]> git.saurik.com Git - apt.git/blob - apt-pkg/tagfile.cc
* cmdline/apt-get.cc:
[apt.git] / apt-pkg / tagfile.cc
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 <apt-pkg/tagfile.h>
15 #include <apt-pkg/error.h>
16 #include <apt-pkg/strutl.h>
17
18 #include <apti18n.h>
19
20 #include <string>
21 #include <stdio.h>
22 #include <ctype.h>
23 /*}}}*/
24
25 using std::string;
26
27 // TagFile::pkgTagFile - Constructor /*{{{*/
28 // ---------------------------------------------------------------------
29 /* */
30 pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long Size) :
31 Fd(*pFd),
32 Size(Size)
33 {
34 if (Fd.IsOpen() == false)
35 {
36 Buffer = 0;
37 Start = End = Buffer = 0;
38 Done = true;
39 iOffset = 0;
40 return;
41 }
42
43 Buffer = new char[Size];
44 Start = End = Buffer;
45 Done = false;
46 iOffset = 0;
47 Fill();
48 }
49 /*}}}*/
50 // TagFile::~pkgTagFile - Destructor /*{{{*/
51 // ---------------------------------------------------------------------
52 /* */
53 pkgTagFile::~pkgTagFile()
54 {
55 delete [] Buffer;
56 }
57 /*}}}*/
58 // TagFile::Resize - Resize the internal buffer /*{{{*/
59 // ---------------------------------------------------------------------
60 /* Resize the internal buffer (double it in size). Fail if a maximum size
61 * size is reached.
62 */
63 bool pkgTagFile::Resize()
64 {
65 char *tmp;
66 unsigned long EndSize = End - Start;
67
68 // fail is the buffer grows too big
69 if(Size > 1024*1024+1)
70 return false;
71
72 // get new buffer and use it
73 tmp = new char[2*Size];
74 memcpy(tmp, Buffer, Size);
75 Size = Size*2;
76 delete [] Buffer;
77 Buffer = tmp;
78
79 // update the start/end pointers to the new buffer
80 Start = Buffer;
81 End = Start + EndSize;
82 return true;
83 }
84 /*}}}*/
85 // TagFile::Step - Advance to the next section /*{{{*/
86 // ---------------------------------------------------------------------
87 /* If the Section Scanner fails we refill the buffer and try again.
88 * If that fails too, double the buffer size and try again until a
89 * maximum buffer is reached.
90 */
91 bool pkgTagFile::Step(pkgTagSection &Tag)
92 {
93 while (Tag.Scan(Start,End - Start) == false)
94 {
95 if (Fill() == false)
96 return false;
97
98 if(Tag.Scan(Start,End - Start))
99 break;
100
101 if (Resize() == false)
102 return _error->Error(_("Unable to parse package file %s (1)"),
103 Fd.Name().c_str());
104 }
105 Start += Tag.size();
106 iOffset += Tag.size();
107
108 Tag.Trim();
109 return true;
110 }
111 /*}}}*/
112 // TagFile::Fill - Top up the buffer /*{{{*/
113 // ---------------------------------------------------------------------
114 /* This takes the bit at the end of the buffer and puts it at the start
115 then fills the rest from the file */
116 bool pkgTagFile::Fill()
117 {
118 unsigned long EndSize = End - Start;
119 unsigned long Actual = 0;
120
121 memmove(Buffer,Start,EndSize);
122 Start = Buffer;
123 End = Buffer + EndSize;
124
125 if (Done == false)
126 {
127 // See if only a bit of the file is left
128 if (Fd.Read(End,Size - (End - Buffer),&Actual) == false)
129 return false;
130 if (Actual != Size - (End - Buffer))
131 Done = true;
132 End += Actual;
133 }
134
135 if (Done == true)
136 {
137 if (EndSize <= 3 && Actual == 0)
138 return false;
139 if (Size - (End - Buffer) < 4)
140 return true;
141
142 // Append a double new line if one does not exist
143 unsigned int LineCount = 0;
144 for (const char *E = End - 1; E - End < 6 && (*E == '\n' || *E == '\r'); E--)
145 if (*E == '\n')
146 LineCount++;
147 for (; LineCount < 2; LineCount++)
148 *End++ = '\n';
149
150 return true;
151 }
152
153 return true;
154 }
155 /*}}}*/
156 // TagFile::Jump - Jump to a pre-recorded location in the file /*{{{*/
157 // ---------------------------------------------------------------------
158 /* This jumps to a pre-recorded file location and reads the record
159 that is there */
160 bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset)
161 {
162 // We are within a buffer space of the next hit..
163 if (Offset >= iOffset && iOffset + (End - Start) > Offset)
164 {
165 unsigned long Dist = Offset - iOffset;
166 Start += Dist;
167 iOffset += Dist;
168 return Step(Tag);
169 }
170
171 // Reposition and reload..
172 iOffset = Offset;
173 Done = false;
174 if (Fd.Seek(Offset) == false)
175 return false;
176 End = Start = Buffer;
177
178 if (Fill() == false)
179 return false;
180
181 if (Tag.Scan(Start,End - Start) == true)
182 return true;
183
184 // This appends a double new line (for the real eof handling)
185 if (Fill() == false)
186 return false;
187
188 if (Tag.Scan(Start,End - Start) == false)
189 return _error->Error(_("Unable to parse package file %s (2)"),Fd.Name().c_str());
190
191 return true;
192 }
193 /*}}}*/
194 // TagSection::Scan - Scan for the end of the header information /*{{{*/
195 // ---------------------------------------------------------------------
196 /* This looks for the first double new line in the data stream.
197 It also indexes the tags in the section. */
198 bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
199 {
200 const char *End = Start + MaxLength;
201 Stop = Section = Start;
202 memset(AlphaIndexes,0,sizeof(AlphaIndexes));
203
204 if (Stop == 0)
205 return false;
206
207 TagCount = 0;
208 while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End)
209 {
210 TrimRecord(true,End);
211
212 // Start a new index and add it to the hash
213 if (isspace(Stop[0]) == 0)
214 {
215 Indexes[TagCount++] = Stop - Section;
216 AlphaIndexes[AlphaHash(Stop,End)] = TagCount;
217 }
218
219 Stop = (const char *)memchr(Stop,'\n',End - Stop);
220
221 if (Stop == 0)
222 return false;
223
224 for (; Stop+1 < End && Stop[1] == '\r'; Stop++);
225
226 // Double newline marks the end of the record
227 if (Stop+1 < End && Stop[1] == '\n')
228 {
229 Indexes[TagCount] = Stop - Section;
230 TrimRecord(false,End);
231 return true;
232 }
233
234 Stop++;
235 }
236
237 return false;
238 }
239 /*}}}*/
240 // TagSection::TrimRecord - Trim off any garbage before/after a record /*{{{*/
241 // ---------------------------------------------------------------------
242 /* There should be exactly 2 newline at the end of the record, no more. */
243 void pkgTagSection::TrimRecord(bool BeforeRecord, const char*& End)
244 {
245 if (BeforeRecord == true)
246 return;
247 for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
248 }
249 /*}}}*/
250 // TagSection::Trim - Trim off any trailing garbage /*{{{*/
251 // ---------------------------------------------------------------------
252 /* There should be exactly 1 newline at the end of the buffer, no more. */
253 void pkgTagSection::Trim()
254 {
255 for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--);
256 }
257 /*}}}*/
258 // TagSection::Find - Locate a tag /*{{{*/
259 // ---------------------------------------------------------------------
260 /* This searches the section for a tag that matches the given string. */
261 bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const
262 {
263 unsigned int Length = strlen(Tag);
264 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
265 if (I == 0)
266 return false;
267 I--;
268
269 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
270 I = (I+1)%TagCount)
271 {
272 const char *St;
273 St = Section + Indexes[I];
274 if (strncasecmp(Tag,St,Length) != 0)
275 continue;
276
277 // Make sure the colon is in the right place
278 const char *C = St + Length;
279 for (; isspace(*C) != 0; C++);
280 if (*C != ':')
281 continue;
282 Pos = I;
283 return true;
284 }
285
286 Pos = 0;
287 return false;
288 }
289 /*}}}*/
290 // TagSection::Find - Locate a tag /*{{{*/
291 // ---------------------------------------------------------------------
292 /* This searches the section for a tag that matches the given string. */
293 bool pkgTagSection::Find(const char *Tag,const char *&Start,
294 const char *&End) const
295 {
296 unsigned int Length = strlen(Tag);
297 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
298 if (I == 0)
299 return false;
300 I--;
301
302 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
303 I = (I+1)%TagCount)
304 {
305 const char *St;
306 St = Section + Indexes[I];
307 if (strncasecmp(Tag,St,Length) != 0)
308 continue;
309
310 // Make sure the colon is in the right place
311 const char *C = St + Length;
312 for (; isspace(*C) != 0; C++);
313 if (*C != ':')
314 continue;
315
316 // Strip off the gunk from the start end
317 Start = C;
318 End = Section + Indexes[I+1];
319 if (Start >= End)
320 return _error->Error("Internal parsing error");
321
322 for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++);
323 for (; isspace(End[-1]) != 0 && End > Start; End--);
324
325 return true;
326 }
327
328 Start = End = 0;
329 return false;
330 }
331 /*}}}*/
332 // TagSection::FindS - Find a string /*{{{*/
333 // ---------------------------------------------------------------------
334 /* */
335 string pkgTagSection::FindS(const char *Tag) const
336 {
337 const char *Start;
338 const char *End;
339 if (Find(Tag,Start,End) == false)
340 return string();
341 return string(Start,End);
342 }
343 /*}}}*/
344 // TagSection::FindI - Find an integer /*{{{*/
345 // ---------------------------------------------------------------------
346 /* */
347 signed int pkgTagSection::FindI(const char *Tag,signed long Default) const
348 {
349 const char *Start;
350 const char *Stop;
351 if (Find(Tag,Start,Stop) == false)
352 return Default;
353
354 // Copy it into a temp buffer so we can use strtol
355 char S[300];
356 if ((unsigned)(Stop - Start) >= sizeof(S))
357 return Default;
358 strncpy(S,Start,Stop-Start);
359 S[Stop - Start] = 0;
360
361 char *End;
362 signed long Result = strtol(S,&End,10);
363 if (S == End)
364 return Default;
365 return Result;
366 }
367 /*}}}*/
368 // TagSection::FindULL - Find an unsigned long long integer /*{{{*/
369 // ---------------------------------------------------------------------
370 /* */
371 unsigned long long pkgTagSection::FindULL(const char *Tag, unsigned long long const &Default) const
372 {
373 const char *Start;
374 const char *Stop;
375 if (Find(Tag,Start,Stop) == false)
376 return Default;
377
378 // Copy it into a temp buffer so we can use strtoull
379 char S[100];
380 if ((unsigned)(Stop - Start) >= sizeof(S))
381 return Default;
382 strncpy(S,Start,Stop-Start);
383 S[Stop - Start] = 0;
384
385 char *End;
386 unsigned long long Result = strtoull(S,&End,10);
387 if (S == End)
388 return Default;
389 return Result;
390 }
391 /*}}}*/
392 // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/
393 // ---------------------------------------------------------------------
394 /* The bits marked in Flag are masked on/off in Flags */
395 bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags,
396 unsigned long Flag) const
397 {
398 const char *Start;
399 const char *Stop;
400 if (Find(Tag,Start,Stop) == false)
401 return true;
402 return FindFlag(Flags, Flag, Start, Stop);
403 }
404 bool const pkgTagSection::FindFlag(unsigned long &Flags, unsigned long Flag,
405 char const* Start, char const* Stop)
406 {
407 switch (StringToBool(string(Start, Stop)))
408 {
409 case 0:
410 Flags &= ~Flag;
411 return true;
412
413 case 1:
414 Flags |= Flag;
415 return true;
416
417 default:
418 _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str());
419 return true;
420 }
421 return true;
422 }
423 /*}}}*/
424 // TFRewrite - Rewrite a control record /*{{{*/
425 // ---------------------------------------------------------------------
426 /* This writes the control record to stdout rewriting it as necessary. The
427 override map item specificies the rewriting rules to follow. This also
428 takes the time to sort the feild list. */
429
430 /* The order of this list is taken from dpkg source lib/parse.c the fieldinfos
431 array. */
432 static const char *iTFRewritePackageOrder[] = {
433 "Package",
434 "Essential",
435 "Status",
436 "Priority",
437 "Section",
438 "Installed-Size",
439 "Maintainer",
440 "Original-Maintainer",
441 "Architecture",
442 "Source",
443 "Version",
444 "Revision", // Obsolete
445 "Config-Version", // Obsolete
446 "Replaces",
447 "Provides",
448 "Depends",
449 "Pre-Depends",
450 "Recommends",
451 "Suggests",
452 "Conflicts",
453 "Breaks",
454 "Conffiles",
455 "Filename",
456 "Size",
457 "MD5Sum",
458 "SHA1",
459 "SHA256",
460 "MSDOS-Filename", // Obsolete
461 "Description",
462 0};
463 static const char *iTFRewriteSourceOrder[] = {"Package",
464 "Source",
465 "Binary",
466 "Version",
467 "Priority",
468 "Section",
469 "Maintainer",
470 "Original-Maintainer",
471 "Build-Depends",
472 "Build-Depends-Indep",
473 "Build-Conflicts",
474 "Build-Conflicts-Indep",
475 "Architecture",
476 "Standards-Version",
477 "Format",
478 "Directory",
479 "Files",
480 0};
481
482 /* Two levels of initialization are used because gcc will set the symbol
483 size of an array to the length of the array, causing dynamic relinking
484 errors. Doing this makes the symbol size constant */
485 const char **TFRewritePackageOrder = iTFRewritePackageOrder;
486 const char **TFRewriteSourceOrder = iTFRewriteSourceOrder;
487
488 bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[],
489 TFRewriteData *Rewrite)
490 {
491 unsigned char Visited[256]; // Bit 1 is Order, Bit 2 is Rewrite
492 for (unsigned I = 0; I != 256; I++)
493 Visited[I] = 0;
494
495 // Set new tag up as necessary.
496 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
497 {
498 if (Rewrite[J].NewTag == 0)
499 Rewrite[J].NewTag = Rewrite[J].Tag;
500 }
501
502 // Write all all of the tags, in order.
503 for (unsigned int I = 0; Order[I] != 0; I++)
504 {
505 bool Rewritten = false;
506
507 // See if this is a field that needs to be rewritten
508 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
509 {
510 if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0)
511 {
512 Visited[J] |= 2;
513 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
514 {
515 if (isspace(Rewrite[J].Rewrite[0]))
516 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
517 else
518 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
519 }
520
521 Rewritten = true;
522 break;
523 }
524 }
525
526 // See if it is in the fragment
527 unsigned Pos;
528 if (Tags.Find(Order[I],Pos) == false)
529 continue;
530 Visited[Pos] |= 1;
531
532 if (Rewritten == true)
533 continue;
534
535 /* Write out this element, taking a moment to rewrite the tag
536 in case of changes of case. */
537 const char *Start;
538 const char *Stop;
539 Tags.Get(Start,Stop,Pos);
540
541 if (fputs(Order[I],Output) < 0)
542 return _error->Errno("fputs","IO Error to output");
543 Start += strlen(Order[I]);
544 if (fwrite(Start,Stop - Start,1,Output) != 1)
545 return _error->Errno("fwrite","IO Error to output");
546 if (Stop[-1] != '\n')
547 fprintf(Output,"\n");
548 }
549
550 // Now write all the old tags that were missed.
551 for (unsigned int I = 0; I != Tags.Count(); I++)
552 {
553 if ((Visited[I] & 1) == 1)
554 continue;
555
556 const char *Start;
557 const char *Stop;
558 Tags.Get(Start,Stop,I);
559 const char *End = Start;
560 for (; End < Stop && *End != ':'; End++);
561
562 // See if this is a field that needs to be rewritten
563 bool Rewritten = false;
564 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
565 {
566 if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0)
567 {
568 Visited[J] |= 2;
569 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
570 {
571 if (isspace(Rewrite[J].Rewrite[0]))
572 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
573 else
574 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
575 }
576
577 Rewritten = true;
578 break;
579 }
580 }
581
582 if (Rewritten == true)
583 continue;
584
585 // Write out this element
586 if (fwrite(Start,Stop - Start,1,Output) != 1)
587 return _error->Errno("fwrite","IO Error to output");
588 if (Stop[-1] != '\n')
589 fprintf(Output,"\n");
590 }
591
592 // Now write all the rewrites that were missed
593 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
594 {
595 if ((Visited[J] & 2) == 2)
596 continue;
597
598 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
599 {
600 if (isspace(Rewrite[J].Rewrite[0]))
601 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
602 else
603 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
604 }
605 }
606
607 return true;
608 }
609 /*}}}*/