🌐 AI搜索 & 代理 主页
blob: 282b12495ae7d4af9db51a76017af6065b924d3a [file] [log] [blame]
Elijah Newrenfc7bd512023-02-24 00:09:341#include "git-compat-util.h"
Calvin Wand88e8102023-09-29 21:20:482#include "hex-ll.h"
Elijah Newrenfc7bd512023-02-24 00:09:343#include "strbuf.h"
Stephen Boydc2e86ad2011-03-22 07:51:054#include "url.h"
Jeff King638794c2010-05-23 09:17:555
6int is_urlschemechar(int first_flag, int ch)
7{
8 /*
9 * The set of valid URL schemes, as per STD66 (RFC3986) is
Elijah Newren15beaaa2019-11-05 17:07:2310 * '[A-Za-z][A-Za-z0-9+.-]*'. But use slightly looser check
Jeff King638794c2010-05-23 09:17:5511 * of '[A-Za-z0-9][A-Za-z0-9+.-]*' because earlier version
12 * of check used '[A-Za-z0-9]+' so not to break any remote
13 * helpers.
14 */
15 int alphanumeric, special;
16 alphanumeric = ch > 0 && isalnum(ch);
17 special = ch == '+' || ch == '-' || ch == '.';
18 return alphanumeric || (!first_flag && special);
19}
20
21int is_url(const char *url)
22{
Junio C Hamanob33a1b92011-10-03 17:56:4223 /* Is "scheme" part reasonable? */
24 if (!url || !is_urlschemechar(1, *url++))
Jeff King638794c2010-05-23 09:17:5525 return 0;
Junio C Hamanob33a1b92011-10-03 17:56:4226 while (*url && *url != ':') {
27 if (!is_urlschemechar(0, *url++))
Jeff King638794c2010-05-23 09:17:5528 return 0;
Jeff King638794c2010-05-23 09:17:5529 }
Junio C Hamanob33a1b92011-10-03 17:56:4230 /* We've seen "scheme"; we want colon-slash-slash */
31 return (url[0] == ':' && url[1] == '/' && url[2] == '/');
Jeff King638794c2010-05-23 09:17:5532}
33
Jeff King66c84482011-07-18 07:48:5134static char *url_decode_internal(const char **query, int len,
35 const char *stop_at, struct strbuf *out,
36 int decode_plus)
Jeff King638794c2010-05-23 09:17:5537{
38 const char *q = *query;
Jeff King638794c2010-05-23 09:17:5539
Jeff King66c84482011-07-18 07:48:5140 while (len) {
Jeff King638794c2010-05-23 09:17:5541 unsigned char c = *q;
42
43 if (!c)
44 break;
45 if (stop_at && strchr(stop_at, c)) {
46 q++;
Jeff King66c84482011-07-18 07:48:5147 len--;
Jeff King638794c2010-05-23 09:17:5548 break;
49 }
50
Matthew DeVore3f6b8a62019-06-04 17:57:0451 if (c == '%' && (len < 0 || len >= 3)) {
René Scharfed2330972016-09-03 15:59:2052 int val = hex2chr(q + 1);
Matthew DeVored37dc232019-06-04 17:57:0553 if (0 < val) {
Junio C Hamanoce83eda2010-06-23 17:27:3954 strbuf_addch(out, val);
Jeff King638794c2010-05-23 09:17:5555 q += 3;
Jeff King66c84482011-07-18 07:48:5156 len -= 3;
Jeff King638794c2010-05-23 09:17:5557 continue;
58 }
59 }
60
Thomas Rast730220d2010-07-24 14:49:0461 if (decode_plus && c == '+')
Junio C Hamanoce83eda2010-06-23 17:27:3962 strbuf_addch(out, ' ');
Jeff King638794c2010-05-23 09:17:5563 else
Junio C Hamanoce83eda2010-06-23 17:27:3964 strbuf_addch(out, c);
Jeff King638794c2010-05-23 09:17:5565 q++;
Jeff King66c84482011-07-18 07:48:5166 len--;
67 }
Jeff King638794c2010-05-23 09:17:5568 *query = q;
Junio C Hamanoce83eda2010-06-23 17:27:3969 return strbuf_detach(out, NULL);
Jeff King638794c2010-05-23 09:17:5570}
71
72char *url_decode(const char *url)
73{
Jeff King66c84482011-07-18 07:48:5174 return url_decode_mem(url, strlen(url));
75}
76
77char *url_decode_mem(const char *url, int len)
78{
Junio C Hamanoce83eda2010-06-23 17:27:3979 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 07:48:5180 const char *colon = memchr(url, ':', len);
Junio C Hamanoce83eda2010-06-23 17:27:3981
82 /* Skip protocol part if present */
Junio C Hamano3c73a1d2010-06-24 20:36:3083 if (colon && url < colon) {
84 strbuf_add(&out, url, colon - url);
Jeff King66c84482011-07-18 07:48:5185 len -= colon - url;
Junio C Hamano3c73a1d2010-06-24 20:36:3086 url = colon;
Junio C Hamanoce83eda2010-06-23 17:27:3987 }
Jeff King66c84482011-07-18 07:48:5188 return url_decode_internal(&url, len, NULL, &out, 0);
Jeff King638794c2010-05-23 09:17:5589}
90
Matthew DeVoree987df52019-06-27 22:54:0891char *url_percent_decode(const char *encoded)
92{
93 struct strbuf out = STRBUF_INIT;
94 return url_decode_internal(&encoded, strlen(encoded), NULL, &out, 0);
95}
96
Jeff King638794c2010-05-23 09:17:5597char *url_decode_parameter_name(const char **query)
98{
Junio C Hamanoce83eda2010-06-23 17:27:3999 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 07:48:51100 return url_decode_internal(query, -1, "&=", &out, 1);
Jeff King638794c2010-05-23 09:17:55101}
102
103char *url_decode_parameter_value(const char **query)
104{
Junio C Hamanoce83eda2010-06-23 17:27:39105 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 07:48:51106 return url_decode_internal(query, -1, "&", &out, 1);
Jeff King638794c2010-05-23 09:17:55107}
Tay Ray Chuan1966d9f2010-11-25 08:21:04108
109void end_url_with_slash(struct strbuf *buf, const char *url)
110{
111 strbuf_addstr(buf, url);
Jeff King00b6c172015-09-24 21:08:35112 strbuf_complete(buf, '/');
Tay Ray Chuan1966d9f2010-11-25 08:21:04113}
Tay Ray Chuan3793a302010-11-25 08:21:05114
Nguyễn Thái Ngọc Duy3b335762018-12-09 10:25:21115void str_end_url_with_slash(const char *url, char **dest)
116{
Tay Ray Chuan3793a302010-11-25 08:21:05117 struct strbuf buf = STRBUF_INIT;
118 end_url_with_slash(&buf, url);
119 free(*dest);
120 *dest = strbuf_detach(&buf, NULL);
121}