]>
Commit | Line | Data |
---|---|---|
43ae1d95 | 1 | /* |
bbc27441 | 2 | * Copyright (C) 1996-2014 The Squid Software Foundation and contributors |
43ae1d95 | 3 | * |
bbc27441 AJ |
4 | * Squid software is distributed under GPLv2+ license and includes |
5 | * contributions from numerous individuals and organizations. | |
6 | * Please see the COPYING and CONTRIBUTORS files for details. | |
43ae1d95 | 7 | */ |
8 | ||
bbc27441 AJ |
9 | /* DEBUG: section 86 ESI processing */ |
10 | ||
582c2af2 | 11 | #include "squid.h" |
582c2af2 | 12 | #include "Debug.h" |
602d9612 | 13 | #include "esi/CustomParser.h" |
c8ea3cc0 | 14 | #include "fatal.h" |
99742866 FC |
15 | #include "libTrie/Trie.h" |
16 | #include "libTrie/TrieCharTransform.h" | |
43ae1d95 | 17 | |
c8ea3cc0 FC |
18 | #include <vector> |
19 | ||
43ae1d95 | 20 | Trie *ESICustomParser::SearchTrie=NULL; |
43ae1d95 | 21 | |
019db3c6 | 22 | EsiParserDefinition(ESICustomParser); |
c2d4889f | 23 | |
43ae1d95 | 24 | Trie * |
25 | ESICustomParser::GetTrie() | |
26 | { | |
27 | if (SearchTrie) | |
28 | return SearchTrie; | |
29 | ||
924f73bc | 30 | SearchTrie = new Trie(new TrieCaseless); |
43ae1d95 | 31 | |
137a13ea | 32 | static const ESITAG_t ESITAG_value = ESITAG; |
33 | ||
43ae1d95 | 34 | assert (SearchTrie->add |
137a13ea | 35 | ("<esi:",5,(void *)&ESITAG_value)); |
36 | ||
37 | static const ESITAG_t ESIENDTAG_value = ESIENDTAG; | |
43ae1d95 | 38 | |
39 | assert (SearchTrie->add | |
137a13ea | 40 | ("</esi:",6,(void *)&ESIENDTAG_value)); |
41 | ||
42 | static const ESITAG_t ESICOMMENT_value = ESICOMMENT; | |
43ae1d95 | 43 | |
44 | assert (SearchTrie->add | |
137a13ea | 45 | ("<!--",4,(void *)&ESICOMMENT_value)); |
43ae1d95 | 46 | |
47 | return SearchTrie; | |
48 | } | |
49 | ||
a5488e5d AJ |
50 | ESICustomParser::ESICustomParser(ESIParserClient *aClient) : |
51 | theClient(aClient), | |
52 | lastTag(ESITAG) | |
43ae1d95 | 53 | {} |
54 | ||
55 | ESICustomParser::~ESICustomParser() | |
56 | { | |
57 | theClient = NULL; | |
58 | } | |
59 | ||
60 | char const * | |
924f73bc | 61 | ESICustomParser::findTag(char const *buffer, size_t bufferLength) |
43ae1d95 | 62 | { |
63 | size_t myOffset (0); | |
137a13ea | 64 | ESITAG_t *resulttype = NULL; |
43ae1d95 | 65 | |
924f73bc | 66 | while (myOffset < bufferLength && |
137a13ea | 67 | (resulttype = static_cast<ESITAG_t *>(GetTrie()->findPrefix (buffer + myOffset, bufferLength - myOffset))) |
68 | == NULL) | |
43ae1d95 | 69 | ++myOffset; |
70 | ||
924f73bc | 71 | if (myOffset == bufferLength) |
43ae1d95 | 72 | return NULL; |
73 | ||
bf8fe701 | 74 | debugs(86, 9, "ESICustomParser::findTag: found " << *resulttype); |
43ae1d95 | 75 | |
137a13ea | 76 | lastTag = *resulttype; |
43ae1d95 | 77 | |
924f73bc | 78 | return buffer + myOffset; |
43ae1d95 | 79 | } |
80 | ||
81 | bool | |
82 | ESICustomParser::parse(char const *dataToParse, size_t const lengthOfData, bool const endOfStream) | |
83 | { | |
bf8fe701 | 84 | debugs(86, 9, "ESICustomParser::parse: Appending data to internal buffer"); |
43ae1d95 | 85 | content.append (dataToParse, lengthOfData); |
86 | ||
87 | if (!endOfStream) { | |
88 | return true; | |
89 | } | |
90 | ||
91 | size_t openESITags (0); | |
a7a42b14 FC |
92 | //erring on the safe side. Probably rawBuf would be ok too |
93 | char const *currentPos = content.termedBuf(); | |
43ae1d95 | 94 | size_t remainingCount = content.size(); |
5aeabf95 | 95 | char const *tag = NULL; |
43ae1d95 | 96 | |
97 | while ((tag = findTag(currentPos, remainingCount))) { | |
98 | if (tag - currentPos) | |
99 | theClient->parserDefault (currentPos,tag - currentPos); | |
100 | ||
101 | switch (lastTag) { | |
102 | ||
103 | case ESITAG: { | |
26ac0430 AJ |
104 | ++openESITags; |
105 | char *tagEnd = strchr(const_cast<char *>(tag), '>'); | |
43ae1d95 | 106 | |
26ac0430 AJ |
107 | if (!tagEnd) { |
108 | error = "Could not find end ('>') of tag"; | |
109 | return false; | |
110 | } | |
43ae1d95 | 111 | |
26ac0430 AJ |
112 | if (tagEnd - tag > (ssize_t)remainingCount) { |
113 | error = "Tag ends beyond the parse buffer."; | |
114 | return false; | |
115 | } | |
43ae1d95 | 116 | |
26ac0430 AJ |
117 | if (*(tagEnd - 1) == '/') |
118 | --openESITags; | |
43ae1d95 | 119 | |
26ac0430 | 120 | char * endofName = strpbrk(const_cast<char *>(tag), w_space); |
43ae1d95 | 121 | |
26ac0430 AJ |
122 | if (endofName > tagEnd) |
123 | endofName = const_cast<char *>(tagEnd); | |
43ae1d95 | 124 | |
26ac0430 | 125 | *endofName = '\0'; |
43ae1d95 | 126 | |
26ac0430 | 127 | *tagEnd = '\0'; |
43ae1d95 | 128 | |
c8ea3cc0 | 129 | std::vector<char *>attributes; |
43ae1d95 | 130 | |
26ac0430 | 131 | char *attribute = const_cast<char *>(endofName + 1); |
43ae1d95 | 132 | |
26ac0430 AJ |
133 | while (attribute > tag && attribute < tagEnd) { |
134 | /* leading spaces */ | |
43ae1d95 | 135 | |
26ac0430 AJ |
136 | while (attribute < tagEnd && (xisspace(*attribute) || (*attribute == '/'))) |
137 | ++attribute; | |
43ae1d95 | 138 | |
26ac0430 AJ |
139 | if (! (attribute < tagEnd)) |
140 | break; | |
43ae1d95 | 141 | |
26ac0430 AJ |
142 | /* attribute name */ |
143 | attributes.push_back(attribute); | |
43ae1d95 | 144 | |
26ac0430 | 145 | char *nextSpace = strpbrk(attribute, w_space); |
43ae1d95 | 146 | |
26ac0430 | 147 | char *equals = strchr(attribute, '='); |
43ae1d95 | 148 | |
26ac0430 AJ |
149 | if (!equals) { |
150 | error = "Missing attribute value."; | |
151 | return false; | |
152 | } | |
43ae1d95 | 153 | |
26ac0430 AJ |
154 | if (nextSpace && nextSpace < equals) |
155 | *nextSpace = '\0'; | |
156 | else | |
157 | *equals = '\0'; | |
43ae1d95 | 158 | |
26ac0430 | 159 | ++equals; |
43ae1d95 | 160 | |
26ac0430 AJ |
161 | while (equals < tagEnd && xisspace(*equals)) |
162 | ++equals; | |
43ae1d95 | 163 | |
26ac0430 | 164 | char sep = *equals; |
43ae1d95 | 165 | |
26ac0430 AJ |
166 | if (sep != '\'' && sep != '"') { |
167 | error = "Unknown identifier ("; | |
168 | error.append (sep); | |
169 | error.append (")"); | |
170 | return false; | |
43ae1d95 | 171 | } |
172 | ||
26ac0430 | 173 | char *value = equals + 1; |
a5488e5d AJ |
174 | char *end = strchr(value, sep); |
175 | ||
176 | if (!end) { | |
177 | error = "Missing attribute ending separator ("; | |
178 | error.append(sep); | |
179 | error.append(")"); | |
180 | return false; | |
181 | } | |
26ac0430 AJ |
182 | attributes.push_back(value); |
183 | *end = '\0'; | |
184 | attribute = end + 1; | |
185 | } | |
43ae1d95 | 186 | |
4814631e FC |
187 | // TODO: after c++11, replace &attributes.front() with attributes.data() |
188 | theClient->start (tag + 1, const_cast<const char **>(&attributes.front()), attributes.size() >> 1); | |
26ac0430 | 189 | /* TODO: attributes */ |
43ae1d95 | 190 | |
26ac0430 AJ |
191 | if (*(tagEnd - 1) == '/') |
192 | theClient->end (tag + 1); | |
43ae1d95 | 193 | |
26ac0430 | 194 | remainingCount -= tagEnd - currentPos + 1; |
43ae1d95 | 195 | |
26ac0430 AJ |
196 | currentPos = tagEnd + 1; |
197 | } | |
198 | ||
199 | break; | |
43ae1d95 | 200 | |
201 | case ESIENDTAG: { | |
26ac0430 AJ |
202 | if (!openESITags) |
203 | return false; | |
43ae1d95 | 204 | |
26ac0430 | 205 | char const *tagEnd = strchr(tag, '>'); |
43ae1d95 | 206 | |
26ac0430 AJ |
207 | if (!tagEnd) |
208 | return false; | |
43ae1d95 | 209 | |
26ac0430 AJ |
210 | if (tagEnd - tag > (ssize_t)remainingCount) |
211 | return false; | |
43ae1d95 | 212 | |
26ac0430 | 213 | char * endofName = strpbrk(const_cast<char *>(tag), w_space); |
43ae1d95 | 214 | |
26ac0430 AJ |
215 | if (endofName > tagEnd) |
216 | endofName = const_cast<char *>(tagEnd); | |
43ae1d95 | 217 | |
26ac0430 | 218 | *endofName = '\0'; |
43ae1d95 | 219 | |
26ac0430 | 220 | theClient->end (tag + 2); |
43ae1d95 | 221 | |
26ac0430 | 222 | --openESITags; |
43ae1d95 | 223 | |
26ac0430 | 224 | remainingCount -= tagEnd - currentPos + 1; |
43ae1d95 | 225 | |
26ac0430 AJ |
226 | currentPos = tagEnd + 1; |
227 | } | |
43ae1d95 | 228 | |
26ac0430 | 229 | break; |
43ae1d95 | 230 | |
231 | case ESICOMMENT: { | |
26ac0430 AJ |
232 | /* Further optimisation potential: |
233 | * 1) recognize end comments for esi and don't callback on | |
234 | * comments. | |
235 | * 2) provide the comment length to the caller. | |
236 | */ | |
237 | /* Comments must not be nested, without CDATA | |
238 | * and we don't support CDATA | |
239 | */ | |
240 | char *commentEnd = strstr (const_cast<char *>(tag), "-->"); | |
241 | ||
242 | if (!commentEnd) { | |
243 | error = "missing end of comment"; | |
244 | return false; | |
245 | } | |
43ae1d95 | 246 | |
26ac0430 AJ |
247 | if (commentEnd - tag > (ssize_t)remainingCount) { |
248 | error = "comment ends beyond parse buffer"; | |
249 | return false; | |
43ae1d95 | 250 | } |
251 | ||
26ac0430 AJ |
252 | *commentEnd = '\0'; |
253 | theClient->parserComment (tag + 4); | |
254 | remainingCount -= commentEnd - currentPos + 3; | |
255 | currentPos = commentEnd + 3; | |
256 | } | |
257 | ||
258 | break; | |
259 | break; | |
43ae1d95 | 260 | |
261 | default: | |
262 | fatal ("unknown ESI tag type found"); | |
263 | }; | |
264 | ||
265 | /* | |
266 | * Find next esi tag (open or closing) or comment | |
267 | * send tag, or full comment text | |
268 | * rinse | |
269 | */ | |
270 | } | |
271 | ||
272 | if (remainingCount) | |
273 | theClient->parserDefault (currentPos,remainingCount); | |
274 | ||
bf8fe701 | 275 | debugs(86, 5, "ESICustomParser::parse: Finished parsing, will return " << !openESITags); |
43ae1d95 | 276 | |
277 | if (openESITags) | |
278 | error = "ESI Tags still open"; | |
279 | ||
280 | return !openESITags; | |
281 | } | |
282 | ||
137a13ea | 283 | long int |
43ae1d95 | 284 | ESICustomParser::lineNumber() const |
285 | { | |
286 | /* We don't track lines in the body */ | |
287 | return 0; | |
288 | } | |
289 | ||
290 | char const * | |
291 | ESICustomParser::errorString() const | |
292 | { | |
293 | if (error.size()) | |
a7a42b14 | 294 | return error.termedBuf(); |
43ae1d95 | 295 | else |
296 | return "Parsing error strings not implemented"; | |
297 | } |