]>
Commit | Line | Data |
---|---|---|
1 | // -*- mode: cpp; mode: fold -*- | |
2 | // Description /*{{{*/ | |
3 | // $Id: tagfile.cc,v 1.37.2.2 2003/12/31 16:02:30 mdz Exp $ | |
4 | /* ###################################################################### | |
5 | ||
6 | Fast scanner for RFC-822 type header information | |
7 | ||
8 | This uses a rotating buffer to load the package information into. | |
9 | The scanner runs over it and isolates and indexes a single section. | |
10 | ||
11 | ##################################################################### */ | |
12 | /*}}}*/ | |
13 | // Include Files /*{{{*/ | |
14 | #include<config.h> | |
15 | ||
16 | #include <apt-pkg/tagfile.h> | |
17 | #include <apt-pkg/error.h> | |
18 | #include <apt-pkg/strutl.h> | |
19 | #include <apt-pkg/fileutl.h> | |
20 | #include <apt-pkg/string_view.h> | |
21 | ||
22 | #include <list> | |
23 | ||
24 | #include <string> | |
25 | #include <stdio.h> | |
26 | #include <ctype.h> | |
27 | #include <stdlib.h> | |
28 | #include <string.h> | |
29 | ||
30 | #include <apti18n.h> | |
31 | /*}}}*/ | |
32 | ||
33 | using std::string; | |
34 | using APT::StringView; | |
35 | ||
36 | class APT_HIDDEN pkgTagFilePrivate /*{{{*/ | |
37 | { | |
38 | public: | |
39 | void Reset(FileFd * const pFd, unsigned long long const pSize, pkgTagFile::Flags const pFlags) | |
40 | { | |
41 | if (Buffer != NULL) | |
42 | free(Buffer); | |
43 | Buffer = NULL; | |
44 | Fd = pFd; | |
45 | Flags = pFlags; | |
46 | Start = NULL; | |
47 | End = NULL; | |
48 | Done = false; | |
49 | iOffset = 0; | |
50 | Size = pSize; | |
51 | isCommentedLine = false; | |
52 | chunks.clear(); | |
53 | } | |
54 | ||
55 | pkgTagFilePrivate(FileFd * const pFd, unsigned long long const Size, pkgTagFile::Flags const pFlags) : Buffer(NULL) | |
56 | { | |
57 | Reset(pFd, Size, pFlags); | |
58 | } | |
59 | FileFd * Fd; | |
60 | pkgTagFile::Flags Flags; | |
61 | char *Buffer; | |
62 | char *Start; | |
63 | char *End; | |
64 | bool Done; | |
65 | unsigned long long iOffset; | |
66 | unsigned long long Size; | |
67 | bool isCommentedLine; | |
68 | struct FileChunk | |
69 | { | |
70 | bool const good; | |
71 | size_t length; | |
72 | FileChunk(bool const pgood, size_t const plength) : good(pgood), length(plength) {} | |
73 | }; | |
74 | std::list<FileChunk> chunks; | |
75 | ||
76 | ~pkgTagFilePrivate() | |
77 | { | |
78 | if (Buffer != NULL) | |
79 | free(Buffer); | |
80 | } | |
81 | }; | |
82 | /*}}}*/ | |
83 | class APT_HIDDEN pkgTagSectionPrivate /*{{{*/ | |
84 | { | |
85 | public: | |
86 | pkgTagSectionPrivate() | |
87 | { | |
88 | } | |
89 | struct TagData { | |
90 | unsigned int StartTag; | |
91 | unsigned int EndTag; | |
92 | unsigned int StartValue; | |
93 | unsigned int NextInBucket; | |
94 | ||
95 | explicit TagData(unsigned int const StartTag) : StartTag(StartTag), EndTag(0), StartValue(0), NextInBucket(0) {} | |
96 | }; | |
97 | std::vector<TagData> Tags; | |
98 | }; | |
99 | /*}}}*/ | |
100 | ||
101 | static unsigned long AlphaHash(const char *Text, size_t Length) /*{{{*/ | |
102 | { | |
103 | /* This very simple hash function for the last 8 letters gives | |
104 | very good performance on the debian package files */ | |
105 | if (Length > 8) | |
106 | { | |
107 | Text += (Length - 8); | |
108 | Length = 8; | |
109 | } | |
110 | unsigned long Res = 0; | |
111 | for (size_t i = 0; i < Length; ++i) | |
112 | Res = ((unsigned long)(Text[i]) & 0xDF) ^ (Res << 1); | |
113 | return Res & 0xFF; | |
114 | } | |
115 | /*}}}*/ | |
116 | ||
117 | // TagFile::pkgTagFile - Constructor /*{{{*/ | |
118 | pkgTagFile::pkgTagFile(FileFd * const pFd,pkgTagFile::Flags const pFlags, unsigned long long const Size) | |
119 | : d(new pkgTagFilePrivate(pFd, Size + 4, pFlags)) | |
120 | { | |
121 | Init(pFd, pFlags, Size); | |
122 | } | |
123 | pkgTagFile::pkgTagFile(FileFd * const pFd,unsigned long long const Size) | |
124 | : pkgTagFile(pFd, pkgTagFile::STRICT, Size) | |
125 | { | |
126 | } | |
127 | void pkgTagFile::Init(FileFd * const pFd, pkgTagFile::Flags const pFlags, unsigned long long Size) | |
128 | { | |
129 | /* The size is increased by 4 because if we start with the Size of the | |
130 | filename we need to try to read 1 char more to see an EOF faster, 1 | |
131 | char the end-pointer can be on and maybe 2 newlines need to be added | |
132 | to the end of the file -> 4 extra chars */ | |
133 | Size += 4; | |
134 | d->Reset(pFd, Size, pFlags); | |
135 | ||
136 | if (d->Fd->IsOpen() == false) | |
137 | d->Start = d->End = d->Buffer = 0; | |
138 | else | |
139 | d->Buffer = (char*)malloc(sizeof(char) * Size); | |
140 | ||
141 | if (d->Buffer == NULL) | |
142 | d->Done = true; | |
143 | else | |
144 | d->Done = false; | |
145 | ||
146 | d->Start = d->End = d->Buffer; | |
147 | d->iOffset = 0; | |
148 | if (d->Done == false) | |
149 | Fill(); | |
150 | } | |
151 | void pkgTagFile::Init(FileFd * const pFd,unsigned long long Size) | |
152 | { | |
153 | Init(pFd, pkgTagFile::STRICT, Size); | |
154 | } | |
155 | /*}}}*/ | |
156 | // TagFile::~pkgTagFile - Destructor /*{{{*/ | |
157 | pkgTagFile::~pkgTagFile() | |
158 | { | |
159 | delete d; | |
160 | } | |
161 | /*}}}*/ | |
162 | // TagFile::Offset - Return the current offset in the buffer /*{{{*/ | |
163 | APT_PURE unsigned long pkgTagFile::Offset() | |
164 | { | |
165 | return d->iOffset; | |
166 | } | |
167 | /*}}}*/ | |
168 | // TagFile::Resize - Resize the internal buffer /*{{{*/ | |
169 | // --------------------------------------------------------------------- | |
170 | /* Resize the internal buffer (double it in size). Fail if a maximum size | |
171 | * size is reached. | |
172 | */ | |
173 | bool pkgTagFile::Resize() | |
174 | { | |
175 | // fail is the buffer grows too big | |
176 | if(d->Size > 1024*1024+1) | |
177 | return false; | |
178 | ||
179 | return Resize(d->Size * 2); | |
180 | } | |
181 | bool pkgTagFile::Resize(unsigned long long const newSize) | |
182 | { | |
183 | unsigned long long const EndSize = d->End - d->Start; | |
184 | ||
185 | // get new buffer and use it | |
186 | char* const newBuffer = static_cast<char*>(realloc(d->Buffer, sizeof(char) * newSize)); | |
187 | if (newBuffer == NULL) | |
188 | return false; | |
189 | d->Buffer = newBuffer; | |
190 | d->Size = newSize; | |
191 | ||
192 | // update the start/end pointers to the new buffer | |
193 | d->Start = d->Buffer; | |
194 | d->End = d->Start + EndSize; | |
195 | return true; | |
196 | } | |
197 | /*}}}*/ | |
198 | // TagFile::Step - Advance to the next section /*{{{*/ | |
199 | // --------------------------------------------------------------------- | |
200 | /* If the Section Scanner fails we refill the buffer and try again. | |
201 | * If that fails too, double the buffer size and try again until a | |
202 | * maximum buffer is reached. | |
203 | */ | |
204 | bool pkgTagFile::Step(pkgTagSection &Tag) | |
205 | { | |
206 | if(Tag.Scan(d->Start,d->End - d->Start) == false) | |
207 | { | |
208 | do | |
209 | { | |
210 | if (Fill() == false) | |
211 | return false; | |
212 | ||
213 | if(Tag.Scan(d->Start,d->End - d->Start, false)) | |
214 | break; | |
215 | ||
216 | if (Resize() == false) | |
217 | return _error->Error(_("Unable to parse package file %s (%d)"), | |
218 | d->Fd->Name().c_str(), 1); | |
219 | ||
220 | } while (Tag.Scan(d->Start,d->End - d->Start, false) == false); | |
221 | } | |
222 | ||
223 | size_t tagSize = Tag.size(); | |
224 | d->Start += tagSize; | |
225 | ||
226 | if ((d->Flags & pkgTagFile::SUPPORT_COMMENTS) == 0) | |
227 | d->iOffset += tagSize; | |
228 | else | |
229 | { | |
230 | auto first = d->chunks.begin(); | |
231 | for (; first != d->chunks.end(); ++first) | |
232 | { | |
233 | if (first->good == false) | |
234 | d->iOffset += first->length; | |
235 | else | |
236 | { | |
237 | if (tagSize < first->length) | |
238 | { | |
239 | first->length -= tagSize; | |
240 | d->iOffset += tagSize; | |
241 | break; | |
242 | } | |
243 | else | |
244 | { | |
245 | tagSize -= first->length; | |
246 | d->iOffset += first->length; | |
247 | } | |
248 | } | |
249 | } | |
250 | d->chunks.erase(d->chunks.begin(), first); | |
251 | } | |
252 | ||
253 | Tag.Trim(); | |
254 | return true; | |
255 | } | |
256 | /*}}}*/ | |
257 | // TagFile::Fill - Top up the buffer /*{{{*/ | |
258 | // --------------------------------------------------------------------- | |
259 | /* This takes the bit at the end of the buffer and puts it at the start | |
260 | then fills the rest from the file */ | |
261 | static bool FillBuffer(pkgTagFilePrivate * const d) | |
262 | { | |
263 | unsigned long long Actual = 0; | |
264 | // See if only a bit of the file is left | |
265 | unsigned long long const dataSize = d->Size - ((d->End - d->Buffer) + 1); | |
266 | if (d->Fd->Read(d->End, dataSize, &Actual) == false) | |
267 | return false; | |
268 | if (Actual != dataSize) | |
269 | d->Done = true; | |
270 | d->End += Actual; | |
271 | return true; | |
272 | } | |
273 | static void RemoveCommentsFromBuffer(pkgTagFilePrivate * const d) | |
274 | { | |
275 | // look for valid comments in the buffer | |
276 | char * good_start = nullptr, * bad_start = nullptr; | |
277 | char * current = d->Start; | |
278 | if (d->isCommentedLine == false) | |
279 | { | |
280 | if (d->Start == d->Buffer) | |
281 | { | |
282 | // the start of the buffer is a newline as a record can't start | |
283 | // in the middle of a line by definition. | |
284 | if (*d->Start == '#') | |
285 | { | |
286 | d->isCommentedLine = true; | |
287 | ++current; | |
288 | if (current > d->End) | |
289 | d->chunks.emplace_back(false, 1); | |
290 | } | |
291 | } | |
292 | if (d->isCommentedLine == false) | |
293 | good_start = d->Start; | |
294 | else | |
295 | bad_start = d->Start; | |
296 | } | |
297 | else | |
298 | bad_start = d->Start; | |
299 | ||
300 | std::vector<std::pair<char*, size_t>> good_parts; | |
301 | while (current <= d->End) | |
302 | { | |
303 | size_t const restLength = (d->End - current) + 1; | |
304 | if (d->isCommentedLine == false) | |
305 | { | |
306 | current = static_cast<char*>(memchr(current, '#', restLength)); | |
307 | if (current == nullptr) | |
308 | { | |
309 | size_t const goodLength = d->End - good_start; | |
310 | d->chunks.emplace_back(true, goodLength); | |
311 | if (good_start != d->Start) | |
312 | good_parts.push_back(std::make_pair(good_start, goodLength)); | |
313 | break; | |
314 | } | |
315 | bad_start = current; | |
316 | --current; | |
317 | // ensure that this is really a comment and not a '#' in the middle of a line | |
318 | if (*current == '\n') | |
319 | { | |
320 | size_t const goodLength = (current - good_start) + 1; | |
321 | d->chunks.emplace_back(true, goodLength); | |
322 | good_parts.push_back(std::make_pair(good_start, goodLength)); | |
323 | good_start = nullptr; | |
324 | d->isCommentedLine = true; | |
325 | } | |
326 | current += 2; | |
327 | } | |
328 | else // the current line is a comment | |
329 | { | |
330 | current = static_cast<char*>(memchr(current, '\n', restLength)); | |
331 | if (current == nullptr) | |
332 | { | |
333 | d->chunks.emplace_back(false, (d->End - bad_start)); | |
334 | break; | |
335 | } | |
336 | ++current; | |
337 | // is the next line a comment, too? | |
338 | if (current > d->End || *current != '#') | |
339 | { | |
340 | d->chunks.emplace_back(false, (current - bad_start)); | |
341 | good_start = current; | |
342 | bad_start = nullptr; | |
343 | d->isCommentedLine = false; | |
344 | } | |
345 | ++current; | |
346 | } | |
347 | } | |
348 | ||
349 | if (good_parts.empty() == false) | |
350 | { | |
351 | // we found comments, so move later parts over them | |
352 | current = d->Start; | |
353 | for (auto const &good: good_parts) | |
354 | { | |
355 | memmove(current, good.first, good.second); | |
356 | current += good.second; | |
357 | } | |
358 | d->End = current; | |
359 | } | |
360 | ||
361 | if (d->isCommentedLine == true) | |
362 | { | |
363 | // deal with a buffer containing only comments | |
364 | // or an (unfinished) comment at the end | |
365 | if (good_parts.empty() == true) | |
366 | d->End = d->Start; | |
367 | else | |
368 | d->Start = d->End; | |
369 | } | |
370 | else | |
371 | { | |
372 | // the buffer was all comment, but ended with the buffer | |
373 | if (good_parts.empty() == true && good_start >= d->End) | |
374 | d->End = d->Start; | |
375 | else | |
376 | d->Start = d->End; | |
377 | } | |
378 | } | |
379 | bool pkgTagFile::Fill() | |
380 | { | |
381 | unsigned long long const EndSize = d->End - d->Start; | |
382 | if (EndSize != 0) | |
383 | { | |
384 | memmove(d->Buffer,d->Start,EndSize); | |
385 | d->Start = d->End = d->Buffer + EndSize; | |
386 | } | |
387 | else | |
388 | d->Start = d->End = d->Buffer; | |
389 | ||
390 | unsigned long long Actual = 0; | |
391 | while (d->Done == false && d->Size > (Actual + 1)) | |
392 | { | |
393 | if (FillBuffer(d) == false) | |
394 | return false; | |
395 | if ((d->Flags & pkgTagFile::SUPPORT_COMMENTS) != 0) | |
396 | RemoveCommentsFromBuffer(d); | |
397 | Actual = d->End - d->Buffer; | |
398 | } | |
399 | d->Start = d->Buffer; | |
400 | ||
401 | if (d->Done == true) | |
402 | { | |
403 | if (EndSize <= 3 && Actual == 0) | |
404 | return false; | |
405 | if (d->Size - (d->End - d->Buffer) < 4) | |
406 | return true; | |
407 | ||
408 | // Append a double new line if one does not exist | |
409 | unsigned int LineCount = 0; | |
410 | for (const char *E = d->End - 1; E - d->End < 6 && (*E == '\n' || *E == '\r'); E--) | |
411 | if (*E == '\n') | |
412 | ++LineCount; | |
413 | if (LineCount < 2) | |
414 | { | |
415 | if (static_cast<unsigned long long>(d->End - d->Buffer) >= d->Size) | |
416 | Resize(d->Size + 3); | |
417 | for (; LineCount < 2; ++LineCount) | |
418 | *d->End++ = '\n'; | |
419 | } | |
420 | } | |
421 | return true; | |
422 | } | |
423 | /*}}}*/ | |
424 | // TagFile::Jump - Jump to a pre-recorded location in the file /*{{{*/ | |
425 | // --------------------------------------------------------------------- | |
426 | /* This jumps to a pre-recorded file location and reads the record | |
427 | that is there */ | |
428 | bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long long Offset) | |
429 | { | |
430 | if ((d->Flags & pkgTagFile::SUPPORT_COMMENTS) == 0 && | |
431 | // We are within a buffer space of the next hit.. | |
432 | Offset >= d->iOffset && d->iOffset + (d->End - d->Start) > Offset) | |
433 | { | |
434 | unsigned long long Dist = Offset - d->iOffset; | |
435 | d->Start += Dist; | |
436 | d->iOffset += Dist; | |
437 | // if we have seen the end, don't ask for more | |
438 | if (d->Done == true) | |
439 | return Tag.Scan(d->Start, d->End - d->Start); | |
440 | else | |
441 | return Step(Tag); | |
442 | } | |
443 | ||
444 | // Reposition and reload.. | |
445 | d->iOffset = Offset; | |
446 | d->Done = false; | |
447 | if (d->Fd->Seek(Offset) == false) | |
448 | return false; | |
449 | d->End = d->Start = d->Buffer; | |
450 | d->isCommentedLine = false; | |
451 | d->chunks.clear(); | |
452 | ||
453 | if (Fill() == false) | |
454 | return false; | |
455 | ||
456 | if (Tag.Scan(d->Start, d->End - d->Start) == true) | |
457 | return true; | |
458 | ||
459 | // This appends a double new line (for the real eof handling) | |
460 | if (Fill() == false) | |
461 | return false; | |
462 | ||
463 | if (Tag.Scan(d->Start, d->End - d->Start, false) == false) | |
464 | return _error->Error(_("Unable to parse package file %s (%d)"),d->Fd->Name().c_str(), 2); | |
465 | ||
466 | return true; | |
467 | } | |
468 | /*}}}*/ | |
469 | // pkgTagSection::pkgTagSection - Constructor /*{{{*/ | |
470 | // --------------------------------------------------------------------- | |
471 | /* */ | |
472 | APT_IGNORE_DEPRECATED_PUSH | |
473 | pkgTagSection::pkgTagSection() | |
474 | : Section(0), d(new pkgTagSectionPrivate()), Stop(0) | |
475 | { | |
476 | memset(&AlphaIndexes, 0, sizeof(AlphaIndexes)); | |
477 | } | |
478 | APT_IGNORE_DEPRECATED_POP | |
479 | /*}}}*/ | |
480 | // TagSection::Scan - Scan for the end of the header information /*{{{*/ | |
481 | bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength, bool const Restart) | |
482 | { | |
483 | Section = Start; | |
484 | const char *End = Start + MaxLength; | |
485 | ||
486 | if (Restart == false && d->Tags.empty() == false) | |
487 | { | |
488 | Stop = Section + d->Tags.back().StartTag; | |
489 | if (End <= Stop) | |
490 | return false; | |
491 | Stop = (const char *)memchr(Stop,'\n',End - Stop); | |
492 | if (Stop == NULL) | |
493 | return false; | |
494 | ++Stop; | |
495 | } | |
496 | else | |
497 | { | |
498 | Stop = Section; | |
499 | if (d->Tags.empty() == false) | |
500 | { | |
501 | memset(&AlphaIndexes, 0, sizeof(AlphaIndexes)); | |
502 | d->Tags.clear(); | |
503 | } | |
504 | d->Tags.reserve(0x100); | |
505 | } | |
506 | unsigned int TagCount = d->Tags.size(); | |
507 | ||
508 | if (Stop == 0) | |
509 | return false; | |
510 | ||
511 | pkgTagSectionPrivate::TagData lastTagData(0); | |
512 | lastTagData.EndTag = 0; | |
513 | unsigned long lastTagHash = 0; | |
514 | while (Stop < End) | |
515 | { | |
516 | TrimRecord(true,End); | |
517 | ||
518 | // this can happen when TrimRecord trims away the entire Record | |
519 | // (e.g. because it just contains comments) | |
520 | if(Stop == End) | |
521 | return true; | |
522 | ||
523 | // Start a new index and add it to the hash | |
524 | if (isspace_ascii(Stop[0]) == 0) | |
525 | { | |
526 | // store the last found tag | |
527 | if (lastTagData.EndTag != 0) | |
528 | { | |
529 | if (AlphaIndexes[lastTagHash] != 0) | |
530 | lastTagData.NextInBucket = AlphaIndexes[lastTagHash]; | |
531 | APT_IGNORE_DEPRECATED_PUSH | |
532 | AlphaIndexes[lastTagHash] = TagCount; | |
533 | APT_IGNORE_DEPRECATED_POP | |
534 | d->Tags.push_back(lastTagData); | |
535 | } | |
536 | ||
537 | APT_IGNORE_DEPRECATED(++TagCount;) | |
538 | lastTagData = pkgTagSectionPrivate::TagData(Stop - Section); | |
539 | // find the colon separating tag and value | |
540 | char const * Colon = (char const *) memchr(Stop, ':', End - Stop); | |
541 | if (Colon == NULL) | |
542 | return false; | |
543 | // find the end of the tag (which might or might not be the colon) | |
544 | char const * EndTag = Colon; | |
545 | --EndTag; | |
546 | for (; EndTag > Stop && isspace_ascii(*EndTag) != 0; --EndTag) | |
547 | ; | |
548 | ++EndTag; | |
549 | lastTagData.EndTag = EndTag - Section; | |
550 | lastTagHash = AlphaHash(Stop, EndTag - Stop); | |
551 | // find the beginning of the value | |
552 | Stop = Colon + 1; | |
553 | for (; Stop < End && isspace_ascii(*Stop) != 0; ++Stop) | |
554 | if (*Stop == '\n' && Stop[1] != ' ') | |
555 | break; | |
556 | if (Stop >= End) | |
557 | return false; | |
558 | lastTagData.StartValue = Stop - Section; | |
559 | } | |
560 | ||
561 | Stop = (const char *)memchr(Stop,'\n',End - Stop); | |
562 | ||
563 | if (Stop == 0) | |
564 | return false; | |
565 | ||
566 | for (; Stop+1 < End && Stop[1] == '\r'; Stop++) | |
567 | /* nothing */ | |
568 | ; | |
569 | ||
570 | // Double newline marks the end of the record | |
571 | if (Stop+1 < End && Stop[1] == '\n') | |
572 | { | |
573 | if (lastTagData.EndTag != 0) | |
574 | { | |
575 | if (AlphaIndexes[lastTagHash] != 0) | |
576 | lastTagData.NextInBucket = AlphaIndexes[lastTagHash]; | |
577 | APT_IGNORE_DEPRECATED(AlphaIndexes[lastTagHash] = TagCount;) | |
578 | d->Tags.push_back(lastTagData); | |
579 | } | |
580 | ||
581 | pkgTagSectionPrivate::TagData const td(Stop - Section); | |
582 | d->Tags.push_back(td); | |
583 | TrimRecord(false,End); | |
584 | return true; | |
585 | } | |
586 | ||
587 | Stop++; | |
588 | } | |
589 | ||
590 | return false; | |
591 | } | |
592 | /*}}}*/ | |
593 | // TagSection::TrimRecord - Trim off any garbage before/after a record /*{{{*/ | |
594 | // --------------------------------------------------------------------- | |
595 | /* There should be exactly 2 newline at the end of the record, no more. */ | |
596 | void pkgTagSection::TrimRecord(bool BeforeRecord, const char*& End) | |
597 | { | |
598 | if (BeforeRecord == true) | |
599 | return; | |
600 | for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++); | |
601 | } | |
602 | /*}}}*/ | |
603 | // TagSection::Trim - Trim off any trailing garbage /*{{{*/ | |
604 | // --------------------------------------------------------------------- | |
605 | /* There should be exactly 1 newline at the end of the buffer, no more. */ | |
606 | void pkgTagSection::Trim() | |
607 | { | |
608 | for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--); | |
609 | } | |
610 | /*}}}*/ | |
611 | // TagSection::Exists - return True if a tag exists /*{{{*/ | |
612 | bool pkgTagSection::Exists(StringView Tag) const | |
613 | { | |
614 | unsigned int tmp; | |
615 | return Find(Tag, tmp); | |
616 | } | |
617 | /*}}}*/ | |
618 | // TagSection::Find - Locate a tag /*{{{*/ | |
619 | // --------------------------------------------------------------------- | |
620 | /* This searches the section for a tag that matches the given string. */ | |
621 | bool pkgTagSection::Find(StringView TagView,unsigned int &Pos) const | |
622 | { | |
623 | const char * const Tag = TagView.data(); | |
624 | size_t const Length = TagView.length(); | |
625 | unsigned int Bucket = AlphaIndexes[AlphaHash(Tag, Length)]; | |
626 | if (Bucket == 0) | |
627 | return false; | |
628 | ||
629 | for (; Bucket != 0; Bucket = d->Tags[Bucket - 1].NextInBucket) | |
630 | { | |
631 | if ((d->Tags[Bucket - 1].EndTag - d->Tags[Bucket - 1].StartTag) != Length) | |
632 | continue; | |
633 | ||
634 | char const * const St = Section + d->Tags[Bucket - 1].StartTag; | |
635 | if (strncasecmp(Tag,St,Length) != 0) | |
636 | continue; | |
637 | ||
638 | Pos = Bucket - 1; | |
639 | return true; | |
640 | } | |
641 | ||
642 | Pos = 0; | |
643 | return false; | |
644 | } | |
645 | ||
646 | bool pkgTagSection::Find(StringView Tag,const char *&Start, | |
647 | const char *&End) const | |
648 | { | |
649 | unsigned int Pos; | |
650 | if (Find(Tag, Pos) == false) | |
651 | return false; | |
652 | ||
653 | Start = Section + d->Tags[Pos].StartValue; | |
654 | // Strip off the gunk from the end | |
655 | End = Section + d->Tags[Pos + 1].StartTag; | |
656 | if (unlikely(Start > End)) | |
657 | return _error->Error("Internal parsing error"); | |
658 | ||
659 | for (; isspace_ascii(End[-1]) != 0 && End > Start; --End); | |
660 | ||
661 | return true; | |
662 | } | |
663 | /*}}}*/ | |
664 | // TagSection::FindS - Find a string /*{{{*/ | |
665 | StringView pkgTagSection::Find(StringView Tag) const | |
666 | { | |
667 | const char *Start; | |
668 | const char *End; | |
669 | if (Find(Tag,Start,End) == false) | |
670 | return StringView(); | |
671 | return StringView(Start, End - Start); | |
672 | } | |
673 | /*}}}*/ | |
674 | // TagSection::FindRawS - Find a string /*{{{*/ | |
675 | StringView pkgTagSection::FindRaw(StringView Tag) const | |
676 | { | |
677 | unsigned int Pos; | |
678 | if (Find(Tag, Pos) == false) | |
679 | return ""; | |
680 | ||
681 | char const *Start = (char const *) memchr(Section + d->Tags[Pos].EndTag, ':', d->Tags[Pos].StartValue - d->Tags[Pos].EndTag); | |
682 | ++Start; | |
683 | char const *End = Section + d->Tags[Pos + 1].StartTag; | |
684 | if (unlikely(Start > End)) | |
685 | return ""; | |
686 | ||
687 | for (; isspace_ascii(End[-1]) != 0 && End > Start; --End); | |
688 | ||
689 | return StringView(Start, End - Start); | |
690 | } | |
691 | /*}}}*/ | |
692 | // TagSection::FindI - Find an integer /*{{{*/ | |
693 | // --------------------------------------------------------------------- | |
694 | /* */ | |
695 | signed int pkgTagSection::FindI(StringView Tag,signed long Default) const | |
696 | { | |
697 | const char *Start; | |
698 | const char *Stop; | |
699 | if (Find(Tag,Start,Stop) == false) | |
700 | return Default; | |
701 | ||
702 | // Copy it into a temp buffer so we can use strtol | |
703 | char S[300]; | |
704 | if ((unsigned)(Stop - Start) >= sizeof(S)) | |
705 | return Default; | |
706 | strncpy(S,Start,Stop-Start); | |
707 | S[Stop - Start] = 0; | |
708 | ||
709 | errno = 0; | |
710 | char *End; | |
711 | signed long Result = strtol(S,&End,10); | |
712 | if (errno == ERANGE || | |
713 | Result < std::numeric_limits<int>::min() || Result > std::numeric_limits<int>::max()) { | |
714 | errno = ERANGE; | |
715 | _error->Error(_("Cannot convert %s to integer: out of range"), S); | |
716 | } | |
717 | if (S == End) | |
718 | return Default; | |
719 | return Result; | |
720 | } | |
721 | /*}}}*/ | |
722 | // TagSection::FindULL - Find an unsigned long long integer /*{{{*/ | |
723 | // --------------------------------------------------------------------- | |
724 | /* */ | |
725 | unsigned long long pkgTagSection::FindULL(StringView Tag, unsigned long long const &Default) const | |
726 | { | |
727 | const char *Start; | |
728 | const char *Stop; | |
729 | if (Find(Tag,Start,Stop) == false) | |
730 | return Default; | |
731 | ||
732 | // Copy it into a temp buffer so we can use strtoull | |
733 | char S[100]; | |
734 | if ((unsigned)(Stop - Start) >= sizeof(S)) | |
735 | return Default; | |
736 | strncpy(S,Start,Stop-Start); | |
737 | S[Stop - Start] = 0; | |
738 | ||
739 | char *End; | |
740 | unsigned long long Result = strtoull(S,&End,10); | |
741 | if (S == End) | |
742 | return Default; | |
743 | return Result; | |
744 | } | |
745 | /*}}}*/ | |
746 | // TagSection::FindB - Find boolean value /*{{{*/ | |
747 | // --------------------------------------------------------------------- | |
748 | /* */ | |
749 | bool pkgTagSection::FindB(StringView Tag, bool Default) const | |
750 | { | |
751 | const char *Start, *Stop; | |
752 | if (Find(Tag, Start, Stop) == false) | |
753 | return Default; | |
754 | return StringToBool(string(Start, Stop)); | |
755 | } | |
756 | /*}}}*/ | |
757 | // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/ | |
758 | // --------------------------------------------------------------------- | |
759 | /* The bits marked in Flag are masked on/off in Flags */ | |
760 | bool pkgTagSection::FindFlag(StringView Tag, uint8_t &Flags, | |
761 | uint8_t const Flag) const | |
762 | { | |
763 | const char *Start; | |
764 | const char *Stop; | |
765 | if (Find(Tag,Start,Stop) == false) | |
766 | return true; | |
767 | return FindFlag(Flags, Flag, Start, Stop); | |
768 | } | |
769 | bool pkgTagSection::FindFlag(uint8_t &Flags, uint8_t const Flag, | |
770 | char const* const Start, char const* const Stop) | |
771 | { | |
772 | switch (StringToBool(string(Start, Stop))) | |
773 | { | |
774 | case 0: | |
775 | Flags &= ~Flag; | |
776 | return true; | |
777 | ||
778 | case 1: | |
779 | Flags |= Flag; | |
780 | return true; | |
781 | ||
782 | default: | |
783 | _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str()); | |
784 | return true; | |
785 | } | |
786 | return true; | |
787 | } | |
788 | bool pkgTagSection::FindFlag(StringView Tag,unsigned long &Flags, | |
789 | unsigned long Flag) const | |
790 | { | |
791 | const char *Start; | |
792 | const char *Stop; | |
793 | if (Find(Tag,Start,Stop) == false) | |
794 | return true; | |
795 | return FindFlag(Flags, Flag, Start, Stop); | |
796 | } | |
797 | bool pkgTagSection::FindFlag(unsigned long &Flags, unsigned long Flag, | |
798 | char const* Start, char const* Stop) | |
799 | { | |
800 | switch (StringToBool(string(Start, Stop))) | |
801 | { | |
802 | case 0: | |
803 | Flags &= ~Flag; | |
804 | return true; | |
805 | ||
806 | case 1: | |
807 | Flags |= Flag; | |
808 | return true; | |
809 | ||
810 | default: | |
811 | _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str()); | |
812 | return true; | |
813 | } | |
814 | return true; | |
815 | } | |
816 | /*}}}*/ | |
817 | void pkgTagSection::Get(const char *&Start,const char *&Stop,unsigned int I) const/*{{{*/ | |
818 | { | |
819 | Start = Section + d->Tags[I].StartTag; | |
820 | Stop = Section + d->Tags[I+1].StartTag; | |
821 | } | |
822 | /*}}}*/ | |
823 | APT_PURE unsigned int pkgTagSection::Count() const { /*{{{*/ | |
824 | if (d->Tags.empty() == true) | |
825 | return 0; | |
826 | // the last element is just marking the end and isn't a real one | |
827 | return d->Tags.size() - 1; | |
828 | } | |
829 | /*}}}*/ | |
830 | // TagSection::Write - Ordered (re)writing of fields /*{{{*/ | |
831 | pkgTagSection::Tag pkgTagSection::Tag::Remove(std::string const &Name) | |
832 | { | |
833 | return Tag(REMOVE, Name, ""); | |
834 | } | |
835 | pkgTagSection::Tag pkgTagSection::Tag::Rename(std::string const &OldName, std::string const &NewName) | |
836 | { | |
837 | return Tag(RENAME, OldName, NewName); | |
838 | } | |
839 | pkgTagSection::Tag pkgTagSection::Tag::Rewrite(std::string const &Name, std::string const &Data) | |
840 | { | |
841 | if (Data.empty() == true) | |
842 | return Tag(REMOVE, Name, ""); | |
843 | else | |
844 | return Tag(REWRITE, Name, Data); | |
845 | } | |
846 | static bool WriteTag(FileFd &File, std::string Tag, StringView Value) | |
847 | { | |
848 | if (Value.empty() || isspace_ascii(Value[0]) != 0) | |
849 | Tag.append(":"); | |
850 | else | |
851 | Tag.append(": "); | |
852 | Tag.append(Value.data(), Value.length()); | |
853 | Tag.append("\n"); | |
854 | return File.Write(Tag.c_str(), Tag.length()); | |
855 | } | |
856 | static bool RewriteTags(FileFd &File, pkgTagSection const * const This, char const * const Tag, | |
857 | std::vector<pkgTagSection::Tag>::const_iterator &R, | |
858 | std::vector<pkgTagSection::Tag>::const_iterator const &REnd) | |
859 | { | |
860 | size_t const TagLen = strlen(Tag); | |
861 | for (; R != REnd; ++R) | |
862 | { | |
863 | std::string data; | |
864 | if (R->Name.length() == TagLen && strncasecmp(R->Name.c_str(), Tag, R->Name.length()) == 0) | |
865 | { | |
866 | if (R->Action != pkgTagSection::Tag::REWRITE) | |
867 | break; | |
868 | data = R->Data; | |
869 | } | |
870 | else if(R->Action == pkgTagSection::Tag::RENAME && R->Data.length() == TagLen && | |
871 | strncasecmp(R->Data.c_str(), Tag, R->Data.length()) == 0) | |
872 | data = This->FindRaw(R->Name.c_str()).to_string(); | |
873 | else | |
874 | continue; | |
875 | ||
876 | return WriteTag(File, Tag, data); | |
877 | } | |
878 | return true; | |
879 | } | |
880 | bool pkgTagSection::Write(FileFd &File, char const * const * const Order, std::vector<Tag> const &Rewrite) const | |
881 | { | |
882 | // first pass: Write everything we have an order for | |
883 | if (Order != NULL) | |
884 | { | |
885 | for (unsigned int I = 0; Order[I] != 0; ++I) | |
886 | { | |
887 | std::vector<Tag>::const_iterator R = Rewrite.begin(); | |
888 | if (RewriteTags(File, this, Order[I], R, Rewrite.end()) == false) | |
889 | return false; | |
890 | if (R != Rewrite.end()) | |
891 | continue; | |
892 | ||
893 | if (Exists(Order[I]) == false) | |
894 | continue; | |
895 | ||
896 | if (WriteTag(File, Order[I], FindRaw(Order[I])) == false) | |
897 | return false; | |
898 | } | |
899 | } | |
900 | // second pass: See if we have tags which aren't ordered | |
901 | if (d->Tags.empty() == false) | |
902 | { | |
903 | for (std::vector<pkgTagSectionPrivate::TagData>::const_iterator T = d->Tags.begin(); T != d->Tags.end() - 1; ++T) | |
904 | { | |
905 | char const * const fieldname = Section + T->StartTag; | |
906 | size_t fieldnamelen = T->EndTag - T->StartTag; | |
907 | if (Order != NULL) | |
908 | { | |
909 | unsigned int I = 0; | |
910 | for (; Order[I] != 0; ++I) | |
911 | { | |
912 | if (fieldnamelen == strlen(Order[I]) && strncasecmp(fieldname, Order[I], fieldnamelen) == 0) | |
913 | break; | |
914 | } | |
915 | if (Order[I] != 0) | |
916 | continue; | |
917 | } | |
918 | ||
919 | std::string const name(fieldname, fieldnamelen); | |
920 | std::vector<Tag>::const_iterator R = Rewrite.begin(); | |
921 | if (RewriteTags(File, this, name.c_str(), R, Rewrite.end()) == false) | |
922 | return false; | |
923 | if (R != Rewrite.end()) | |
924 | continue; | |
925 | ||
926 | if (WriteTag(File, name, FindRaw(name)) == false) | |
927 | return false; | |
928 | } | |
929 | } | |
930 | // last pass: see if there are any rewrites remaining we haven't done yet | |
931 | for (std::vector<Tag>::const_iterator R = Rewrite.begin(); R != Rewrite.end(); ++R) | |
932 | { | |
933 | if (R->Action == Tag::REMOVE) | |
934 | continue; | |
935 | std::string const name = ((R->Action == Tag::RENAME) ? R->Data : R->Name); | |
936 | if (Exists(name.c_str())) | |
937 | continue; | |
938 | if (Order != NULL) | |
939 | { | |
940 | unsigned int I = 0; | |
941 | for (; Order[I] != 0; ++I) | |
942 | { | |
943 | if (strncasecmp(name.c_str(), Order[I], name.length()) == 0 && name.length() == strlen(Order[I])) | |
944 | break; | |
945 | } | |
946 | if (Order[I] != 0) | |
947 | continue; | |
948 | } | |
949 | ||
950 | if (WriteTag(File, name, ((R->Action == Tag::RENAME) ? FindRaw(R->Name) : R->Data)) == false) | |
951 | return false; | |
952 | } | |
953 | return true; | |
954 | } | |
955 | /*}}}*/ | |
956 | ||
957 | void pkgUserTagSection::TrimRecord(bool /*BeforeRecord*/, const char* &End)/*{{{*/ | |
958 | { | |
959 | for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r' || Stop[0] == '#'); Stop++) | |
960 | if (Stop[0] == '#') | |
961 | Stop = (const char*) memchr(Stop,'\n',End-Stop); | |
962 | } | |
963 | /*}}}*/ | |
964 | ||
965 | #include "tagfile-order.c" | |
966 | ||
967 | // TFRewrite - Rewrite a control record /*{{{*/ | |
968 | // --------------------------------------------------------------------- | |
969 | /* This writes the control record to stdout rewriting it as necessary. The | |
970 | override map item specificies the rewriting rules to follow. This also | |
971 | takes the time to sort the feild list. */ | |
972 | APT_IGNORE_DEPRECATED_PUSH | |
973 | bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[], | |
974 | TFRewriteData *Rewrite) | |
975 | { | |
976 | unsigned char Visited[256]; // Bit 1 is Order, Bit 2 is Rewrite | |
977 | for (unsigned I = 0; I != 256; I++) | |
978 | Visited[I] = 0; | |
979 | ||
980 | // Set new tag up as necessary. | |
981 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) | |
982 | { | |
983 | if (Rewrite[J].NewTag == 0) | |
984 | Rewrite[J].NewTag = Rewrite[J].Tag; | |
985 | } | |
986 | ||
987 | // Write all all of the tags, in order. | |
988 | if (Order != NULL) | |
989 | { | |
990 | for (unsigned int I = 0; Order[I] != 0; I++) | |
991 | { | |
992 | bool Rewritten = false; | |
993 | ||
994 | // See if this is a field that needs to be rewritten | |
995 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) | |
996 | { | |
997 | if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0) | |
998 | { | |
999 | Visited[J] |= 2; | |
1000 | if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) | |
1001 | { | |
1002 | if (isspace_ascii(Rewrite[J].Rewrite[0])) | |
1003 | fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); | |
1004 | else | |
1005 | fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); | |
1006 | } | |
1007 | Rewritten = true; | |
1008 | break; | |
1009 | } | |
1010 | } | |
1011 | ||
1012 | // See if it is in the fragment | |
1013 | unsigned Pos; | |
1014 | if (Tags.Find(StringView(Order[I]),Pos) == false) | |
1015 | continue; | |
1016 | Visited[Pos] |= 1; | |
1017 | ||
1018 | if (Rewritten == true) | |
1019 | continue; | |
1020 | ||
1021 | /* Write out this element, taking a moment to rewrite the tag | |
1022 | in case of changes of case. */ | |
1023 | const char *Start; | |
1024 | const char *Stop; | |
1025 | Tags.Get(Start,Stop,Pos); | |
1026 | ||
1027 | if (fputs(Order[I],Output) < 0) | |
1028 | return _error->Errno("fputs","IO Error to output"); | |
1029 | Start += strlen(Order[I]); | |
1030 | if (fwrite(Start,Stop - Start,1,Output) != 1) | |
1031 | return _error->Errno("fwrite","IO Error to output"); | |
1032 | if (Stop[-1] != '\n') | |
1033 | fprintf(Output,"\n"); | |
1034 | } | |
1035 | } | |
1036 | ||
1037 | // Now write all the old tags that were missed. | |
1038 | for (unsigned int I = 0; I != Tags.Count(); I++) | |
1039 | { | |
1040 | if ((Visited[I] & 1) == 1) | |
1041 | continue; | |
1042 | ||
1043 | const char *Start; | |
1044 | const char *Stop; | |
1045 | Tags.Get(Start,Stop,I); | |
1046 | const char *End = Start; | |
1047 | for (; End < Stop && *End != ':'; End++); | |
1048 | ||
1049 | // See if this is a field that needs to be rewritten | |
1050 | bool Rewritten = false; | |
1051 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) | |
1052 | { | |
1053 | if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0) | |
1054 | { | |
1055 | Visited[J] |= 2; | |
1056 | if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) | |
1057 | { | |
1058 | if (isspace_ascii(Rewrite[J].Rewrite[0])) | |
1059 | fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); | |
1060 | else | |
1061 | fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); | |
1062 | } | |
1063 | ||
1064 | Rewritten = true; | |
1065 | break; | |
1066 | } | |
1067 | } | |
1068 | ||
1069 | if (Rewritten == true) | |
1070 | continue; | |
1071 | ||
1072 | // Write out this element | |
1073 | if (fwrite(Start,Stop - Start,1,Output) != 1) | |
1074 | return _error->Errno("fwrite","IO Error to output"); | |
1075 | if (Stop[-1] != '\n') | |
1076 | fprintf(Output,"\n"); | |
1077 | } | |
1078 | ||
1079 | // Now write all the rewrites that were missed | |
1080 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) | |
1081 | { | |
1082 | if ((Visited[J] & 2) == 2) | |
1083 | continue; | |
1084 | ||
1085 | if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) | |
1086 | { | |
1087 | if (isspace_ascii(Rewrite[J].Rewrite[0])) | |
1088 | fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); | |
1089 | else | |
1090 | fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); | |
1091 | } | |
1092 | } | |
1093 | ||
1094 | return true; | |
1095 | } | |
1096 | APT_IGNORE_DEPRECATED_POP | |
1097 | /*}}}*/ | |
1098 | ||
1099 | pkgTagSection::~pkgTagSection() { delete d; } |