Merge with Michael
[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 || Fd.Size() == 0)
39 {
40 Buffer = 0;
41 Start = End = Buffer = 0;
42 iOffset = 0;
43 Map = NULL;
44 return;
45 }
46
47 Map = new MMap (Fd, MMap::Public | MMap::ReadOnly);
48 Buffer = (char *) Map->Data ();
49 Start = Buffer;
50 End = Buffer + Map->Size ();
51 iOffset = 0;
52 }
53 /*}}}*/
54 // TagFile::~pkgTagFile - Destructor /*{{{*/
55 // ---------------------------------------------------------------------
56 /* */
57 pkgTagFile::~pkgTagFile()
58 {
59 delete Map;
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 (Start == End)
68 return false;
69
70 if (Tag.Scan(Start,End - Start) == false)
71 {
72 return _error->Error(_("Unable to parse package file %s (1)"),
73 Fd.Name().c_str());
74 }
75 Start += Tag.size();
76 iOffset += Tag.size();
77
78 Tag.Trim();
79 return true;
80 }
81 /*}}}*/
82 // TagFile::Jump - Jump to a pre-recorded location in the file /*{{{*/
83 // ---------------------------------------------------------------------
84 /* This jumps to a pre-recorded file location and reads the record
85 that is there */
86 bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset)
87 {
88 // We are within a buffer space of the next hit..
89 if (Offset >= iOffset && iOffset + (End - Start) > Offset)
90 {
91 unsigned long Dist = Offset - iOffset;
92 Start += Dist;
93 iOffset += Dist;
94 return Step(Tag);
95 }
96
97 // Reposition and reload..
98 iOffset = Offset;
99 Start = Buffer + iOffset;
100
101 // Start != End is a special case to not fail on empty TagFiles
102 if (Start != End && Tag.Scan(Start,End - Start) == false)
103 return _error->Error(_("Unable to parse package file %s (2)"),Fd.Name().c_str());
104
105 return true;
106 }
107 /*}}}*/
108 // TagSection::Scan - Scan for the end of the header information /*{{{*/
109 // ---------------------------------------------------------------------
110 /* This looks for the first double new line in the data stream. It also
111 indexes the tags in the section. This very simple hash function for the
112 first 3 letters gives very good performance on the debian package files */
113 inline static unsigned long AlphaHash(const char *Text, const char *End = 0)
114 {
115 unsigned long Res = 0;
116 for (; Text != End && *Text != ':' && *Text != 0; Text++)
117 Res = (unsigned long)(*Text) ^ (Res << 2);
118 return Res & 0xFF;
119 }
120
121 bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
122 {
123 const char *End = Start + MaxLength;
124 Stop = Section = Start;
125 memset(AlphaIndexes,0,sizeof(AlphaIndexes));
126
127 if (Stop == 0 || MaxLength == 0)
128 return false;
129
130 TagCount = 0;
131 while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End)
132 {
133 // Start a new index and add it to the hash
134 if (isspace(Stop[0]) == 0)
135 {
136 Indexes[TagCount++] = Stop - Section;
137 AlphaIndexes[AlphaHash(Stop,End)] = TagCount;
138 }
139
140 Stop = (const char *)memchr(Stop,'\n',End - Stop);
141
142 if (Stop == 0)
143 return false;
144
145 for (; Stop+1 < End && Stop[1] == '\r'; Stop++);
146
147 // Double newline marks the end of the record
148 if (Stop+1 < End && Stop[1] == '\n')
149 {
150 Indexes[TagCount] = Stop - Section;
151 for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
152 return true;
153 }
154
155 Stop++;
156 }
157
158 if ((Stop+1 >= End) && (End[-1] == '\n' || End[-1] == '\r'))
159 {
160 Indexes[TagCount] = (End - 1) - Section;
161 return true;
162 }
163
164 return false;
165 }
166 /*}}}*/
167 // TagSection::Trim - Trim off any trailing garbage /*{{{*/
168 // ---------------------------------------------------------------------
169 /* There should be exactly 1 newline at the end of the buffer, no more. */
170 void pkgTagSection::Trim()
171 {
172 for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--);
173 }
174 /*}}}*/
175 // TagSection::Find - Locate a tag /*{{{*/
176 // ---------------------------------------------------------------------
177 /* This searches the section for a tag that matches the given string. */
178 bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const
179 {
180 unsigned int Length = strlen(Tag);
181 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
182 if (I == 0)
183 return false;
184 I--;
185
186 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
187 I = (I+1)%TagCount)
188 {
189 const char *St;
190 St = Section + Indexes[I];
191 if (strncasecmp(Tag,St,Length) != 0)
192 continue;
193
194 // Make sure the colon is in the right place
195 const char *C = St + Length;
196 for (; isspace(*C) != 0; C++);
197 if (*C != ':')
198 continue;
199 Pos = I;
200 return true;
201 }
202
203 Pos = 0;
204 return false;
205 }
206 /*}}}*/
207 // TagSection::Find - Locate a tag /*{{{*/
208 // ---------------------------------------------------------------------
209 /* This searches the section for a tag that matches the given string. */
210 bool pkgTagSection::Find(const char *Tag,const char *&Start,
211 const char *&End) const
212 {
213 unsigned int Length = strlen(Tag);
214 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
215 if (I == 0)
216 return false;
217 I--;
218
219 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
220 I = (I+1)%TagCount)
221 {
222 const char *St;
223 St = Section + Indexes[I];
224 if (strncasecmp(Tag,St,Length) != 0)
225 continue;
226
227 // Make sure the colon is in the right place
228 const char *C = St + Length;
229 for (; isspace(*C) != 0; C++);
230 if (*C != ':')
231 continue;
232
233 // Strip off the gunk from the start end
234 Start = C;
235 End = Section + Indexes[I+1];
236 if (Start >= End)
237 return _error->Error("Internal parsing error");
238
239 for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++);
240 for (; isspace(End[-1]) != 0 && End > Start; End--);
241
242 return true;
243 }
244
245 Start = End = 0;
246 return false;
247 }
248 /*}}}*/
249 // TagSection::FindS - Find a string /*{{{*/
250 // ---------------------------------------------------------------------
251 /* */
252 string pkgTagSection::FindS(const char *Tag) const
253 {
254 const char *Start;
255 const char *End;
256 if (Find(Tag,Start,End) == false)
257 return string();
258 return string(Start,End);
259 }
260 /*}}}*/
261 // TagSection::FindI - Find an integer /*{{{*/
262 // ---------------------------------------------------------------------
263 /* */
264 signed int pkgTagSection::FindI(const char *Tag,signed long Default) const
265 {
266 const char *Start;
267 const char *Stop;
268 if (Find(Tag,Start,Stop) == false)
269 return Default;
270
271 // Copy it into a temp buffer so we can use strtol
272 char S[300];
273 if ((unsigned)(Stop - Start) >= sizeof(S))
274 return Default;
275 strncpy(S,Start,Stop-Start);
276 S[Stop - Start] = 0;
277
278 char *End;
279 signed long Result = strtol(S,&End,10);
280 if (S == End)
281 return Default;
282 return Result;
283 }
284 /*}}}*/
285 // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/
286 // ---------------------------------------------------------------------
287 /* The bits marked in Flag are masked on/off in Flags */
288 bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags,
289 unsigned long Flag) const
290 {
291 const char *Start;
292 const char *Stop;
293 if (Find(Tag,Start,Stop) == false)
294 return true;
295
296 switch (StringToBool(string(Start,Stop)))
297 {
298 case 0:
299 Flags &= ~Flag;
300 return true;
301
302 case 1:
303 Flags |= Flag;
304 return true;
305
306 default:
307 _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str());
308 return true;
309 }
310 return true;
311 }
312 /*}}}*/
313
314 // TFRewrite - Rewrite a control record /*{{{*/
315 // ---------------------------------------------------------------------
316 /* This writes the control record to stdout rewriting it as necessary. The
317 override map item specificies the rewriting rules to follow. This also
318 takes the time to sort the feild list. */
319
320 /* The order of this list is taken from dpkg source lib/parse.c the fieldinfos
321 array. */
322 static const char *iTFRewritePackageOrder[] = {
323 "Package",
324 "Essential",
325 "Status",
326 "Priority",
327 "Section",
328 "Installed-Size",
329 "Maintainer",
330 "Architecture",
331 "Source",
332 "Version",
333 "Revision", // Obsolete
334 "Config-Version", // Obsolete
335 "Replaces",
336 "Provides",
337 "Depends",
338 "Pre-Depends",
339 "Recommends",
340 "Suggests",
341 "Conflicts",
342 "Conffiles",
343 "Filename",
344 "Size",
345 "MD5Sum",
346 "SHA1Sum",
347 "MSDOS-Filename", // Obsolete
348 "Description",
349 0};
350 static const char *iTFRewriteSourceOrder[] = {"Package",
351 "Source",
352 "Binary",
353 "Version",
354 "Priority",
355 "Section",
356 "Maintainer",
357 "Build-Depends",
358 "Build-Depends-Indep",
359 "Build-Conflicts",
360 "Build-Conflicts-Indep",
361 "Architecture",
362 "Standards-Version",
363 "Format",
364 "Directory",
365 "Files",
366 0};
367
368 /* Two levels of initialization are used because gcc will set the symbol
369 size of an array to the length of the array, causing dynamic relinking
370 errors. Doing this makes the symbol size constant */
371 const char **TFRewritePackageOrder = iTFRewritePackageOrder;
372 const char **TFRewriteSourceOrder = iTFRewriteSourceOrder;
373
374 bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[],
375 TFRewriteData *Rewrite)
376 {
377 unsigned char Visited[256]; // Bit 1 is Order, Bit 2 is Rewrite
378 for (unsigned I = 0; I != 256; I++)
379 Visited[I] = 0;
380
381 // Set new tag up as necessary.
382 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
383 {
384 if (Rewrite[J].NewTag == 0)
385 Rewrite[J].NewTag = Rewrite[J].Tag;
386 }
387
388 // Write all all of the tags, in order.
389 for (unsigned int I = 0; Order[I] != 0; I++)
390 {
391 bool Rewritten = false;
392
393 // See if this is a field that needs to be rewritten
394 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
395 {
396 if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0)
397 {
398 Visited[J] |= 2;
399 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
400 {
401 if (isspace(Rewrite[J].Rewrite[0]))
402 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
403 else
404 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
405 }
406
407 Rewritten = true;
408 break;
409 }
410 }
411
412 // See if it is in the fragment
413 unsigned Pos;
414 if (Tags.Find(Order[I],Pos) == false)
415 continue;
416 Visited[Pos] |= 1;
417
418 if (Rewritten == true)
419 continue;
420
421 /* Write out this element, taking a moment to rewrite the tag
422 in case of changes of case. */
423 const char *Start;
424 const char *Stop;
425 Tags.Get(Start,Stop,Pos);
426
427 if (fputs(Order[I],Output) < 0)
428 return _error->Errno("fputs","IO Error to output");
429 Start += strlen(Order[I]);
430 if (fwrite(Start,Stop - Start,1,Output) != 1)
431 return _error->Errno("fwrite","IO Error to output");
432 if (Stop[-1] != '\n')
433 fprintf(Output,"\n");
434 }
435
436 // Now write all the old tags that were missed.
437 for (unsigned int I = 0; I != Tags.Count(); I++)
438 {
439 if ((Visited[I] & 1) == 1)
440 continue;
441
442 const char *Start;
443 const char *Stop;
444 Tags.Get(Start,Stop,I);
445 const char *End = Start;
446 for (; End < Stop && *End != ':'; End++);
447
448 // See if this is a field that needs to be rewritten
449 bool Rewritten = false;
450 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
451 {
452 if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0)
453 {
454 Visited[J] |= 2;
455 if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
456 {
457 if (isspace(Rewrite[J].Rewrite[0]))
458 fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
459 else
460 fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
461 }
462
463 Rewritten = true;
464 break;
465 }
466 }
467
468 if (Rewritten == true)
469 continue;
470
471 // Write out this element
472 if (fwrite(Start,Stop - Start,1,Output) != 1)
473 return _error->Errno("fwrite","IO Error to output");
474 if (Stop[-1] != '\n')
475 fprintf(Output,"\n");
476 }
477
478 // Now write all the rewrites that were missed
479 for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
480 {
481 if ((Visited[J] & 2) == 2)
482 continue;
483
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
493 return true;
494 }
495 /*}}}*/