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