blob: 2d89ad190cfe1c57cd20194661329551ff229993 [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
8 * '[A-Za-z][A-Za-z0-9+.-]*'. But use sightly looser check
9 * 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
32static int url_decode_char(const char *q)
33{
34 int i;
35 unsigned char val = 0;
36 for (i = 0; i < 2; i++) {
37 unsigned char c = *q++;
38 val <<= 4;
39 if (c >= '0' && c <= '9')
40 val += c - '0';
41 else if (c >= 'a' && c <= 'f')
42 val += c - 'a' + 10;
43 else if (c >= 'A' && c <= 'F')
44 val += c - 'A' + 10;
45 else
46 return -1;
47 }
48 return val;
49}
50
Jeff King66c84482011-07-18 03:48:51 -040051static char *url_decode_internal(const char **query, int len,
52 const char *stop_at, struct strbuf *out,
53 int decode_plus)
Jeff King638794c2010-05-23 05:17:55 -040054{
55 const char *q = *query;
Jeff King638794c2010-05-23 05:17:55 -040056
Jeff King66c84482011-07-18 03:48:51 -040057 while (len) {
Jeff King638794c2010-05-23 05:17:55 -040058 unsigned char c = *q;
59
60 if (!c)
61 break;
62 if (stop_at && strchr(stop_at, c)) {
63 q++;
Jeff King66c84482011-07-18 03:48:51 -040064 len--;
Jeff King638794c2010-05-23 05:17:55 -040065 break;
66 }
67
68 if (c == '%') {
69 int val = url_decode_char(q + 1);
70 if (0 <= val) {
Junio C Hamanoce83eda2010-06-23 10:27:39 -070071 strbuf_addch(out, val);
Jeff King638794c2010-05-23 05:17:55 -040072 q += 3;
Jeff King66c84482011-07-18 03:48:51 -040073 len -= 3;
Jeff King638794c2010-05-23 05:17:55 -040074 continue;
75 }
76 }
77
Thomas Rast730220d2010-07-24 16:49:04 +020078 if (decode_plus && c == '+')
Junio C Hamanoce83eda2010-06-23 10:27:39 -070079 strbuf_addch(out, ' ');
Jeff King638794c2010-05-23 05:17:55 -040080 else
Junio C Hamanoce83eda2010-06-23 10:27:39 -070081 strbuf_addch(out, c);
Jeff King638794c2010-05-23 05:17:55 -040082 q++;
Jeff King66c84482011-07-18 03:48:51 -040083 len--;
84 }
Jeff King638794c2010-05-23 05:17:55 -040085 *query = q;
Junio C Hamanoce83eda2010-06-23 10:27:39 -070086 return strbuf_detach(out, NULL);
Jeff King638794c2010-05-23 05:17:55 -040087}
88
89char *url_decode(const char *url)
90{
Jeff King66c84482011-07-18 03:48:51 -040091 return url_decode_mem(url, strlen(url));
92}
93
94char *url_decode_mem(const char *url, int len)
95{
Junio C Hamanoce83eda2010-06-23 10:27:39 -070096 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 03:48:51 -040097 const char *colon = memchr(url, ':', len);
Junio C Hamanoce83eda2010-06-23 10:27:39 -070098
99 /* Skip protocol part if present */
Junio C Hamano3c73a1d2010-06-24 13:36:30 -0700100 if (colon && url < colon) {
101 strbuf_add(&out, url, colon - url);
Jeff King66c84482011-07-18 03:48:51 -0400102 len -= colon - url;
Junio C Hamano3c73a1d2010-06-24 13:36:30 -0700103 url = colon;
Junio C Hamanoce83eda2010-06-23 10:27:39 -0700104 }
Jeff King66c84482011-07-18 03:48:51 -0400105 return url_decode_internal(&url, len, NULL, &out, 0);
Jeff King638794c2010-05-23 05:17:55 -0400106}
107
108char *url_decode_parameter_name(const char **query)
109{
Junio C Hamanoce83eda2010-06-23 10:27:39 -0700110 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 03:48:51 -0400111 return url_decode_internal(query, -1, "&=", &out, 1);
Jeff King638794c2010-05-23 05:17:55 -0400112}
113
114char *url_decode_parameter_value(const char **query)
115{
Junio C Hamanoce83eda2010-06-23 10:27:39 -0700116 struct strbuf out = STRBUF_INIT;
Jeff King66c84482011-07-18 03:48:51 -0400117 return url_decode_internal(query, -1, "&", &out, 1);
Jeff King638794c2010-05-23 05:17:55 -0400118}
Tay Ray Chuan1966d9f2010-11-25 16:21:04 +0800119
120void end_url_with_slash(struct strbuf *buf, const char *url)
121{
122 strbuf_addstr(buf, url);
Jeff King00b6c172015-09-24 17:08:35 -0400123 strbuf_complete(buf, '/');
Tay Ray Chuan1966d9f2010-11-25 16:21:04 +0800124}
Tay Ray Chuan3793a302010-11-25 16:21:05 +0800125
126void str_end_url_with_slash(const char *url, char **dest) {
127 struct strbuf buf = STRBUF_INIT;
128 end_url_with_slash(&buf, url);
129 free(*dest);
130 *dest = strbuf_detach(&buf, NULL);
131}