* merged from the apt--auto-mark branch
[ntk/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 #ifdef __GNUG__
15 #pragma implementation "apt-pkg/tagfile.h"
16 #endif
17
18 #include <apt-pkg/tagfile.h>
19 #include <apt-pkg/error.h>
20 #include <apt-pkg/strutl.h>
21
22 #include <apti18n.h>
23
24 #include <string>
25 #include <stdio.h>
26 #include <ctype.h>
27 /*}}}*/
28
29 using std::string;
30
31 // TagFile::pkgTagFile - Constructor /*{{{*/
32 // ---------------------------------------------------------------------
33 /* */
34 pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long Size) :
35 Fd(*pFd),
36 Size(Size)
37 {
38 if (Fd.IsOpen() == false)
39 {
40 Buffer = 0;
41 Start = End = Buffer = 0;
42 Done = true;
43 iOffset = 0;
44 return;
45 }
46
47 Buffer = new char[Size];
48 Start = End = Buffer;
49 Done = false;
50 iOffset = 0;
51 Fill();
52 }
53 /*}}}*/
54 // TagFile::~pkgTagFile - Destructor /*{{{*/
55 // ---------------------------------------------------------------------
56 /* */
57 pkgTagFile::~pkgTagFile()
58 {
59 delete [] Buffer;
60 }
61 /*}}}*/
62 // TagFile::Resize - Resize the internal buffer /*{{{*/
63 // ---------------------------------------------------------------------
64 /* Resize the internal buffer (double it in size). Fail if a maximum size
65 * size is reached.
66 */
67 bool pkgTagFile::Resize()
68 {
69 char *tmp;
70 unsigned long EndSize = End - Start;
71
72 // fail is the buffer grows too big
73 if(Size > 1024*1024+1)
74 return false;
75
76 // get new buffer and use it
77 tmp = new char[2*Size];
78 memcpy(tmp, Buffer, Size);
79 Size = Size*2;
80 delete [] Buffer;
81 Buffer = tmp;
82
83 // update the start/end pointers to the new buffer
84 Start = Buffer;
85 End = Start + EndSize;
86 return true;
87 }
88
89 // TagFile::Step - Advance to the next section /*{{{*/
90 // ---------------------------------------------------------------------
91 /* If the Section Scanner fails we refill the buffer and try again.
92 * If that fails too, double the buffer size and try again until a
93 * maximum buffer is reached.
94 */
95 bool pkgTagFile::Step(pkgTagSection &Tag)
96 {
97 while (Tag.Scan(Start,End - Start) == false)
98 {
99 if (Fill() == false)
100 return false;
101
102 if(Tag.Scan(Start,End - Start))
103 break;
104
105 if (Resize() == false)
106 return _error->Error(_("Unable to parse package file %s (1)"),
107 Fd.Name().c_str());
108 }
109 Start += Tag.size();
110 iOffset += Tag.size();
111
112 Tag.Trim();
113 return true;
114 }
115 /*}}}*/
116 // TagFile::Fill - Top up the buffer /*{{{*/
117 // ---------------------------------------------------------------------
118 /* This takes the bit at the end of the buffer and puts it at the start
119 then fills the rest from the file */
120 bool pkgTagFile::Fill()
121 {
122 unsigned long EndSize = End - Start;
123 unsigned long Actual = 0;
124
125 memmove(Buffer,Start,EndSize);
126 Start = Buffer;
127 End = Buffer + EndSize;
128
129 if (Done == false)
130 {
131 // See if only a bit of the file is left
132 if (Fd.Read(End,Size - (End - Buffer),&Actual) == false)
133 return false;
134 if (Actual != Size - (End - Buffer))
135 Done = true;
136 End += Actual;
137 }
138
139 if (Done == true)
140 {
141 if (EndSize <= 3 && Actual == 0)
142 return false;
143 if (Size - (End - Buffer) < 4)
144 return true;
145
146 // Append a double new line if one does not exist
147 unsigned int LineCount = 0;
148 for (const char *E = End - 1; E - End < 6 && (*E == '\n' || *E == '\r'); E--)
149 if (*E == '\n')
150 LineCount++;
151 for (; LineCount < 2; LineCount++)
152 *End++ = '\n';
153
154 return true;
155 }
156
157 return true;
158 }
159 /*}}}*/
160 // TagFile::Jump - Jump to a pre-recorded location in the file /*{{{*/
161 // ---------------------------------------------------------------------
162 /* This jumps to a pre-recorded file location and reads the record
163 that is there */
164 bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset)
165 {
166 // We are within a buffer space of the next hit..
167 if (Offset >= iOffset && iOffset + (End - Start) > Offset)
168 {
169 unsigned long Dist = Offset - iOffset;
170 Start += Dist;
171 iOffset += Dist;
172 return Step(Tag);
173 }
174
175 // Reposition and reload..
176 iOffset = Offset;
177 Done = false;
178 if (Fd.Seek(Offset) == false)
179 return false;
180 End = Start = Buffer;
181
182 if (Fill() == false)
183 return false;
184
185 if (Tag.Scan(Start,End - Start) == true)
186 return true;
187
188 // This appends a double new line (for the real eof handling)
189 if (Fill() == false)
190 return false;
191
192 if (Tag.Scan(Start,End - Start) == false)
193 return _error->Error(_("Unable to parse package file %s (2)"),Fd.Name().c_str());
194
195 return true;
196 }
197 /*}}}*/
198 // TagSection::Scan - Scan for the end of the header information /*{{{*/
199 // ---------------------------------------------------------------------
200 /* This looks for the first double new line in the data stream. It also
201 indexes the tags in the section. This very simple hash function for the
202 last 8 letters gives very good performance on the debian package files */
203 inline static unsigned long AlphaHash(const char *Text, const char *End = 0)
204 {
205 unsigned long Res = 0;
206 for (; Text != End && *Text != ':' && *Text != 0; Text++)
207 Res = ((unsigned long)(*Text) & 0xDF) ^ (Res << 1);
208 return Res & 0xFF;
209 }
210
211 bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
212 {
213 const char *End = Start + MaxLength;
214 Stop = Section = Start;
215 memset(AlphaIndexes,0,sizeof(AlphaIndexes));
216
217 if (Stop == 0)
218 return false;
219
220 TagCount = 0;
221 while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End)
222 {
223 // Start a new index and add it to the hash
224 if (isspace(Stop[0]) == 0)
225 {
226 Indexes[TagCount++] = Stop - Section;
227 AlphaIndexes[AlphaHash(Stop,End)] = TagCount;
228 }
229
230 Stop = (const char *)memchr(Stop,'\n',End - Stop);
231
232 if (Stop == 0)
233 return false;
234
235 for (; Stop+1 < End && Stop[1] == '\r'; Stop++);
236
237 // Double newline marks the end of the record
238 if (Stop+1 < End && Stop[1] == '\n')
239 {
240 Indexes[TagCount] = Stop - Section;
241 for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
242 return true;
243 }
244
245 Stop++;
246 }
247
248 return false;
249 }
250 /*}}}*/
251 // TagSection::Trim - Trim off any trailing garbage /*{{{*/
252 // ---------------------------------------------------------------------
253 /* There should be exactly 1 newline at the end of the buffer, no more. */
254 void pkgTagSection::Trim()
255 {
256 for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--);
257 }
258 /*}}}*/
259 // TagSection::Find - Locate a tag /*{{{*/
260 // ---------------------------------------------------------------------
261 /* This searches the section for a tag that matches the given string. */
262 bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const
263 {
264 unsigned int Length = strlen(Tag);
265 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
266 if (I == 0)
267 return false;
268 I--;
269
270 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
271 I = (I+1)%TagCount)
272 {
273 const char *St;
274 St = Section + Indexes[I];
275 if (strncasecmp(Tag,St,Length) != 0)
276 continue;
277
278 // Make sure the colon is in the right place
279 const char *C = St + Length;
280 for (; isspace(*C) != 0; C++);
281 if (*C != ':')
282 continue;
283 Pos = I;
284 return true;
285 }
286
287 Pos = 0;
288 return false;
289 }
290 /*}}}*/
291 // TagSection::Find - Locate a tag /*{{{*/
292 // ---------------------------------------------------------------------
293 /* This searches the section for a tag that matches the given string. */
294 bool pkgTagSection::Find(const char *Tag,const char *&Start,
295 const char *&End) const
296 {
297 unsigned int Length = strlen(Tag);
298 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
299 if (I == 0)
300 return false;
301 I--;
302
303 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
304 I = (I+1)%TagCount)
305 {
306 const char *St;
307 St = Section + Indexes[I];
308 if (strncasecmp(Tag,St,Length) != 0)
309 continue;
310
311 // Make sure the colon is in the right place
312 const char *C = St + Length;
313 for (; isspace(*C) != 0; C++);
314 if (*C != ':')
315 continue;
316
317 // Strip off the gunk from the start end
318 Start = C;
319 End = Section + Indexes[I+1];
320 if (Start >= End)
321 return _error->Error("Internal parsing error");
322
323 for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++);
324 for (; isspace(End[-1]) != 0 && End > Start; End--);
325
326 return true;
327 }
328
329 Start = End = 0;
330 return false;
331 }
332 /*}}}*/
333 // TagSection::FindS - Find a string /*{{{*/
334 // ---------------------------------------------------------------------
335 /* */
336 string pkgTagSection::FindS(const char *Tag) const
337 {
338 const char *Start;
339 const char *End;
340 if (Find(Tag,Start,End) == false)
341 return string();
342 return string(Start,End);
343 }
344 /*}}}*/
345 // TagSection::FindI - Find an integer /*{{{*/
346 // ---------------------------------------------------------------------
347 /* */
348 signed int pkgTagSection::FindI(const char *Tag,signed long Default) const
349 {
350 const char *Start;
351 const char *Stop;
352 if (Find(Tag,Start,Stop) == false)
353 return Default;
354
355 // Copy it into a temp buffer so we can use strtol
356 char S[300];
357 if ((unsigned)(Stop - Start) >= sizeof(S))
358 return Default;
359 strncpy(S,Start,Stop-Start);
360 S[Stop - Start] = 0;
361
362 char *End;
363 signed long Result = strtol(S,&End,10);
364 if (S == End)
365 return Default;
366 return Result;
367 }
368 /*}}}*/
369 // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/
370 // ---------------------------------------------------------------------
371 /* The bits marked in Flag are masked on/off in Flags */
372 bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags,
373 unsigned long Flag) const
374 {
375 const char *Start;
376 const char *Stop;
377 if (Find(Tag,Start,Stop) == false)
378 return true;
379
380 switch (StringToBool(string(Start,Stop)))
381 {
382 case 0:
383 Flags &= ~Flag;
384 return true;
385
386 case 1:
387 Flags |= Flag;
388 return true;
389
390 default:
391 _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str());
392 return true;
393 }
394 return true;
395 }
396 /*}}}*/
397
398 // TFRewrite - Rewrite a control record /*{{{*/
399 // ---------------------------------------------------------------------
400 /* This writes the control record to stdout rewriting it as necessary. The
401 override map item specificies the rewriting rules to follow. This also
402 takes the time to sort the feild list. */
403
404 /* The order of this list is taken from dpkg source lib/parse.c the fieldinfos
405 array. */
406 static const char *iTFRewritePackageOrder[] = {
407 "Package",
408 "Essential",
409 "Status",
410 "Priority",
411 "Section",
412 "Installed-Size",
413 "Maintainer",
414 "Architecture",
415 "Source",
416 "Version",
417 "Revision", // Obsolete
418 "Config-Version", // Obsolete
419 "Replaces",
420 "Provides",
421 "Depends",
422 "Pre-Depends",
423 "Recommends",
424 "Suggests",
425 "Conflicts",
426 "Conffiles",
427 "Filename",
428 "Size",
429 "MD5Sum",
430 "SHA1",
431 "SHA256",
432 "MSDOS-Filename", // Obsolete
433 "Description",
434 0};
435 static const char *iTFRewriteSourceOrder[] = {"Package",
436 "Source",
437 "Binary",
438 "Version",
439 "Priority",
440 "Section",
441 "Maintainer",
442 "Build-Depends",
443 "Build-Depends-Indep",
444 "Build-Conflicts",
445 "Build-Conflicts-Indep",
446 "Architecture",
447 "Standards-Version",
448 "Format",
449 "Directory",
450 "Files",
451 0};
452
453 /* Two levels of initialization are used because gcc will set the symbol
454 size of an array to the length of the array, causing dynamic relinking
455 errors. Doing this makes the symbol size constant */
456 const char **TFRewritePackageOrder = iTFRewritePackageOrder;
457 const char **TFRewriteSourceOrder = iTFRewriteSourceOrder;
458
459 bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[],
460 TFRewriteData *Rewrite)
461 {
462 unsigned char Visited[256]; // Bit 1 is Order, Bit 2 is Rewrite
463 for (unsigned I = 0; I != 256; I++)
464 Visited[I] = 0;
465
466 // Set new tag up as necessary.
467 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
468 {
469 if (Rewrite[J].NewTag == 0)
470 Rewrite[J].NewTag = Rewrite[J].Tag;
471 }
472
473 // Write all all of the tags, in order.
474 for (unsigned int I = 0; Order[I] != 0; I++)
475 {
476 bool Rewritten = false;
477
478 // See if this is a field that needs to be rewritten
479 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
480 {
481 if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0)
482 {
483 Visited[J] |= 2;
484 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
485 {
486 if (isspace(Rewrite[J].Rewrite[0]))
487 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
488 else
489 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
490 }
491
492 Rewritten = true;
493 break;
494 }
495 }
496
497 // See if it is in the fragment
498 unsigned Pos;
499 if (Tags.Find(Order[I],Pos) == false)
500 continue;
501 Visited[Pos] |= 1;
502
503 if (Rewritten == true)
504 continue;
505
506 /* Write out this element, taking a moment to rewrite the tag
507 in case of changes of case. */
508 const char *Start;
509 const char *Stop;
510 Tags.Get(Start,Stop,Pos);
511
512 if (fputs(Order[I],Output) < 0)
513 return _error->Errno("fputs","IO Error to output");
514 Start += strlen(Order[I]);
515 if (fwrite(Start,Stop - Start,1,Output) != 1)
516 return _error->Errno("fwrite","IO Error to output");
517 if (Stop[-1] != '\n')
518 fprintf(Output,"\n");
519 }
520
521 // Now write all the old tags that were missed.
522 for (unsigned int I = 0; I != Tags.Count(); I++)
523 {
524 if ((Visited[I] & 1) == 1)
525 continue;
526
527 const char *Start;
528 const char *Stop;
529 Tags.Get(Start,Stop,I);
530 const char *End = Start;
531 for (; End < Stop && *End != ':'; End++);
532
533 // See if this is a field that needs to be rewritten
534 bool Rewritten = false;
535 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
536 {
537 if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0)
538 {
539 Visited[J] |= 2;
540 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
541 {
542 if (isspace(Rewrite[J].Rewrite[0]))
543 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
544 else
545 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
546 }
547
548 Rewritten = true;
549 break;
550 }
551 }
552
553 if (Rewritten == true)
554 continue;
555
556 // Write out this element
557 if (fwrite(Start,Stop - Start,1,Output) != 1)
558 return _error->Errno("fwrite","IO Error to output");
559 if (Stop[-1] != '\n')
560 fprintf(Output,"\n");
561 }
562
563 // Now write all the rewrites that were missed
564 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
565 {
566 if ((Visited[J] & 2) == 2)
567 continue;
568
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
578 return true;
579 }
580 /*}}}*/