Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 1 | #include "cache.h" |
Stephen Boyd | c2e86ad | 2011-03-22 00:51:05 -0700 | [diff] [blame] | 2 | #include "url.h" |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 3 | |
| 4 | int 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 | |
| 19 | int is_url(const char *url) |
| 20 | { |
Junio C Hamano | b33a1b9 | 2011-10-03 10:56:42 -0700 | [diff] [blame] | 21 | /* Is "scheme" part reasonable? */ |
| 22 | if (!url || !is_urlschemechar(1, *url++)) |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 23 | return 0; |
Junio C Hamano | b33a1b9 | 2011-10-03 10:56:42 -0700 | [diff] [blame] | 24 | while (*url && *url != ':') { |
| 25 | if (!is_urlschemechar(0, *url++)) |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 26 | return 0; |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 27 | } |
Junio C Hamano | b33a1b9 | 2011-10-03 10:56:42 -0700 | [diff] [blame] | 28 | /* We've seen "scheme"; we want colon-slash-slash */ |
| 29 | return (url[0] == ':' && url[1] == '/' && url[2] == '/'); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 30 | } |
| 31 | |
| 32 | static 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 King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 51 | static char *url_decode_internal(const char **query, int len, |
| 52 | const char *stop_at, struct strbuf *out, |
| 53 | int decode_plus) |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 54 | { |
| 55 | const char *q = *query; |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 56 | |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 57 | while (len) { |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 58 | unsigned char c = *q; |
| 59 | |
| 60 | if (!c) |
| 61 | break; |
| 62 | if (stop_at && strchr(stop_at, c)) { |
| 63 | q++; |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 64 | len--; |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 65 | break; |
| 66 | } |
| 67 | |
| 68 | if (c == '%') { |
| 69 | int val = url_decode_char(q + 1); |
| 70 | if (0 <= val) { |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 71 | strbuf_addch(out, val); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 72 | q += 3; |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 73 | len -= 3; |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 74 | continue; |
| 75 | } |
| 76 | } |
| 77 | |
Thomas Rast | 730220d | 2010-07-24 16:49:04 +0200 | [diff] [blame] | 78 | if (decode_plus && c == '+') |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 79 | strbuf_addch(out, ' '); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 80 | else |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 81 | strbuf_addch(out, c); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 82 | q++; |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 83 | len--; |
| 84 | } |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 85 | *query = q; |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 86 | return strbuf_detach(out, NULL); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 87 | } |
| 88 | |
| 89 | char *url_decode(const char *url) |
| 90 | { |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 91 | return url_decode_mem(url, strlen(url)); |
| 92 | } |
| 93 | |
| 94 | char *url_decode_mem(const char *url, int len) |
| 95 | { |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 96 | struct strbuf out = STRBUF_INIT; |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 97 | const char *colon = memchr(url, ':', len); |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 98 | |
| 99 | /* Skip protocol part if present */ |
Junio C Hamano | 3c73a1d | 2010-06-24 13:36:30 -0700 | [diff] [blame] | 100 | if (colon && url < colon) { |
| 101 | strbuf_add(&out, url, colon - url); |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 102 | len -= colon - url; |
Junio C Hamano | 3c73a1d | 2010-06-24 13:36:30 -0700 | [diff] [blame] | 103 | url = colon; |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 104 | } |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 105 | return url_decode_internal(&url, len, NULL, &out, 0); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 106 | } |
| 107 | |
| 108 | char *url_decode_parameter_name(const char **query) |
| 109 | { |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 110 | struct strbuf out = STRBUF_INIT; |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 111 | return url_decode_internal(query, -1, "&=", &out, 1); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 112 | } |
| 113 | |
| 114 | char *url_decode_parameter_value(const char **query) |
| 115 | { |
Junio C Hamano | ce83eda | 2010-06-23 10:27:39 -0700 | [diff] [blame] | 116 | struct strbuf out = STRBUF_INIT; |
Jeff King | 66c8448 | 2011-07-18 03:48:51 -0400 | [diff] [blame] | 117 | return url_decode_internal(query, -1, "&", &out, 1); |
Jeff King | 638794c | 2010-05-23 05:17:55 -0400 | [diff] [blame] | 118 | } |
Tay Ray Chuan | 1966d9f | 2010-11-25 16:21:04 +0800 | [diff] [blame] | 119 | |
| 120 | void end_url_with_slash(struct strbuf *buf, const char *url) |
| 121 | { |
| 122 | strbuf_addstr(buf, url); |
Jeff King | 00b6c17 | 2015-09-24 17:08:35 -0400 | [diff] [blame] | 123 | strbuf_complete(buf, '/'); |
Tay Ray Chuan | 1966d9f | 2010-11-25 16:21:04 +0800 | [diff] [blame] | 124 | } |
Tay Ray Chuan | 3793a30 | 2010-11-25 16:21:05 +0800 | [diff] [blame] | 125 | |
| 126 | void 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 | } |