| 1 | #define CHARSET_MAX 41 |
| 2 | |
| 3 | static const char * |
| 4 | getTok(const char **pp) |
| 5 | { |
| 6 | enum { inAtom, inString, init, inComment }; |
| 7 | int state = init; |
| 8 | const char *tokStart = 0; |
| 9 | for (;;) { |
| 10 | switch (**pp) { |
| 11 | case '\0': |
| 12 | return 0; |
| 13 | case ' ': |
| 14 | case '\r': |
| 15 | case '\t': |
| 16 | case '\n': |
| 17 | if (state == inAtom) |
| 18 | return tokStart; |
| 19 | break; |
| 20 | case '(': |
| 21 | if (state == inAtom) |
| 22 | return tokStart; |
| 23 | if (state != inString) |
| 24 | state++; |
| 25 | break; |
| 26 | case ')': |
| 27 | if (state > init) |
| 28 | --state; |
| 29 | else if (state != inString) |
| 30 | return 0; |
| 31 | break; |
| 32 | case ';': |
| 33 | case '/': |
| 34 | case '=': |
| 35 | if (state == inAtom) |
| 36 | return tokStart; |
| 37 | if (state == init) |
| 38 | return (*pp)++; |
| 39 | break; |
| 40 | case '\\': |
| 41 | ++*pp; |
| 42 | if (**pp == '\0') |
| 43 | return 0; |
| 44 | break; |
| 45 | case '"': |
| 46 | switch (state) { |
| 47 | case inString: |
| 48 | ++*pp; |
| 49 | return tokStart; |
| 50 | case inAtom: |
| 51 | return tokStart; |
| 52 | case init: |
| 53 | tokStart = *pp; |
| 54 | state = inString; |
| 55 | break; |
| 56 | } |
| 57 | break; |
| 58 | default: |
| 59 | if (state == init) { |
| 60 | tokStart = *pp; |
| 61 | state = inAtom; |
| 62 | } |
| 63 | break; |
| 64 | } |
| 65 | ++*pp; |
| 66 | } |
| 67 | /* not reached */ |
| 68 | } |
| 69 | |
| 70 | /* key must be lowercase ASCII */ |
| 71 | |
| 72 | static int |
| 73 | matchkey(const char *start, const char *end, const char *key) |
| 74 | { |
| 75 | if (!start) |
| 76 | return 0; |
| 77 | for (; start != end; start++, key++) |
| 78 | if (*start != *key && *start != 'A' + (*key - 'a')) |
| 79 | return 0; |
| 80 | return *key == '\0'; |
| 81 | } |
| 82 | |
| 83 | void |
| 84 | getXMLCharset(const char *buf, char *charset) |
| 85 | { |
| 86 | const char *next, *p; |
| 87 | |
| 88 | charset[0] = '\0'; |
| 89 | next = buf; |
| 90 | p = getTok(&next); |
| 91 | if (matchkey(p, next, "text")) |
| 92 | strcpy(charset, "us-ascii"); |
| 93 | else if (!matchkey(p, next, "application")) |
| 94 | return; |
| 95 | p = getTok(&next); |
| 96 | if (!p || *p != '/') |
| 97 | return; |
| 98 | p = getTok(&next); |
| 99 | if (matchkey(p, next, "xml")) |
| 100 | isXml = 1; |
| 101 | p = getTok(&next); |
| 102 | while (p) { |
| 103 | if (*p == ';') { |
| 104 | p = getTok(&next); |
| 105 | if (matchkey(p, next, "charset")) { |
| 106 | p = getTok(&next); |
| 107 | if (p && *p == '=') { |
| 108 | p = getTok(&next); |
| 109 | if (p) { |
| 110 | char *s = charset; |
| 111 | if (*p == '"') { |
| 112 | while (++p != next - 1) { |
| 113 | if (*p == '\\') |
| 114 | ++p; |
| 115 | if (s == charset + CHARSET_MAX - 1) { |
| 116 | charset[0] = '\0'; |
| 117 | break; |
| 118 | } |
| 119 | *s++ = *p; |
| 120 | } |
| 121 | *s++ = '\0'; |
| 122 | } |
| 123 | else { |
| 124 | if (next - p > CHARSET_MAX - 1) |
| 125 | break; |
| 126 | while (p != next) |
| 127 | *s++ = *p++; |
| 128 | *s = 0; |
| 129 | break; |
| 130 | } |
| 131 | } |
| 132 | } |
| 133 | } |
| 134 | } |
| 135 | else |
| 136 | p = getTok(&next); |
| 137 | } |
| 138 | } |
| 139 | |
| 140 | int |
| 141 | main(int argc, char **argv) |
| 142 | { |
| 143 | char buf[CHARSET_MAX]; |
| 144 | getXMLCharset(argv[1], buf); |
| 145 | printf("charset = \"%s\"\n", buf); |
| 146 | return 0; |
| 147 | } |