are recognized by the IUri interface as of IE 7. */
181 static const struct {
183 WCHAR scheme_name[16];
184 } recognized_schemes[] = {
185 {URL_SCHEME_FTP, {'f','t','p',0}},
186 {URL_SCHEME_HTTP, {'h','t','t','p',0}},
187 {URL_SCHEME_GOPHER, {'g','o','p','h','e','r',0}},
188 {URL_SCHEME_MAILTO, {'m','a','i','l','t','o',0}},
189 {URL_SCHEME_NEWS, {'n','e','w','s',0}},
190 {URL_SCHEME_NNTP, {'n','n','t','p',0}},
191 {URL_SCHEME_TELNET, {'t','e','l','n','e','t',0}},
192 {URL_SCHEME_WAIS, {'w','a','i','s',0}},
193 {URL_SCHEME_FILE, {'f','i','l','e',0}},
194 {URL_SCHEME_MK, {'m','k',0}},
195 {URL_SCHEME_HTTPS, {'h','t','t','p','s',0}},
196 {URL_SCHEME_SHELL, {'s','h','e','l','l',0}},
197 {URL_SCHEME_SNEWS, {'s','n','e','w','s',0}},
198 {URL_SCHEME_LOCAL, {'l','o','c','a','l',0}},
199 {URL_SCHEME_JAVASCRIPT, {'j','a','v','a','s','c','r','i','p','t',0}},
200 {URL_SCHEME_VBSCRIPT, {'v','b','s','c','r','i','p','t',0}},
201 {URL_SCHEME_ABOUT, {'a','b','o','u','t',0}},
202 {URL_SCHEME_RES, {'r','e','s',0}},
203 {URL_SCHEME_MSSHELLROOTED, {'m','s','-','s','h','e','l','l','-','r','o','o','t','e','d',0}},
204 {URL_SCHEME_MSSHELLIDLIST, {'m','s','-','s','h','e','l','l','-','i','d','l','i','s','t',0}},
205 {URL_SCHEME_MSHELP, {'h','c','p',0}},
206 {URL_SCHEME_WILDCARD, {'*',0}}
209 /* List of default ports Windows recognizes. */
210 static const struct {
213 } default_ports[] = {
214 {URL_SCHEME_FTP, 21},
215 {URL_SCHEME_HTTP, 80},
216 {URL_SCHEME_GOPHER, 70},
217 {URL_SCHEME_NNTP, 119},
218 {URL_SCHEME_TELNET, 23},
219 {URL_SCHEME_WAIS, 210},
220 {URL_SCHEME_HTTPS, 443},
223 /* List of 3 character top level domain names Windows seems to recognize.
224 * There might be more, but, these are the only ones I've found so far.
226 static const struct {
228 } recognized_tlds[] = {
238 static Uri *get_uri_obj(IUri *uri)
243 hres = IUri_QueryInterface(uri, &IID_IUriObj, (void**)&ret);
244 return SUCCEEDED(hres) ? ret : NULL;
247 static inline BOOL is_alpha(WCHAR val) {
248 return ((val >= 'a' && val <= 'z') || (val >= 'A' && val <= 'Z'));
251 static inline BOOL is_num(WCHAR val) {
252 return (val >= '0' && val <= '9');
255 static inline BOOL is_drive_path(const WCHAR *str) {
256 return (is_alpha(str[0]) && (str[1] == ':' || str[1] == '|'));
259 static inline BOOL is_unc_path(const WCHAR *str) {
260 return (str[0] == '\\' && str[0] == '\\');
263 static inline BOOL is_forbidden_dos_path_char(WCHAR val) {
264 return (val == '>' || val == '<' || val == '\"');
267 /* A URI is implicitly a file path if it begins with
268 * a drive letter (eg X:) or starts with "\\" (UNC path).
270 static inline BOOL is_implicit_file_path(const WCHAR *str) {
271 return (is_unc_path(str) || (is_alpha(str[0]) && str[1] == ':'));
274 /* Checks if the URI is a hierarchical URI. A hierarchical
275 * URI is one that has "//" after the scheme.
277 static BOOL check_hierarchical(const WCHAR **ptr) {
278 const WCHAR *start = *ptr;
293 /* unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~" */
294 static inline BOOL is_unreserved(WCHAR val) {
295 return (is_alpha(val) || is_num(val) || val == '-' || val == '.' ||
296 val == '_' || val == '~');
299 /* sub-delims = "!" / "$" / "&" / "'" / "(" / ")"
300 * / "*" / "+" / "," / ";" / "="
302 static inline BOOL is_subdelim(WCHAR val) {
303 return (val == '!' || val == '$' || val == '&' ||
304 val == '\'' || val == '(' || val == ')' ||
305 val == '*' || val == '+' || val == ',' ||
306 val == ';' || val == '=');
309 /* gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@" */
310 static inline BOOL is_gendelim(WCHAR val) {
311 return (val == ':' || val == '/' || val == '?' ||
312 val == '#' || val == '[' || val == ']' ||
316 /* Characters that delimit the end of the authority
317 * section of a URI. Sometimes a '\\' is considered
318 * an authority delimeter.
320 static inline BOOL is_auth_delim(WCHAR val, BOOL acceptSlash) {
321 return (val == '#' || val == '/' || val == '?' ||
322 val == '\0' || (acceptSlash && val == '\\'));
325 /* reserved = gen-delims / sub-delims */
326 static inline BOOL is_reserved(WCHAR val) {
327 return (is_subdelim(val) || is_gendelim(val));
330 static inline BOOL is_hexdigit(WCHAR val) {
331 return ((val >= 'a' && val <= 'f') ||
332 (val >= 'A' && val <= 'F') ||
333 (val >= '0' && val <= '9'));
336 static inline BOOL is_path_delim(WCHAR val) {
337 return (!val || val == '#' || val == '?');
340 /* List of schemes types Windows seems to expect to be hierarchical. */
341 static inline BOOL is_hierarchical_scheme(URL_SCHEME type) {
342 return(type == URL_SCHEME_HTTP || type == URL_SCHEME_FTP ||
343 type == URL_SCHEME_GOPHER || type == URL_SCHEME_NNTP ||
344 type == URL_SCHEME_TELNET || type == URL_SCHEME_WAIS ||
345 type == URL_SCHEME_FILE || type == URL_SCHEME_HTTPS ||
346 type == URL_SCHEME_RES);
349 /* Checks if 'flags' contains an invalid combination of Uri_CREATE flags. */
350 static inline BOOL has_invalid_flag_combination(DWORD flags) {
351 return((flags & Uri_CREATE_DECODE_EXTRA_INFO && flags & Uri_CREATE_NO_DECODE_EXTRA_INFO) ||
352 (flags & Uri_CREATE_CANONICALIZE && flags & Uri_CREATE_NO_CANONICALIZE) ||
353 (flags & Uri_CREATE_CRACK_UNKNOWN_SCHEMES && flags & Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES) ||
354 (flags & Uri_CREATE_PRE_PROCESS_HTML_URI && flags & Uri_CREATE_NO_PRE_PROCESS_HTML_URI) ||
355 (flags & Uri_CREATE_IE_SETTINGS && flags & Uri_CREATE_NO_IE_SETTINGS));
358 /* Applies each default Uri_CREATE flags to 'flags' if it
359 * doesn't cause a flag conflict.
361 static void apply_default_flags(DWORD *flags) {
362 if(!(*flags & Uri_CREATE_NO_CANONICALIZE))
363 *flags |= Uri_CREATE_CANONICALIZE;
364 if(!(*flags & Uri_CREATE_NO_DECODE_EXTRA_INFO))
365 *flags |= Uri_CREATE_DECODE_EXTRA_INFO;
366 if(!(*flags & Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES))
367 *flags |= Uri_CREATE_CRACK_UNKNOWN_SCHEMES;
368 if(!(*flags & Uri_CREATE_NO_PRE_PROCESS_HTML_URI))
369 *flags |= Uri_CREATE_PRE_PROCESS_HTML_URI;
370 if(!(*flags & Uri_CREATE_IE_SETTINGS))
371 *flags |= Uri_CREATE_NO_IE_SETTINGS;
374 /* Determines if the URI is hierarchical using the information already parsed into
375 * data and using the current location of parsing in the URI string.
377 * Windows considers a URI hierarchical if on of the following is true:
378 * A.) It's a wildcard scheme.
379 * B.) It's an implicit file scheme.
380 * C.) It's a known hierarchical scheme and it has two '\\' after the scheme name.
381 * (the '\\' will be converted into "//" during canonicalization).
382 * D.) It's not a relative URI and "//" appears after the scheme name.
384 static inline BOOL is_hierarchical_uri(const WCHAR **ptr, const parse_data *data) {
385 const WCHAR *start = *ptr;
387 if(data->scheme_type == URL_SCHEME_WILDCARD)
389 else if(data->scheme_type == URL_SCHEME_FILE && data->has_implicit_scheme)
391 else if(is_hierarchical_scheme(data->scheme_type) && (*ptr)[0] == '\\' && (*ptr)[1] == '\\') {
394 } else if(!data->is_relative && check_hierarchical(ptr))
401 /* Checks if the two Uri's are logically equivalent. It's a simple
402 * comparison, since they are both of type Uri, and it can access
403 * the properties of each Uri directly without the need to go
404 * through the "IUri_Get*" interface calls.
406 static BOOL are_equal_simple(const Uri *a, const Uri *b) {
407 if(a->scheme_type == b->scheme_type) {
408 const BOOL known_scheme = a->scheme_type != URL_SCHEME_UNKNOWN;
409 const BOOL are_hierarchical =
410 (a->authority_start > -1 && b->authority_start > -1);
412 if(a->scheme_type == URL_SCHEME_FILE) {
413 if(a->canon_len == b->canon_len)
414 return !StrCmpIW(a->canon_uri, b->canon_uri);
417 /* Only compare the scheme names (if any) if their unknown scheme types. */
419 if((a->scheme_start > -1 && b->scheme_start > -1) &&
420 (a->scheme_len == b->scheme_len)) {
421 /* Make sure the schemes are the same. */
422 if(StrCmpNW(a->canon_uri+a->scheme_start, b->canon_uri+b->scheme_start, a->scheme_len))
424 } else if(a->scheme_len != b->scheme_len)
425 /* One of the Uri's has a scheme name, while the other doesn't. */
429 /* If they have a userinfo component, perform case sensitive compare. */
430 if((a->userinfo_start > -1 && b->userinfo_start > -1) &&
431 (a->userinfo_len == b->userinfo_len)) {
432 if(StrCmpNW(a->canon_uri+a->userinfo_start, b->canon_uri+b->userinfo_start, a->userinfo_len))
434 } else if(a->userinfo_len != b->userinfo_len)
435 /* One of the Uri's had a userinfo, while the other one doesn't. */
438 /* Check if they have a host name. */
439 if((a->host_start > -1 && b->host_start > -1) &&
440 (a->host_len == b->host_len)) {
441 /* Perform a case insensitive compare if they are a known scheme type. */
443 if(StrCmpNIW(a->canon_uri+a->host_start, b->canon_uri+b->host_start, a->host_len))
445 } else if(StrCmpNW(a->canon_uri+a->host_start, b->canon_uri+b->host_start, a->host_len))
447 } else if(a->host_len != b->host_len)
448 /* One of the Uri's had a host, while the other one didn't. */
451 if(a->has_port && b->has_port) {
452 if(a->port != b->port)
454 } else if(a->has_port || b->has_port)
455 /* One had a port, while the other one didn't. */
458 /* Windows is weird with how it handles paths. For example
459 * One URI could be "http://google.com" (after canonicalization)
460 * and one could be "http://google.com/" and the IsEqual function
461 * would still evaluate to TRUE, but, only if they are both hierarchical
464 if((a->path_start > -1 && b->path_start > -1) &&
465 (a->path_len == b->path_len)) {
466 if(StrCmpNW(a->canon_uri+a->path_start, b->canon_uri+b->path_start, a->path_len))
468 } else if(are_hierarchical && a->path_len == -1 && b->path_len == 0) {
469 if(*(a->canon_uri+a->path_start) != '/')
471 } else if(are_hierarchical && b->path_len == 1 && a->path_len == 0) {
472 if(*(b->canon_uri+b->path_start) != '/')
474 } else if(a->path_len != b->path_len)
477 /* Compare the query strings of the two URIs. */
478 if((a->query_start > -1 && b->query_start > -1) &&
479 (a->query_len == b->query_len)) {
480 if(StrCmpNW(a->canon_uri+a->query_start, b->canon_uri+b->query_start, a->query_len))
482 } else if(a->query_len != b->query_len)
485 if((a->fragment_start > -1 && b->fragment_start > -1) &&
486 (a->fragment_len == b->fragment_len)) {
487 if(StrCmpNW(a->canon_uri+a->fragment_start, b->canon_uri+b->fragment_start, a->fragment_len))
489 } else if(a->fragment_len != b->fragment_len)
492 /* If we get here, the two URIs are equivalent. */
499 /* Computes the size of the given IPv6 address.
500 * Each h16 component is 16bits, if there is an IPv4 address, it's
501 * 32bits. If there's an elision it can be 16bits to 128bits, depending
502 * on the number of other components.
504 * Modeled after google-url's CheckIPv6ComponentsSize function
506 static void compute_ipv6_comps_size(ipv6_address *address) {
507 address->components_size = address->h16_count * 2;
510 /* IPv4 address is 4 bytes. */
511 address->components_size += 4;
513 if(address->elision) {
514 /* An elision can be anywhere from 2 bytes up to 16 bytes.
515 * It size depends on the size of the h16 and IPv4 components.
517 address->elision_size = 16 - address->components_size;
518 if(address->elision_size < 2)
519 address->elision_size = 2;
521 address->elision_size = 0;
524 /* Taken from dlls/jscript/lex.c */
525 static int hex_to_int(WCHAR val) {
526 if(val >= '0' && val <= '9')
528 else if(val >= 'a' && val <= 'f')
529 return val - 'a' + 10;
530 else if(val >= 'A' && val <= 'F')
531 return val - 'A' + 10;
536 /* Helper function for converting a percent encoded string
537 * representation of a WCHAR value into its actual WCHAR value. If
538 * the two characters following the '%' aren't valid hex values then
539 * this function returns the NULL character.
542 * "%2E" will result in '.' being returned by this function.
544 static WCHAR decode_pct_val(const WCHAR *ptr) {
547 if(*ptr == '%' && is_hexdigit(*(ptr + 1)) && is_hexdigit(*(ptr + 2))) {
548 INT a = hex_to_int(*(ptr + 1));
549 INT b = hex_to_int(*(ptr + 2));
558 /* Helper function for percent encoding a given character
559 * and storing the encoded value into a given buffer (dest).
561 * It's up to the calling function to ensure that there is
562 * at least enough space in 'dest' for the percent encoded
563 * value to be stored (so dest + 3 spaces available).
565 static inline void pct_encode_val(WCHAR val, WCHAR *dest) {
567 dest[1] = hexDigits[(val >> 4) & 0xf];
568 dest[2] = hexDigits[val & 0xf];
571 /* Scans the range of characters [str, end] and returns the last occurrence
572 * of 'ch' or returns NULL.
574 static const WCHAR *str_last_of(const WCHAR *str, const WCHAR *end, WCHAR ch) {
575 const WCHAR *ptr = end;
586 /* Attempts to parse the domain name from the host.
588 * This function also includes the Top-level Domain (TLD) name
589 * of the host when it tries to find the domain name. If it finds
590 * a valid domain name it will assign 'domain_start' the offset
591 * into 'host' where the domain name starts.
593 * It's implied that if a domain name its range is implied to be
594 * [host+domain_start, host+host_len).
596 static void find_domain_name(const WCHAR *host, DWORD host_len,
598 const WCHAR *last_tld, *sec_last_tld, *end;
600 end = host+host_len-1;
604 /* There has to be at least enough room for a '.' followed by a
605 * 3 character TLD for a domain to even exist in the host name.
610 last_tld = str_last_of(host, end, '.');
612 /* http://hostname -> has no domain name. */
615 sec_last_tld = str_last_of(host, last_tld-1, '.');
617 /* If the '.' is at the beginning of the host there
618 * has to be at least 3 characters in the TLD for it
620 * Ex: .com -> .com as the domain name.
621 * .co -> has no domain name.
623 if(last_tld-host == 0) {
624 if(end-(last_tld-1) < 3)
626 } else if(last_tld-host == 3) {
629 /* If there's three characters in front of last_tld and
630 * they are on the list of recognized TLDs, then this
631 * host doesn't have a domain (since the host only contains
633 * Ex: edu.uk -> has no domain name.
634 * foo.uk -> foo.uk as the domain name.
636 for(i = 0; i < sizeof(recognized_tlds)/sizeof(recognized_tlds[0]); ++i) {
637 if(!StrCmpNIW(host, recognized_tlds[i].tld_name, 3))
640 } else if(last_tld-host < 3)
641 /* Anything less than 3 characters is considered part
643 * Ex: ak.uk -> Has no domain name.
647 /* Otherwise the domain name is the whole host name. */
649 } else if(end+1-last_tld > 3) {
650 /* If the last_tld has more than 3 characters, then it's automatically
651 * considered the TLD of the domain name.
652 * Ex: www.winehq.org.uk.test -> uk.test as the domain name.
654 *domain_start = (sec_last_tld+1)-host;
655 } else if(last_tld - (sec_last_tld+1) < 4) {
657 /* If the sec_last_tld is 3 characters long it HAS to be on the list of
658 * recognized to still be considered part of the TLD name, otherwise
659 * its considered the domain name.
660 * Ex: www.google.com.uk -> google.com.uk as the domain name.
661 * www.google.foo.uk -> foo.uk as the domain name.
663 if(last_tld - (sec_last_tld+1) == 3) {
664 for(i = 0; i < sizeof(recognized_tlds)/sizeof(recognized_tlds[0]); ++i) {
665 if(!StrCmpNIW(sec_last_tld+1, recognized_tlds[i].tld_name, 3)) {
666 const WCHAR *domain = str_last_of(host, sec_last_tld-1, '.');
671 *domain_start = (domain+1) - host;
672 TRACE("Found domain name %s\n", debugstr_wn(host+*domain_start,
673 (host+host_len)-(host+*domain_start)));
678 *domain_start = (sec_last_tld+1)-host;
680 /* Since the sec_last_tld is less than 3 characters it's considered
682 * Ex: www.google.fo.uk -> google.fo.uk as the domain name.
684 const WCHAR *domain = str_last_of(host, sec_last_tld-1, '.');
689 *domain_start = (domain+1) - host;
692 /* The second to last TLD has more than 3 characters making it
694 * Ex: www.google.test.us -> test.us as the domain name.
696 *domain_start = (sec_last_tld+1)-host;
699 TRACE("Found domain name %s\n", debugstr_wn(host+*domain_start,
700 (host+host_len)-(host+*domain_start)));
703 /* Removes the dot segments from a hierarchical URIs path component. This
704 * function performs the removal in place.
706 * This is a modified version of Qt's QUrl function "removeDotsFromPath".
708 * This function returns the new length of the path string.
710 static DWORD remove_dot_segments(WCHAR *path, DWORD path_len) {
712 const WCHAR *in = out;
713 const WCHAR *end = out + path_len;
717 /* A. if the input buffer begins with a prefix of "/./" or "/.",
718 * where "." is a complete path segment, then replace that
719 * prefix with "/" in the input buffer; otherwise,
721 if(in <= end - 3 && in[0] == '/' && in[1] == '.' && in[2] == '/') {
724 } else if(in == end - 2 && in[0] == '/' && in[1] == '.') {
730 /* B. if the input buffer begins with a prefix of "/../" or "/..",
731 * where ".." is a complete path segment, then replace that
732 * prefix with "/" in the input buffer and remove the last
733 * segment and its preceding "/" (if any) from the output
736 if(in <= end - 4 && in[0] == '/' && in[1] == '.' && in[2] == '.' && in[3] == '/') {
737 while(out > path && *(--out) != '/');
741 } else if(in == end - 3 && in[0] == '/' && in[1] == '.' && in[2] == '.') {
742 while(out > path && *(--out) != '/');
751 /* C. move the first path segment in the input buffer to the end of
752 * the output buffer, including the initial "/" character (if
753 * any) and any subsequent characters up to, but not including,
754 * the next "/" character or the end of the input buffer.
757 while(in < end && *in != '/')
762 TRACE("(%p %d): Path after dot segments removed %s len=%d\n", path, path_len,
763 debugstr_wn(path, len), len);
767 /* Attempts to find the file extension in a given path. */
768 static INT find_file_extension(const WCHAR *path, DWORD path_len) {
771 for(end = path+path_len-1; end >= path && *end != '/' && *end != '\\'; --end) {
779 /* Computes the location where the elision should occur in the IPv6
780 * address using the numerical values of each component stored in
781 * 'values'. If the address shouldn't contain an elision then 'index'
782 * is assigned -1 as it's value. Otherwise 'index' will contain the
783 * starting index (into values) where the elision should be, and 'count'
784 * will contain the number of cells the elision covers.
787 * Windows will expand an elision if the elision only represents 1 h16
788 * component of the URI.
790 * Ex: [1::2:3:4:5:6:7] -> [1:0:2:3:4:5:6:7]
792 * If the IPv6 address contains an IPv4 address, the IPv4 address is also
793 * considered for being included as part of an elision if all it's components
796 * Ex: [1:2:3:4:5:6:0.0.0.0] -> [1:2:3:4:5:6::]
798 static void compute_elision_location(const ipv6_address *address, const USHORT values[8],
799 INT *index, DWORD *count) {
800 DWORD i, max_len, cur_len;
801 INT max_index, cur_index;
803 max_len = cur_len = 0;
804 max_index = cur_index = -1;
805 for(i = 0; i < 8; ++i) {
806 BOOL check_ipv4 = (address->ipv4 && i == 6);
807 BOOL is_end = (check_ipv4 || i == 7);
810 /* Check if the IPv4 address contains only zeros. */
811 if(values[i] == 0 && values[i+1] == 0) {
818 } else if(values[i] == 0) {
825 if(is_end || values[i] != 0) {
826 /* We only consider it for an elision if it's
827 * more than 1 component long.
829 if(cur_len > 1 && cur_len > max_len) {
830 /* Found the new elision location. */
832 max_index = cur_index;
835 /* Reset the current range for the next range of zeros. */
845 /* Removes all the leading and trailing white spaces or
846 * control characters from the URI and removes all control
847 * characters inside of the URI string.
849 static BSTR pre_process_uri(LPCWSTR uri) {
852 const WCHAR *start, *end;
858 /* Skip leading controls and whitespace. */
859 while(iscntrlW(*start) || isspaceW(*start)) ++start;
863 /* URI consisted only of control/whitespace. */
864 ret = SysAllocStringLen(NULL, 0);
866 while(iscntrlW(*end) || isspaceW(*end)) --end;
868 buf = heap_alloc(((end+1)-start)*sizeof(WCHAR));
872 for(ptr = buf; start < end+1; ++start) {
873 if(!iscntrlW(*start))
877 ret = SysAllocStringLen(buf, ptr-buf);
884 /* Converts the specified IPv4 address into an uint value.
886 * This function assumes that the IPv4 address has already been validated.
888 static UINT ipv4toui(const WCHAR *ip, DWORD len) {
890 DWORD comp_value = 0;
893 for(ptr = ip; ptr < ip+len; ++ptr) {
899 comp_value = comp_value*10 + (*ptr-'0');
908 /* Converts an IPv4 address in numerical form into it's fully qualified
909 * string form. This function returns the number of characters written
910 * to 'dest'. If 'dest' is NULL this function will return the number of
911 * characters that would have been written.
913 * It's up to the caller to ensure there's enough space in 'dest' for the
916 static DWORD ui2ipv4(WCHAR *dest, UINT address) {
917 static const WCHAR formatW[] =
918 {'%','u','.','%','u','.','%','u','.','%','u',0};
922 digits[0] = (address >> 24) & 0xff;
923 digits[1] = (address >> 16) & 0xff;
924 digits[2] = (address >> 8) & 0xff;
925 digits[3] = address & 0xff;
929 ret = sprintfW(tmp, formatW, digits[0], digits[1], digits[2], digits[3]);
931 ret = sprintfW(dest, formatW, digits[0], digits[1], digits[2], digits[3]);
936 static DWORD ui2str(WCHAR *dest, UINT value) {
937 static const WCHAR formatW[] = {'%','u',0};
942 ret = sprintfW(tmp, formatW, value);
944 ret = sprintfW(dest, formatW, value);
949 /* Converts an h16 component (from an IPv6 address) into it's
952 * This function assumes that the h16 component has already been validated.
954 static USHORT h16tous(h16 component) {
958 for(i = 0; i < component.len; ++i) {
960 ret += hex_to_int(component.str[i]);
966 /* Converts an IPv6 address into it's 128 bits (16 bytes) numerical value.
968 * This function assumes that the ipv6_address has already been validated.
970 static BOOL ipv6_to_number(const ipv6_address *address, USHORT number[8]) {
971 DWORD i, cur_component = 0;
972 BOOL already_passed_elision = FALSE;
974 for(i = 0; i < address->h16_count; ++i) {
975 if(address->elision) {
976 if(address->components[i].str > address->elision && !already_passed_elision) {
977 /* Means we just passed the elision and need to add it's values to
978 * 'number' before we do anything else.
981 for(j = 0; j < address->elision_size; j+=2)
982 number[cur_component++] = 0;
984 already_passed_elision = TRUE;
988 number[cur_component++] = h16tous(address->components[i]);
991 /* Case when the elision appears after the h16 components. */
992 if(!already_passed_elision && address->elision) {
993 for(i = 0; i < address->elision_size; i+=2)
994 number[cur_component++] = 0;
995 already_passed_elision = TRUE;
999 UINT value = ipv4toui(address->ipv4, address->ipv4_len);
1001 if(cur_component != 6) {
1002 ERR("(%p %p): Failed sanity check with %d\n", address, number, cur_component);
1006 number[cur_component++] = (value >> 16) & 0xffff;
1007 number[cur_component] = value & 0xffff;
1013 /* Checks if the characters pointed to by 'ptr' are
1014 * a percent encoded data octet.
1016 * pct-encoded = "%" HEXDIG HEXDIG
1018 static BOOL check_pct_encoded(const WCHAR **ptr) {
1019 const WCHAR *start = *ptr;
1025 if(!is_hexdigit(**ptr)) {
1031 if(!is_hexdigit(**ptr)) {
1040 /* dec-octet = DIGIT ; 0-9
1041 * / %x31-39 DIGIT ; 10-99
1042 * / "1" 2DIGIT ; 100-199
1043 * / "2" %x30-34 DIGIT ; 200-249
1044 * / "25" %x30-35 ; 250-255
1046 static BOOL check_dec_octet(const WCHAR **ptr) {
1047 const WCHAR *c1, *c2, *c3;
1050 /* A dec-octet must be at least 1 digit long. */
1051 if(*c1 < '0' || *c1 > '9')
1057 /* Since the 1 digit requirment was meet, it doesn't
1058 * matter if this is a DIGIT value, it's considered a
1061 if(*c2 < '0' || *c2 > '9')
1067 /* Same explanation as above. */
1068 if(*c3 < '0' || *c3 > '9')
1071 /* Anything > 255 isn't a valid IP dec-octet. */
1072 if(*c1 >= '2' && *c2 >= '5' && *c3 >= '5') {
1081 /* Checks if there is an implicit IPv4 address in the host component of the URI.
1082 * The max value of an implicit IPv4 address is UINT_MAX.
1085 * "234567" would be considered an implicit IPv4 address.
1087 static BOOL check_implicit_ipv4(const WCHAR **ptr, UINT *val) {
1088 const WCHAR *start = *ptr;
1092 while(is_num(**ptr)) {
1093 ret = ret*10 + (**ptr - '0');
1095 if(ret > UINT_MAX) {
1109 /* Checks if the string contains an IPv4 address.
1111 * This function has a strict mode or a non-strict mode of operation
1112 * When 'strict' is set to FALSE this function will return TRUE if
1113 * the string contains at least 'dec-octet "." dec-octet' since partial
1114 * IPv4 addresses will be normalized out into full IPv4 addresses. When
1115 * 'strict' is set this function expects there to be a full IPv4 address.
1117 * IPv4address = dec-octet "." dec-octet "." dec-octet "." dec-octet
1119 static BOOL check_ipv4address(const WCHAR **ptr, BOOL strict) {
1120 const WCHAR *start = *ptr;
1122 if(!check_dec_octet(ptr)) {
1133 if(!check_dec_octet(ptr)) {
1147 if(!check_dec_octet(ptr)) {
1161 if(!check_dec_octet(ptr)) {
1166 /* Found a four digit ip address. */
1169 /* Tries to parse the scheme name of the URI.
1171 * scheme = ALPHA *(ALPHA | NUM | '+' | '-' | '.') as defined by RFC 3896.
1172 * NOTE: Windows accepts a number as the first character of a scheme.
1174 static BOOL parse_scheme_name(const WCHAR **ptr, parse_data *data, DWORD extras) {
1175 const WCHAR *start = *ptr;
1177 data->scheme = NULL;
1178 data->scheme_len = 0;
1181 if(**ptr == '*' && *ptr == start) {
1182 /* Might have found a wildcard scheme. If it is the next
1183 * char has to be a ':' for it to be a valid URI
1187 } else if(!is_num(**ptr) && !is_alpha(**ptr) && **ptr != '+' &&
1188 **ptr != '-' && **ptr != '.')
1197 /* Schemes must end with a ':' */
1198 if(**ptr != ':' && !((extras & ALLOW_NULL_TERM_SCHEME) && !**ptr)) {
1203 data->scheme = start;
1204 data->scheme_len = *ptr - start;
1210 /* Tries to deduce the corresponding URL_SCHEME for the given URI. Stores
1211 * the deduced URL_SCHEME in data->scheme_type.
1213 static BOOL parse_scheme_type(parse_data *data) {
1214 /* If there's scheme data then see if it's a recognized scheme. */
1215 if(data->scheme && data->scheme_len) {
1218 for(i = 0; i < sizeof(recognized_schemes)/sizeof(recognized_schemes[0]); ++i) {
1219 if(lstrlenW(recognized_schemes[i].scheme_name) == data->scheme_len) {
1220 /* Has to be a case insensitive compare. */
1221 if(!StrCmpNIW(recognized_schemes[i].scheme_name, data->scheme, data->scheme_len)) {
1222 data->scheme_type = recognized_schemes[i].scheme;
1228 /* If we get here it means it's not a recognized scheme. */
1229 data->scheme_type = URL_SCHEME_UNKNOWN;
1231 } else if(data->is_relative) {
1232 /* Relative URI's have no scheme. */
1233 data->scheme_type = URL_SCHEME_UNKNOWN;
1236 /* Should never reach here! what happened... */
1237 FIXME("(%p): Unable to determine scheme type for URI %s\n", data, debugstr_w(data->uri));
1242 /* Tries to parse (or deduce) the scheme_name of a URI. If it can't
1243 * parse a scheme from the URI it will try to deduce the scheme_name and scheme_type
1244 * using the flags specified in 'flags' (if any). Flags that affect how this function
1245 * operates are the Uri_CREATE_ALLOW_* flags.
1247 * All parsed/deduced information will be stored in 'data' when the function returns.
1249 * Returns TRUE if it was able to successfully parse the information.
1251 static BOOL parse_scheme(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
1252 static const WCHAR fileW[] = {'f','i','l','e',0};
1253 static const WCHAR wildcardW[] = {'*',0};
1255 /* First check to see if the uri could implicitly be a file path. */
1256 if(is_implicit_file_path(*ptr)) {
1257 if(flags & Uri_CREATE_ALLOW_IMPLICIT_FILE_SCHEME) {
1258 data->scheme = fileW;
1259 data->scheme_len = lstrlenW(fileW);
1260 data->has_implicit_scheme = TRUE;
1262 TRACE("(%p %p %x): URI is an implicit file path.\n", ptr, data, flags);
1264 /* Window's does not consider anything that can implicitly be a file
1265 * path to be a valid URI if the ALLOW_IMPLICIT_FILE_SCHEME flag is not set...
1267 TRACE("(%p %p %x): URI is implicitly a file path, but, the ALLOW_IMPLICIT_FILE_SCHEME flag wasn't set.\n",
1271 } else if(!parse_scheme_name(ptr, data, extras)) {
1272 /* No Scheme was found, this means it could be:
1273 * a) an implicit Wildcard scheme
1277 if(flags & Uri_CREATE_ALLOW_IMPLICIT_WILDCARD_SCHEME) {
1278 data->scheme = wildcardW;
1279 data->scheme_len = lstrlenW(wildcardW);
1280 data->has_implicit_scheme = TRUE;
1282 TRACE("(%p %p %x): URI is an implicit wildcard scheme.\n", ptr, data, flags);
1283 } else if (flags & Uri_CREATE_ALLOW_RELATIVE) {
1284 data->is_relative = TRUE;
1285 TRACE("(%p %p %x): URI is relative.\n", ptr, data, flags);
1287 TRACE("(%p %p %x): Malformed URI found. Unable to deduce scheme name.\n", ptr, data, flags);
1292 if(!data->is_relative)
1293 TRACE("(%p %p %x): Found scheme=%s scheme_len=%d\n", ptr, data, flags,
1294 debugstr_wn(data->scheme, data->scheme_len), data->scheme_len);
1296 if(!parse_scheme_type(data))
1299 TRACE("(%p %p %x): Assigned %d as the URL_SCHEME.\n", ptr, data, flags, data->scheme_type);
1303 static BOOL parse_username(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
1304 data->username = *ptr;
1306 while(**ptr != ':' && **ptr != '@') {
1308 if(!check_pct_encoded(ptr)) {
1309 if(data->scheme_type != URL_SCHEME_UNKNOWN) {
1310 *ptr = data->username;
1311 data->username = NULL;
1316 } else if(extras & ALLOW_NULL_TERM_USER_NAME && !**ptr)
1318 else if(is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)) {
1319 *ptr = data->username;
1320 data->username = NULL;
1327 data->username_len = *ptr - data->username;
1331 static BOOL parse_password(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
1332 const WCHAR *start = *ptr;
1338 data->password = *ptr;
1340 while(**ptr != '@') {
1342 if(!check_pct_encoded(ptr)) {
1343 if(data->scheme_type != URL_SCHEME_UNKNOWN) {
1345 data->password = NULL;
1350 } else if(extras & ALLOW_NULL_TERM_PASSWORD && !**ptr)
1352 else if(is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)) {
1354 data->password = NULL;
1361 data->password_len = *ptr - data->password;
1365 /* Parses the userinfo part of the URI (if it exists). The userinfo field of
1366 * a URI can consist of "username:password@", or just "username@".
1369 * userinfo = *( unreserved / pct-encoded / sub-delims / ":" )
1372 * 1) If there is more than one ':' in the userinfo part of the URI Windows
1373 * uses the first occurrence of ':' to delimit the username and password
1377 * ftp://user:pass:word@winehq.org
1379 * Would yield, "user" as the username and "pass:word" as the password.
1381 * 2) Windows allows any character to appear in the "userinfo" part of
1382 * a URI, as long as it's not an authority delimeter character set.
1384 static void parse_userinfo(const WCHAR **ptr, parse_data *data, DWORD flags) {
1385 const WCHAR *start = *ptr;
1387 if(!parse_username(ptr, data, flags, 0)) {
1388 TRACE("(%p %p %x): URI contained no userinfo.\n", ptr, data, flags);
1392 if(!parse_password(ptr, data, flags, 0)) {
1394 data->username = NULL;
1395 data->username_len = 0;
1396 TRACE("(%p %p %x): URI contained no userinfo.\n", ptr, data, flags);
1402 data->username = NULL;
1403 data->username_len = 0;
1404 data->password = NULL;
1405 data->password_len = 0;
1407 TRACE("(%p %p %x): URI contained no userinfo.\n", ptr, data, flags);
1412 TRACE("(%p %p %x): Found username %s len=%d.\n", ptr, data, flags,
1413 debugstr_wn(data->username, data->username_len), data->username_len);
1416 TRACE("(%p %p %x): Found password %s len=%d.\n", ptr, data, flags,
1417 debugstr_wn(data->password, data->password_len), data->password_len);
1422 /* Attempts to parse a port from the URI.
1425 * Windows seems to have a cap on what the maximum value
1426 * for a port can be. The max value is USHORT_MAX.
1430 static BOOL parse_port(const WCHAR **ptr, parse_data *data, DWORD flags) {
1434 while(!is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)) {
1435 if(!is_num(**ptr)) {
1441 port = port*10 + (**ptr-'0');
1443 if(port > USHORT_MAX) {
1452 data->has_port = TRUE;
1453 data->port_value = port;
1454 data->port_len = *ptr - data->port;
1456 TRACE("(%p %p %x): Found port %s len=%d value=%u\n", ptr, data, flags,
1457 debugstr_wn(data->port, data->port_len), data->port_len, data->port_value);
1461 /* Attempts to parse a IPv4 address from the URI.
1464 * Window's normalizes IPv4 addresses, This means there's three
1465 * possibilities for the URI to contain an IPv4 address.
1466 * 1) A well formed address (ex. 192.2.2.2).
1467 * 2) A partially formed address. For example "192.0" would
1468 * normalize to "192.0.0.0" during canonicalization.
1469 * 3) An implicit IPv4 address. For example "256" would
1470 * normalize to "0.0.1.0" during canonicalization. Also
1471 * note that the maximum value for an implicit IP address
1472 * is UINT_MAX, if the value in the URI exceeds this then
1473 * it is not considered an IPv4 address.
1475 static BOOL parse_ipv4address(const WCHAR **ptr, parse_data *data, DWORD flags) {
1476 const BOOL is_unknown = data->scheme_type == URL_SCHEME_UNKNOWN;
1479 if(!check_ipv4address(ptr, FALSE)) {
1480 if(!check_implicit_ipv4(ptr, &data->implicit_ipv4)) {
1481 TRACE("(%p %p %x): URI didn't contain anything looking like an IPv4 address.\n",
1487 data->has_implicit_ip = TRUE;
1490 /* Check if what we found is the only part of the host name (if it isn't
1491 * we don't have an IPv4 address).
1495 if(!parse_port(ptr, data, flags)) {
1500 } else if(!is_auth_delim(**ptr, !is_unknown)) {
1501 /* Found more data which belongs the host, so this isn't an IPv4. */
1504 data->has_implicit_ip = FALSE;
1508 data->host_len = *ptr - data->host;
1509 data->host_type = Uri_HOST_IPV4;
1511 TRACE("(%p %p %x): IPv4 address found. host=%s host_len=%d host_type=%d\n",
1512 ptr, data, flags, debugstr_wn(data->host, data->host_len),
1513 data->host_len, data->host_type);
1517 /* Attempts to parse the reg-name from the URI.
1519 * Because of the way Windows handles ':' this function also
1520 * handles parsing the port.
1522 * reg-name = *( unreserved / pct-encoded / sub-delims )
1525 * Windows allows everything, but, the characters in "auth_delims" and ':'
1526 * to appear in a reg-name, unless it's an unknown scheme type then ':' is
1527 * allowed to appear (even if a valid port isn't after it).
1529 * Windows doesn't like host names which start with '[' and end with ']'
1530 * and don't contain a valid IP literal address in between them.
1532 * On Windows if an '[' is encountered in the host name the ':' no longer
1533 * counts as a delimiter until you reach the next ']' or an "authority delimeter".
1535 * A reg-name CAN be empty.
1537 static BOOL parse_reg_name(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
1538 const BOOL has_start_bracket = **ptr == '[';
1539 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
1540 BOOL inside_brackets = has_start_bracket;
1541 BOOL ignore_col = extras & IGNORE_PORT_DELIMITER;
1543 /* We have to be careful with file schemes. */
1544 if(data->scheme_type == URL_SCHEME_FILE) {
1545 /* This is because an implicit file scheme could be "C:\\test" and it
1546 * would trick this function into thinking the host is "C", when after
1547 * canonicalization the host would end up being an empty string. A drive
1548 * path can also have a '|' instead of a ':' after the drive letter.
1550 if(is_drive_path(*ptr)) {
1551 /* Regular old drive paths don't have a host type (or host name). */
1552 data->host_type = Uri_HOST_UNKNOWN;
1556 } else if(is_unc_path(*ptr))
1557 /* Skip past the "\\" of a UNC path. */
1563 while(!is_auth_delim(**ptr, known_scheme)) {
1564 if(**ptr == ':' && !ignore_col) {
1565 /* We can ignore ':' if were inside brackets.*/
1566 if(!inside_brackets) {
1567 const WCHAR *tmp = (*ptr)++;
1569 /* Attempt to parse the port. */
1570 if(!parse_port(ptr, data, flags)) {
1571 /* Windows expects there to be a valid port for known scheme types. */
1572 if(data->scheme_type != URL_SCHEME_UNKNOWN) {
1575 TRACE("(%p %p %x %x): Expected valid port\n", ptr, data, flags, extras);
1578 /* Windows gives up on trying to parse a port when it
1579 * encounters 1 invalid port.
1583 data->host_len = tmp - data->host;
1587 } else if(**ptr == '%' && known_scheme) {
1588 /* Has to be a legit % encoded value. */
1589 if(!check_pct_encoded(ptr)) {
1595 } else if(**ptr == ']')
1596 inside_brackets = FALSE;
1597 else if(**ptr == '[')
1598 inside_brackets = TRUE;
1603 if(has_start_bracket) {
1604 /* Make sure the last character of the host wasn't a ']'. */
1605 if(*(*ptr-1) == ']') {
1606 TRACE("(%p %p %x %x): Expected an IP literal inside of the host\n",
1607 ptr, data, flags, extras);
1614 /* Don't overwrite our length if we found a port earlier. */
1616 data->host_len = *ptr - data->host;
1618 /* If the host is empty, then it's an unknown host type. */
1619 if(data->host_len == 0)
1620 data->host_type = Uri_HOST_UNKNOWN;
1622 data->host_type = Uri_HOST_DNS;
1624 TRACE("(%p %p %x %x): Parsed reg-name. host=%s len=%d\n", ptr, data, flags, extras,
1625 debugstr_wn(data->host, data->host_len), data->host_len);
1629 /* Attempts to parse an IPv6 address out of the URI.
1631 * IPv6address = 6( h16 ":" ) ls32
1632 * / "::" 5( h16 ":" ) ls32
1633 * / [ h16 ] "::" 4( h16 ":" ) ls32
1634 * / [ *1( h16 ":" ) h16 ] "::" 3( h16 ":" ) ls32
1635 * / [ *2( h16 ":" ) h16 ] "::" 2( h16 ":" ) ls32
1636 * / [ *3( h16 ":" ) h16 ] "::" h16 ":" ls32
1637 * / [ *4( h16 ":" ) h16 ] "::" ls32
1638 * / [ *5( h16 ":" ) h16 ] "::" h16
1639 * / [ *6( h16 ":" ) h16 ] "::"
1641 * ls32 = ( h16 ":" h16 ) / IPv4address
1642 * ; least-significant 32 bits of address.
1645 * ; 16 bits of address represented in hexadecimal.
1647 * Modeled after google-url's 'DoParseIPv6' function.
1649 static BOOL parse_ipv6address(const WCHAR **ptr, parse_data *data, DWORD flags) {
1650 const WCHAR *start, *cur_start;
1653 start = cur_start = *ptr;
1654 memset(&ip, 0, sizeof(ipv6_address));
1657 /* Check if we're on the last character of the host. */
1658 BOOL is_end = (is_auth_delim(**ptr, data->scheme_type != URL_SCHEME_UNKNOWN)
1661 BOOL is_split = (**ptr == ':');
1662 BOOL is_elision = (is_split && !is_end && *(*ptr+1) == ':');
1664 /* Check if we're at the end of a component, or
1665 * if we're at the end of the IPv6 address.
1667 if(is_split || is_end) {
1670 cur_len = *ptr - cur_start;
1672 /* h16 can't have a length > 4. */
1676 TRACE("(%p %p %x): h16 component to long.\n",
1682 /* An h16 component can't have the length of 0 unless
1683 * the elision is at the beginning of the address, or
1684 * at the end of the address.
1686 if(!((*ptr == start && is_elision) ||
1687 (is_end && (*ptr-2) == ip.elision))) {
1689 TRACE("(%p %p %x): IPv6 component cannot have a length of 0.\n",
1696 /* An IPv6 address can have no more than 8 h16 components. */
1697 if(ip.h16_count >= 8) {
1699 TRACE("(%p %p %x): Not a IPv6 address, to many h16 components.\n",
1704 ip.components[ip.h16_count].str = cur_start;
1705 ip.components[ip.h16_count].len = cur_len;
1707 TRACE("(%p %p %x): Found h16 component %s, len=%d, h16_count=%d\n",
1708 ptr, data, flags, debugstr_wn(cur_start, cur_len), cur_len,
1718 /* A IPv6 address can only have 1 elision ('::'). */
1722 TRACE("(%p %p %x): IPv6 address cannot have 2 elisions.\n",
1734 if(!check_ipv4address(ptr, TRUE)) {
1735 if(!is_hexdigit(**ptr)) {
1736 /* Not a valid character for an IPv6 address. */
1741 /* Found an IPv4 address. */
1742 ip.ipv4 = cur_start;
1743 ip.ipv4_len = *ptr - cur_start;
1745 TRACE("(%p %p %x): Found an attached IPv4 address %s len=%d.\n",
1746 ptr, data, flags, debugstr_wn(ip.ipv4, ip.ipv4_len),
1749 /* IPv4 addresses can only appear at the end of a IPv6. */
1755 compute_ipv6_comps_size(&ip);
1757 /* Make sure the IPv6 address adds up to 16 bytes. */
1758 if(ip.components_size + ip.elision_size != 16) {
1760 TRACE("(%p %p %x): Invalid IPv6 address, did not add up to 16 bytes.\n",
1765 if(ip.elision_size == 2) {
1766 /* For some reason on Windows if an elision that represents
1767 * only 1 h16 component is encountered at the very begin or
1768 * end of an IPv6 address, Windows does not consider it a
1769 * valid IPv6 address.
1771 * Ex: [::2:3:4:5:6:7] is not valid, even though the sum
1772 * of all the components == 128bits.
1774 if(ip.elision < ip.components[0].str ||
1775 ip.elision > ip.components[ip.h16_count-1].str) {
1777 TRACE("(%p %p %x): Invalid IPv6 address. Detected elision of 2 bytes at the beginning or end of the address.\n",
1783 data->host_type = Uri_HOST_IPV6;
1784 data->has_ipv6 = TRUE;
1785 data->ipv6_address = ip;
1787 TRACE("(%p %p %x): Found valid IPv6 literal %s len=%d\n",
1788 ptr, data, flags, debugstr_wn(start, *ptr-start),
1793 /* IPvFuture = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" ) */
1794 static BOOL parse_ipvfuture(const WCHAR **ptr, parse_data *data, DWORD flags) {
1795 const WCHAR *start = *ptr;
1797 /* IPvFuture has to start with a 'v' or 'V'. */
1798 if(**ptr != 'v' && **ptr != 'V')
1801 /* Following the v there must be at least 1 hex digit. */
1803 if(!is_hexdigit(**ptr)) {
1809 while(is_hexdigit(**ptr))
1812 /* End of the hexdigit sequence must be a '.' */
1819 if(!is_unreserved(**ptr) && !is_subdelim(**ptr) && **ptr != ':') {
1825 while(is_unreserved(**ptr) || is_subdelim(**ptr) || **ptr == ':')
1828 data->host_type = Uri_HOST_UNKNOWN;
1830 TRACE("(%p %p %x): Parsed IPvFuture address %s len=%d\n", ptr, data, flags,
1831 debugstr_wn(start, *ptr-start), *ptr-start);
1836 /* IP-literal = "[" ( IPv6address / IPvFuture ) "]" */
1837 static BOOL parse_ip_literal(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
1840 if(**ptr != '[' && !(extras & ALLOW_BRACKETLESS_IP_LITERAL)) {
1843 } else if(**ptr == '[')
1846 if(!parse_ipv6address(ptr, data, flags)) {
1847 if(extras & SKIP_IP_FUTURE_CHECK || !parse_ipvfuture(ptr, data, flags)) {
1854 if(**ptr != ']' && !(extras & ALLOW_BRACKETLESS_IP_LITERAL)) {
1858 } else if(!**ptr && extras & ALLOW_BRACKETLESS_IP_LITERAL) {
1859 /* The IP literal didn't contain brackets and was followed by
1860 * a NULL terminator, so no reason to even check the port.
1862 data->host_len = *ptr - data->host;
1869 /* If a valid port is not found, then let it trickle down to
1872 if(!parse_port(ptr, data, flags)) {
1878 data->host_len = *ptr - data->host;
1883 /* Parses the host information from the URI.
1885 * host = IP-literal / IPv4address / reg-name
1887 static BOOL parse_host(const WCHAR **ptr, parse_data *data, DWORD flags, DWORD extras) {
1888 if(!parse_ip_literal(ptr, data, flags, extras)) {
1889 if(!parse_ipv4address(ptr, data, flags)) {
1890 if(!parse_reg_name(ptr, data, flags, extras)) {
1891 TRACE("(%p %p %x %x): Malformed URI, Unknown host type.\n",
1892 ptr, data, flags, extras);
1901 /* Parses the authority information from the URI.
1903 * authority = [ userinfo "@" ] host [ ":" port ]
1905 static BOOL parse_authority(const WCHAR **ptr, parse_data *data, DWORD flags) {
1906 parse_userinfo(ptr, data, flags);
1908 /* Parsing the port will happen during one of the host parsing
1909 * routines (if the URI has a port).
1911 if(!parse_host(ptr, data, flags, 0))
1917 /* Attempts to parse the path information of a hierarchical URI. */
1918 static BOOL parse_path_hierarchical(const WCHAR **ptr, parse_data *data, DWORD flags) {
1919 const WCHAR *start = *ptr;
1920 static const WCHAR slash[] = {'/',0};
1921 const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
1923 if(is_path_delim(**ptr)) {
1924 if(data->scheme_type == URL_SCHEME_WILDCARD) {
1925 /* Wildcard schemes don't get a '/' attached if their path is
1930 } else if(!(flags & Uri_CREATE_NO_CANONICALIZE)) {
1931 /* If the path component is empty, then a '/' is added. */
1936 while(!is_path_delim(**ptr)) {
1937 if(**ptr == '%' && data->scheme_type != URL_SCHEME_UNKNOWN && !is_file) {
1938 if(!check_pct_encoded(ptr)) {
1943 } else if(is_forbidden_dos_path_char(**ptr) && is_file &&
1944 (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
1945 /* File schemes with USE_DOS_PATH set aren't allowed to have
1946 * a '<' or '>' or '\"' appear in them.
1950 } else if(**ptr == '\\') {
1951 /* Not allowed to have a backslash if NO_CANONICALIZE is set
1952 * and the scheme is known type (but not a file scheme).
1954 if(flags & Uri_CREATE_NO_CANONICALIZE) {
1955 if(data->scheme_type != URL_SCHEME_FILE &&
1956 data->scheme_type != URL_SCHEME_UNKNOWN) {
1966 /* The only time a URI doesn't have a path is when
1967 * the NO_CANONICALIZE flag is set and the raw URI
1968 * didn't contain one.
1975 data->path_len = *ptr - start;
1980 TRACE("(%p %p %x): Parsed path %s len=%d\n", ptr, data, flags,
1981 debugstr_wn(data->path, data->path_len), data->path_len);
1983 TRACE("(%p %p %x): The URI contained no path\n", ptr, data, flags);
1988 /* Parses the path of a opaque URI (much less strict then the parser
1989 * for a hierarchical URI).
1992 * Windows allows invalid % encoded data to appear in opaque URI paths
1993 * for unknown scheme types.
1995 * File schemes with USE_DOS_PATH set aren't allowed to have '<', '>', or '\"'
1998 static BOOL parse_path_opaque(const WCHAR **ptr, parse_data *data, DWORD flags) {
1999 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
2000 const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
2004 while(!is_path_delim(**ptr)) {
2005 if(**ptr == '%' && known_scheme) {
2006 if(!check_pct_encoded(ptr)) {
2012 } else if(is_forbidden_dos_path_char(**ptr) && is_file &&
2013 (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
2022 data->path_len = *ptr - data->path;
2023 TRACE("(%p %p %x): Parsed opaque URI path %s len=%d\n", ptr, data, flags,
2024 debugstr_wn(data->path, data->path_len), data->path_len);
2028 /* Determines how the URI should be parsed after the scheme information.
2030 * If the scheme is followed, by "//" then, it is treated as an hierarchical URI
2031 * which then the authority and path information will be parsed out. Otherwise, the
2032 * URI will be treated as an opaque URI which the authority information is not parsed
2035 * RFC 3896 definition of hier-part:
2037 * hier-part = "//" authority path-abempty
2042 * MSDN opaque URI definition:
2043 * scheme ":" path [ "#" fragment ]
2046 * If the URI is of an unknown scheme type and has a "//" following the scheme then it
2047 * is treated as a hierarchical URI, but, if the CREATE_NO_CRACK_UNKNOWN_SCHEMES flag is
2048 * set then it is considered an opaque URI reguardless of what follows the scheme information
2049 * (per MSDN documentation).
2051 static BOOL parse_hierpart(const WCHAR **ptr, parse_data *data, DWORD flags) {
2052 const WCHAR *start = *ptr;
2054 /* Checks if the authority information needs to be parsed. */
2055 if(is_hierarchical_uri(ptr, data)) {
2056 /* Only treat it as a hierarchical URI if the scheme_type is known or
2057 * the Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES flag is not set.
2059 if(data->scheme_type != URL_SCHEME_UNKNOWN ||
2060 !(flags & Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES)) {
2061 TRACE("(%p %p %x): Treating URI as an hierarchical URI.\n", ptr, data, flags);
2062 data->is_opaque = FALSE;
2064 /* TODO: Handle hierarchical URI's, parse authority then parse the path. */
2065 if(!parse_authority(ptr, data, flags))
2068 return parse_path_hierarchical(ptr, data, flags);
2070 /* Reset ptr to it's starting position so opaque path parsing
2071 * begins at the correct location.
2076 /* If it reaches here, then the URI will be treated as an opaque
2080 TRACE("(%p %p %x): Treating URI as an opaque URI.\n", ptr, data, flags);
2082 data->is_opaque = TRUE;
2083 if(!parse_path_opaque(ptr, data, flags))
2089 /* Attempts to parse the query string from the URI.
2092 * If NO_DECODE_EXTRA_INFO flag is set, then invalid percent encoded
2093 * data is allowed appear in the query string. For unknown scheme types
2094 * invalid percent encoded data is allowed to appear reguardless.
2096 static BOOL parse_query(const WCHAR **ptr, parse_data *data, DWORD flags) {
2097 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
2100 TRACE("(%p %p %x): URI didn't contain a query string.\n", ptr, data, flags);
2107 while(**ptr && **ptr != '#') {
2108 if(**ptr == '%' && known_scheme &&
2109 !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
2110 if(!check_pct_encoded(ptr)) {
2121 data->query_len = *ptr - data->query;
2123 TRACE("(%p %p %x): Parsed query string %s len=%d\n", ptr, data, flags,
2124 debugstr_wn(data->query, data->query_len), data->query_len);
2128 /* Attempts to parse the fragment from the URI.
2131 * If NO_DECODE_EXTRA_INFO flag is set, then invalid percent encoded
2132 * data is allowed appear in the query string. For unknown scheme types
2133 * invalid percent encoded data is allowed to appear reguardless.
2135 static BOOL parse_fragment(const WCHAR **ptr, parse_data *data, DWORD flags) {
2136 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
2139 TRACE("(%p %p %x): URI didn't contain a fragment.\n", ptr, data, flags);
2143 data->fragment = *ptr;
2147 if(**ptr == '%' && known_scheme &&
2148 !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
2149 if(!check_pct_encoded(ptr)) {
2150 *ptr = data->fragment;
2151 data->fragment = NULL;
2160 data->fragment_len = *ptr - data->fragment;
2162 TRACE("(%p %p %x): Parsed fragment %s len=%d\n", ptr, data, flags,
2163 debugstr_wn(data->fragment, data->fragment_len), data->fragment_len);
2167 /* Parses and validates the components of the specified by data->uri
2168 * and stores the information it parses into 'data'.
2170 * Returns TRUE if it successfully parsed the URI. False otherwise.
2172 static BOOL parse_uri(parse_data *data, DWORD flags) {
2179 TRACE("(%p %x): BEGINNING TO PARSE URI %s.\n", data, flags, debugstr_w(data->uri));
2181 if(!parse_scheme(pptr, data, flags, 0))
2184 if(!parse_hierpart(pptr, data, flags))
2187 if(!parse_query(pptr, data, flags))
2190 if(!parse_fragment(pptr, data, flags))
2193 TRACE("(%p %x): FINISHED PARSING URI.\n", data, flags);
2197 static BOOL canonicalize_username(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2200 if(!data->username) {
2201 uri->userinfo_start = -1;
2205 uri->userinfo_start = uri->canon_len;
2206 for(ptr = data->username; ptr < data->username+data->username_len; ++ptr) {
2208 /* Only decode % encoded values for known scheme types. */
2209 if(data->scheme_type != URL_SCHEME_UNKNOWN) {
2210 /* See if the value really needs decoded. */
2211 WCHAR val = decode_pct_val(ptr);
2212 if(is_unreserved(val)) {
2214 uri->canon_uri[uri->canon_len] = val;
2218 /* Move pass the hex characters. */
2223 } else if(!is_reserved(*ptr) && !is_unreserved(*ptr) && *ptr != '\\') {
2224 /* Only percent encode forbidden characters if the NO_ENCODE_FORBIDDEN_CHARACTERS flag
2227 if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS)) {
2229 pct_encode_val(*ptr, uri->canon_uri + uri->canon_len);
2231 uri->canon_len += 3;
2237 /* Nothing special, so just copy the character over. */
2238 uri->canon_uri[uri->canon_len] = *ptr;
2245 static BOOL canonicalize_password(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2248 if(!data->password) {
2249 uri->userinfo_split = -1;
2253 if(uri->userinfo_start == -1)
2254 /* Has a password, but, doesn't have a username. */
2255 uri->userinfo_start = uri->canon_len;
2257 uri->userinfo_split = uri->canon_len - uri->userinfo_start;
2259 /* Add the ':' to the userinfo component. */
2261 uri->canon_uri[uri->canon_len] = ':';
2264 for(ptr = data->password; ptr < data->password+data->password_len; ++ptr) {
2266 /* Only decode % encoded values for known scheme types. */
2267 if(data->scheme_type != URL_SCHEME_UNKNOWN) {
2268 /* See if the value really needs decoded. */
2269 WCHAR val = decode_pct_val(ptr);
2270 if(is_unreserved(val)) {
2272 uri->canon_uri[uri->canon_len] = val;
2276 /* Move pass the hex characters. */
2281 } else if(!is_reserved(*ptr) && !is_unreserved(*ptr) && *ptr != '\\') {
2282 /* Only percent encode forbidden characters if the NO_ENCODE_FORBIDDEN_CHARACTERS flag
2285 if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS)) {
2287 pct_encode_val(*ptr, uri->canon_uri + uri->canon_len);
2289 uri->canon_len += 3;
2295 /* Nothing special, so just copy the character over. */
2296 uri->canon_uri[uri->canon_len] = *ptr;
2303 /* Canonicalizes the userinfo of the URI represented by the parse_data.
2305 * Canonicalization of the userinfo is a simple process. If there are any percent
2306 * encoded characters that fall in the "unreserved" character set, they are decoded
2307 * to their actual value. If a character is not in the "unreserved" or "reserved" sets
2308 * then it is percent encoded. Other than that the characters are copied over without
2311 static BOOL canonicalize_userinfo(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2312 uri->userinfo_start = uri->userinfo_split = -1;
2313 uri->userinfo_len = 0;
2315 if(!data->username && !data->password)
2316 /* URI doesn't have userinfo, so nothing to do here. */
2319 if(!canonicalize_username(data, uri, flags, computeOnly))
2322 if(!canonicalize_password(data, uri, flags, computeOnly))
2325 uri->userinfo_len = uri->canon_len - uri->userinfo_start;
2327 TRACE("(%p %p %x %d): Canonicalized userinfo, userinfo_start=%d, userinfo=%s, userinfo_split=%d userinfo_len=%d.\n",
2328 data, uri, flags, computeOnly, uri->userinfo_start, debugstr_wn(uri->canon_uri + uri->userinfo_start, uri->userinfo_len),
2329 uri->userinfo_split, uri->userinfo_len);
2331 /* Now insert the '@' after the userinfo. */
2333 uri->canon_uri[uri->canon_len] = '@';
2339 /* Attempts to canonicalize a reg_name.
2341 * Things that happen:
2342 * 1) If Uri_CREATE_NO_CANONICALIZE flag is not set, then the reg_name is
2343 * lower cased. Unless it's an unknown scheme type, which case it's
2344 * no lower cased reguardless.
2346 * 2) Unreserved % encoded characters are decoded for known
2349 * 3) Forbidden characters are % encoded as long as
2350 * Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS flag is not set and
2351 * it isn't an unknown scheme type.
2353 * 4) If it's a file scheme and the host is "localhost" it's removed.
2355 static BOOL canonicalize_reg_name(const parse_data *data, Uri *uri,
2356 DWORD flags, BOOL computeOnly) {
2357 static const WCHAR localhostW[] =
2358 {'l','o','c','a','l','h','o','s','t',0};
2360 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
2362 uri->host_start = uri->canon_len;
2364 if(data->scheme_type == URL_SCHEME_FILE &&
2365 data->host_len == lstrlenW(localhostW)) {
2366 if(!StrCmpNIW(data->host, localhostW, data->host_len)) {
2367 uri->host_start = -1;
2369 uri->host_type = Uri_HOST_UNKNOWN;
2374 for(ptr = data->host; ptr < data->host+data->host_len; ++ptr) {
2375 if(*ptr == '%' && known_scheme) {
2376 WCHAR val = decode_pct_val(ptr);
2377 if(is_unreserved(val)) {
2378 /* If NO_CANONICALZE is not set, then windows lower cases the
2381 if(!(flags & Uri_CREATE_NO_CANONICALIZE) && isupperW(val)) {
2383 uri->canon_uri[uri->canon_len] = tolowerW(val);
2386 uri->canon_uri[uri->canon_len] = val;
2390 /* Skip past the % encoded character. */
2394 /* Just copy the % over. */
2396 uri->canon_uri[uri->canon_len] = *ptr;
2399 } else if(*ptr == '\\') {
2400 /* Only unknown scheme types could have made it here with a '\\' in the host name. */
2402 uri->canon_uri[uri->canon_len] = *ptr;
2404 } else if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) &&
2405 !is_unreserved(*ptr) && !is_reserved(*ptr) && known_scheme) {
2407 pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
2409 /* The percent encoded value gets lower cased also. */
2410 if(!(flags & Uri_CREATE_NO_CANONICALIZE)) {
2411 uri->canon_uri[uri->canon_len+1] = tolowerW(uri->canon_uri[uri->canon_len+1]);
2412 uri->canon_uri[uri->canon_len+2] = tolowerW(uri->canon_uri[uri->canon_len+2]);
2416 uri->canon_len += 3;
2419 if(!(flags & Uri_CREATE_NO_CANONICALIZE) && known_scheme)
2420 uri->canon_uri[uri->canon_len] = tolowerW(*ptr);
2422 uri->canon_uri[uri->canon_len] = *ptr;
2429 uri->host_len = uri->canon_len - uri->host_start;
2432 TRACE("(%p %p %x %d): Canonicalize reg_name=%s len=%d\n", data, uri, flags,
2433 computeOnly, debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
2437 find_domain_name(uri->canon_uri+uri->host_start, uri->host_len,
2438 &(uri->domain_offset));
2443 /* Attempts to canonicalize an implicit IPv4 address. */
2444 static BOOL canonicalize_implicit_ipv4address(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2445 uri->host_start = uri->canon_len;
2447 TRACE("%u\n", data->implicit_ipv4);
2448 /* For unknown scheme types Window's doesn't convert
2449 * the value into an IP address, but, it still considers
2450 * it an IPv4 address.
2452 if(data->scheme_type == URL_SCHEME_UNKNOWN) {
2454 memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
2455 uri->canon_len += data->host_len;
2458 uri->canon_len += ui2ipv4(uri->canon_uri+uri->canon_len, data->implicit_ipv4);
2460 uri->canon_len += ui2ipv4(NULL, data->implicit_ipv4);
2463 uri->host_len = uri->canon_len - uri->host_start;
2464 uri->host_type = Uri_HOST_IPV4;
2467 TRACE("%p %p %x %d): Canonicalized implicit IP address=%s len=%d\n",
2468 data, uri, flags, computeOnly,
2469 debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
2475 /* Attempts to canonicalize an IPv4 address.
2477 * If the parse_data represents a URI that has an implicit IPv4 address
2478 * (ex. http://256/, this function will convert 256 into 0.0.1.0). If
2479 * the implicit IP address exceeds the value of UINT_MAX (maximum value
2480 * for an IPv4 address) it's canonicalized as if were a reg-name.
2482 * If the parse_data contains a partial or full IPv4 address it normalizes it.
2483 * A partial IPv4 address is something like "192.0" and would be normalized to
2484 * "192.0.0.0". With a full (or partial) IPv4 address like "192.002.01.003" would
2485 * be normalized to "192.2.1.3".
2488 * Window's ONLY normalizes IPv4 address for known scheme types (one that isn't
2489 * URL_SCHEME_UNKNOWN). For unknown scheme types, it simply copies the data from
2490 * the original URI into the canonicalized URI, but, it still recognizes URI's
2491 * host type as HOST_IPV4.
2493 static BOOL canonicalize_ipv4address(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2494 if(data->has_implicit_ip)
2495 return canonicalize_implicit_ipv4address(data, uri, flags, computeOnly);
2497 uri->host_start = uri->canon_len;
2499 /* Windows only normalizes for known scheme types. */
2500 if(data->scheme_type != URL_SCHEME_UNKNOWN) {
2501 /* parse_data contains a partial or full IPv4 address, so normalize it. */
2502 DWORD i, octetDigitCount = 0, octetCount = 0;
2503 BOOL octetHasDigit = FALSE;
2505 for(i = 0; i < data->host_len; ++i) {
2506 if(data->host[i] == '0' && !octetHasDigit) {
2507 /* Can ignore leading zeros if:
2508 * 1) It isn't the last digit of the octet.
2509 * 2) i+1 != data->host_len
2512 if(octetDigitCount == 2 ||
2513 i+1 == data->host_len ||
2514 data->host[i+1] == '.') {
2516 uri->canon_uri[uri->canon_len] = data->host[i];
2518 TRACE("Adding zero\n");
2520 } else if(data->host[i] == '.') {
2522 uri->canon_uri[uri->canon_len] = data->host[i];
2525 octetDigitCount = 0;
2526 octetHasDigit = FALSE;
2530 uri->canon_uri[uri->canon_len] = data->host[i];
2534 octetHasDigit = TRUE;
2538 /* Make sure the canonicalized IP address has 4 dec-octets.
2539 * If doesn't add "0" ones until there is 4;
2541 for( ; octetCount < 3; ++octetCount) {
2543 uri->canon_uri[uri->canon_len] = '.';
2544 uri->canon_uri[uri->canon_len+1] = '0';
2547 uri->canon_len += 2;
2550 /* Windows doesn't normalize addresses in unknown schemes. */
2552 memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
2553 uri->canon_len += data->host_len;
2556 uri->host_len = uri->canon_len - uri->host_start;
2558 TRACE("(%p %p %x %d): Canonicalized IPv4 address, ip=%s len=%d\n",
2559 data, uri, flags, computeOnly,
2560 debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
2567 /* Attempts to canonicalize the IPv6 address of the URI.
2569 * Multiple things happen during the canonicalization of an IPv6 address:
2570 * 1) Any leading zero's in an h16 component are removed.
2571 * Ex: [0001:0022::] -> [1:22::]
2573 * 2) The longest sequence of zero h16 components are compressed
2574 * into a "::" (elision). If there's a tie, the first is choosen.
2576 * Ex: [0:0:0:0:1:6:7:8] -> [::1:6:7:8]
2577 * [0:0:0:0:1:2::] -> [::1:2:0:0]
2578 * [0:0:1:2:0:0:7:8] -> [::1:2:0:0:7:8]
2580 * 3) If an IPv4 address is attached to the IPv6 address, it's
2582 * Ex: [::001.002.022.000] -> [::1.2.22.0]
2584 * 4) If an elision is present, but, only represents 1 h16 component
2587 * Ex: [1::2:3:4:5:6:7] -> [1:0:2:3:4:5:6:7]
2589 * 5) If the IPv6 address contains an IPv4 address and there exists
2590 * at least 1 non-zero h16 component the IPv4 address is converted
2591 * into two h16 components, otherwise it's normalized and kept as is.
2593 * Ex: [::192.200.003.4] -> [::192.200.3.4]
2594 * [ffff::192.200.003.4] -> [ffff::c0c8:3041]
2597 * For unknown scheme types Windows simply copies the address over without any
2600 * IPv4 address can be included in an elision if all its components are 0's.
2602 static BOOL canonicalize_ipv6address(const parse_data *data, Uri *uri,
2603 DWORD flags, BOOL computeOnly) {
2604 uri->host_start = uri->canon_len;
2606 if(data->scheme_type == URL_SCHEME_UNKNOWN) {
2608 memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
2609 uri->canon_len += data->host_len;
2613 DWORD i, elision_len;
2615 if(!ipv6_to_number(&(data->ipv6_address), values)) {
2616 TRACE("(%p %p %x %d): Failed to compute numerical value for IPv6 address.\n",
2617 data, uri, flags, computeOnly);
2622 uri->canon_uri[uri->canon_len] = '[';
2625 /* Find where the elision should occur (if any). */
2626 compute_elision_location(&(data->ipv6_address), values, &elision_start, &elision_len);
2628 TRACE("%p %p %x %d): Elision starts at %d, len=%u\n", data, uri, flags,
2629 computeOnly, elision_start, elision_len);
2631 for(i = 0; i < 8; ++i) {
2632 BOOL in_elision = (elision_start > -1 && i >= elision_start &&
2633 i < elision_start+elision_len);
2634 BOOL do_ipv4 = (i == 6 && data->ipv6_address.ipv4 && !in_elision &&
2635 data->ipv6_address.h16_count == 0);
2637 if(i == elision_start) {
2639 uri->canon_uri[uri->canon_len] = ':';
2640 uri->canon_uri[uri->canon_len+1] = ':';
2642 uri->canon_len += 2;
2645 /* We can ignore the current component if we're in the elision. */
2649 /* We only add a ':' if we're not at i == 0, or when we're at
2650 * the very end of elision range since the ':' colon was handled
2651 * earlier. Otherwise we would end up with ":::" after elision.
2653 if(i != 0 && !(elision_start > -1 && i == elision_start+elision_len)) {
2655 uri->canon_uri[uri->canon_len] = ':';
2663 /* Combine the two parts of the IPv4 address values. */
2669 len = ui2ipv4(uri->canon_uri+uri->canon_len, val);
2671 len = ui2ipv4(NULL, val);
2673 uri->canon_len += len;
2676 /* Write a regular h16 component to the URI. */
2678 /* Short circuit for the trivial case. */
2679 if(values[i] == 0) {
2681 uri->canon_uri[uri->canon_len] = '0';
2684 static const WCHAR formatW[] = {'%','x',0};
2687 uri->canon_len += sprintfW(uri->canon_uri+uri->canon_len,
2688 formatW, values[i]);
2691 uri->canon_len += sprintfW(tmp, formatW, values[i]);
2697 /* Add the closing ']'. */
2699 uri->canon_uri[uri->canon_len] = ']';
2703 uri->host_len = uri->canon_len - uri->host_start;
2706 TRACE("(%p %p %x %d): Canonicalized IPv6 address %s, len=%d\n", data, uri, flags,
2707 computeOnly, debugstr_wn(uri->canon_uri+uri->host_start, uri->host_len),
2713 /* Attempts to canonicalize the host of the URI (if any). */
2714 static BOOL canonicalize_host(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2715 uri->host_start = -1;
2717 uri->domain_offset = -1;
2720 switch(data->host_type) {
2722 uri->host_type = Uri_HOST_DNS;
2723 if(!canonicalize_reg_name(data, uri, flags, computeOnly))
2728 uri->host_type = Uri_HOST_IPV4;
2729 if(!canonicalize_ipv4address(data, uri, flags, computeOnly))
2734 if(!canonicalize_ipv6address(data, uri, flags, computeOnly))
2737 uri->host_type = Uri_HOST_IPV6;
2739 case Uri_HOST_UNKNOWN:
2740 if(data->host_len > 0 || data->scheme_type != URL_SCHEME_FILE) {
2741 uri->host_start = uri->canon_len;
2743 /* Nothing happens to unknown host types. */
2745 memcpy(uri->canon_uri+uri->canon_len, data->host, data->host_len*sizeof(WCHAR));
2746 uri->canon_len += data->host_len;
2747 uri->host_len = data->host_len;
2750 uri->host_type = Uri_HOST_UNKNOWN;
2753 FIXME("(%p %p %x %d): Canonicalization for host type %d not supported.\n", data,
2754 uri, flags, computeOnly, data->host_type);
2762 static BOOL canonicalize_port(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2763 BOOL has_default_port = FALSE;
2764 USHORT default_port = 0;
2767 uri->has_port = FALSE;
2769 /* Check if the scheme has a default port. */
2770 for(i = 0; i < sizeof(default_ports)/sizeof(default_ports[0]); ++i) {
2771 if(default_ports[i].scheme == data->scheme_type) {
2772 has_default_port = TRUE;
2773 default_port = default_ports[i].port;
2778 if(data->port || has_default_port)
2779 uri->has_port = TRUE;
2782 * 1) Has a port which is the default port.
2783 * 2) Has a port (not the default).
2784 * 3) Doesn't have a port, but, scheme has a default port.
2787 if(has_default_port && data->port && data->port_value == default_port) {
2788 /* If it's the default port and this flag isn't set, don't do anything. */
2789 if(flags & Uri_CREATE_NO_CANONICALIZE) {
2790 /* Copy the original port over. */
2792 uri->canon_uri[uri->canon_len] = ':';
2793 memcpy(uri->canon_uri+uri->canon_len+1, data->port, data->port_len*sizeof(WCHAR));
2795 uri->canon_len += data->port_len+1;
2798 uri->port = default_port;
2799 } else if(data->port) {
2801 uri->canon_uri[uri->canon_len] = ':';
2804 if(flags & Uri_CREATE_NO_CANONICALIZE) {
2805 /* Copy the original over without changes. */
2807 memcpy(uri->canon_uri+uri->canon_len, data->port, data->port_len*sizeof(WCHAR));
2808 uri->canon_len += data->port_len;
2810 const WCHAR formatW[] = {'%','u',0};
2813 len = sprintfW(uri->canon_uri+uri->canon_len, formatW, data->port_value);
2816 len = sprintfW(tmp, formatW, data->port_value);
2818 uri->canon_len += len;
2821 uri->port = data->port_value;
2822 } else if(has_default_port)
2823 uri->port = default_port;
2828 /* Canonicalizes the authority of the URI represented by the parse_data. */
2829 static BOOL canonicalize_authority(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
2830 uri->authority_start = uri->canon_len;
2831 uri->authority_len = 0;
2833 if(!canonicalize_userinfo(data, uri, flags, computeOnly))
2836 if(!canonicalize_host(data, uri, flags, computeOnly))
2839 if(!canonicalize_port(data, uri, flags, computeOnly))
2842 if(uri->host_start != -1)
2843 uri->authority_len = uri->canon_len - uri->authority_start;
2845 uri->authority_start = -1;
2850 /* Attempts to canonicalize the path of a hierarchical URI.
2852 * Things that happen:
2853 * 1). Forbidden characters are percent encoded, unless the NO_ENCODE_FORBIDDEN
2854 * flag is set or it's a file URI. Forbidden characters are always encoded
2855 * for file schemes reguardless and forbidden characters are never encoded
2856 * for unknown scheme types.
2858 * 2). For known scheme types '\\' are changed to '/'.
2860 * 3). Percent encoded, unreserved characters are decoded to their actual values.
2861 * Unless the scheme type is unknown. For file schemes any percent encoded
2862 * character in the unreserved or reserved set is decoded.
2864 * 4). For File schemes if the path is starts with a drive letter and doesn't
2865 * start with a '/' then one is appended.
2866 * Ex: file://c:/test.mp3 -> file:///c:/test.mp3
2868 * 5). Dot segments are removed from the path for all scheme types
2869 * unless NO_CANONICALIZE flag is set. Dot segments aren't removed
2870 * for wildcard scheme types.
2873 * file://c:/test%20test -> file:///c:/test%2520test
2874 * file://c:/test%3Etest -> file:///c:/test%253Etest
2875 * file:///c:/test%20test -> file:///c:/test%20test
2876 * file:///c:/test%test -> file:///c:/test%25test
2878 static BOOL canonicalize_path_hierarchical(const parse_data *data, Uri *uri,
2879 DWORD flags, BOOL computeOnly) {
2881 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
2882 const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
2884 BOOL escape_pct = FALSE;
2887 uri->path_start = -1;
2892 uri->path_start = uri->canon_len;
2895 if(is_file && uri->host_start == -1) {
2896 /* Check if a '/' needs to be appended for the file scheme. */
2897 if(data->path_len > 1 && is_drive_path(ptr) && !(flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
2899 uri->canon_uri[uri->canon_len] = '/';
2902 } else if(*ptr == '/') {
2903 if(!(flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
2904 /* Copy the extra '/' over. */
2906 uri->canon_uri[uri->canon_len] = '/';
2912 if(is_drive_path(ptr)) {
2914 uri->canon_uri[uri->canon_len] = *ptr;
2915 /* If theres a '|' after the drive letter, convert it to a ':'. */
2916 uri->canon_uri[uri->canon_len+1] = ':';
2919 uri->canon_len += 2;
2923 for(; ptr < data->path+data->path_len; ++ptr) {
2925 const WCHAR *tmp = ptr;
2928 /* Check if the % represents a valid encoded char, or if it needs encoded. */
2929 BOOL force_encode = !check_pct_encoded(&tmp) && is_file;
2930 val = decode_pct_val(ptr);
2932 if(force_encode || escape_pct) {
2933 /* Escape the percent sign in the file URI. */
2935 pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
2936 uri->canon_len += 3;
2937 } else if((is_unreserved(val) && known_scheme) ||
2938 (is_file && (is_unreserved(val) || is_reserved(val)))) {
2940 uri->canon_uri[uri->canon_len] = val;
2947 uri->canon_uri[uri->canon_len] = *ptr;
2950 } else if(*ptr == '/' && is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
2951 /* Convert the '/' back to a '\\'. */
2953 uri->canon_uri[uri->canon_len] = '\\';
2955 } else if(*ptr == '\\' && known_scheme) {
2956 if(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
2957 /* Don't convert the '\\' to a '/'. */
2959 uri->canon_uri[uri->canon_len] = *ptr;
2963 uri->canon_uri[uri->canon_len] = '/';
2966 } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr) &&
2967 (!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) || is_file)) {
2968 if(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
2969 /* Don't escape the character. */
2971 uri->canon_uri[uri->canon_len] = *ptr;
2974 /* Escape the forbidden character. */
2976 pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
2977 uri->canon_len += 3;
2981 uri->canon_uri[uri->canon_len] = *ptr;
2986 uri->path_len = uri->canon_len - uri->path_start;
2988 /* Removing the dot segments only happens when it's not in
2989 * computeOnly mode and it's not a wildcard scheme. File schemes
2990 * with USE_DOS_PATH set don't get dot segments removed.
2992 if(!(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) &&
2993 data->scheme_type != URL_SCHEME_WILDCARD) {
2994 if(!(flags & Uri_CREATE_NO_CANONICALIZE) && !computeOnly) {
2995 /* Remove the dot segments (if any) and reset everything to the new
2998 DWORD new_len = remove_dot_segments(uri->canon_uri+uri->path_start, uri->path_len);
2999 uri->canon_len -= uri->path_len-new_len;
3000 uri->path_len = new_len;
3005 TRACE("Canonicalized path %s len=%d\n",
3006 debugstr_wn(uri->canon_uri+uri->path_start, uri->path_len),
3012 /* Attempts to canonicalize the path for an opaque URI.
3014 * For known scheme types:
3015 * 1) forbidden characters are percent encoded if
3016 * NO_ENCODE_FORBIDDEN_CHARACTERS isn't set.
3018 * 2) Percent encoded, unreserved characters are decoded
3019 * to their actual values, for known scheme types.
3021 * 3) '\\' are changed to '/' for known scheme types
3022 * except for mailto schemes.
3024 * 4) For file schemes, if USE_DOS_PATH is set all '/'
3025 * are converted to backslashes.
3027 * 5) For file schemes, if USE_DOS_PATH isn't set all '\'
3028 * are converted to forward slashes.
3030 static BOOL canonicalize_path_opaque(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
3032 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
3033 const BOOL is_file = data->scheme_type == URL_SCHEME_FILE;
3036 uri->path_start = -1;
3041 uri->path_start = uri->canon_len;
3043 /* Windows doesn't allow a "//" to appear after the scheme
3044 * of a URI, if it's an opaque URI.
3046 if(data->scheme && *(data->path) == '/' && *(data->path+1) == '/') {
3047 /* So it inserts a "/." before the "//" if it exists. */
3049 uri->canon_uri[uri->canon_len] = '/';
3050 uri->canon_uri[uri->canon_len+1] = '.';
3053 uri->canon_len += 2;
3056 for(ptr = data->path; ptr < data->path+data->path_len; ++ptr) {
3057 if(*ptr == '%' && known_scheme) {
3058 WCHAR val = decode_pct_val(ptr);
3060 if(is_unreserved(val)) {
3062 uri->canon_uri[uri->canon_len] = val;
3069 uri->canon_uri[uri->canon_len] = *ptr;
3072 } else if(*ptr == '/' && is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
3074 uri->canon_uri[uri->canon_len] = '\\';
3076 } else if(*ptr == '\\' && is_file) {
3077 if(!(flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
3078 /* Convert to a '/'. */
3080 uri->canon_uri[uri->canon_len] = '/';
3083 /* Just copy it over. */
3085 uri->canon_uri[uri->canon_len] = *ptr;
3088 } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr) &&
3089 !(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS)) {
3090 if(is_file && (flags & Uri_CREATE_FILE_USE_DOS_PATH)) {
3091 /* Forbidden characters aren't percent encoded for file schemes
3092 * with USE_DOS_PATH set.
3095 uri->canon_uri[uri->canon_len] = *ptr;
3097 } else if(data->scheme_type == URL_SCHEME_MK && *ptr == '\\') {
3098 /* MK URIs don't get '\\' percent encoded. */
3100 uri->canon_uri[uri->canon_len] = *ptr;
3104 pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
3105 uri->canon_len += 3;
3109 uri->canon_uri[uri->canon_len] = *ptr;
3114 uri->path_len = uri->canon_len - uri->path_start;
3116 TRACE("(%p %p %x %d): Canonicalized opaque URI path %s len=%d\n", data, uri, flags, computeOnly,
3117 debugstr_wn(uri->canon_uri+uri->path_start, uri->path_len), uri->path_len);
3121 /* Determines how the URI represented by the parse_data should be canonicalized.
3123 * Essentially, if the parse_data represents an hierarchical URI then it calls
3124 * canonicalize_authority and the canonicalization functions for the path. If the
3125 * URI is opaque it canonicalizes the path of the URI.
3127 static BOOL canonicalize_hierpart(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
3128 uri->display_absolute = TRUE;
3130 if(!data->is_opaque) {
3131 /* "//" is only added for non-wildcard scheme types. */
3132 if(data->scheme_type != URL_SCHEME_WILDCARD) {
3134 INT pos = uri->canon_len;
3136 uri->canon_uri[pos] = '/';
3137 uri->canon_uri[pos+1] = '/';
3139 uri->canon_len += 2;
3142 if(!canonicalize_authority(data, uri, flags, computeOnly))
3145 /* TODO: Canonicalize the path of the URI. */
3146 if(!canonicalize_path_hierarchical(data, uri, flags, computeOnly))
3150 /* Opaque URI's don't have an authority. */
3151 uri->userinfo_start = uri->userinfo_split = -1;
3152 uri->userinfo_len = 0;
3153 uri->host_start = -1;
3155 uri->host_type = Uri_HOST_UNKNOWN;
3156 uri->has_port = FALSE;
3157 uri->authority_start = -1;
3158 uri->authority_len = 0;
3159 uri->domain_offset = -1;
3161 if(is_hierarchical_scheme(data->scheme_type)) {
3164 /* Absolute URIs aren't displayed for known scheme types
3165 * which should be hierarchical URIs.
3167 uri->display_absolute = FALSE;
3169 /* Windows also sets the port for these (if they have one). */
3170 for(i = 0; i < sizeof(default_ports)/sizeof(default_ports[0]); ++i) {
3171 if(data->scheme_type == default_ports[i].scheme) {
3172 uri->has_port = TRUE;
3173 uri->port = default_ports[i].port;
3179 if(!canonicalize_path_opaque(data, uri, flags, computeOnly))
3183 if(uri->path_start > -1 && !computeOnly)
3184 /* Finding file extensions happens for both types of URIs. */
3185 uri->extension_offset = find_file_extension(uri->canon_uri+uri->path_start, uri->path_len);
3187 uri->extension_offset = -1;
3192 /* Attempts to canonicalize the query string of the URI.
3194 * Things that happen:
3195 * 1) For known scheme types forbidden characters
3196 * are percent encoded, unless the NO_DECODE_EXTRA_INFO flag is set
3197 * or NO_ENCODE_FORBIDDEN_CHARACTERS is set.
3199 * 2) For known scheme types, percent encoded, unreserved characters
3200 * are decoded as long as the NO_DECODE_EXTRA_INFO flag isn't set.
3202 static BOOL canonicalize_query(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
3203 const WCHAR *ptr, *end;
3204 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
3207 uri->query_start = -1;
3212 uri->query_start = uri->canon_len;
3214 end = data->query+data->query_len;
3215 for(ptr = data->query; ptr < end; ++ptr) {
3217 if(known_scheme && !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
3218 WCHAR val = decode_pct_val(ptr);
3219 if(is_unreserved(val)) {
3221 uri->canon_uri[uri->canon_len] = val;
3228 } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr)) {
3229 if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) &&
3230 !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
3232 pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
3233 uri->canon_len += 3;
3239 uri->canon_uri[uri->canon_len] = *ptr;
3243 uri->query_len = uri->canon_len - uri->query_start;
3246 TRACE("(%p %p %x %d): Canonicalized query string %s len=%d\n", data, uri, flags,
3247 computeOnly, debugstr_wn(uri->canon_uri+uri->query_start, uri->query_len),
3252 static BOOL canonicalize_fragment(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
3253 const WCHAR *ptr, *end;
3254 const BOOL known_scheme = data->scheme_type != URL_SCHEME_UNKNOWN;
3256 if(!data->fragment) {
3257 uri->fragment_start = -1;
3258 uri->fragment_len = 0;
3262 uri->fragment_start = uri->canon_len;
3264 end = data->fragment + data->fragment_len;
3265 for(ptr = data->fragment; ptr < end; ++ptr) {
3267 if(known_scheme && !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
3268 WCHAR val = decode_pct_val(ptr);
3269 if(is_unreserved(val)) {
3271 uri->canon_uri[uri->canon_len] = val;
3278 } else if(known_scheme && !is_unreserved(*ptr) && !is_reserved(*ptr)) {
3279 if(!(flags & Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS) &&
3280 !(flags & Uri_CREATE_NO_DECODE_EXTRA_INFO)) {
3282 pct_encode_val(*ptr, uri->canon_uri+uri->canon_len);
3283 uri->canon_len += 3;
3289 uri->canon_uri[uri->canon_len] = *ptr;
3293 uri->fragment_len = uri->canon_len - uri->fragment_start;
3296 TRACE("(%p %p %x %d): Canonicalized fragment %s len=%d\n", data, uri, flags,
3297 computeOnly, debugstr_wn(uri->canon_uri+uri->fragment_start, uri->fragment_len),
3302 /* Canonicalizes the scheme information specified in the parse_data using the specified flags. */
3303 static BOOL canonicalize_scheme(const parse_data *data, Uri *uri, DWORD flags, BOOL computeOnly) {
3304 uri->scheme_start = -1;
3305 uri->scheme_len = 0;
3308 /* The only type of URI that doesn't have to have a scheme is a relative
3311 if(!data->is_relative) {
3312 FIXME("(%p %p %x): Unable to determine the scheme type of %s.\n", data,
3313 uri, flags, debugstr_w(data->uri));
3319 INT pos = uri->canon_len;
3321 for(i = 0; i < data->scheme_len; ++i) {
3322 /* Scheme name must be lower case after canonicalization. */
3323 uri->canon_uri[i + pos] = tolowerW(data->scheme[i]);
3326 uri->canon_uri[i + pos] = ':';
3327 uri->scheme_start = pos;
3329 TRACE("(%p %p %x): Canonicalized scheme=%s, len=%d.\n", data, uri, flags,
3330 debugstr_wn(uri->canon_uri, uri->scheme_len), data->scheme_len);
3333 /* This happens in both computation modes. */
3334 uri->canon_len += data->scheme_len + 1;
3335 uri->scheme_len = data->scheme_len;
3340 /* Compute's what the length of the URI specified by the parse_data will be
3341 * after canonicalization occurs using the specified flags.
3343 * This function will return a non-zero value indicating the length of the canonicalized
3344 * URI, or -1 on error.
3346 static int compute_canonicalized_length(const parse_data *data, DWORD flags) {
3349 memset(&uri, 0, sizeof(Uri));
3351 TRACE("(%p %x): Beginning to compute canonicalized length for URI %s\n", data, flags,
3352 debugstr_w(data->uri));
3354 if(!canonicalize_scheme(data, &uri, flags, TRUE)) {
3355 ERR("(%p %x): Failed to compute URI scheme length.\n", data, flags);
3359 if(!canonicalize_hierpart(data, &uri, flags, TRUE)) {
3360 ERR("(%p %x): Failed to compute URI hierpart length.\n", data, flags);
3364 if(!canonicalize_query(data, &uri, flags, TRUE)) {
3365 ERR("(%p %x): Failed to compute query string length.\n", data, flags);
3369 if(!canonicalize_fragment(data, &uri, flags, TRUE)) {
3370 ERR("(%p %x): Failed to compute fragment length.\n", data, flags);
3374 TRACE("(%p %x): Finished computing canonicalized URI length. length=%d\n", data, flags, uri.canon_len);
3376 return uri.canon_len;
3379 /* Canonicalizes the URI data specified in the parse_data, using the given flags. If the
3380 * canonicalization succeededs it will store all the canonicalization information
3381 * in the pointer to the Uri.
3383 * To canonicalize a URI this function first computes what the length of the URI
3384 * specified by the parse_data will be. Once this is done it will then perfom the actual
3385 * canonicalization of the URI.
3387 static HRESULT canonicalize_uri(const parse_data *data, Uri *uri, DWORD flags) {
3390 uri->canon_uri = NULL;
3391 len = uri->canon_size = uri->canon_len = 0;
3393 TRACE("(%p %p %x): beginning to canonicalize URI %s.\n", data, uri, flags, debugstr_w(data->uri));
3395 /* First try to compute the length of the URI. */
3396 len = compute_canonicalized_length(data, flags);
3398 ERR("(%p %p %x): Could not compute the canonicalized length of %s.\n", data, uri, flags,
3399 debugstr_w(data->uri));
3400 return E_INVALIDARG;
3403 uri->canon_uri = heap_alloc((len+1)*sizeof(WCHAR));
3405 return E_OUTOFMEMORY;
3407 uri->canon_size = len;
3408 if(!canonicalize_scheme(data, uri, flags, FALSE)) {
3409 ERR("(%p %p %x): Unable to canonicalize the scheme of the URI.\n", data, uri, flags);
3410 heap_free(uri->canon_uri);
3411 return E_INVALIDARG;
3413 uri->scheme_type = data->scheme_type;
3415 if(!canonicalize_hierpart(data, uri, flags, FALSE)) {
3416 ERR("(%p %p %x): Unable to canonicalize the heirpart of the URI\n", data, uri, flags);
3417 heap_free(uri->canon_uri);
3418 return E_INVALIDARG;
3421 if(!canonicalize_query(data, uri, flags, FALSE)) {
3422 ERR("(%p %p %x): Unable to canonicalize query string of the URI.\n",
3424 return E_INVALIDARG;
3427 if(!canonicalize_fragment(data, uri, flags, FALSE)) {
3428 ERR("(%p %p %x): Unable to canonicalize fragment of the URI.\n",
3430 return E_INVALIDARG;
3433 /* There's a possibility we didn't use all the space we allocated
3436 if(uri->canon_len < uri->canon_size) {
3437 /* This happens if the URI is hierarchical and dot
3438 * segments were removed from it's path.
3440 WCHAR *tmp = heap_realloc(uri->canon_uri, (uri->canon_len+1)*sizeof(WCHAR));
3442 return E_OUTOFMEMORY;
3444 uri->canon_uri = tmp;
3445 uri->canon_size = uri->canon_len;
3448 uri->canon_uri[uri->canon_len] = '\0';
3449 TRACE("(%p %p %x): finished canonicalizing the URI. uri=%s\n", data, uri, flags, debugstr_w(uri->canon_uri));
3454 static HRESULT get_builder_component(LPWSTR *component, DWORD *component_len,
3455 LPCWSTR source, DWORD source_len,
3456 LPCWSTR *output, DWORD *output_len)
3469 if(!(*component) && source) {
3470 /* Allocate 'component', and copy the contents from 'source'
3471 * into the new allocation.
3473 *component = heap_alloc((source_len+1)*sizeof(WCHAR));
3475 return E_OUTOFMEMORY;
3477 memcpy(*component, source, source_len*sizeof(WCHAR));
3478 (*component)[source_len] = '\0';
3479 *component_len = source_len;
3482 *output = *component;
3483 *output_len = *component_len;
3484 return *output ? S_OK : S_FALSE;
3487 /* Allocates 'component' and copies the string from 'new_value' into 'component'.
3488 * If 'prefix' is set and 'new_value' isn't NULL, then it checks if 'new_value'
3489 * starts with 'prefix'. If it doesn't then 'prefix' is prepended to 'component'.
3491 * If everything is successful, then will set 'success_flag' in 'flags'.
3493 static HRESULT set_builder_component(LPWSTR *component, DWORD *component_len, LPCWSTR new_value,
3494 WCHAR prefix, DWORD *flags, DWORD success_flag)
3496 heap_free(*component);
3502 BOOL add_prefix = FALSE;
3503 DWORD len = lstrlenW(new_value);
3506 if(prefix && *new_value != prefix) {
3508 *component = heap_alloc((len+2)*sizeof(WCHAR));
3510 *component = heap_alloc((len+1)*sizeof(WCHAR));
3513 return E_OUTOFMEMORY;
3516 (*component)[pos++] = prefix;
3518 memcpy(*component+pos, new_value, (len+1)*sizeof(WCHAR));
3519 *component_len = len+pos;
3522 *flags |= success_flag;
3526 #define URI(x) ((IUri*) &(x)->lpIUriVtbl)
3527 #define URIBUILDER(x) ((IUriBuilder*) &(x)->lpIUriBuilderVtbl)
3529 static void reset_builder(UriBuilder *builder) {
3531 IUri_Release(URI(builder->uri));
3532 builder->uri = NULL;
3534 heap_free(builder->fragment);
3535 builder->fragment = NULL;
3536 builder->fragment_len = 0;
3538 heap_free(builder->host);
3539 builder->host = NULL;
3540 builder->host_len = 0;
3542 heap_free(builder->password);
3543 builder->password = NULL;
3544 builder->password_len = 0;
3546 heap_free(builder->path);
3547 builder->path = NULL;
3548 builder->path_len = 0;
3550 heap_free(builder->query);
3551 builder->query = NULL;
3552 builder->query_len = 0;
3554 heap_free(builder->scheme);
3555 builder->scheme = NULL;
3556 builder->scheme_len = 0;
3558 heap_free(builder->username);
3559 builder->username = NULL;
3560 builder->username_len = 0;
3562 builder->has_port = FALSE;
3564 builder->modified_props = 0;
3567 static HRESULT validate_scheme_name(const UriBuilder *builder, parse_data *data, DWORD flags) {
3568 const WCHAR *component;
3573 if(builder->scheme) {
3574 ptr = builder->scheme;
3575 expected_len = builder->scheme_len;
3576 } else if(builder->uri && builder->uri->scheme_start > -1) {
3577 ptr = builder->uri->canon_uri+builder->uri->scheme_start;
3578 expected_len = builder->uri->scheme_len;
3580 static const WCHAR nullW[] = {0};
3587 if(parse_scheme(pptr, data, flags, ALLOW_NULL_TERM_SCHEME) &&
3588 data->scheme_len == expected_len) {
3590 TRACE("(%p %p %x): Found valid scheme component %s len=%d.\n", builder, data, flags,
3591 debugstr_wn(data->scheme, data->scheme_len), data->scheme_len);
3593 TRACE("(%p %p %x): Invalid scheme component found %s.\n", builder, data, flags,
3594 debugstr_wn(component, expected_len));
3595 return INET_E_INVALID_URL;
3601 static HRESULT validate_username(const UriBuilder *builder, parse_data *data, DWORD flags) {
3606 if(builder->username) {
3607 ptr = builder->username;
3608 expected_len = builder->username_len;
3609 } else if(!(builder->modified_props & Uri_HAS_USER_NAME) && builder->uri &&
3610 builder->uri->userinfo_start > -1 && builder->uri->userinfo_split != 0) {
3611 /* Just use the username from the base Uri. */
3612 data->username = builder->uri->canon_uri+builder->uri->userinfo_start;
3613 data->username_len = (builder->uri->userinfo_split > -1) ?
3614 builder->uri->userinfo_split : builder->uri->userinfo_len;
3622 const WCHAR *component = ptr;
3624 if(parse_username(pptr, data, flags, ALLOW_NULL_TERM_USER_NAME) &&
3625 data->username_len == expected_len)
3626 TRACE("(%p %p %x): Found valid username component %s len=%d.\n", builder, data, flags,
3627 debugstr_wn(data->username, data->username_len), data->username_len);
3629 TRACE("(%p %p %x): Invalid username component found %s.\n", builder, data, flags,
3630 debugstr_wn(component, expected_len));
3631 return INET_E_INVALID_URL;
3638 static HRESULT validate_password(const UriBuilder *builder, parse_data *data, DWORD flags) {
3643 if(builder->password) {
3644 ptr = builder->password;
3645 expected_len = builder->password_len;
3646 } else if(!(builder->modified_props & Uri_HAS_PASSWORD) && builder->uri &&
3647 builder->uri->userinfo_split > -1) {
3648 data->password = builder->uri->canon_uri+builder->uri->userinfo_start+builder->uri->userinfo_split+1;
3649 data->password_len = builder->uri->userinfo_len-builder->uri->userinfo_split-1;
3657 const WCHAR *component = ptr;
3659 if(parse_password(pptr, data, flags, ALLOW_NULL_TERM_PASSWORD) &&
3660 data->password_len == expected_len)
3661 TRACE("(%p %p %x): Found valid password component %s len=%d.\n", builder, data, flags,
3662 debugstr_wn(data->password, data->password_len), data->password_len);
3664 TRACE("(%p %p %x): Invalid password component found %s.\n", builder, data, flags,
3665 debugstr_wn(component, expected_len));
3666 return INET_E_INVALID_URL;
3673 static HRESULT validate_userinfo(const UriBuilder *builder, parse_data *data, DWORD flags) {
3676 hr = validate_username(builder, data, flags);
3680 hr = validate_password(builder, data, flags);
3687 static HRESULT validate_host(const UriBuilder *builder, parse_data *data, DWORD flags) {
3693 ptr = builder->host;
3694 expected_len = builder->host_len;
3695 } else if(!(builder->modified_props & Uri_HAS_HOST) && builder->uri && builder->uri->host_start > -1) {
3696 ptr = builder->uri->canon_uri + builder->uri->host_start;
3697 expected_len = builder->uri->host_len;
3702 const WCHAR *component = ptr;
3703 DWORD extras = ALLOW_BRACKETLESS_IP_LITERAL|IGNORE_PORT_DELIMITER|SKIP_IP_FUTURE_CHECK;
3706 if(parse_host(pptr, data, flags, extras) && data->host_len == expected_len)
3707 TRACE("(%p %p %x): Found valid host name %s len=%d type=%d.\n", builder, data, flags,
3708 debugstr_wn(data->host, data->host_len), data->host_len, data->host_type);
3710 TRACE("(%p %p %x): Invalid host name found %s.\n", builder, data, flags,
3711 debugstr_wn(component, expected_len));
3712 return INET_E_INVALID_URL;
3719 static void setup_port(const UriBuilder *builder, parse_data *data, DWORD flags) {
3720 if(builder->modified_props & Uri_HAS_PORT) {
3721 if(builder->has_port) {
3722 data->has_port = TRUE;
3723 data->port_value = builder->port;
3725 } else if(builder->uri && builder->uri->has_port) {
3726 data->has_port = TRUE;
3727 data->port_value = builder->uri->port;
3731 TRACE("(%p %p %x): Using %u as port for IUri.\n", builder, data, flags, data->port_value);
3734 static HRESULT validate_path(const UriBuilder *builder, parse_data *data, DWORD flags) {
3735 const WCHAR *ptr = NULL;
3740 ptr = builder->path;
3741 expected_len = builder->path_len;
3742 } else if(!(builder->modified_props & Uri_HAS_PATH) &&
3743 builder->uri && builder->uri->path_start > -1) {
3744 ptr = builder->uri->canon_uri+builder->uri->path_start;
3745 expected_len = builder->uri->path_len;
3750 const WCHAR *component = ptr;
3753 /* How the path is validated depends on what type of
3756 valid = data->is_opaque ?
3757 parse_path_opaque(pptr, data, flags) : parse_path_hierarchical(pptr, data, flags);
3759 if(!valid || expected_len != data->path_len) {
3760 TRACE("(%p %p %x): Invalid path componet %s.\n", builder, data, flags,
3761 debugstr_wn(component, expected_len));
3762 return INET_E_INVALID_URL;
3765 TRACE("(%p %p %x): Valid path component %s len=%d.\n", builder, data, flags,
3766 debugstr_wn(data->path, data->path_len), data->path_len);
3772 static HRESULT validate_query(const UriBuilder *builder, parse_data *data, DWORD flags) {
3773 const WCHAR *ptr = NULL;
3777 if(builder->query) {
3778 ptr = builder->query;
3779 expected_len = builder->query_len;
3780 } else if(!(builder->modified_props & Uri_HAS_QUERY) && builder->uri &&
3781 builder->uri->query_start > -1) {
3782 ptr = builder->uri->canon_uri+builder->uri->query_start;
3783 expected_len = builder->uri->query_len;
3787 const WCHAR *component = ptr;
3790 if(parse_query(pptr, data, flags) && expected_len == data->query_len)
3791 TRACE("(%p %p %x): Valid query component %s len=%d.\n", builder, data, flags,
3792 debugstr_wn(data->query, data->query_len), data->query_len);
3794 TRACE("(%p %p %x): Invalid query component %s.\n", builder, data, flags,
3795 debugstr_wn(component, expected_len));
3796 return INET_E_INVALID_URL;
3803 static HRESULT validate_fragment(const UriBuilder *builder, parse_data *data, DWORD flags) {
3804 const WCHAR *ptr = NULL;
3808 if(builder->fragment) {
3809 ptr = builder->fragment;
3810 expected_len = builder->fragment_len;
3811 } else if(!(builder->modified_props & Uri_HAS_FRAGMENT) && builder->uri &&
3812 builder->uri->fragment_start > -1) {
3813 ptr = builder->uri->canon_uri+builder->uri->fragment_start;
3814 expected_len = builder->uri->fragment_len;
3818 const WCHAR *component = ptr;
3821 if(parse_fragment(pptr, data, flags) && expected_len == data->fragment_len)
3822 TRACE("(%p %p %x): Valid fragment component %s len=%d.\n", builder, data, flags,
3823 debugstr_wn(data->fragment, data->fragment_len), data->fragment_len);
3825 TRACE("(%p %p %x): Invalid fragment component %s.\n", builder, data, flags,
3826 debugstr_wn(component, expected_len));
3827 return INET_E_INVALID_URL;
3834 static HRESULT validate_components(const UriBuilder *builder, parse_data *data, DWORD flags) {
3837 memset(data, 0, sizeof(parse_data));
3839 TRACE("(%p %p %x): Beginning to validate builder components.\n", builder, data, flags);
3841 hr = validate_scheme_name(builder, data, flags);
3845 /* Extra validation for file schemes. */
3846 if(data->scheme_type == URL_SCHEME_FILE) {
3847 if((builder->password || (builder->uri && builder->uri->userinfo_split > -1)) ||
3848 (builder->username || (builder->uri && builder->uri->userinfo_start > -1))) {
3849 TRACE("(%p %p %x): File schemes can't contain a username or password.\n",
3850 builder, data, flags);
3851 return INET_E_INVALID_URL;
3855 hr = validate_userinfo(builder, data, flags);
3859 hr = validate_host(builder, data, flags);
3863 setup_port(builder, data, flags);
3865 /* The URI is opaque if it doesn't have an authority component. */
3866 if(!data->is_relative)
3867 data->is_opaque = !data->username && !data->password && !data->host && !data->has_port;
3869 data->is_opaque = !data->host && !data->has_port;
3871 hr = validate_path(builder, data, flags);
3875 hr = validate_query(builder, data, flags);
3879 hr = validate_fragment(builder, data, flags);
3883 TRACE("(%p %p %x): Finished validating builder components.\n", builder, data, flags);
3888 /* Generates a raw uri string using the parse_data. */
3889 static DWORD generate_raw_uri(const parse_data *data, BSTR uri) {
3894 memcpy(uri, data->scheme, data->scheme_len*sizeof(WCHAR));
3895 uri[data->scheme_len] = ':';
3897 length += data->scheme_len+1;
3900 if(!data->is_opaque) {
3901 /* For the "//" which appears before the authority component. */
3904 uri[length+1] = '/';
3909 if(data->username) {
3911 memcpy(uri+length, data->username, data->username_len*sizeof(WCHAR));
3912 length += data->username_len;
3915 if(data->password) {
3918 memcpy(uri+length+1, data->password, data->password_len*sizeof(WCHAR));
3920 length += data->password_len+1;
3923 if(data->password || data->username) {
3930 /* IPv6 addresses get the brackets added around them if they don't already
3933 const BOOL add_brackets = data->host_type == Uri_HOST_IPV6 && *(data->host) != '[';
3941 memcpy(uri+length, data->host, data->host_len*sizeof(WCHAR));
3942 length += data->host_len;
3951 if(data->has_port) {
3952 /* The port isn't included in the raw uri if it's the default
3953 * port for the scheme type.
3956 BOOL is_default = FALSE;
3958 for(i = 0; i < sizeof(default_ports)/sizeof(default_ports[0]); ++i) {
3959 if(data->scheme_type == default_ports[i].scheme &&
3960 data->port_value == default_ports[i].port)
3970 length += ui2str(uri+length, data->port_value);
3972 length += ui2str(NULL, data->port_value);
3976 /* Check if a '/' should be added before the path for hierarchical URIs. */
3977 if(!data->is_opaque && data->path && *(data->path) != '/') {
3985 memcpy(uri+length, data->path, data->path_len*sizeof(WCHAR));
3986 length += data->path_len;
3991 memcpy(uri+length, data->query, data->query_len*sizeof(WCHAR));
3992 length += data->query_len;
3995 if(data->fragment) {
3997 memcpy(uri+length, data->fragment, data->fragment_len*sizeof(WCHAR));
3998 length += data->fragment_len;
4002 TRACE("(%p %p): Generated raw uri=%s len=%d\n", data, uri, debugstr_wn(uri, length), length);
4004 TRACE("(%p %p): Computed raw uri len=%d\n", data, uri, length);
4009 static HRESULT generate_uri(const UriBuilder *builder, const parse_data *data, Uri *uri, DWORD flags) {
4010 DWORD length = generate_raw_uri(data, NULL);
4011 uri->raw_uri = SysAllocStringLen(NULL, length);
4013 return E_OUTOFMEMORY;
4015 generate_raw_uri(data, uri->raw_uri);
4020 static HRESULT build_uri(const UriBuilder *builder, IUri **uri, DWORD create_flags,
4021 DWORD use_orig_flags, DWORD encoding_mask)
4030 if(encoding_mask && (!builder->uri || builder->modified_props)) {
4035 /* Decide what flags should be used when creating the Uri. */
4036 if((use_orig_flags & UriBuilder_USE_ORIGINAL_FLAGS) && builder->uri)
4037 create_flags = builder->uri->create_flags;
4039 if(has_invalid_flag_combination(create_flags)) {
4041 return E_INVALIDARG;
4044 /* Set the default flags if they don't cause a conflict. */
4045 apply_default_flags(&create_flags);
4048 /* Return the base IUri if no changes have been made and the create_flags match. */
4049 if(builder->uri && !builder->modified_props && builder->uri->create_flags == create_flags) {
4050 *uri = URI(builder->uri);
4055 hr = validate_components(builder, &data, create_flags);
4061 ret = heap_alloc_zero(sizeof(Uri));
4064 return E_OUTOFMEMORY;
4067 hr = generate_uri(builder, &data, ret, create_flags);
4069 SysFreeString(ret->raw_uri);
4070 heap_free(ret->canon_uri);
4079 #define URI_THIS(iface) DEFINE_THIS(Uri, IUri, iface)
4081 static HRESULT WINAPI Uri_QueryInterface(IUri *iface, REFIID riid, void **ppv)
4083 Uri *This = URI_THIS(iface);
4085 if(IsEqualGUID(&IID_IUnknown, riid)) {
4086 TRACE("(%p)->(IID_IUnknown %p)\n", This, ppv);
4088 }else if(IsEqualGUID(&IID_IUri, riid)) {
4089 TRACE("(%p)->(IID_IUri %p)\n", This, ppv);
4091 }else if(IsEqualGUID(&IID_IUriObj, riid)) {
4092 TRACE("(%p)->(IID_IUriObj %p)\n", This, ppv);
4096 TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppv);
4098 return E_NOINTERFACE;
4101 IUnknown_AddRef((IUnknown*)*ppv);
4105 static ULONG WINAPI Uri_AddRef(IUri *iface)
4107 Uri *This = URI_THIS(iface);
4108 LONG ref = InterlockedIncrement(&This->ref);
4110 TRACE("(%p) ref=%d\n", This, ref);
4115 static ULONG WINAPI Uri_Release(IUri *iface)
4117 Uri *This = URI_THIS(iface);
4118 LONG ref = InterlockedDecrement(&This->ref);
4120 TRACE("(%p) ref=%d\n", This, ref);
4123 SysFreeString(This->raw_uri);
4124 heap_free(This->canon_uri);
4131 static HRESULT WINAPI Uri_GetPropertyBSTR(IUri *iface, Uri_PROPERTY uriProp, BSTR *pbstrProperty, DWORD dwFlags)
4133 Uri *This = URI_THIS(iface);
4135 TRACE("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
4140 if(uriProp > Uri_PROPERTY_STRING_LAST) {
4141 /* Windows allocates an empty BSTR for invalid Uri_PROPERTY's. */
4142 *pbstrProperty = SysAllocStringLen(NULL, 0);
4143 if(!(*pbstrProperty))
4144 return E_OUTOFMEMORY;
4146 /* It only returns S_FALSE for the ZONE property... */
4147 if(uriProp == Uri_PROPERTY_ZONE)
4153 /* Don't have support for flags yet. */
4155 FIXME("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
4160 case Uri_PROPERTY_ABSOLUTE_URI:
4161 if(!This->display_absolute) {
4162 *pbstrProperty = SysAllocStringLen(NULL, 0);
4165 if(This->scheme_type != URL_SCHEME_UNKNOWN && This->userinfo_start > -1) {
4166 if(This->userinfo_len == 0) {
4167 /* Don't include the '@' after the userinfo component. */
4168 *pbstrProperty = SysAllocStringLen(NULL, This->canon_len-1);
4170 if(*pbstrProperty) {
4171 /* Copy everything before it. */
4172 memcpy(*pbstrProperty, This->canon_uri, This->userinfo_start*sizeof(WCHAR));
4174 /* And everything after it. */
4175 memcpy(*pbstrProperty+This->userinfo_start, This->canon_uri+This->userinfo_start+1,
4176 (This->canon_len-This->userinfo_start-1)*sizeof(WCHAR));
4178 } else if(This->userinfo_split == 0 && This->userinfo_len == 1) {
4179 /* Don't include the ":@" */
4180 *pbstrProperty = SysAllocStringLen(NULL, This->canon_len-2);
4182 if(*pbstrProperty) {
4183 memcpy(*pbstrProperty, This->canon_uri, This->userinfo_start*sizeof(WCHAR));
4184 memcpy(*pbstrProperty+This->userinfo_start, This->canon_uri+This->userinfo_start+2,
4185 (This->canon_len-This->userinfo_start-2)*sizeof(WCHAR));
4188 *pbstrProperty = SysAllocString(This->canon_uri);
4192 *pbstrProperty = SysAllocString(This->canon_uri);
4197 if(!(*pbstrProperty))
4198 hres = E_OUTOFMEMORY;
4201 case Uri_PROPERTY_AUTHORITY:
4202 if(This->authority_start > -1) {
4203 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->authority_start, This->authority_len);
4206 *pbstrProperty = SysAllocStringLen(NULL, 0);
4210 if(!(*pbstrProperty))
4211 hres = E_OUTOFMEMORY;
4214 case Uri_PROPERTY_DISPLAY_URI:
4215 /* The Display URI contains everything except for the userinfo for known
4218 if(This->scheme_type != URL_SCHEME_UNKNOWN && This->userinfo_start > -1) {
4219 *pbstrProperty = SysAllocStringLen(NULL, This->canon_len-This->userinfo_len);
4221 if(*pbstrProperty) {
4222 /* Copy everything before the userinfo over. */
4223 memcpy(*pbstrProperty, This->canon_uri, This->userinfo_start*sizeof(WCHAR));
4224 /* Copy everything after the userinfo over. */
4225 memcpy(*pbstrProperty+This->userinfo_start,
4226 This->canon_uri+This->userinfo_start+This->userinfo_len+1,
4227 (This->canon_len-(This->userinfo_start+This->userinfo_len+1))*sizeof(WCHAR));
4230 *pbstrProperty = SysAllocString(This->canon_uri);
4232 if(!(*pbstrProperty))
4233 hres = E_OUTOFMEMORY;
4238 case Uri_PROPERTY_DOMAIN:
4239 if(This->domain_offset > -1) {
4240 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->host_start+This->domain_offset,
4241 This->host_len-This->domain_offset);
4244 *pbstrProperty = SysAllocStringLen(NULL, 0);
4248 if(!(*pbstrProperty))
4249 hres = E_OUTOFMEMORY;
4252 case Uri_PROPERTY_EXTENSION:
4253 if(This->extension_offset > -1) {
4254 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->path_start+This->extension_offset,
4255 This->path_len-This->extension_offset);
4258 *pbstrProperty = SysAllocStringLen(NULL, 0);
4262 if(!(*pbstrProperty))
4263 hres = E_OUTOFMEMORY;
4266 case Uri_PROPERTY_FRAGMENT:
4267 if(This->fragment_start > -1) {
4268 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->fragment_start, This->fragment_len);
4271 *pbstrProperty = SysAllocStringLen(NULL, 0);
4275 if(!(*pbstrProperty))
4276 hres = E_OUTOFMEMORY;
4279 case Uri_PROPERTY_HOST:
4280 if(This->host_start > -1) {
4281 /* The '[' and ']' aren't included for IPv6 addresses. */
4282 if(This->host_type == Uri_HOST_IPV6)
4283 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->host_start+1, This->host_len-2);
4285 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->host_start, This->host_len);
4289 *pbstrProperty = SysAllocStringLen(NULL, 0);
4293 if(!(*pbstrProperty))
4294 hres = E_OUTOFMEMORY;
4297 case Uri_PROPERTY_PASSWORD:
4298 if(This->userinfo_split > -1) {
4299 *pbstrProperty = SysAllocStringLen(
4300 This->canon_uri+This->userinfo_start+This->userinfo_split+1,
4301 This->userinfo_len-This->userinfo_split-1);
4304 *pbstrProperty = SysAllocStringLen(NULL, 0);
4308 if(!(*pbstrProperty))
4309 return E_OUTOFMEMORY;
4312 case Uri_PROPERTY_PATH:
4313 if(This->path_start > -1) {
4314 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->path_start, This->path_len);
4317 *pbstrProperty = SysAllocStringLen(NULL, 0);
4321 if(!(*pbstrProperty))
4322 hres = E_OUTOFMEMORY;
4325 case Uri_PROPERTY_PATH_AND_QUERY:
4326 if(This->path_start > -1) {
4327 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->path_start, This->path_len+This->query_len);
4329 } else if(This->query_start > -1) {
4330 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->query_start, This->query_len);
4333 *pbstrProperty = SysAllocStringLen(NULL, 0);
4337 if(!(*pbstrProperty))
4338 hres = E_OUTOFMEMORY;
4341 case Uri_PROPERTY_QUERY:
4342 if(This->query_start > -1) {
4343 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->query_start, This->query_len);
4346 *pbstrProperty = SysAllocStringLen(NULL, 0);
4350 if(!(*pbstrProperty))
4351 hres = E_OUTOFMEMORY;
4354 case Uri_PROPERTY_RAW_URI:
4355 *pbstrProperty = SysAllocString(This->raw_uri);
4356 if(!(*pbstrProperty))
4357 hres = E_OUTOFMEMORY;
4361 case Uri_PROPERTY_SCHEME_NAME:
4362 if(This->scheme_start > -1) {
4363 *pbstrProperty = SysAllocStringLen(This->canon_uri + This->scheme_start, This->scheme_len);
4366 *pbstrProperty = SysAllocStringLen(NULL, 0);
4370 if(!(*pbstrProperty))
4371 hres = E_OUTOFMEMORY;
4374 case Uri_PROPERTY_USER_INFO:
4375 if(This->userinfo_start > -1) {
4376 *pbstrProperty = SysAllocStringLen(This->canon_uri+This->userinfo_start, This->userinfo_len);
4379 *pbstrProperty = SysAllocStringLen(NULL, 0);
4383 if(!(*pbstrProperty))
4384 hres = E_OUTOFMEMORY;
4387 case Uri_PROPERTY_USER_NAME:
4388 if(This->userinfo_start > -1 && This->userinfo_split != 0) {
4389 /* If userinfo_split is set, that means a password exists
4390 * so the username is only from userinfo_start to userinfo_split.
4392 if(This->userinfo_split > -1) {
4393 *pbstrProperty = SysAllocStringLen(This->canon_uri + This->userinfo_start, This->userinfo_split);
4396 *pbstrProperty = SysAllocStringLen(This->canon_uri + This->userinfo_start, This->userinfo_len);
4400 *pbstrProperty = SysAllocStringLen(NULL, 0);
4404 if(!(*pbstrProperty))
4405 return E_OUTOFMEMORY;
4409 FIXME("(%p)->(%d %p %x)\n", This, uriProp, pbstrProperty, dwFlags);
4416 static HRESULT WINAPI Uri_GetPropertyLength(IUri *iface, Uri_PROPERTY uriProp, DWORD *pcchProperty, DWORD dwFlags)
4418 Uri *This = URI_THIS(iface);
4420 TRACE("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
4423 return E_INVALIDARG;
4425 /* Can only return a length for a property if it's a string. */
4426 if(uriProp > Uri_PROPERTY_STRING_LAST)
4427 return E_INVALIDARG;
4429 /* Don't have support for flags yet. */
4431 FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
4436 case Uri_PROPERTY_ABSOLUTE_URI:
4437 if(!This->display_absolute) {
4441 if(This->scheme_type != URL_SCHEME_UNKNOWN) {
4442 if(This->userinfo_start > -1 && This->userinfo_len == 0)
4443 /* Don't include the '@' in the length. */
4444 *pcchProperty = This->canon_len-1;
4445 else if(This->userinfo_start > -1 && This->userinfo_len == 1 &&
4446 This->userinfo_split == 0)
4447 /* Don't include the ":@" in the length. */
4448 *pcchProperty = This->canon_len-2;
4450 *pcchProperty = This->canon_len;
4452 *pcchProperty = This->canon_len;
4458 case Uri_PROPERTY_AUTHORITY:
4459 *pcchProperty = This->authority_len;
4460 hres = (This->authority_start > -1) ? S_OK : S_FALSE;
4462 case Uri_PROPERTY_DISPLAY_URI:
4463 if(This->scheme_type != URL_SCHEME_UNKNOWN && This->userinfo_start > -1)
4464 *pcchProperty = This->canon_len-This->userinfo_len-1;
4466 *pcchProperty = This->canon_len;
4470 case Uri_PROPERTY_DOMAIN:
4471 if(This->domain_offset > -1)
4472 *pcchProperty = This->host_len - This->domain_offset;
4476 hres = (This->domain_offset > -1) ? S_OK : S_FALSE;
4478 case Uri_PROPERTY_EXTENSION:
4479 if(This->extension_offset > -1) {
4480 *pcchProperty = This->path_len - This->extension_offset;
4488 case Uri_PROPERTY_FRAGMENT:
4489 *pcchProperty = This->fragment_len;
4490 hres = (This->fragment_start > -1) ? S_OK : S_FALSE;
4492 case Uri_PROPERTY_HOST:
4493 *pcchProperty = This->host_len;
4495 /* '[' and ']' aren't included in the length. */
4496 if(This->host_type == Uri_HOST_IPV6)
4499 hres = (This->host_start > -1) ? S_OK : S_FALSE;
4501 case Uri_PROPERTY_PASSWORD:
4502 *pcchProperty = (This->userinfo_split > -1) ? This->userinfo_len-This->userinfo_split-1 : 0;
4503 hres = (This->userinfo_split > -1) ? S_OK : S_FALSE;
4505 case Uri_PROPERTY_PATH:
4506 *pcchProperty = This->path_len;
4507 hres = (This->path_start > -1) ? S_OK : S_FALSE;
4509 case Uri_PROPERTY_PATH_AND_QUERY:
4510 *pcchProperty = This->path_len+This->query_len;
4511 hres = (This->path_start > -1 || This->query_start > -1) ? S_OK : S_FALSE;
4513 case Uri_PROPERTY_QUERY:
4514 *pcchProperty = This->query_len;
4515 hres = (This->query_start > -1) ? S_OK : S_FALSE;
4517 case Uri_PROPERTY_RAW_URI:
4518 *pcchProperty = SysStringLen(This->raw_uri);
4521 case Uri_PROPERTY_SCHEME_NAME:
4522 *pcchProperty = This->scheme_len;
4523 hres = (This->scheme_start > -1) ? S_OK : S_FALSE;
4525 case Uri_PROPERTY_USER_INFO:
4526 *pcchProperty = This->userinfo_len;
4527 hres = (This->userinfo_start > -1) ? S_OK : S_FALSE;
4529 case Uri_PROPERTY_USER_NAME:
4530 *pcchProperty = (This->userinfo_split > -1) ? This->userinfo_split : This->userinfo_len;
4531 if(This->userinfo_split == 0)
4534 hres = (This->userinfo_start > -1) ? S_OK : S_FALSE;
4537 FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
4544 static HRESULT WINAPI Uri_GetPropertyDWORD(IUri *iface, Uri_PROPERTY uriProp, DWORD *pcchProperty, DWORD dwFlags)
4546 Uri *This = URI_THIS(iface);
4549 TRACE("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
4552 return E_INVALIDARG;
4554 /* Microsoft's implementation for the ZONE property of a URI seems to be lacking...
4555 * From what I can tell, instead of checking which URLZONE the URI belongs to it
4556 * simply assigns URLZONE_INVALID and returns E_NOTIMPL. This also applies to the GetZone
4559 if(uriProp == Uri_PROPERTY_ZONE) {
4560 *pcchProperty = URLZONE_INVALID;
4564 if(uriProp < Uri_PROPERTY_DWORD_START) {
4566 return E_INVALIDARG;
4570 case Uri_PROPERTY_HOST_TYPE:
4571 *pcchProperty = This->host_type;
4574 case Uri_PROPERTY_PORT:
4575 if(!This->has_port) {
4579 *pcchProperty = This->port;
4584 case Uri_PROPERTY_SCHEME:
4585 *pcchProperty = This->scheme_type;
4589 FIXME("(%p)->(%d %p %x)\n", This, uriProp, pcchProperty, dwFlags);
4596 static HRESULT WINAPI Uri_HasProperty(IUri *iface, Uri_PROPERTY uriProp, BOOL *pfHasProperty)
4598 Uri *This = URI_THIS(iface);
4599 TRACE("(%p)->(%d %p)\n", This, uriProp, pfHasProperty);
4602 return E_INVALIDARG;
4605 case Uri_PROPERTY_ABSOLUTE_URI:
4606 *pfHasProperty = This->display_absolute;
4608 case Uri_PROPERTY_AUTHORITY:
4609 *pfHasProperty = This->authority_start > -1;
4611 case Uri_PROPERTY_DISPLAY_URI:
4612 *pfHasProperty = TRUE;
4614 case Uri_PROPERTY_DOMAIN:
4615 *pfHasProperty = This->domain_offset > -1;
4617 case Uri_PROPERTY_EXTENSION:
4618 *pfHasProperty = This->extension_offset > -1;
4620 case Uri_PROPERTY_FRAGMENT:
4621 *pfHasProperty = This->fragment_start > -1;
4623 case Uri_PROPERTY_HOST:
4624 *pfHasProperty = This->host_start > -1;
4626 case Uri_PROPERTY_PASSWORD:
4627 *pfHasProperty = This->userinfo_split > -1;
4629 case Uri_PROPERTY_PATH:
4630 *pfHasProperty = This->path_start > -1;
4632 case Uri_PROPERTY_PATH_AND_QUERY:
4633 *pfHasProperty = (This->path_start > -1 || This->query_start > -1);
4635 case Uri_PROPERTY_QUERY:
4636 *pfHasProperty = This->query_start > -1;
4638 case Uri_PROPERTY_RAW_URI:
4639 *pfHasProperty = TRUE;
4641 case Uri_PROPERTY_SCHEME_NAME:
4642 *pfHasProperty = This->scheme_start > -1;
4644 case Uri_PROPERTY_USER_INFO:
4645 *pfHasProperty = This->userinfo_start > -1;
4647 case Uri_PROPERTY_USER_NAME:
4648 if(This->userinfo_split == 0)
4649 *pfHasProperty = FALSE;
4651 *pfHasProperty = This->userinfo_start > -1;
4653 case Uri_PROPERTY_HOST_TYPE:
4654 *pfHasProperty = TRUE;
4656 case Uri_PROPERTY_PORT:
4657 *pfHasProperty = This->has_port;
4659 case Uri_PROPERTY_SCHEME:
4660 *pfHasProperty = TRUE;
4662 case Uri_PROPERTY_ZONE:
4663 *pfHasProperty = FALSE;
4666 FIXME("(%p)->(%d %p): Unsupported property type.\n", This, uriProp, pfHasProperty);
4673 static HRESULT WINAPI Uri_GetAbsoluteUri(IUri *iface, BSTR *pstrAbsoluteUri)
4675 TRACE("(%p)->(%p)\n", iface, pstrAbsoluteUri);
4676 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_ABSOLUTE_URI, pstrAbsoluteUri, 0);
4679 static HRESULT WINAPI Uri_GetAuthority(IUri *iface, BSTR *pstrAuthority)
4681 TRACE("(%p)->(%p)\n", iface, pstrAuthority);
4682 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_AUTHORITY, pstrAuthority, 0);
4685 static HRESULT WINAPI Uri_GetDisplayUri(IUri *iface, BSTR *pstrDisplayUri)
4687 TRACE("(%p)->(%p)\n", iface, pstrDisplayUri);
4688 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_DISPLAY_URI, pstrDisplayUri, 0);
4691 static HRESULT WINAPI Uri_GetDomain(IUri *iface, BSTR *pstrDomain)
4693 TRACE("(%p)->(%p)\n", iface, pstrDomain);
4694 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_DOMAIN, pstrDomain, 0);
4697 static HRESULT WINAPI Uri_GetExtension(IUri *iface, BSTR *pstrExtension)
4699 TRACE("(%p)->(%p)\n", iface, pstrExtension);
4700 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_EXTENSION, pstrExtension, 0);
4703 static HRESULT WINAPI Uri_GetFragment(IUri *iface, BSTR *pstrFragment)
4705 TRACE("(%p)->(%p)\n", iface, pstrFragment);
4706 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_FRAGMENT, pstrFragment, 0);
4709 static HRESULT WINAPI Uri_GetHost(IUri *iface, BSTR *pstrHost)
4711 TRACE("(%p)->(%p)\n", iface, pstrHost);
4712 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_HOST, pstrHost, 0);
4715 static HRESULT WINAPI Uri_GetPassword(IUri *iface, BSTR *pstrPassword)
4717 TRACE("(%p)->(%p)\n", iface, pstrPassword);
4718 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_PASSWORD, pstrPassword, 0);
4721 static HRESULT WINAPI Uri_GetPath(IUri *iface, BSTR *pstrPath)
4723 TRACE("(%p)->(%p)\n", iface, pstrPath);
4724 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_PATH, pstrPath, 0);
4727 static HRESULT WINAPI Uri_GetPathAndQuery(IUri *iface, BSTR *pstrPathAndQuery)
4729 TRACE("(%p)->(%p)\n", iface, pstrPathAndQuery);
4730 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_PATH_AND_QUERY, pstrPathAndQuery, 0);
4733 static HRESULT WINAPI Uri_GetQuery(IUri *iface, BSTR *pstrQuery)
4735 TRACE("(%p)->(%p)\n", iface, pstrQuery);
4736 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_QUERY, pstrQuery, 0);
4739 static HRESULT WINAPI Uri_GetRawUri(IUri *iface, BSTR *pstrRawUri)
4741 TRACE("(%p)->(%p)\n", iface, pstrRawUri);
4742 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_RAW_URI, pstrRawUri, 0);
4745 static HRESULT WINAPI Uri_GetSchemeName(IUri *iface, BSTR *pstrSchemeName)
4747 TRACE("(%p)->(%p)\n", iface, pstrSchemeName);
4748 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_SCHEME_NAME, pstrSchemeName, 0);
4751 static HRESULT WINAPI Uri_GetUserInfo(IUri *iface, BSTR *pstrUserInfo)
4753 TRACE("(%p)->(%p)\n", iface, pstrUserInfo);
4754 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_USER_INFO, pstrUserInfo, 0);
4757 static HRESULT WINAPI Uri_GetUserName(IUri *iface, BSTR *pstrUserName)
4759 TRACE("(%p)->(%p)\n", iface, pstrUserName);
4760 return Uri_GetPropertyBSTR(iface, Uri_PROPERTY_USER_NAME, pstrUserName, 0);
4763 static HRESULT WINAPI Uri_GetHostType(IUri *iface, DWORD *pdwHostType)
4765 TRACE("(%p)->(%p)\n", iface, pdwHostType);
4766 return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_HOST_TYPE, pdwHostType, 0);
4769 static HRESULT WINAPI Uri_GetPort(IUri *iface, DWORD *pdwPort)
4771 TRACE("(%p)->(%p)\n", iface, pdwPort);
4772 return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_PORT, pdwPort, 0);
4775 static HRESULT WINAPI Uri_GetScheme(IUri *iface, DWORD *pdwScheme)
4777 Uri *This = URI_THIS(iface);
4778 TRACE("(%p)->(%p)\n", This, pdwScheme);
4779 return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_SCHEME, pdwScheme, 0);
4782 static HRESULT WINAPI Uri_GetZone(IUri *iface, DWORD *pdwZone)
4784 TRACE("(%p)->(%p)\n", iface, pdwZone);
4785 return Uri_GetPropertyDWORD(iface, Uri_PROPERTY_ZONE,pdwZone, 0);
4788 static HRESULT WINAPI Uri_GetProperties(IUri *iface, DWORD *pdwProperties)
4790 Uri *This = URI_THIS(iface);
4791 TRACE("(%p)->(%p)\n", This, pdwProperties);
4794 return E_INVALIDARG;
4796 /* All URIs have these. */
4797 *pdwProperties = Uri_HAS_DISPLAY_URI|Uri_HAS_RAW_URI|Uri_HAS_SCHEME|Uri_HAS_HOST_TYPE;
4799 if(This->display_absolute)
4800 *pdwProperties |= Uri_HAS_ABSOLUTE_URI;
4802 if(This->scheme_start > -1)
4803 *pdwProperties |= Uri_HAS_SCHEME_NAME;
4805 if(This->authority_start > -1) {
4806 *pdwProperties |= Uri_HAS_AUTHORITY;
4807 if(This->userinfo_start > -1) {
4808 *pdwProperties |= Uri_HAS_USER_INFO;
4809 if(This->userinfo_split != 0)
4810 *pdwProperties |= Uri_HAS_USER_NAME;
4812 if(This->userinfo_split > -1)
4813 *pdwProperties |= Uri_HAS_PASSWORD;
4814 if(This->host_start > -1)
4815 *pdwProperties |= Uri_HAS_HOST;
4816 if(This->domain_offset > -1)
4817 *pdwProperties |= Uri_HAS_DOMAIN;
4821 *pdwProperties |= Uri_HAS_PORT;
4822 if(This->path_start > -1)
4823 *pdwProperties |= Uri_HAS_PATH|Uri_HAS_PATH_AND_QUERY;
4824 if(This->query_start > -1)
4825 *pdwProperties |= Uri_HAS_QUERY|Uri_HAS_PATH_AND_QUERY;
4827 if(This->extension_offset > -1)
4828 *pdwProperties |= Uri_HAS_EXTENSION;
4830 if(This->fragment_start > -1)
4831 *pdwProperties |= Uri_HAS_FRAGMENT;
4836 static HRESULT WINAPI Uri_IsEqual(IUri *iface, IUri *pUri, BOOL *pfEqual)
4838 Uri *This = URI_THIS(iface);
4841 TRACE("(%p)->(%p %p)\n", This, pUri, pfEqual);
4849 /* For some reason Windows returns S_OK here... */
4853 /* Try to convert it to a Uri (allows for a more simple comparison). */
4854 if((other = get_uri_obj(pUri)))
4855 *pfEqual = are_equal_simple(This, other);
4857 /* Do it the hard way. */
4858 FIXME("(%p)->(%p %p) No support for unknown IUri's yet.\n", iface, pUri, pfEqual);
4867 static const IUriVtbl UriVtbl = {
4871 Uri_GetPropertyBSTR,
4872 Uri_GetPropertyLength,
4873 Uri_GetPropertyDWORD,
4884 Uri_GetPathAndQuery,
4898 /***********************************************************************
4899 * CreateUri (urlmon.@)
4901 * Creates a new IUri object using the URI represented by pwzURI. This function
4902 * parses and validates the components of pwzURI and then canonicalizes the
4903 * parsed components.
4906 * pwzURI [I] The URI to parse, validate, and canonicalize.
4907 * dwFlags [I] Flags which can affect how the parsing/canonicalization is performed.
4908 * dwReserved [I] Reserved (not used).
4909 * ppURI [O] The resulting IUri after parsing/canonicalization occurs.
4912 * Success: Returns S_OK. ppURI contains the pointer to the newly allocated IUri.
4913 * Failure: E_INVALIDARG if there's invalid flag combinations in dwFlags, or an
4914 * invalid parameters, or pwzURI doesn't represnt a valid URI.
4915 * E_OUTOFMEMORY if any memory allocation fails.
4919 * Uri_CREATE_CANONICALIZE, Uri_CREATE_DECODE_EXTRA_INFO, Uri_CREATE_CRACK_UNKNOWN_SCHEMES,
4920 * Uri_CREATE_PRE_PROCESS_HTML_URI, Uri_CREATE_NO_IE_SETTINGS.
4922 HRESULT WINAPI CreateUri(LPCWSTR pwzURI, DWORD dwFlags, DWORD_PTR dwReserved, IUri **ppURI)
4924 const DWORD supported_flags = Uri_CREATE_ALLOW_RELATIVE|Uri_CREATE_ALLOW_IMPLICIT_WILDCARD_SCHEME|
4925 Uri_CREATE_ALLOW_IMPLICIT_FILE_SCHEME|Uri_CREATE_NO_CANONICALIZE|Uri_CREATE_CANONICALIZE|
4926 Uri_CREATE_DECODE_EXTRA_INFO|Uri_CREATE_NO_DECODE_EXTRA_INFO|Uri_CREATE_CRACK_UNKNOWN_SCHEMES|
4927 Uri_CREATE_NO_CRACK_UNKNOWN_SCHEMES|Uri_CREATE_PRE_PROCESS_HTML_URI|Uri_CREATE_NO_PRE_PROCESS_HTML_URI|
4928 Uri_CREATE_NO_IE_SETTINGS|Uri_CREATE_NO_ENCODE_FORBIDDEN_CHARACTERS|Uri_CREATE_FILE_USE_DOS_PATH;
4933 TRACE("(%s %x %x %p)\n", debugstr_w(pwzURI), dwFlags, (DWORD)dwReserved, ppURI);
4936 return E_INVALIDARG;
4938 if(!pwzURI || !*pwzURI) {
4940 return E_INVALIDARG;
4943 /* Check for invalid flags. */
4944 if(has_invalid_flag_combination(dwFlags)) {
4946 return E_INVALIDARG;
4949 /* Currently unsupported. */
4950 if(dwFlags & ~supported_flags)
4951 FIXME("Ignoring unsupported flag(s) %x\n", dwFlags & ~supported_flags);
4953 ret = heap_alloc(sizeof(Uri));
4955 return E_OUTOFMEMORY;
4957 ret->lpIUriVtbl = &UriVtbl;
4960 /* Explicitly set the default flags if it doesn't cause a flag conflict. */
4961 apply_default_flags(&dwFlags);
4963 /* Pre process the URI, unless told otherwise. */
4964 if(!(dwFlags & Uri_CREATE_NO_PRE_PROCESS_HTML_URI))
4965 ret->raw_uri = pre_process_uri(pwzURI);
4967 ret->raw_uri = SysAllocString(pwzURI);
4971 return E_OUTOFMEMORY;
4974 memset(&data, 0, sizeof(parse_data));
4975 data.uri = ret->raw_uri;
4977 /* Validate and parse the URI into it's components. */
4978 if(!parse_uri(&data, dwFlags)) {
4979 /* Encountered an unsupported or invalid URI */
4980 SysFreeString(ret->raw_uri);
4983 return E_INVALIDARG;
4986 /* Canonicalize the URI. */
4987 hr = canonicalize_uri(&data, ret, dwFlags);
4989 SysFreeString(ret->raw_uri);
4995 ret->create_flags = dwFlags;
5001 /***********************************************************************
5002 * CreateUriWithFragment (urlmon.@)
5004 * Creates a new IUri object. This is almost the same as CreateUri, expect that
5005 * it allows you to explicitly specify a fragment (pwzFragment) for pwzURI.
5008 * pwzURI [I] The URI to parse and perform canonicalization on.
5009 * pwzFragment [I] The explict fragment string which should be added to pwzURI.
5010 * dwFlags [I] The flags which will be passed to CreateUri.
5011 * dwReserved [I] Reserved (not used).
5012 * ppURI [O] The resulting IUri after parsing/canonicalization.
5015 * Success: S_OK. ppURI contains the pointer to the newly allocated IUri.
5016 * Failure: E_INVALIDARG if pwzURI already contains a fragment and pwzFragment
5017 * isn't NULL. Will also return E_INVALIDARG for the same reasons as
5018 * CreateUri will. E_OUTOFMEMORY if any allocations fail.
5020 HRESULT WINAPI CreateUriWithFragment(LPCWSTR pwzURI, LPCWSTR pwzFragment, DWORD dwFlags,
5021 DWORD_PTR dwReserved, IUri **ppURI)
5024 TRACE("(%s %s %x %x %p)\n", debugstr_w(pwzURI), debugstr_w(pwzFragment), dwFlags, (DWORD)dwReserved, ppURI);
5027 return E_INVALIDARG;
5031 return E_INVALIDARG;
5034 /* Check if a fragment should be appended to the URI string. */
5037 DWORD uri_len, frag_len;
5040 /* Check if the original URI already has a fragment component. */
5041 if(StrChrW(pwzURI, '#')) {
5043 return E_INVALIDARG;
5046 uri_len = lstrlenW(pwzURI);
5047 frag_len = lstrlenW(pwzFragment);
5049 /* If the fragment doesn't start with a '#', one will be added. */
5050 add_pound = *pwzFragment != '#';
5053 uriW = heap_alloc((uri_len+frag_len+2)*sizeof(WCHAR));
5055 uriW = heap_alloc((uri_len+frag_len+1)*sizeof(WCHAR));
5058 return E_OUTOFMEMORY;
5060 memcpy(uriW, pwzURI, uri_len*sizeof(WCHAR));
5062 uriW[uri_len++] = '#';
5063 memcpy(uriW+uri_len, pwzFragment, (frag_len+1)*sizeof(WCHAR));
5065 hres = CreateUri(uriW, dwFlags, 0, ppURI);
5069 /* A fragment string wasn't specified, so just forward the call. */
5070 hres = CreateUri(pwzURI, dwFlags, 0, ppURI);
5075 #define URIBUILDER_THIS(iface) DEFINE_THIS(UriBuilder, IUriBuilder, iface)
5077 static HRESULT WINAPI UriBuilder_QueryInterface(IUriBuilder *iface, REFIID riid, void **ppv)
5079 UriBuilder *This = URIBUILDER_THIS(iface);
5081 if(IsEqualGUID(&IID_IUnknown, riid)) {
5082 TRACE("(%p)->(IID_IUnknown %p)\n", This, ppv);
5083 *ppv = URIBUILDER(This);
5084 }else if(IsEqualGUID(&IID_IUriBuilder, riid)) {
5085 TRACE("(%p)->(IID_IUri %p)\n", This, ppv);
5086 *ppv = URIBUILDER(This);
5088 TRACE("(%p)->(%s %p)\n", This, debugstr_guid(riid), ppv);
5090 return E_NOINTERFACE;
5093 IUnknown_AddRef((IUnknown*)*ppv);
5097 static ULONG WINAPI UriBuilder_AddRef(IUriBuilder *iface)
5099 UriBuilder *This = URIBUILDER_THIS(iface);
5100 LONG ref = InterlockedIncrement(&This->ref);
5102 TRACE("(%p) ref=%d\n", This, ref);
5107 static ULONG WINAPI UriBuilder_Release(IUriBuilder *iface)
5109 UriBuilder *This = URIBUILDER_THIS(iface);
5110 LONG ref = InterlockedDecrement(&This->ref);
5112 TRACE("(%p) ref=%d\n", This, ref);
5115 if(This->uri) IUri_Release(URI(This->uri));
5116 heap_free(This->fragment);
5117 heap_free(This->host);
5118 heap_free(This->password);
5119 heap_free(This->path);
5120 heap_free(This->query);
5121 heap_free(This->scheme);
5122 heap_free(This->username);
5129 static HRESULT WINAPI UriBuilder_CreateUriSimple(IUriBuilder *iface,
5130 DWORD dwAllowEncodingPropertyMask,
5131 DWORD_PTR dwReserved,
5134 UriBuilder *This = URIBUILDER_THIS(iface);
5136 TRACE("(%p)->(%d %d %p)\n", This, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
5138 hr = build_uri(This, ppIUri, 0, UriBuilder_USE_ORIGINAL_FLAGS, dwAllowEncodingPropertyMask);
5140 FIXME("(%p)->(%d %d %p)\n", This, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
5144 static HRESULT WINAPI UriBuilder_CreateUri(IUriBuilder *iface,
5145 DWORD dwCreateFlags,
5146 DWORD dwAllowEncodingPropertyMask,
5147 DWORD_PTR dwReserved,
5150 UriBuilder *This = URIBUILDER_THIS(iface);
5152 TRACE("(%p)->(0x%08x %d %d %p)\n", This, dwCreateFlags, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
5154 if(dwCreateFlags == -1)
5155 hr = build_uri(This, ppIUri, 0, UriBuilder_USE_ORIGINAL_FLAGS, dwAllowEncodingPropertyMask);
5157 hr = build_uri(This, ppIUri, dwCreateFlags, 0, dwAllowEncodingPropertyMask);
5160 FIXME("(%p)->(0x%08x %d %d %p)\n", This, dwCreateFlags, dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
5164 static HRESULT WINAPI UriBuilder_CreateUriWithFlags(IUriBuilder *iface,
5165 DWORD dwCreateFlags,
5166 DWORD dwUriBuilderFlags,
5167 DWORD dwAllowEncodingPropertyMask,
5168 DWORD_PTR dwReserved,
5171 UriBuilder *This = URIBUILDER_THIS(iface);
5173 TRACE("(%p)->(0x%08x 0x%08x %d %d %p)\n", This, dwCreateFlags, dwUriBuilderFlags,
5174 dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
5176 hr = build_uri(This, ppIUri, dwCreateFlags, dwUriBuilderFlags, dwAllowEncodingPropertyMask);
5178 FIXME("(%p)->(0x%08x 0x%08x %d %d %p)\n", This, dwCreateFlags, dwUriBuilderFlags,
5179 dwAllowEncodingPropertyMask, (DWORD)dwReserved, ppIUri);
5183 static HRESULT WINAPI UriBuilder_GetIUri(IUriBuilder *iface, IUri **ppIUri)
5185 UriBuilder *This = URIBUILDER_THIS(iface);
5186 TRACE("(%p)->(%p)\n", This, ppIUri);
5192 IUri *uri = URI(This->uri);
5201 static HRESULT WINAPI UriBuilder_SetIUri(IUriBuilder *iface, IUri *pIUri)
5203 UriBuilder *This = URIBUILDER_THIS(iface);
5204 TRACE("(%p)->(%p)\n", This, pIUri);
5209 if((uri = get_uri_obj(pIUri))) {
5210 /* Only reset the builder if it's Uri isn't the same as
5211 * the Uri passed to the function.
5213 if(This->uri != uri) {
5214 reset_builder(This);
5218 This->port = uri->port;
5223 FIXME("(%p)->(%p) Unknown IUri types not supported yet.\n", This, pIUri);
5226 } else if(This->uri)
5227 /* Only reset the builder if it's Uri isn't NULL. */
5228 reset_builder(This);
5233 static HRESULT WINAPI UriBuilder_GetFragment(IUriBuilder *iface, DWORD *pcchFragment, LPCWSTR *ppwzFragment)
5235 UriBuilder *This = URIBUILDER_THIS(iface);
5236 TRACE("(%p)->(%p %p)\n", This, pcchFragment, ppwzFragment);
5238 if(!This->uri || This->uri->fragment_start == -1 || This->modified_props & Uri_HAS_FRAGMENT)
5239 return get_builder_component(&This->fragment, &This->fragment_len, NULL, 0, ppwzFragment, pcchFragment);
5241 return get_builder_component(&This->fragment, &This->fragment_len, This->uri->canon_uri+This->uri->fragment_start,
5242 This->uri->fragment_len, ppwzFragment, pcchFragment);
5245 static HRESULT WINAPI UriBuilder_GetHost(IUriBuilder *iface, DWORD *pcchHost, LPCWSTR *ppwzHost)
5247 UriBuilder *This = URIBUILDER_THIS(iface);
5248 TRACE("(%p)->(%p %p)\n", This, pcchHost, ppwzHost);
5250 if(!This->uri || This->uri->host_start == -1 || This->modified_props & Uri_HAS_HOST)
5251 return get_builder_component(&This->host, &This->host_len, NULL, 0, ppwzHost, pcchHost);
5253 if(This->uri->host_type == Uri_HOST_IPV6)
5254 /* Don't include the '[' and ']' around the address. */
5255 return get_builder_component(&This->host, &This->host_len, This->uri->canon_uri+This->uri->host_start+1,
5256 This->uri->host_len-2, ppwzHost, pcchHost);
5258 return get_builder_component(&This->host, &This->host_len, This->uri->canon_uri+This->uri->host_start,
5259 This->uri->host_len, ppwzHost, pcchHost);
5263 static HRESULT WINAPI UriBuilder_GetPassword(IUriBuilder *iface, DWORD *pcchPassword, LPCWSTR *ppwzPassword)
5265 UriBuilder *This = URIBUILDER_THIS(iface);
5266 TRACE("(%p)->(%p %p)\n", This, pcchPassword, ppwzPassword);
5268 if(!This->uri || This->uri->userinfo_split == -1 || This->modified_props & Uri_HAS_PASSWORD)
5269 return get_builder_component(&This->password, &This->password_len, NULL, 0, ppwzPassword, pcchPassword);
5271 const WCHAR *start = This->uri->canon_uri+This->uri->userinfo_start+This->uri->userinfo_split+1;
5272 DWORD len = This->uri->userinfo_len-This->uri->userinfo_split-1;
5273 return get_builder_component(&This->password, &This->password_len, start, len, ppwzPassword, pcchPassword);
5277 static HRESULT WINAPI UriBuilder_GetPath(IUriBuilder *iface, DWORD *pcchPath, LPCWSTR *ppwzPath)
5279 UriBuilder *This = URIBUILDER_THIS(iface);
5280 TRACE("(%p)->(%p %p)\n", This, pcchPath, ppwzPath);
5282 if(!This->uri || This->uri->path_start == -1 || This->modified_props & Uri_HAS_PATH)
5283 return get_builder_component(&This->path, &This->path_len, NULL, 0, ppwzPath, pcchPath);
5285 return get_builder_component(&This->path, &This->path_len, This->uri->canon_uri+This->uri->path_start,
5286 This->uri->path_len, ppwzPath, pcchPath);
5289 static HRESULT WINAPI UriBuilder_GetPort(IUriBuilder *iface, BOOL *pfHasPort, DWORD *pdwPort)
5291 UriBuilder *This = URIBUILDER_THIS(iface);
5292 TRACE("(%p)->(%p %p)\n", This, pfHasPort, pdwPort);
5305 *pfHasPort = This->has_port;
5306 *pdwPort = This->port;
5310 static HRESULT WINAPI UriBuilder_GetQuery(IUriBuilder *iface, DWORD *pcchQuery, LPCWSTR *ppwzQuery)
5312 UriBuilder *This = URIBUILDER_THIS(iface);
5313 TRACE("(%p)->(%p %p)\n", This, pcchQuery, ppwzQuery);
5315 if(!This->uri || This->uri->query_start == -1 || This->modified_props & Uri_HAS_QUERY)
5316 return get_builder_component(&This->query, &This->query_len, NULL, 0, ppwzQuery, pcchQuery);
5318 return get_builder_component(&This->query, &This->query_len, This->uri->canon_uri+This->uri->query_start,
5319 This->uri->query_len, ppwzQuery, pcchQuery);
5322 static HRESULT WINAPI UriBuilder_GetSchemeName(IUriBuilder *iface, DWORD *pcchSchemeName, LPCWSTR *ppwzSchemeName)
5324 UriBuilder *This = URIBUILDER_THIS(iface);
5325 TRACE("(%p)->(%p %p)\n", This, pcchSchemeName, ppwzSchemeName);
5327 if(!This->uri || This->uri->scheme_start == -1 || This->modified_props & Uri_HAS_SCHEME_NAME)
5328 return get_builder_component(&This->scheme, &This->scheme_len, NULL, 0, ppwzSchemeName, pcchSchemeName);
5330 return get_builder_component(&This->scheme, &This->scheme_len, This->uri->canon_uri+This->uri->scheme_start,
5331 This->uri->scheme_len, ppwzSchemeName, pcchSchemeName);
5334 static HRESULT WINAPI UriBuilder_GetUserName(IUriBuilder *iface, DWORD *pcchUserName, LPCWSTR *ppwzUserName)
5336 UriBuilder *This = URIBUILDER_THIS(iface);
5337 TRACE("(%p)->(%p %p)\n", This, pcchUserName, ppwzUserName);
5339 if(!This->uri || This->uri->userinfo_start == -1 || This->uri->userinfo_split == 0 ||
5340 This->modified_props & Uri_HAS_USER_NAME)
5341 return get_builder_component(&This->username, &This->username_len, NULL, 0, ppwzUserName, pcchUserName);
5343 const WCHAR *start = This->uri->canon_uri+This->uri->userinfo_start;
5345 /* Check if there's a password in the userinfo section. */
5346 if(This->uri->userinfo_split > -1)
5347 /* Don't include the password. */
5348 return get_builder_component(&This->username, &This->username_len, start,
5349 This->uri->userinfo_split, ppwzUserName, pcchUserName);
5351 return get_builder_component(&This->username, &This->username_len, start,
5352 This->uri->userinfo_len, ppwzUserName, pcchUserName);
5356 static HRESULT WINAPI UriBuilder_SetFragment(IUriBuilder *iface, LPCWSTR pwzNewValue)
5358 UriBuilder *This = URIBUILDER_THIS(iface);
5359 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5360 return set_builder_component(&This->fragment, &This->fragment_len, pwzNewValue, '#',
5361 &This->modified_props, Uri_HAS_FRAGMENT);
5364 static HRESULT WINAPI UriBuilder_SetHost(IUriBuilder *iface, LPCWSTR pwzNewValue)
5366 UriBuilder *This = URIBUILDER_THIS(iface);
5367 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5369 /* Host name can't be set to NULL. */
5371 return E_INVALIDARG;
5373 return set_builder_component(&This->host, &This->host_len, pwzNewValue, 0,
5374 &This->modified_props, Uri_HAS_HOST);
5377 static HRESULT WINAPI UriBuilder_SetPassword(IUriBuilder *iface, LPCWSTR pwzNewValue)
5379 UriBuilder *This = URIBUILDER_THIS(iface);
5380 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5381 return set_builder_component(&This->password, &This->password_len, pwzNewValue, 0,
5382 &This->modified_props, Uri_HAS_PASSWORD);
5385 static HRESULT WINAPI UriBuilder_SetPath(IUriBuilder *iface, LPCWSTR pwzNewValue)
5387 UriBuilder *This = URIBUILDER_THIS(iface);
5388 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5389 return set_builder_component(&This->path, &This->path_len, pwzNewValue, 0,
5390 &This->modified_props, Uri_HAS_PATH);
5393 static HRESULT WINAPI UriBuilder_SetPort(IUriBuilder *iface, BOOL fHasPort, DWORD dwNewValue)
5395 UriBuilder *This = URIBUILDER_THIS(iface);
5396 TRACE("(%p)->(%d %d)\n", This, fHasPort, dwNewValue);
5398 This->has_port = fHasPort;
5399 This->port = dwNewValue;
5400 This->modified_props |= Uri_HAS_PORT;
5404 static HRESULT WINAPI UriBuilder_SetQuery(IUriBuilder *iface, LPCWSTR pwzNewValue)
5406 UriBuilder *This = URIBUILDER_THIS(iface);
5407 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5408 return set_builder_component(&This->query, &This->query_len, pwzNewValue, '?',
5409 &This->modified_props, Uri_HAS_QUERY);
5412 static HRESULT WINAPI UriBuilder_SetSchemeName(IUriBuilder *iface, LPCWSTR pwzNewValue)
5414 UriBuilder *This = URIBUILDER_THIS(iface);
5415 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5417 /* Only set the scheme name if it's not NULL or empty. */
5418 if(!pwzNewValue || !*pwzNewValue)
5419 return E_INVALIDARG;
5421 return set_builder_component(&This->scheme, &This->scheme_len, pwzNewValue, 0,
5422 &This->modified_props, Uri_HAS_SCHEME_NAME);
5425 static HRESULT WINAPI UriBuilder_SetUserName(IUriBuilder *iface, LPCWSTR pwzNewValue)
5427 UriBuilder *This = URIBUILDER_THIS(iface);
5428 TRACE("(%p)->(%s)\n", This, debugstr_w(pwzNewValue));
5429 return set_builder_component(&This->username, &This->username_len, pwzNewValue, 0,
5430 &This->modified_props, Uri_HAS_USER_NAME);
5433 static HRESULT WINAPI UriBuilder_RemoveProperties(IUriBuilder *iface, DWORD dwPropertyMask)
5435 const DWORD accepted_flags = Uri_HAS_AUTHORITY|Uri_HAS_DOMAIN|Uri_HAS_EXTENSION|Uri_HAS_FRAGMENT|Uri_HAS_HOST|
5436 Uri_HAS_PASSWORD|Uri_HAS_PATH|Uri_HAS_PATH_AND_QUERY|Uri_HAS_QUERY|
5437 Uri_HAS_USER_INFO|Uri_HAS_USER_NAME;
5439 UriBuilder *This = URIBUILDER_THIS(iface);
5440 TRACE("(%p)->(0x%08x)\n", This, dwPropertyMask);
5442 if(dwPropertyMask & ~accepted_flags)
5443 return E_INVALIDARG;
5445 if(dwPropertyMask & Uri_HAS_FRAGMENT)
5446 UriBuilder_SetFragment(iface, NULL);
5448 /* Even though you can't set the host name to NULL or an
5449 * empty string, you can still remove it... for some reason.
5451 if(dwPropertyMask & Uri_HAS_HOST)
5452 set_builder_component(&This->host, &This->host_len, NULL, 0,
5453 &This->modified_props, Uri_HAS_HOST);
5455 if(dwPropertyMask & Uri_HAS_PASSWORD)
5456 UriBuilder_SetPassword(iface, NULL);
5458 if(dwPropertyMask & Uri_HAS_PATH)
5459 UriBuilder_SetPath(iface, NULL);
5461 if(dwPropertyMask & Uri_HAS_PORT)
5462 UriBuilder_SetPort(iface, FALSE, 0);
5464 if(dwPropertyMask & Uri_HAS_QUERY)
5465 UriBuilder_SetQuery(iface, NULL);
5467 if(dwPropertyMask & Uri_HAS_USER_NAME)
5468 UriBuilder_SetUserName(iface, NULL);
5473 static HRESULT WINAPI UriBuilder_HasBeenModified(IUriBuilder *iface, BOOL *pfModified)
5475 UriBuilder *This = URIBUILDER_THIS(iface);
5476 TRACE("(%p)->(%p)\n", This, pfModified);
5481 *pfModified = This->modified_props > 0;
5485 #undef URIBUILDER_THIS
5487 static const IUriBuilderVtbl UriBuilderVtbl = {
5488 UriBuilder_QueryInterface,
5491 UriBuilder_CreateUriSimple,
5492 UriBuilder_CreateUri,
5493 UriBuilder_CreateUriWithFlags,
5496 UriBuilder_GetFragment,
5498 UriBuilder_GetPassword,
5501 UriBuilder_GetQuery,
5502 UriBuilder_GetSchemeName,
5503 UriBuilder_GetUserName,
5504 UriBuilder_SetFragment,
5506 UriBuilder_SetPassword,
5509 UriBuilder_SetQuery,
5510 UriBuilder_SetSchemeName,
5511 UriBuilder_SetUserName,
5512 UriBuilder_RemoveProperties,
5513 UriBuilder_HasBeenModified,
5516 /***********************************************************************
5517 * CreateIUriBuilder (urlmon.@)
5519 HRESULT WINAPI CreateIUriBuilder(IUri *pIUri, DWORD dwFlags, DWORD_PTR dwReserved, IUriBuilder **ppIUriBuilder)
5523 TRACE("(%p %x %x %p)\n", pIUri, dwFlags, (DWORD)dwReserved, ppIUriBuilder);
5528 ret = heap_alloc_zero(sizeof(UriBuilder));
5530 return E_OUTOFMEMORY;
5532 ret->lpIUriBuilderVtbl = &UriBuilderVtbl;
5538 if((uri = get_uri_obj(pIUri))) {
5543 /* Windows doesn't set 'has_port' to TRUE in this case. */
5544 ret->port = uri->port;
5548 *ppIUriBuilder = NULL;
5549 FIXME("(%p %x %x %p): Unknown IUri types not supported yet.\n", pIUri, dwFlags,
5550 (DWORD)dwReserved, ppIUriBuilder);
5555 *ppIUriBuilder = URIBUILDER(ret);