* fix languages like pt_PT, pt_BR, zh_CN by special casing
[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())
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 if (Tag.Scan(Start,End - Start) == false)
102 return _error->Error(_("Unable to parse package file %s (2)"),Fd.Name().c_str());
103
104 return true;
105 }
106 /*}}}*/
107 // TagSection::Scan - Scan for the end of the header information /*{{{*/
108 // ---------------------------------------------------------------------
109 /* This looks for the first double new line in the data stream. It also
110 indexes the tags in the section. This very simple hash function for the
111 first 3 letters gives very good performance on the debian package files */
112 inline static unsigned long AlphaHash(const char *Text, const char *End = 0)
113 {
114 unsigned long Res = 0;
115 for (; Text != End && *Text != ':' && *Text != 0; Text++)
116 Res = (unsigned long)(*Text) ^ (Res << 2);
117 return Res & 0xFF;
118 }
119
120 bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
121 {
122 const char *End = Start + MaxLength;
123 Stop = Section = Start;
124 memset(AlphaIndexes,0,sizeof(AlphaIndexes));
125
126 if (Stop == 0 || MaxLength == 0)
127 return false;
128
129 TagCount = 0;
130 while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End)
131 {
132 // Start a new index and add it to the hash
133 if (isspace(Stop[0]) == 0)
134 {
135 Indexes[TagCount++] = Stop - Section;
136 AlphaIndexes[AlphaHash(Stop,End)] = TagCount;
137 }
138
139 Stop = (const char *)memchr(Stop,'\n',End - Stop);
140
141 if (Stop == 0)
142 return false;
143
144 for (; Stop+1 < End && Stop[1] == '\r'; Stop++);
145
146 // Double newline marks the end of the record
147 if (Stop+1 < End && Stop[1] == '\n')
148 {
149 Indexes[TagCount] = Stop - Section;
150 for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
151 return true;
152 }
153
154 Stop++;
155 }
156
157 if ((Stop+1 >= End) && (End[-1] == '\n' || End[-1] == '\r'))
158 {
159 Indexes[TagCount] = (End - 1) - Section;
160 return true;
161 }
162
163 return false;
164 }
165 /*}}}*/
166 // TagSection::Trim - Trim off any trailing garbage /*{{{*/
167 // ---------------------------------------------------------------------
168 /* There should be exactly 1 newline at the end of the buffer, no more. */
169 void pkgTagSection::Trim()
170 {
171 for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--);
172 }
173 /*}}}*/
174 // TagSection::Find - Locate a tag /*{{{*/
175 // ---------------------------------------------------------------------
176 /* This searches the section for a tag that matches the given string. */
177 bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const
178 {
179 unsigned int Length = strlen(Tag);
180 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
181 if (I == 0)
182 return false;
183 I--;
184
185 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
186 I = (I+1)%TagCount)
187 {
188 const char *St;
189 St = Section + Indexes[I];
190 if (strncasecmp(Tag,St,Length) != 0)
191 continue;
192
193 // Make sure the colon is in the right place
194 const char *C = St + Length;
195 for (; isspace(*C) != 0; C++);
196 if (*C != ':')
197 continue;
198 Pos = I;
199 return true;
200 }
201
202 Pos = 0;
203 return false;
204 }
205 /*}}}*/
206 // TagSection::Find - Locate a tag /*{{{*/
207 // ---------------------------------------------------------------------
208 /* This searches the section for a tag that matches the given string. */
209 bool pkgTagSection::Find(const char *Tag,const char *&Start,
210 const char *&End) const
211 {
212 unsigned int Length = strlen(Tag);
213 unsigned int I = AlphaIndexes[AlphaHash(Tag)];
214 if (I == 0)
215 return false;
216 I--;
217
218 for (unsigned int Counter = 0; Counter != TagCount; Counter++,
219 I = (I+1)%TagCount)
220 {
221 const char *St;
222 St = Section + Indexes[I];
223 if (strncasecmp(Tag,St,Length) != 0)
224 continue;
225
226 // Make sure the colon is in the right place
227 const char *C = St + Length;
228 for (; isspace(*C) != 0; C++);
229 if (*C != ':')
230 continue;
231
232 // Strip off the gunk from the start end
233 Start = C;
234 End = Section + Indexes[I+1];
235 if (Start >= End)
236 return _error->Error("Internal parsing error");
237
238 for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++);
239 for (; isspace(End[-1]) != 0 && End > Start; End--);
240
241 return true;
242 }
243
244 Start = End = 0;
245 return false;
246 }
247 /*}}}*/
248 // TagSection::FindS - Find a string /*{{{*/
249 // ---------------------------------------------------------------------
250 /* */
251 string pkgTagSection::FindS(const char *Tag) const
252 {
253 const char *Start;
254 const char *End;
255 if (Find(Tag,Start,End) == false)
256 return string();
257 return string(Start,End);
258 }
259 /*}}}*/
260 // TagSection::FindI - Find an integer /*{{{*/
261 // ---------------------------------------------------------------------
262 /* */
263 signed int pkgTagSection::FindI(const char *Tag,signed long Default) const
264 {
265 const char *Start;
266 const char *Stop;
267 if (Find(Tag,Start,Stop) == false)
268 return Default;
269
270 // Copy it into a temp buffer so we can use strtol
271 char S[300];
272 if ((unsigned)(Stop - Start) >= sizeof(S))
273 return Default;
274 strncpy(S,Start,Stop-Start);
275 S[Stop - Start] = 0;
276
277 char *End;
278 signed long Result = strtol(S,&End,10);
279 if (S == End)
280 return Default;
281 return Result;
282 }
283 /*}}}*/
284 // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/
285 // ---------------------------------------------------------------------
286 /* The bits marked in Flag are masked on/off in Flags */
287 bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags,
288 unsigned long Flag) const
289 {
290 const char *Start;
291 const char *Stop;
292 if (Find(Tag,Start,Stop) == false)
293 return true;
294
295 switch (StringToBool(string(Start,Stop)))
296 {
297 case 0:
298 Flags &= ~Flag;
299 return true;
300
301 case 1:
302 Flags |= Flag;
303 return true;
304
305 default:
306 _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str());
307 return true;
308 }
309 return true;
310 }
311 /*}}}*/
312
313 // TFRewrite - Rewrite a control record /*{{{*/
314 // ---------------------------------------------------------------------
315 /* This writes the control record to stdout rewriting it as necessary. The
316 override map item specificies the rewriting rules to follow. This also
317 takes the time to sort the feild list. */
318
319 /* The order of this list is taken from dpkg source lib/parse.c the fieldinfos
320 array. */
321 static const char *iTFRewritePackageOrder[] = {
322 "Package",
323 "Essential",
324 "Status",
325 "Priority",
326 "Section",
327 "Installed-Size",
328 "Maintainer",
329 "Architecture",
330 "Source",
331 "Version",
332 "Revision", // Obsolete
333 "Config-Version", // Obsolete
334 "Replaces",
335 "Provides",
336 "Depends",
337 "Pre-Depends",
338 "Recommends",
339 "Suggests",
340 "Conflicts",
341 "Conffiles",
342 "Filename",
343 "Size",
344 "MD5Sum",
345 "SHA1",
346 "SHA256",
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 /*}}}*/