config: add helper to normalize and match URLs
[git/git.git] / urlmatch.c
CommitLineData
3402a8dc
KM
1#include "cache.h"
2#include "urlmatch.h"
3
4#define URL_ALPHA "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"
5#define URL_DIGIT "0123456789"
6#define URL_ALPHADIGIT URL_ALPHA URL_DIGIT
7#define URL_SCHEME_CHARS URL_ALPHADIGIT "+.-"
8#define URL_HOST_CHARS URL_ALPHADIGIT ".-[:]" /* IPv6 literals need [:] */
9#define URL_UNSAFE_CHARS " <>\"%{}|\\^`" /* plus 0x00-0x1F,0x7F-0xFF */
10#define URL_GEN_RESERVED ":/?#[]@"
11#define URL_SUB_RESERVED "!$&'()*+,;="
12#define URL_RESERVED URL_GEN_RESERVED URL_SUB_RESERVED /* only allowed delims */
13
14static int append_normalized_escapes(struct strbuf *buf,
15 const char *from,
16 size_t from_len,
17 const char *esc_extra,
18 const char *esc_ok)
19{
20 /*
21 * Append to strbuf 'buf' characters from string 'from' with length
22 * 'from_len' while unescaping characters that do not need to be escaped
23 * and escaping characters that do. The set of characters to escape
24 * (the complement of which is unescaped) starts out as the RFC 3986
25 * unsafe characters (0x00-0x1F,0x7F-0xFF," <>\"#%{}|\\^`"). If
26 * 'esc_extra' is not NULL, those additional characters will also always
27 * be escaped. If 'esc_ok' is not NULL, those characters will be left
28 * escaped if found that way, but will not be unescaped otherwise (used
29 * for delimiters). If a %-escape sequence is encountered that is not
30 * followed by 2 hexadecimal digits, the sequence is invalid and
31 * false (0) will be returned. Otherwise true (1) will be returned for
32 * success.
33 *
34 * Note that all %-escape sequences will be normalized to UPPERCASE
35 * as indicated in RFC 3986. Unless included in esc_extra or esc_ok
36 * alphanumerics and "-._~" will always be unescaped as per RFC 3986.
37 */
38
39 while (from_len) {
40 int ch = *from++;
41 int was_esc = 0;
42
43 from_len--;
44 if (ch == '%') {
45 if (from_len < 2 ||
46 !isxdigit((unsigned char)from[0]) ||
47 !isxdigit((unsigned char)from[1]))
48 return 0;
49 ch = hexval_table[(unsigned char)*from++] << 4;
50 ch |= hexval_table[(unsigned char)*from++];
51 from_len -= 2;
52 was_esc = 1;
53 }
54 if ((unsigned char)ch <= 0x1F || (unsigned char)ch >= 0x7F ||
55 strchr(URL_UNSAFE_CHARS, ch) ||
56 (esc_extra && strchr(esc_extra, ch)) ||
57 (was_esc && strchr(esc_ok, ch)))
58 strbuf_addf(buf, "%%%02X", (unsigned char)ch);
59 else
60 strbuf_addch(buf, ch);
61 }
62
63 return 1;
64}
65
66char *url_normalize(const char *url, struct url_info *out_info)
67{
68 /*
69 * Normalize NUL-terminated url using the following rules:
70 *
71 * 1. Case-insensitive parts of url will be converted to lower case
72 * 2. %-encoded characters that do not need to be will be unencoded
73 * 3. Characters that are not %-encoded and must be will be encoded
74 * 4. All %-encodings will be converted to upper case hexadecimal
75 * 5. Leading 0s are removed from port numbers
76 * 6. If the default port for the scheme is given it will be removed
77 * 7. A path part (including empty) not starting with '/' has one added
78 * 8. Any dot segments (. or ..) in the path are resolved and removed
79 * 9. IPv6 host literals are allowed (but not normalized or validated)
80 *
81 * The rules are based on information in RFC 3986.
82 *
83 * Please note this function requires a full URL including a scheme
84 * and host part (except for file: URLs which may have an empty host).
85 *
86 * The return value is a newly allocated string that must be freed
87 * or NULL if the url is not valid.
88 *
89 * If out_info is non-NULL, the url and err fields therein will always
90 * be set. If a non-NULL value is returned, it will be stored in
91 * out_info->url as well, out_info->err will be set to NULL and the
92 * other fields of *out_info will also be filled in. If a NULL value
93 * is returned, NULL will be stored in out_info->url and out_info->err
94 * will be set to a brief, translated, error message, but no other
95 * fields will be filled in.
96 *
97 * This is NOT a URL validation function. Full URL validation is NOT
98 * performed. Some invalid host names are passed through this function
99 * undetected. However, most all other problems that make a URL invalid
100 * will be detected (including a missing host for non file: URLs).
101 */
102
103 size_t url_len = strlen(url);
104 struct strbuf norm;
105 size_t spanned;
106 size_t scheme_len, user_off=0, user_len=0, passwd_off=0, passwd_len=0;
107 size_t host_off=0, host_len=0, port_len=0, path_off, path_len, result_len;
108 const char *slash_ptr, *at_ptr, *colon_ptr, *path_start;
109 char *result;
110
111 /*
112 * Copy lowercased scheme and :// suffix, %-escapes are not allowed
113 * First character of scheme must be URL_ALPHA
114 */
115 spanned = strspn(url, URL_SCHEME_CHARS);
116 if (!spanned || !isalpha(url[0]) || spanned + 3 > url_len ||
117 url[spanned] != ':' || url[spanned+1] != '/' || url[spanned+2] != '/') {
118 if (out_info) {
119 out_info->url = NULL;
120 out_info->err = _("invalid URL scheme name or missing '://' suffix");
121 }
122 return NULL; /* Bad scheme and/or missing "://" part */
123 }
124 strbuf_init(&norm, url_len);
125 scheme_len = spanned;
126 spanned += 3;
127 url_len -= spanned;
128 while (spanned--)
129 strbuf_addch(&norm, tolower(*url++));
130
131
132 /*
133 * Copy any username:password if present normalizing %-escapes
134 */
135 at_ptr = strchr(url, '@');
136 slash_ptr = url + strcspn(url, "/?#");
137 if (at_ptr && at_ptr < slash_ptr) {
138 user_off = norm.len;
139 if (at_ptr > url) {
140 if (!append_normalized_escapes(&norm, url, at_ptr - url,
141 "", URL_RESERVED)) {
142 if (out_info) {
143 out_info->url = NULL;
144 out_info->err = _("invalid %XX escape sequence");
145 }
146 strbuf_release(&norm);
147 return NULL;
148 }
149 colon_ptr = strchr(norm.buf + scheme_len + 3, ':');
150 if (colon_ptr) {
151 passwd_off = (colon_ptr + 1) - norm.buf;
152 passwd_len = norm.len - passwd_off;
153 user_len = (passwd_off - 1) - (scheme_len + 3);
154 } else {
155 user_len = norm.len - (scheme_len + 3);
156 }
157 }
158 strbuf_addch(&norm, '@');
159 url_len -= (++at_ptr - url);
160 url = at_ptr;
161 }
162
163
164 /*
165 * Copy the host part excluding any port part, no %-escapes allowed
166 */
167 if (!url_len || strchr(":/?#", *url)) {
168 /* Missing host invalid for all URL schemes except file */
169 if (strncmp(norm.buf, "file:", 5)) {
170 if (out_info) {
171 out_info->url = NULL;
172 out_info->err = _("missing host and scheme is not 'file:'");
173 }
174 strbuf_release(&norm);
175 return NULL;
176 }
177 } else {
178 host_off = norm.len;
179 }
180 colon_ptr = slash_ptr - 1;
181 while (colon_ptr > url && *colon_ptr != ':' && *colon_ptr != ']')
182 colon_ptr--;
183 if (*colon_ptr != ':') {
184 colon_ptr = slash_ptr;
185 } else if (!host_off && colon_ptr < slash_ptr && colon_ptr + 1 != slash_ptr) {
186 /* file: URLs may not have a port number */
187 if (out_info) {
188 out_info->url = NULL;
189 out_info->err = _("a 'file:' URL may not have a port number");
190 }
191 strbuf_release(&norm);
192 return NULL;
193 }
194 spanned = strspn(url, URL_HOST_CHARS);
195 if (spanned < colon_ptr - url) {
196 /* Host name has invalid characters */
197 if (out_info) {
198 out_info->url = NULL;
199 out_info->err = _("invalid characters in host name");
200 }
201 strbuf_release(&norm);
202 return NULL;
203 }
204 while (url < colon_ptr) {
205 strbuf_addch(&norm, tolower(*url++));
206 url_len--;
207 }
208
209
210 /*
211 * Check the port part and copy if not the default (after removing any
212 * leading 0s); no %-escapes allowed
213 */
214 if (colon_ptr < slash_ptr) {
215 /* skip the ':' and leading 0s but not the last one if all 0s */
216 url++;
217 url += strspn(url, "0");
218 if (url == slash_ptr && url[-1] == '0')
219 url--;
220 if (url == slash_ptr) {
221 /* Skip ":" port with no number, it's same as default */
222 } else if (slash_ptr - url == 2 &&
223 !strncmp(norm.buf, "http:", 5) &&
224 !strncmp(url, "80", 2)) {
225 /* Skip http :80 as it's the default */
226 } else if (slash_ptr - url == 3 &&
227 !strncmp(norm.buf, "https:", 6) &&
228 !strncmp(url, "443", 3)) {
229 /* Skip https :443 as it's the default */
230 } else {
231 /*
232 * Port number must be all digits with leading 0s removed
233 * and since all the protocols we deal with have a 16-bit
234 * port number it must also be in the range 1..65535
235 * 0 is not allowed because that means "next available"
236 * on just about every system and therefore cannot be used
237 */
238 unsigned long pnum = 0;
239 spanned = strspn(url, URL_DIGIT);
240 if (spanned < slash_ptr - url) {
241 /* port number has invalid characters */
242 if (out_info) {
243 out_info->url = NULL;
244 out_info->err = _("invalid port number");
245 }
246 strbuf_release(&norm);
247 return NULL;
248 }
249 if (slash_ptr - url <= 5)
250 pnum = strtoul(url, NULL, 10);
251 if (pnum == 0 || pnum > 65535) {
252 /* port number not in range 1..65535 */
253 if (out_info) {
254 out_info->url = NULL;
255 out_info->err = _("invalid port number");
256 }
257 strbuf_release(&norm);
258 return NULL;
259 }
260 strbuf_addch(&norm, ':');
261 strbuf_add(&norm, url, slash_ptr - url);
262 port_len = slash_ptr - url;
263 }
264 url_len -= slash_ptr - colon_ptr;
265 url = slash_ptr;
266 }
267 if (host_off)
268 host_len = norm.len - host_off;
269
270
271 /*
272 * Now copy the path resolving any . and .. segments being careful not
273 * to corrupt the URL by unescaping any delimiters, but do add an
274 * initial '/' if it's missing and do normalize any %-escape sequences.
275 */
276 path_off = norm.len;
277 path_start = norm.buf + path_off;
278 strbuf_addch(&norm, '/');
279 if (*url == '/') {
280 url++;
281 url_len--;
282 }
283 for (;;) {
284 const char *seg_start = norm.buf + norm.len;
285 const char *next_slash = url + strcspn(url, "/?#");
286 int skip_add_slash = 0;
287 /*
288 * RFC 3689 indicates that any . or .. segments should be
289 * unescaped before being checked for.
290 */
291 if (!append_normalized_escapes(&norm, url, next_slash - url, "",
292 URL_RESERVED)) {
293 if (out_info) {
294 out_info->url = NULL;
295 out_info->err = _("invalid %XX escape sequence");
296 }
297 strbuf_release(&norm);
298 return NULL;
299 }
300 if (!strcmp(seg_start, ".")) {
301 /* ignore a . segment; be careful not to remove initial '/' */
302 if (seg_start == path_start + 1) {
303 strbuf_setlen(&norm, norm.len - 1);
304 skip_add_slash = 1;
305 } else {
306 strbuf_setlen(&norm, norm.len - 2);
307 }
308 } else if (!strcmp(seg_start, "..")) {
309 /*
310 * ignore a .. segment and remove the previous segment;
311 * be careful not to remove initial '/' from path
312 */
313 const char *prev_slash = norm.buf + norm.len - 3;
314 if (prev_slash == path_start) {
315 /* invalid .. because no previous segment to remove */
316 if (out_info) {
317 out_info->url = NULL;
318 out_info->err = _("invalid '..' path segment");
319 }
320 strbuf_release(&norm);
321 return NULL;
322 }
323 while (*--prev_slash != '/') {}
324 if (prev_slash == path_start) {
325 strbuf_setlen(&norm, prev_slash - norm.buf + 1);
326 skip_add_slash = 1;
327 } else {
328 strbuf_setlen(&norm, prev_slash - norm.buf);
329 }
330 }
331 url_len -= next_slash - url;
332 url = next_slash;
333 /* if the next char is not '/' done with the path */
334 if (*url != '/')
335 break;
336 url++;
337 url_len--;
338 if (!skip_add_slash)
339 strbuf_addch(&norm, '/');
340 }
341 path_len = norm.len - path_off;
342
343
344 /*
345 * Now simply copy the rest, if any, only normalizing %-escapes and
346 * being careful not to corrupt the URL by unescaping any delimiters.
347 */
348 if (*url) {
349 if (!append_normalized_escapes(&norm, url, url_len, "", URL_RESERVED)) {
350 if (out_info) {
351 out_info->url = NULL;
352 out_info->err = _("invalid %XX escape sequence");
353 }
354 strbuf_release(&norm);
355 return NULL;
356 }
357 }
358
359
360 result = strbuf_detach(&norm, &result_len);
361 if (out_info) {
362 out_info->url = result;
363 out_info->err = NULL;
364 out_info->url_len = result_len;
365 out_info->scheme_len = scheme_len;
366 out_info->user_off = user_off;
367 out_info->user_len = user_len;
368 out_info->passwd_off = passwd_off;
369 out_info->passwd_len = passwd_len;
370 out_info->host_off = host_off;
371 out_info->host_len = host_len;
372 out_info->port_len = port_len;
373 out_info->path_off = path_off;
374 out_info->path_len = path_len;
375 }
376 return result;
377}
378
379static size_t url_match_prefix(const char *url,
380 const char *url_prefix,
381 size_t url_prefix_len)
382{
383 /*
384 * url_prefix matches url if url_prefix is an exact match for url or it
385 * is a prefix of url and the match ends on a path component boundary.
386 * Both url and url_prefix are considered to have an implicit '/' on the
387 * end for matching purposes if they do not already.
388 *
389 * url must be NUL terminated. url_prefix_len is the length of
390 * url_prefix which need not be NUL terminated.
391 *
392 * The return value is the length of the match in characters (including
393 * the final '/' even if it's implicit) or 0 for no match.
394 *
395 * Passing NULL as url and/or url_prefix will always cause 0 to be
396 * returned without causing any faults.
397 */
398 if (!url || !url_prefix)
399 return 0;
400 if (!url_prefix_len || (url_prefix_len == 1 && *url_prefix == '/'))
401 return (!*url || *url == '/') ? 1 : 0;
402 if (url_prefix[url_prefix_len - 1] == '/')
403 url_prefix_len--;
404 if (strncmp(url, url_prefix, url_prefix_len))
405 return 0;
406 if ((strlen(url) == url_prefix_len) || (url[url_prefix_len] == '/'))
407 return url_prefix_len + 1;
408 return 0;
409}
410
411int match_urls(const struct url_info *url,
412 const struct url_info *url_prefix,
413 int *exactusermatch)
414{
415 /*
416 * url_prefix matches url if the scheme, host and port of url_prefix
417 * are the same as those of url and the path portion of url_prefix
418 * is the same as the path portion of url or it is a prefix that
419 * matches at a '/' boundary. If url_prefix contains a user name,
420 * that must also exactly match the user name in url.
421 *
422 * If the user, host, port and path match in this fashion, the returned
423 * value is the length of the path match including any implicit
424 * final '/'. For example, "http://me@example.com/path" is matched by
425 * "http://example.com" with a path length of 1.
426 *
427 * If there is a match and exactusermatch is not NULL, then
428 * *exactusermatch will be set to true if both url and url_prefix
429 * contained a user name or false if url_prefix did not have a
430 * user name. If there is no match *exactusermatch is left untouched.
431 */
432 int usermatched = 0;
433 int pathmatchlen;
434
435 if (!url || !url_prefix || !url->url || !url_prefix->url)
436 return 0;
437
438 /* check the scheme */
439 if (url_prefix->scheme_len != url->scheme_len ||
440 strncmp(url->url, url_prefix->url, url->scheme_len))
441 return 0; /* schemes do not match */
442
443 /* check the user name if url_prefix has one */
444 if (url_prefix->user_off) {
445 if (!url->user_off || url->user_len != url_prefix->user_len ||
446 strncmp(url->url + url->user_off,
447 url_prefix->url + url_prefix->user_off,
448 url->user_len))
449 return 0; /* url_prefix has a user but it's not a match */
450 usermatched = 1;
451 }
452
453 /* check the host and port */
454 if (url_prefix->host_len != url->host_len ||
455 strncmp(url->url + url->host_off,
456 url_prefix->url + url_prefix->host_off, url->host_len))
457 return 0; /* host names and/or ports do not match */
458
459 /* check the path */
460 pathmatchlen = url_match_prefix(
461 url->url + url->path_off,
462 url_prefix->url + url_prefix->path_off,
463 url_prefix->url_len - url_prefix->path_off);
464
465 if (pathmatchlen && exactusermatch)
466 *exactusermatch = usermatched;
467 return pathmatchlen;
468}