blob: e04bd60b6bead493e3236949b5b4a837c729c146 [file] [log] [blame]
Jeff King638794c2010-05-23 05:17:55 -04001#include "cache.h"
Stephen Boydc2e86ad2011-03-22 00:51:05 -07002#include "url.h"
Jeff King638794c2010-05-23 05:17:55 -04003
4int is_urlschemechar(int first_flag, int ch)
5{
6 /*
7 * The set of valid URL schemes, as per STD66 (RFC3986) is
Elijah Newren15beaaa2019-11-05 17:07:23 +00008 * '[A-Za-z][A-Za-z0-9+.-]*'. But use slightly looser check
Jeff King638794c2010-05-23 05:17:55 -04009 * of '[A-Za-z0-9][A-Za-z0-9+.-]*' because earlier version
10 * of check used '[A-Za-z0-9]+' so not to break any remote
11 * helpers.
12 */
13 int alphanumeric, special;
14 alphanumeric = ch > 0 && isalnum(ch);
15 special = ch == '+' || ch == '-' || ch == '.';
16 return alphanumeric || (!first_flag && special);
17}
18
19int is_url(const char *url)
20{
Junio C Hamanob33a1b92011-10-03 10:56:42 -070021 /* Is "scheme" part reasonable? */
22 if (!url || !is_urlschemechar(1, *url++))
Jeff King638794c2010-05-23 05:17:55 -040023 return 0;
Junio C Hamanob33a1b92011-10-03 10:56:42 -070024 while (*url && *url != ':') {
25 if (!is_urlschemechar(0, *url++))
Jeff King638794c2010-05-23 05:17:55 -040026 return 0;
Jeff King638794c2010-05-23 05:17:55 -040027 }
Junio C Hamanob33a1b92011-10-03 10:56:42 -070028 /* We've seen "scheme"; we want colon-slash-slash */
29 return (url[0] == ':' && url[1] == '/' && url[2] == '/');
Jeff King638794c2010-05-23 05:17:55 -040030}
31
Jeff King66c84482011-07-18 03:48:51 -040032static char *url_decode_internal(const char **query, int len,
33 const char *stop_at, struct strbuf *out,
34 int decode_plus)
Jeff King638794c2010-05-23 05:17:55 -040035{
36 const char *q = *query;
Jeff King638794c2010-05-23 05:17:55 -040037
Jeff King66c84482011-07-18 03:48:51 -040038 while (len) {
Jeff King638794c2010-05-23 05:17:55 -040039 unsigned char c = *q;
40
41 if (!c)
42 break;
43 if (stop_at && strchr(stop_at, c)) {
44 q++;
Jeff King66c84482011-07-18 03:48:51 -040045 len--;
Jeff King638794c2010-05-23 05:17:55 -040046 break;
47 }
48
Matthew DeVore3f6b8a62019-06-04 10:57:04 -070049 if (c == '%' && (len < 0 || len >= 3)) {
René Scharfed2330972016-09-03 17:59:20 +020050 int val = hex2chr(q + 1);
Matthew DeVored37dc232019-06-04 10:57:05 -070051 if (0 < val) {
Junio C Hamanoce83eda2010-06-23 10:27:39 -070052 strbuf_addch(out, val);
Jeff King638794c2010-05-23 05:17:55 -040053 q += 3;
Jeff King66c84482011-07-18 03:48:51 -040054 len -= 3;
Jeff King638794c2010-05-23 05:17:55 -040055 continue;
56 }
57 }
58
Thomas Rast730220d2010-07-24 16:49:04 +020059 if (decode_plus && c == '+')
Junio C Hamanoce83eda2010-06-23 10:27:39 -070060 strbuf_addch(out, ' ');
Jeff King638794c2010-05-23 05:17:55 -040061 else
Junio C Hamanoce83eda2010-06-23 10:27:39 -070062 strbuf_addch(out, c);
Jeff King638794c2010-05-23 05:17:55 -040063 q++;
Jeff King66c84482011-07-18 03:48:51 -040064 len--;
65 }
Jeff King638794c2010-05-23 05:17:55 -040066 *query = q;
Junio C Hamanoce83eda2010-06-23 10:27:39 -070067 return strbuf_detach(out, NULL);
Jeff King638794c2010-05-23 05:17:55 -040068}
69
70char *url_decode(const char *url)
71{
Jeff King66c84482011-07-18 03:48:51 -040072 return url_decode_mem(url, strlen(url));
73}
74
75char *url_decode_mem(const char *url, int len)
76{
Junio C Hamanoce83eda2010-06-23 10:27:39 -070077 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 03:48:51 -040078 const char *colon = memchr(url, ':', len);
Junio C Hamanoce83eda2010-06-23 10:27:39 -070079
80 /* Skip protocol part if present */
Junio C Hamano3c73a1d2010-06-24 13:36:30 -070081 if (colon && url < colon) {
82 strbuf_add(&out, url, colon - url);
Jeff King66c84482011-07-18 03:48:51 -040083 len -= colon - url;
Junio C Hamano3c73a1d2010-06-24 13:36:30 -070084 url = colon;
Junio C Hamanoce83eda2010-06-23 10:27:39 -070085 }
Jeff King66c84482011-07-18 03:48:51 -040086 return url_decode_internal(&url, len, NULL, &out, 0);
Jeff King638794c2010-05-23 05:17:55 -040087}
88
Matthew DeVoree987df52019-06-27 15:54:08 -070089char *url_percent_decode(const char *encoded)
90{
91 struct strbuf out = STRBUF_INIT;
92 return url_decode_internal(&encoded, strlen(encoded), NULL, &out, 0);
93}
94
Jeff King638794c2010-05-23 05:17:55 -040095char *url_decode_parameter_name(const char **query)
96{
Junio C Hamanoce83eda2010-06-23 10:27:39 -070097 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 03:48:51 -040098 return url_decode_internal(query, -1, "&=", &out, 1);
Jeff King638794c2010-05-23 05:17:55 -040099}
100
101char *url_decode_parameter_value(const char **query)
102{
Junio C Hamanoce83eda2010-06-23 10:27:39 -0700103 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 03:48:51 -0400104 return url_decode_internal(query, -1, "&", &out, 1);
Jeff King638794c2010-05-23 05:17:55 -0400105}
Tay Ray Chuan1966d9f2010-11-25 16:21:04 +0800106
107void end_url_with_slash(struct strbuf *buf, const char *url)
108{
109 strbuf_addstr(buf, url);
Jeff King00b6c172015-09-24 17:08:35 -0400110 strbuf_complete(buf, '/');
Tay Ray Chuan1966d9f2010-11-25 16:21:04 +0800111}
Tay Ray Chuan3793a302010-11-25 16:21:05 +0800112
Nguyễn Thái Ngọc Duy3b335762018-12-09 11:25:21 +0100113void str_end_url_with_slash(const char *url, char **dest)
114{
Tay Ray Chuan3793a302010-11-25 16:21:05 +0800115 struct strbuf buf = STRBUF_INIT;
116 end_url_with_slash(&buf, url);
117 free(*dest);
118 *dest = strbuf_detach(&buf, NULL);
119}