merge 'after squeeze release'-stuff
[ntk/apt.git] / apt-pkg / contrib / strutl.cc
1 // -*- mode: cpp; mode: fold -*-
2 // Description /*{{{*/
3 // $Id: strutl.cc,v 1.48 2003/07/18 14:15:11 mdz Exp $
4 /* ######################################################################
5
6 String Util - Some useful string functions.
7
8 These have been collected from here and there to do all sorts of useful
9 things to strings. They are useful in file parsers, URI handlers and
10 especially in APT methods.
11
12 This source is placed in the Public Domain, do with it what you will
13 It was originally written by Jason Gunthorpe <jgg@gpu.srv.ualberta.ca>
14
15 ##################################################################### */
16 /*}}}*/
17 // Includes /*{{{*/
18 #include <apt-pkg/strutl.h>
19 #include <apt-pkg/fileutl.h>
20 #include <apt-pkg/error.h>
21
22 #include <apti18n.h>
23
24 #include <ctype.h>
25 #include <string.h>
26 #include <stdio.h>
27 #include <algorithm>
28 #include <unistd.h>
29 #include <regex.h>
30 #include <errno.h>
31 #include <stdarg.h>
32 #include <iconv.h>
33
34 #include "config.h"
35
36 using namespace std;
37 /*}}}*/
38
39 // UTF8ToCodeset - Convert some UTF-8 string for some codeset /*{{{*/
40 // ---------------------------------------------------------------------
41 /* This is handy to use before display some information for enduser */
42 bool UTF8ToCodeset(const char *codeset, const string &orig, string *dest)
43 {
44 iconv_t cd;
45 const char *inbuf;
46 char *inptr, *outbuf;
47 size_t insize, bufsize;
48 dest->clear();
49
50 cd = iconv_open(codeset, "UTF-8");
51 if (cd == (iconv_t)(-1)) {
52 // Something went wrong
53 if (errno == EINVAL)
54 _error->Error("conversion from 'UTF-8' to '%s' not available",
55 codeset);
56 else
57 perror("iconv_open");
58
59 return false;
60 }
61
62 insize = bufsize = orig.size();
63 inbuf = orig.data();
64 inptr = (char *)inbuf;
65 outbuf = new char[bufsize];
66 size_t lastError = -1;
67
68 while (insize != 0)
69 {
70 char *outptr = outbuf;
71 size_t outsize = bufsize;
72 size_t const err = iconv(cd, &inptr, &insize, &outptr, &outsize);
73 dest->append(outbuf, outptr - outbuf);
74 if (err == (size_t)(-1))
75 {
76 switch (errno)
77 {
78 case EILSEQ:
79 insize--;
80 inptr++;
81 // replace a series of unknown multibytes with a single "?"
82 if (lastError != insize) {
83 lastError = insize - 1;
84 dest->append("?");
85 }
86 break;
87 case EINVAL:
88 insize = 0;
89 break;
90 case E2BIG:
91 if (outptr == outbuf)
92 {
93 bufsize *= 2;
94 delete[] outbuf;
95 outbuf = new char[bufsize];
96 }
97 break;
98 }
99 }
100 }
101
102 delete[] outbuf;
103
104 iconv_close(cd);
105
106 return true;
107 }
108 /*}}}*/
109 // strstrip - Remove white space from the front and back of a string /*{{{*/
110 // ---------------------------------------------------------------------
111 /* This is handy to use when parsing a file. It also removes \n's left
112 over from fgets and company */
113 char *_strstrip(char *String)
114 {
115 for (;*String != 0 && (*String == ' ' || *String == '\t'); String++);
116
117 if (*String == 0)
118 return String;
119
120 char *End = String + strlen(String) - 1;
121 for (;End != String - 1 && (*End == ' ' || *End == '\t' || *End == '\n' ||
122 *End == '\r'); End--);
123 End++;
124 *End = 0;
125 return String;
126 };
127 /*}}}*/
128 // strtabexpand - Converts tabs into 8 spaces /*{{{*/
129 // ---------------------------------------------------------------------
130 /* */
131 char *_strtabexpand(char *String,size_t Len)
132 {
133 for (char *I = String; I != I + Len && *I != 0; I++)
134 {
135 if (*I != '\t')
136 continue;
137 if (I + 8 > String + Len)
138 {
139 *I = 0;
140 return String;
141 }
142
143 /* Assume the start of the string is 0 and find the next 8 char
144 division */
145 int Len;
146 if (String == I)
147 Len = 1;
148 else
149 Len = 8 - ((String - I) % 8);
150 Len -= 2;
151 if (Len <= 0)
152 {
153 *I = ' ';
154 continue;
155 }
156
157 memmove(I + Len,I + 1,strlen(I) + 1);
158 for (char *J = I; J + Len != I; *I = ' ', I++);
159 }
160 return String;
161 }
162 /*}}}*/
163 // ParseQuoteWord - Parse a single word out of a string /*{{{*/
164 // ---------------------------------------------------------------------
165 /* This grabs a single word, converts any % escaped characters to their
166 proper values and advances the pointer. Double quotes are understood
167 and striped out as well. This is for URI/URL parsing. It also can
168 understand [] brackets.*/
169 bool ParseQuoteWord(const char *&String,string &Res)
170 {
171 // Skip leading whitespace
172 const char *C = String;
173 for (;*C != 0 && *C == ' '; C++);
174 if (*C == 0)
175 return false;
176
177 // Jump to the next word
178 for (;*C != 0 && isspace(*C) == 0; C++)
179 {
180 if (*C == '"')
181 {
182 for (C++; *C != 0 && *C != '"'; C++);
183 if (*C == 0)
184 return false;
185 }
186 if (*C == '[')
187 {
188 for (C++; *C != 0 && *C != ']'; C++);
189 if (*C == 0)
190 return false;
191 }
192 }
193
194 // Now de-quote characters
195 char Buffer[1024];
196 char Tmp[3];
197 const char *Start = String;
198 char *I;
199 for (I = Buffer; I < Buffer + sizeof(Buffer) && Start != C; I++)
200 {
201 if (*Start == '%' && Start + 2 < C &&
202 isxdigit(Start[1]) && isxdigit(Start[2]))
203 {
204 Tmp[0] = Start[1];
205 Tmp[1] = Start[2];
206 Tmp[2] = 0;
207 *I = (char)strtol(Tmp,0,16);
208 Start += 3;
209 continue;
210 }
211 if (*Start != '"')
212 *I = *Start;
213 else
214 I--;
215 Start++;
216 }
217 *I = 0;
218 Res = Buffer;
219
220 // Skip ending white space
221 for (;*C != 0 && isspace(*C) != 0; C++);
222 String = C;
223 return true;
224 }
225 /*}}}*/
226 // ParseCWord - Parses a string like a C "" expression /*{{{*/
227 // ---------------------------------------------------------------------
228 /* This expects a series of space separated strings enclosed in ""'s.
229 It concatenates the ""'s into a single string. */
230 bool ParseCWord(const char *&String,string &Res)
231 {
232 // Skip leading whitespace
233 const char *C = String;
234 for (;*C != 0 && *C == ' '; C++);
235 if (*C == 0)
236 return false;
237
238 char Buffer[1024];
239 char *Buf = Buffer;
240 if (strlen(String) >= sizeof(Buffer))
241 return false;
242
243 for (; *C != 0; C++)
244 {
245 if (*C == '"')
246 {
247 for (C++; *C != 0 && *C != '"'; C++)
248 *Buf++ = *C;
249
250 if (*C == 0)
251 return false;
252
253 continue;
254 }
255
256 if (C != String && isspace(*C) != 0 && isspace(C[-1]) != 0)
257 continue;
258 if (isspace(*C) == 0)
259 return false;
260 *Buf++ = ' ';
261 }
262 *Buf = 0;
263 Res = Buffer;
264 String = C;
265 return true;
266 }
267 /*}}}*/
268 // QuoteString - Convert a string into quoted from /*{{{*/
269 // ---------------------------------------------------------------------
270 /* */
271 string QuoteString(const string &Str, const char *Bad)
272 {
273 string Res;
274 for (string::const_iterator I = Str.begin(); I != Str.end(); I++)
275 {
276 if (strchr(Bad,*I) != 0 || isprint(*I) == 0 ||
277 *I == 0x25 || // percent '%' char
278 *I <= 0x20 || *I >= 0x7F) // control chars
279 {
280 char Buf[10];
281 sprintf(Buf,"%%%02x",(int)*I);
282 Res += Buf;
283 }
284 else
285 Res += *I;
286 }
287 return Res;
288 }
289 /*}}}*/
290 // DeQuoteString - Convert a string from quoted from /*{{{*/
291 // ---------------------------------------------------------------------
292 /* This undoes QuoteString */
293 string DeQuoteString(const string &Str)
294 {
295 return DeQuoteString(Str.begin(),Str.end());
296 }
297 string DeQuoteString(string::const_iterator const &begin,
298 string::const_iterator const &end)
299 {
300 string Res;
301 for (string::const_iterator I = begin; I != end; I++)
302 {
303 if (*I == '%' && I + 2 < end &&
304 isxdigit(I[1]) && isxdigit(I[2]))
305 {
306 char Tmp[3];
307 Tmp[0] = I[1];
308 Tmp[1] = I[2];
309 Tmp[2] = 0;
310 Res += (char)strtol(Tmp,0,16);
311 I += 2;
312 continue;
313 }
314 else
315 Res += *I;
316 }
317 return Res;
318 }
319
320 /*}}}*/
321 // SizeToStr - Convert a long into a human readable size /*{{{*/
322 // ---------------------------------------------------------------------
323 /* A max of 4 digits are shown before conversion to the next highest unit.
324 The max length of the string will be 5 chars unless the size is > 10
325 YottaBytes (E24) */
326 string SizeToStr(double Size)
327 {
328 char S[300];
329 double ASize;
330 if (Size >= 0)
331 ASize = Size;
332 else
333 ASize = -1*Size;
334
335 /* bytes, KiloBytes, MegaBytes, GigaBytes, TeraBytes, PetaBytes,
336 ExaBytes, ZettaBytes, YottaBytes */
337 char Ext[] = {'\0','k','M','G','T','P','E','Z','Y'};
338 int I = 0;
339 while (I <= 8)
340 {
341 if (ASize < 100 && I != 0)
342 {
343 sprintf(S,"%'.1f %c",ASize,Ext[I]);
344 break;
345 }
346
347 if (ASize < 10000)
348 {
349 sprintf(S,"%'.0f %c",ASize,Ext[I]);
350 break;
351 }
352 ASize /= 1000.0;
353 I++;
354 }
355
356 return S;
357 }
358 /*}}}*/
359 // TimeToStr - Convert the time into a string /*{{{*/
360 // ---------------------------------------------------------------------
361 /* Converts a number of seconds to a hms format */
362 string TimeToStr(unsigned long Sec)
363 {
364 char S[300];
365
366 while (1)
367 {
368 if (Sec > 60*60*24)
369 {
370 //d means days, h means hours, min means minutes, s means seconds
371 sprintf(S,_("%lid %lih %limin %lis"),Sec/60/60/24,(Sec/60/60) % 24,(Sec/60) % 60,Sec % 60);
372 break;
373 }
374
375 if (Sec > 60*60)
376 {
377 //h means hours, min means minutes, s means seconds
378 sprintf(S,_("%lih %limin %lis"),Sec/60/60,(Sec/60) % 60,Sec % 60);
379 break;
380 }
381
382 if (Sec > 60)
383 {
384 //min means minutes, s means seconds
385 sprintf(S,_("%limin %lis"),Sec/60,Sec % 60);
386 break;
387 }
388
389 //s means seconds
390 sprintf(S,_("%lis"),Sec);
391 break;
392 }
393
394 return S;
395 }
396 /*}}}*/
397 // SubstVar - Substitute a string for another string /*{{{*/
398 // ---------------------------------------------------------------------
399 /* This replaces all occurances of Subst with Contents in Str. */
400 string SubstVar(const string &Str,const string &Subst,const string &Contents)
401 {
402 string::size_type Pos = 0;
403 string::size_type OldPos = 0;
404 string Temp;
405
406 while (OldPos < Str.length() &&
407 (Pos = Str.find(Subst,OldPos)) != string::npos)
408 {
409 Temp += string(Str,OldPos,Pos) + Contents;
410 OldPos = Pos + Subst.length();
411 }
412
413 if (OldPos == 0)
414 return Str;
415
416 return Temp + string(Str,OldPos);
417 }
418
419 string SubstVar(string Str,const struct SubstVar *Vars)
420 {
421 for (; Vars->Subst != 0; Vars++)
422 Str = SubstVar(Str,Vars->Subst,*Vars->Contents);
423 return Str;
424 }
425 /*}}}*/
426 // OutputInDepth - return a string with separator multiplied with depth /*{{{*/
427 // ---------------------------------------------------------------------
428 /* Returns a string with the supplied separator depth + 1 times in it */
429 std::string OutputInDepth(const unsigned long Depth, const char* Separator)
430 {
431 std::string output = "";
432 for(unsigned long d=Depth+1; d > 0; d--)
433 output.append(Separator);
434 return output;
435 }
436 /*}}}*/
437 // URItoFileName - Convert the uri into a unique file name /*{{{*/
438 // ---------------------------------------------------------------------
439 /* This converts a URI into a safe filename. It quotes all unsafe characters
440 and converts / to _ and removes the scheme identifier. The resulting
441 file name should be unique and never occur again for a different file */
442 string URItoFileName(const string &URI)
443 {
444 // Nuke 'sensitive' items
445 ::URI U(URI);
446 U.User.clear();
447 U.Password.clear();
448 U.Access.clear();
449
450 // "\x00-\x20{}|\\\\^\\[\\]<>\"\x7F-\xFF";
451 string NewURI = QuoteString(U,"\\|{}[]<>\"^~_=!@#$%^&*");
452 replace(NewURI.begin(),NewURI.end(),'/','_');
453 return NewURI;
454 }
455 /*}}}*/
456 // Base64Encode - Base64 Encoding routine for short strings /*{{{*/
457 // ---------------------------------------------------------------------
458 /* This routine performs a base64 transformation on a string. It was ripped
459 from wget and then patched and bug fixed.
460
461 This spec can be found in rfc2045 */
462 string Base64Encode(const string &S)
463 {
464 // Conversion table.
465 static char tbl[64] = {'A','B','C','D','E','F','G','H',
466 'I','J','K','L','M','N','O','P',
467 'Q','R','S','T','U','V','W','X',
468 'Y','Z','a','b','c','d','e','f',
469 'g','h','i','j','k','l','m','n',
470 'o','p','q','r','s','t','u','v',
471 'w','x','y','z','0','1','2','3',
472 '4','5','6','7','8','9','+','/'};
473
474 // Pre-allocate some space
475 string Final;
476 Final.reserve((4*S.length() + 2)/3 + 2);
477
478 /* Transform the 3x8 bits to 4x6 bits, as required by
479 base64. */
480 for (string::const_iterator I = S.begin(); I < S.end(); I += 3)
481 {
482 char Bits[3] = {0,0,0};
483 Bits[0] = I[0];
484 if (I + 1 < S.end())
485 Bits[1] = I[1];
486 if (I + 2 < S.end())
487 Bits[2] = I[2];
488
489 Final += tbl[Bits[0] >> 2];
490 Final += tbl[((Bits[0] & 3) << 4) + (Bits[1] >> 4)];
491
492 if (I + 1 >= S.end())
493 break;
494
495 Final += tbl[((Bits[1] & 0xf) << 2) + (Bits[2] >> 6)];
496
497 if (I + 2 >= S.end())
498 break;
499
500 Final += tbl[Bits[2] & 0x3f];
501 }
502
503 /* Apply the padding elements, this tells how many bytes the remote
504 end should discard */
505 if (S.length() % 3 == 2)
506 Final += '=';
507 if (S.length() % 3 == 1)
508 Final += "==";
509
510 return Final;
511 }
512 /*}}}*/
513 // stringcmp - Arbitrary string compare /*{{{*/
514 // ---------------------------------------------------------------------
515 /* This safely compares two non-null terminated strings of arbitrary
516 length */
517 int stringcmp(const char *A,const char *AEnd,const char *B,const char *BEnd)
518 {
519 for (; A != AEnd && B != BEnd; A++, B++)
520 if (*A != *B)
521 break;
522
523 if (A == AEnd && B == BEnd)
524 return 0;
525 if (A == AEnd)
526 return 1;
527 if (B == BEnd)
528 return -1;
529 if (*A < *B)
530 return -1;
531 return 1;
532 }
533
534 #if __GNUC__ >= 3
535 int stringcmp(string::const_iterator A,string::const_iterator AEnd,
536 const char *B,const char *BEnd)
537 {
538 for (; A != AEnd && B != BEnd; A++, B++)
539 if (*A != *B)
540 break;
541
542 if (A == AEnd && B == BEnd)
543 return 0;
544 if (A == AEnd)
545 return 1;
546 if (B == BEnd)
547 return -1;
548 if (*A < *B)
549 return -1;
550 return 1;
551 }
552 int stringcmp(string::const_iterator A,string::const_iterator AEnd,
553 string::const_iterator B,string::const_iterator BEnd)
554 {
555 for (; A != AEnd && B != BEnd; A++, B++)
556 if (*A != *B)
557 break;
558
559 if (A == AEnd && B == BEnd)
560 return 0;
561 if (A == AEnd)
562 return 1;
563 if (B == BEnd)
564 return -1;
565 if (*A < *B)
566 return -1;
567 return 1;
568 }
569 #endif
570 /*}}}*/
571 // stringcasecmp - Arbitrary case insensitive string compare /*{{{*/
572 // ---------------------------------------------------------------------
573 /* */
574 int stringcasecmp(const char *A,const char *AEnd,const char *B,const char *BEnd)
575 {
576 for (; A != AEnd && B != BEnd; A++, B++)
577 if (tolower_ascii(*A) != tolower_ascii(*B))
578 break;
579
580 if (A == AEnd && B == BEnd)
581 return 0;
582 if (A == AEnd)
583 return 1;
584 if (B == BEnd)
585 return -1;
586 if (tolower_ascii(*A) < tolower_ascii(*B))
587 return -1;
588 return 1;
589 }
590 #if __GNUC__ >= 3
591 int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
592 const char *B,const char *BEnd)
593 {
594 for (; A != AEnd && B != BEnd; A++, B++)
595 if (tolower_ascii(*A) != tolower_ascii(*B))
596 break;
597
598 if (A == AEnd && B == BEnd)
599 return 0;
600 if (A == AEnd)
601 return 1;
602 if (B == BEnd)
603 return -1;
604 if (tolower_ascii(*A) < tolower_ascii(*B))
605 return -1;
606 return 1;
607 }
608 int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
609 string::const_iterator B,string::const_iterator BEnd)
610 {
611 for (; A != AEnd && B != BEnd; A++, B++)
612 if (tolower_ascii(*A) != tolower_ascii(*B))
613 break;
614
615 if (A == AEnd && B == BEnd)
616 return 0;
617 if (A == AEnd)
618 return 1;
619 if (B == BEnd)
620 return -1;
621 if (tolower_ascii(*A) < tolower_ascii(*B))
622 return -1;
623 return 1;
624 }
625 #endif
626 /*}}}*/
627 // LookupTag - Lookup the value of a tag in a taged string /*{{{*/
628 // ---------------------------------------------------------------------
629 /* The format is like those used in package files and the method
630 communication system */
631 string LookupTag(const string &Message,const char *Tag,const char *Default)
632 {
633 // Look for a matching tag.
634 int Length = strlen(Tag);
635 for (string::const_iterator I = Message.begin(); I + Length < Message.end(); I++)
636 {
637 // Found the tag
638 if (I[Length] == ':' && stringcasecmp(I,I+Length,Tag) == 0)
639 {
640 // Find the end of line and strip the leading/trailing spaces
641 string::const_iterator J;
642 I += Length + 1;
643 for (; isspace(*I) != 0 && I < Message.end(); I++);
644 for (J = I; *J != '\n' && J < Message.end(); J++);
645 for (; J > I && isspace(J[-1]) != 0; J--);
646
647 return string(I,J);
648 }
649
650 for (; *I != '\n' && I < Message.end(); I++);
651 }
652
653 // Failed to find a match
654 if (Default == 0)
655 return string();
656 return Default;
657 }
658 /*}}}*/
659 // StringToBool - Converts a string into a boolean /*{{{*/
660 // ---------------------------------------------------------------------
661 /* This inspects the string to see if it is true or if it is false and
662 then returns the result. Several varients on true/false are checked. */
663 int StringToBool(const string &Text,int Default)
664 {
665 char *End;
666 int Res = strtol(Text.c_str(),&End,0);
667 if (End != Text.c_str() && Res >= 0 && Res <= 1)
668 return Res;
669
670 // Check for positives
671 if (strcasecmp(Text.c_str(),"no") == 0 ||
672 strcasecmp(Text.c_str(),"false") == 0 ||
673 strcasecmp(Text.c_str(),"without") == 0 ||
674 strcasecmp(Text.c_str(),"off") == 0 ||
675 strcasecmp(Text.c_str(),"disable") == 0)
676 return 0;
677
678 // Check for negatives
679 if (strcasecmp(Text.c_str(),"yes") == 0 ||
680 strcasecmp(Text.c_str(),"true") == 0 ||
681 strcasecmp(Text.c_str(),"with") == 0 ||
682 strcasecmp(Text.c_str(),"on") == 0 ||
683 strcasecmp(Text.c_str(),"enable") == 0)
684 return 1;
685
686 return Default;
687 }
688 /*}}}*/
689 // TimeRFC1123 - Convert a time_t into RFC1123 format /*{{{*/
690 // ---------------------------------------------------------------------
691 /* This converts a time_t into a string time representation that is
692 year 2000 complient and timezone neutral */
693 string TimeRFC1123(time_t Date)
694 {
695 struct tm Conv = *gmtime(&Date);
696 char Buf[300];
697
698 const char *Day[] = {"Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
699 const char *Month[] = {"Jan","Feb","Mar","Apr","May","Jun","Jul",
700 "Aug","Sep","Oct","Nov","Dec"};
701
702 sprintf(Buf,"%s, %02i %s %i %02i:%02i:%02i GMT",Day[Conv.tm_wday],
703 Conv.tm_mday,Month[Conv.tm_mon],Conv.tm_year+1900,Conv.tm_hour,
704 Conv.tm_min,Conv.tm_sec);
705 return Buf;
706 }
707 /*}}}*/
708 // ReadMessages - Read messages from the FD /*{{{*/
709 // ---------------------------------------------------------------------
710 /* This pulls full messages from the input FD into the message buffer.
711 It assumes that messages will not pause during transit so no
712 fancy buffering is used.
713
714 In particular: this reads blocks from the input until it believes
715 that it's run out of input text. Each block is terminated by a
716 double newline ('\n' followed by '\n'). As noted below, there is a
717 bug in this code: it assumes that all the blocks have been read if
718 it doesn't see additional text in the buffer after the last one is
719 parsed, which will cause it to lose blocks if the last block
720 coincides with the end of the buffer.
721 */
722 bool ReadMessages(int Fd, vector<string> &List)
723 {
724 char Buffer[64000];
725 char *End = Buffer;
726 // Represents any left-over from the previous iteration of the
727 // parse loop. (i.e., if a message is split across the end
728 // of the buffer, it goes here)
729 string PartialMessage;
730
731 while (1)
732 {
733 int Res = read(Fd,End,sizeof(Buffer) - (End-Buffer));
734 if (Res < 0 && errno == EINTR)
735 continue;
736
737 // Process is dead, this is kind of bad..
738 if (Res == 0)
739 return false;
740
741 // No data
742 if (Res < 0 && errno == EAGAIN)
743 return true;
744 if (Res < 0)
745 return false;
746
747 End += Res;
748
749 // Look for the end of the message
750 for (char *I = Buffer; I + 1 < End; I++)
751 {
752 if (I[0] != '\n' || I[1] != '\n')
753 continue;
754
755 // Pull the message out
756 string Message(Buffer,I-Buffer);
757 PartialMessage += Message;
758
759 // Fix up the buffer
760 for (; I < End && *I == '\n'; I++);
761 End -= I-Buffer;
762 memmove(Buffer,I,End-Buffer);
763 I = Buffer;
764
765 List.push_back(PartialMessage);
766 PartialMessage.clear();
767 }
768 if (End != Buffer)
769 {
770 // If there's text left in the buffer, store it
771 // in PartialMessage and throw the rest of the buffer
772 // away. This allows us to handle messages that
773 // are longer than the static buffer size.
774 PartialMessage += string(Buffer, End);
775 End = Buffer;
776 }
777 else
778 {
779 // BUG ALERT: if a message block happens to end at a
780 // multiple of 64000 characters, this will cause it to
781 // terminate early, leading to a badly formed block and
782 // probably crashing the method. However, this is the only
783 // way we have to find the end of the message block. I have
784 // an idea of how to fix this, but it will require changes
785 // to the protocol (essentially to mark the beginning and
786 // end of the block).
787 //
788 // -- dburrows 2008-04-02
789 return true;
790 }
791
792 if (WaitFd(Fd) == false)
793 return false;
794 }
795 }
796 /*}}}*/
797 // MonthConv - Converts a month string into a number /*{{{*/
798 // ---------------------------------------------------------------------
799 /* This was lifted from the boa webserver which lifted it from 'wn-v1.07'
800 Made it a bit more robust with a few tolower_ascii though. */
801 static int MonthConv(char *Month)
802 {
803 switch (tolower_ascii(*Month))
804 {
805 case 'a':
806 return tolower_ascii(Month[1]) == 'p'?3:7;
807 case 'd':
808 return 11;
809 case 'f':
810 return 1;
811 case 'j':
812 if (tolower_ascii(Month[1]) == 'a')
813 return 0;
814 return tolower_ascii(Month[2]) == 'n'?5:6;
815 case 'm':
816 return tolower_ascii(Month[2]) == 'r'?2:4;
817 case 'n':
818 return 10;
819 case 'o':
820 return 9;
821 case 's':
822 return 8;
823
824 // Pretend it is January..
825 default:
826 return 0;
827 }
828 }
829 /*}}}*/
830 // timegm - Internal timegm if the gnu version is not available /*{{{*/
831 // ---------------------------------------------------------------------
832 /* Converts struct tm to time_t, assuming the data in tm is UTC rather
833 than local timezone (mktime assumes the latter).
834
835 This function is a nonstandard GNU extension that is also present on
836 the BSDs and maybe other systems. For others we follow the advice of
837 the manpage of timegm and use his portable replacement. */
838 #ifndef HAVE_TIMEGM
839 static time_t timegm(struct tm *t)
840 {
841 char *tz = getenv("TZ");
842 setenv("TZ", "", 1);
843 tzset();
844 time_t ret = mktime(t);
845 if (tz)
846 setenv("TZ", tz, 1);
847 else
848 unsetenv("TZ");
849 tzset();
850 return ret;
851 }
852 #endif
853 /*}}}*/
854 // FullDateToTime - Converts a HTTP1.1 full date strings into a time_t /*{{{*/
855 // ---------------------------------------------------------------------
856 /* tries to parses a full date as specified in RFC2616 Section 3.3.1
857 with one exception: All timezones (%Z) are accepted but the protocol
858 says that it MUST be GMT, but this one is equal to UTC which we will
859 encounter from time to time (e.g. in Release files) so we accept all
860 here and just assume it is GMT (or UTC) later on */
861 bool RFC1123StrToTime(const char* const str,time_t &time)
862 {
863 struct tm Tm;
864 setlocale (LC_ALL,"C");
865 bool const invalid =
866 // Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
867 (strptime(str, "%a, %d %b %Y %H:%M:%S %Z", &Tm) == NULL &&
868 // Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
869 strptime(str, "%A, %d-%b-%y %H:%M:%S %Z", &Tm) == NULL &&
870 // Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
871 strptime(str, "%a %b %d %H:%M:%S %Y", &Tm) == NULL);
872 setlocale (LC_ALL,"");
873 if (invalid == true)
874 return false;
875
876 time = timegm(&Tm);
877 return true;
878 }
879 /*}}}*/
880 // FTPMDTMStrToTime - Converts a ftp modification date into a time_t /*{{{*/
881 // ---------------------------------------------------------------------
882 /* */
883 bool FTPMDTMStrToTime(const char* const str,time_t &time)
884 {
885 struct tm Tm;
886 // MDTM includes no whitespaces but recommend and ignored by strptime
887 if (strptime(str, "%Y %m %d %H %M %S", &Tm) == NULL)
888 return false;
889
890 time = timegm(&Tm);
891 return true;
892 }
893 /*}}}*/
894 // StrToTime - Converts a string into a time_t /*{{{*/
895 // ---------------------------------------------------------------------
896 /* This handles all 3 populare time formats including RFC 1123, RFC 1036
897 and the C library asctime format. It requires the GNU library function
898 'timegm' to convert a struct tm in UTC to a time_t. For some bizzar
899 reason the C library does not provide any such function :< This also
900 handles the weird, but unambiguous FTP time format*/
901 bool StrToTime(const string &Val,time_t &Result)
902 {
903 struct tm Tm;
904 char Month[10];
905 const char *I = Val.c_str();
906
907 // Skip the day of the week
908 for (;*I != 0 && *I != ' '; I++);
909
910 // Handle RFC 1123 time
911 Month[0] = 0;
912 if (sscanf(I," %d %3s %d %d:%d:%d GMT",&Tm.tm_mday,Month,&Tm.tm_year,
913 &Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) != 6)
914 {
915 // Handle RFC 1036 time
916 if (sscanf(I," %d-%3s-%d %d:%d:%d GMT",&Tm.tm_mday,Month,
917 &Tm.tm_year,&Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) == 6)
918 Tm.tm_year += 1900;
919 else
920 {
921 // asctime format
922 if (sscanf(I," %3s %d %d:%d:%d %d",Month,&Tm.tm_mday,
923 &Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec,&Tm.tm_year) != 6)
924 {
925 // 'ftp' time
926 if (sscanf(Val.c_str(),"%4d%2d%2d%2d%2d%2d",&Tm.tm_year,&Tm.tm_mon,
927 &Tm.tm_mday,&Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) != 6)
928 return false;
929 Tm.tm_mon--;
930 }
931 }
932 }
933
934 Tm.tm_isdst = 0;
935 if (Month[0] != 0)
936 Tm.tm_mon = MonthConv(Month);
937 Tm.tm_year -= 1900;
938
939 // Convert to local time and then to GMT
940 Result = timegm(&Tm);
941 return true;
942 }
943 /*}}}*/
944 // StrToNum - Convert a fixed length string to a number /*{{{*/
945 // ---------------------------------------------------------------------
946 /* This is used in decoding the crazy fixed length string headers in
947 tar and ar files. */
948 bool StrToNum(const char *Str,unsigned long &Res,unsigned Len,unsigned Base)
949 {
950 char S[30];
951 if (Len >= sizeof(S))
952 return false;
953 memcpy(S,Str,Len);
954 S[Len] = 0;
955
956 // All spaces is a zero
957 Res = 0;
958 unsigned I;
959 for (I = 0; S[I] == ' '; I++);
960 if (S[I] == 0)
961 return true;
962
963 char *End;
964 Res = strtoul(S,&End,Base);
965 if (End == S)
966 return false;
967
968 return true;
969 }
970 /*}}}*/
971 // Base256ToNum - Convert a fixed length binary to a number /*{{{*/
972 // ---------------------------------------------------------------------
973 /* This is used in decoding the 256bit encoded fixed length fields in
974 tar files */
975 bool Base256ToNum(const char *Str,unsigned long &Res,unsigned int Len)
976 {
977 if ((Str[0] & 0x80) == 0)
978 return false;
979 else
980 {
981 Res = Str[0] & 0x7F;
982 for(unsigned int i = 1; i < Len; ++i)
983 Res = (Res<<8) + Str[i];
984 return true;
985 }
986 }
987 /*}}}*/
988 // HexDigit - Convert a hex character into an integer /*{{{*/
989 // ---------------------------------------------------------------------
990 /* Helper for Hex2Num */
991 static int HexDigit(int c)
992 {
993 if (c >= '0' && c <= '9')
994 return c - '0';
995 if (c >= 'a' && c <= 'f')
996 return c - 'a' + 10;
997 if (c >= 'A' && c <= 'F')
998 return c - 'A' + 10;
999 return 0;
1000 }
1001 /*}}}*/
1002 // Hex2Num - Convert a long hex number into a buffer /*{{{*/
1003 // ---------------------------------------------------------------------
1004 /* The length of the buffer must be exactly 1/2 the length of the string. */
1005 bool Hex2Num(const string &Str,unsigned char *Num,unsigned int Length)
1006 {
1007 if (Str.length() != Length*2)
1008 return false;
1009
1010 // Convert each digit. We store it in the same order as the string
1011 int J = 0;
1012 for (string::const_iterator I = Str.begin(); I != Str.end();J++, I += 2)
1013 {
1014 if (isxdigit(*I) == 0 || isxdigit(I[1]) == 0)
1015 return false;
1016
1017 Num[J] = HexDigit(I[0]) << 4;
1018 Num[J] += HexDigit(I[1]);
1019 }
1020
1021 return true;
1022 }
1023 /*}}}*/
1024 // TokSplitString - Split a string up by a given token /*{{{*/
1025 // ---------------------------------------------------------------------
1026 /* This is intended to be a faster splitter, it does not use dynamic
1027 memories. Input is changed to insert nulls at each token location. */
1028 bool TokSplitString(char Tok,char *Input,char **List,
1029 unsigned long ListMax)
1030 {
1031 // Strip any leading spaces
1032 char *Start = Input;
1033 char *Stop = Start + strlen(Start);
1034 for (; *Start != 0 && isspace(*Start) != 0; Start++);
1035
1036 unsigned long Count = 0;
1037 char *Pos = Start;
1038 while (Pos != Stop)
1039 {
1040 // Skip to the next Token
1041 for (; Pos != Stop && *Pos != Tok; Pos++);
1042
1043 // Back remove spaces
1044 char *End = Pos;
1045 for (; End > Start && (End[-1] == Tok || isspace(End[-1]) != 0); End--);
1046 *End = 0;
1047
1048 List[Count++] = Start;
1049 if (Count >= ListMax)
1050 {
1051 List[Count-1] = 0;
1052 return false;
1053 }
1054
1055 // Advance pos
1056 for (; Pos != Stop && (*Pos == Tok || isspace(*Pos) != 0 || *Pos == 0); Pos++);
1057 Start = Pos;
1058 }
1059
1060 List[Count] = 0;
1061 return true;
1062 }
1063 /*}}}*/
1064 // VectorizeString - Split a string up into a vector of strings /*{{{*/
1065 // ---------------------------------------------------------------------
1066 /* This can be used to split a given string up into a vector, so the
1067 propose is the same as in the method above and this one is a bit slower
1068 also, but the advantage is that we have an iteratable vector */
1069 vector<string> VectorizeString(string const &haystack, char const &split)
1070 {
1071 string::const_iterator start = haystack.begin();
1072 string::const_iterator end = start;
1073 vector<string> exploded;
1074 do {
1075 for (; end != haystack.end() && *end != split; ++end);
1076 exploded.push_back(string(start, end));
1077 start = end + 1;
1078 } while (end != haystack.end() && (++end) != haystack.end());
1079 return exploded;
1080 }
1081 /*}}}*/
1082 // RegexChoice - Simple regex list/list matcher /*{{{*/
1083 // ---------------------------------------------------------------------
1084 /* */
1085 unsigned long RegexChoice(RxChoiceList *Rxs,const char **ListBegin,
1086 const char **ListEnd)
1087 {
1088 for (RxChoiceList *R = Rxs; R->Str != 0; R++)
1089 R->Hit = false;
1090
1091 unsigned long Hits = 0;
1092 for (; ListBegin != ListEnd; ListBegin++)
1093 {
1094 // Check if the name is a regex
1095 const char *I;
1096 bool Regex = true;
1097 for (I = *ListBegin; *I != 0; I++)
1098 if (*I == '.' || *I == '?' || *I == '*' || *I == '|')
1099 break;
1100 if (*I == 0)
1101 Regex = false;
1102
1103 // Compile the regex pattern
1104 regex_t Pattern;
1105 if (Regex == true)
1106 if (regcomp(&Pattern,*ListBegin,REG_EXTENDED | REG_ICASE |
1107 REG_NOSUB) != 0)
1108 Regex = false;
1109
1110 // Search the list
1111 bool Done = false;
1112 for (RxChoiceList *R = Rxs; R->Str != 0; R++)
1113 {
1114 if (R->Str[0] == 0)
1115 continue;
1116
1117 if (strcasecmp(R->Str,*ListBegin) != 0)
1118 {
1119 if (Regex == false)
1120 continue;
1121 if (regexec(&Pattern,R->Str,0,0,0) != 0)
1122 continue;
1123 }
1124 Done = true;
1125
1126 if (R->Hit == false)
1127 Hits++;
1128
1129 R->Hit = true;
1130 }
1131
1132 if (Regex == true)
1133 regfree(&Pattern);
1134
1135 if (Done == false)
1136 _error->Warning(_("Selection %s not found"),*ListBegin);
1137 }
1138
1139 return Hits;
1140 }
1141 /*}}}*/
1142 // ioprintf - C format string outputter to C++ iostreams /*{{{*/
1143 // ---------------------------------------------------------------------
1144 /* This is used to make the internationalization strings easier to translate
1145 and to allow reordering of parameters */
1146 void ioprintf(ostream &out,const char *format,...)
1147 {
1148 va_list args;
1149 va_start(args,format);
1150
1151 // sprintf the description
1152 char S[4096];
1153 vsnprintf(S,sizeof(S),format,args);
1154 out << S;
1155 }
1156 /*}}}*/
1157 // strprintf - C format string outputter to C++ strings /*{{{*/
1158 // ---------------------------------------------------------------------
1159 /* This is used to make the internationalization strings easier to translate
1160 and to allow reordering of parameters */
1161 void strprintf(string &out,const char *format,...)
1162 {
1163 va_list args;
1164 va_start(args,format);
1165
1166 // sprintf the description
1167 char S[4096];
1168 vsnprintf(S,sizeof(S),format,args);
1169 out = string(S);
1170 }
1171 /*}}}*/
1172 // safe_snprintf - Safer snprintf /*{{{*/
1173 // ---------------------------------------------------------------------
1174 /* This is a snprintf that will never (ever) go past 'End' and returns a
1175 pointer to the end of the new string. The returned string is always null
1176 terminated unless Buffer == end. This is a better alterantive to using
1177 consecutive snprintfs. */
1178 char *safe_snprintf(char *Buffer,char *End,const char *Format,...)
1179 {
1180 va_list args;
1181 int Did;
1182
1183 va_start(args,Format);
1184
1185 if (End <= Buffer)
1186 return End;
1187
1188 Did = vsnprintf(Buffer,End - Buffer,Format,args);
1189 if (Did < 0 || Buffer + Did > End)
1190 return End;
1191 return Buffer + Did;
1192 }
1193 /*}}}*/
1194 // StripEpoch - Remove the version "epoch" from a version string /*{{{*/
1195 // ---------------------------------------------------------------------
1196 string StripEpoch(const string &VerStr)
1197 {
1198 size_t i = VerStr.find(":");
1199 if (i == string::npos)
1200 return VerStr;
1201 return VerStr.substr(i+1);
1202 }
1203
1204 // tolower_ascii - tolower() function that ignores the locale /*{{{*/
1205 // ---------------------------------------------------------------------
1206 /* This little function is the most called method we have and tries
1207 therefore to do the absolut minimum - and is noteable faster than
1208 standard tolower/toupper and as a bonus avoids problems with different
1209 locales - we only operate on ascii chars anyway. */
1210 int tolower_ascii(int const c)
1211 {
1212 if (c >= 'A' && c <= 'Z')
1213 return c + 32;
1214 return c;
1215 }
1216 /*}}}*/
1217
1218 // CheckDomainList - See if Host is in a , seperate list /*{{{*/
1219 // ---------------------------------------------------------------------
1220 /* The domain list is a comma seperate list of domains that are suffix
1221 matched against the argument */
1222 bool CheckDomainList(const string &Host,const string &List)
1223 {
1224 string::const_iterator Start = List.begin();
1225 for (string::const_iterator Cur = List.begin(); Cur <= List.end(); Cur++)
1226 {
1227 if (Cur < List.end() && *Cur != ',')
1228 continue;
1229
1230 // Match the end of the string..
1231 if ((Host.size() >= (unsigned)(Cur - Start)) &&
1232 Cur - Start != 0 &&
1233 stringcasecmp(Host.end() - (Cur - Start),Host.end(),Start,Cur) == 0)
1234 return true;
1235
1236 Start = Cur + 1;
1237 }
1238 return false;
1239 }
1240 /*}}}*/
1241
1242 // URI::CopyFrom - Copy from an object /*{{{*/
1243 // ---------------------------------------------------------------------
1244 /* This parses the URI into all of its components */
1245 void URI::CopyFrom(const string &U)
1246 {
1247 string::const_iterator I = U.begin();
1248
1249 // Locate the first colon, this separates the scheme
1250 for (; I < U.end() && *I != ':' ; I++);
1251 string::const_iterator FirstColon = I;
1252
1253 /* Determine if this is a host type URI with a leading double //
1254 and then search for the first single / */
1255 string::const_iterator SingleSlash = I;
1256 if (I + 3 < U.end() && I[1] == '/' && I[2] == '/')
1257 SingleSlash += 3;
1258
1259 /* Find the / indicating the end of the hostname, ignoring /'s in the
1260 square brackets */
1261 bool InBracket = false;
1262 for (; SingleSlash < U.end() && (*SingleSlash != '/' || InBracket == true); SingleSlash++)
1263 {
1264 if (*SingleSlash == '[')
1265 InBracket = true;
1266 if (InBracket == true && *SingleSlash == ']')
1267 InBracket = false;
1268 }
1269
1270 if (SingleSlash > U.end())
1271 SingleSlash = U.end();
1272
1273 // We can now write the access and path specifiers
1274 Access.assign(U.begin(),FirstColon);
1275 if (SingleSlash != U.end())
1276 Path.assign(SingleSlash,U.end());
1277 if (Path.empty() == true)
1278 Path = "/";
1279
1280 // Now we attempt to locate a user:pass@host fragment
1281 if (FirstColon + 2 <= U.end() && FirstColon[1] == '/' && FirstColon[2] == '/')
1282 FirstColon += 3;
1283 else
1284 FirstColon += 1;
1285 if (FirstColon >= U.end())
1286 return;
1287
1288 if (FirstColon > SingleSlash)
1289 FirstColon = SingleSlash;
1290
1291 // Find the colon...
1292 I = FirstColon + 1;
1293 if (I > SingleSlash)
1294 I = SingleSlash;
1295 for (; I < SingleSlash && *I != ':'; I++);
1296 string::const_iterator SecondColon = I;
1297
1298 // Search for the @ after the colon
1299 for (; I < SingleSlash && *I != '@'; I++);
1300 string::const_iterator At = I;
1301
1302 // Now write the host and user/pass
1303 if (At == SingleSlash)
1304 {
1305 if (FirstColon < SingleSlash)
1306 Host.assign(FirstColon,SingleSlash);
1307 }
1308 else
1309 {
1310 Host.assign(At+1,SingleSlash);
1311 // username and password must be encoded (RFC 3986)
1312 User.assign(DeQuoteString(FirstColon,SecondColon));
1313 if (SecondColon < At)
1314 Password.assign(DeQuoteString(SecondColon+1,At));
1315 }
1316
1317 // Now we parse the RFC 2732 [] hostnames.
1318 unsigned long PortEnd = 0;
1319 InBracket = false;
1320 for (unsigned I = 0; I != Host.length();)
1321 {
1322 if (Host[I] == '[')
1323 {
1324 InBracket = true;
1325 Host.erase(I,1);
1326 continue;
1327 }
1328
1329 if (InBracket == true && Host[I] == ']')
1330 {
1331 InBracket = false;
1332 Host.erase(I,1);
1333 PortEnd = I;
1334 continue;
1335 }
1336 I++;
1337 }
1338
1339 // Tsk, weird.
1340 if (InBracket == true)
1341 {
1342 Host.clear();
1343 return;
1344 }
1345
1346 // Now we parse off a port number from the hostname
1347 Port = 0;
1348 string::size_type Pos = Host.rfind(':');
1349 if (Pos == string::npos || Pos < PortEnd)
1350 return;
1351
1352 Port = atoi(string(Host,Pos+1).c_str());
1353 Host.assign(Host,0,Pos);
1354 }
1355 /*}}}*/
1356 // URI::operator string - Convert the URI to a string /*{{{*/
1357 // ---------------------------------------------------------------------
1358 /* */
1359 URI::operator string()
1360 {
1361 string Res;
1362
1363 if (Access.empty() == false)
1364 Res = Access + ':';
1365
1366 if (Host.empty() == false)
1367 {
1368 if (Access.empty() == false)
1369 Res += "//";
1370
1371 if (User.empty() == false)
1372 {
1373 Res += User;
1374 if (Password.empty() == false)
1375 Res += ":" + Password;
1376 Res += "@";
1377 }
1378
1379 // Add RFC 2732 escaping characters
1380 if (Access.empty() == false &&
1381 (Host.find('/') != string::npos || Host.find(':') != string::npos))
1382 Res += '[' + Host + ']';
1383 else
1384 Res += Host;
1385
1386 if (Port != 0)
1387 {
1388 char S[30];
1389 sprintf(S,":%u",Port);
1390 Res += S;
1391 }
1392 }
1393
1394 if (Path.empty() == false)
1395 {
1396 if (Path[0] != '/')
1397 Res += "/" + Path;
1398 else
1399 Res += Path;
1400 }
1401
1402 return Res;
1403 }
1404 /*}}}*/
1405 // URI::SiteOnly - Return the schema and site for the URI /*{{{*/
1406 // ---------------------------------------------------------------------
1407 /* */
1408 string URI::SiteOnly(const string &URI)
1409 {
1410 ::URI U(URI);
1411 U.User.clear();
1412 U.Password.clear();
1413 U.Path.clear();
1414 U.Port = 0;
1415 return U;
1416 }
1417 /*}}}*/
1418 // URI::NoUserPassword - Return the schema, site and path for the URI /*{{{*/
1419 // ---------------------------------------------------------------------
1420 /* */
1421 string URI::NoUserPassword(const string &URI)
1422 {
1423 ::URI U(URI);
1424 U.User.clear();
1425 U.Password.clear();
1426 U.Port = 0;
1427 return U;
1428 }
1429 /*}}}*/