reorder includes: add <config.h> if needed and include it at first
[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 <config.h>
19
20 #include <apt-pkg/strutl.h>
21 #include <apt-pkg/fileutl.h>
22 #include <apt-pkg/error.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 <apti18n.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;
696 if (gmtime_r(&Date, &Conv) == NULL)
697 return "";
698
699 char Buf[300];
700 const char *Day[] = {"Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
701 const char *Month[] = {"Jan","Feb","Mar","Apr","May","Jun","Jul",
702 "Aug","Sep","Oct","Nov","Dec"};
703
704 snprintf(Buf, sizeof(Buf), "%s, %02i %s %i %02i:%02i:%02i GMT",Day[Conv.tm_wday],
705 Conv.tm_mday,Month[Conv.tm_mon],Conv.tm_year+1900,Conv.tm_hour,
706 Conv.tm_min,Conv.tm_sec);
707 return Buf;
708 }
709 /*}}}*/
710 // ReadMessages - Read messages from the FD /*{{{*/
711 // ---------------------------------------------------------------------
712 /* This pulls full messages from the input FD into the message buffer.
713 It assumes that messages will not pause during transit so no
714 fancy buffering is used.
715
716 In particular: this reads blocks from the input until it believes
717 that it's run out of input text. Each block is terminated by a
718 double newline ('\n' followed by '\n'). As noted below, there is a
719 bug in this code: it assumes that all the blocks have been read if
720 it doesn't see additional text in the buffer after the last one is
721 parsed, which will cause it to lose blocks if the last block
722 coincides with the end of the buffer.
723 */
724 bool ReadMessages(int Fd, vector<string> &List)
725 {
726 char Buffer[64000];
727 char *End = Buffer;
728 // Represents any left-over from the previous iteration of the
729 // parse loop. (i.e., if a message is split across the end
730 // of the buffer, it goes here)
731 string PartialMessage;
732
733 while (1)
734 {
735 int Res = read(Fd,End,sizeof(Buffer) - (End-Buffer));
736 if (Res < 0 && errno == EINTR)
737 continue;
738
739 // Process is dead, this is kind of bad..
740 if (Res == 0)
741 return false;
742
743 // No data
744 if (Res < 0 && errno == EAGAIN)
745 return true;
746 if (Res < 0)
747 return false;
748
749 End += Res;
750
751 // Look for the end of the message
752 for (char *I = Buffer; I + 1 < End; I++)
753 {
754 if (I[0] != '\n' || I[1] != '\n')
755 continue;
756
757 // Pull the message out
758 string Message(Buffer,I-Buffer);
759 PartialMessage += Message;
760
761 // Fix up the buffer
762 for (; I < End && *I == '\n'; I++);
763 End -= I-Buffer;
764 memmove(Buffer,I,End-Buffer);
765 I = Buffer;
766
767 List.push_back(PartialMessage);
768 PartialMessage.clear();
769 }
770 if (End != Buffer)
771 {
772 // If there's text left in the buffer, store it
773 // in PartialMessage and throw the rest of the buffer
774 // away. This allows us to handle messages that
775 // are longer than the static buffer size.
776 PartialMessage += string(Buffer, End);
777 End = Buffer;
778 }
779 else
780 {
781 // BUG ALERT: if a message block happens to end at a
782 // multiple of 64000 characters, this will cause it to
783 // terminate early, leading to a badly formed block and
784 // probably crashing the method. However, this is the only
785 // way we have to find the end of the message block. I have
786 // an idea of how to fix this, but it will require changes
787 // to the protocol (essentially to mark the beginning and
788 // end of the block).
789 //
790 // -- dburrows 2008-04-02
791 return true;
792 }
793
794 if (WaitFd(Fd) == false)
795 return false;
796 }
797 }
798 /*}}}*/
799 // MonthConv - Converts a month string into a number /*{{{*/
800 // ---------------------------------------------------------------------
801 /* This was lifted from the boa webserver which lifted it from 'wn-v1.07'
802 Made it a bit more robust with a few tolower_ascii though. */
803 static int MonthConv(char *Month)
804 {
805 switch (tolower_ascii(*Month))
806 {
807 case 'a':
808 return tolower_ascii(Month[1]) == 'p'?3:7;
809 case 'd':
810 return 11;
811 case 'f':
812 return 1;
813 case 'j':
814 if (tolower_ascii(Month[1]) == 'a')
815 return 0;
816 return tolower_ascii(Month[2]) == 'n'?5:6;
817 case 'm':
818 return tolower_ascii(Month[2]) == 'r'?2:4;
819 case 'n':
820 return 10;
821 case 'o':
822 return 9;
823 case 's':
824 return 8;
825
826 // Pretend it is January..
827 default:
828 return 0;
829 }
830 }
831 /*}}}*/
832 // timegm - Internal timegm if the gnu version is not available /*{{{*/
833 // ---------------------------------------------------------------------
834 /* Converts struct tm to time_t, assuming the data in tm is UTC rather
835 than local timezone (mktime assumes the latter).
836
837 This function is a nonstandard GNU extension that is also present on
838 the BSDs and maybe other systems. For others we follow the advice of
839 the manpage of timegm and use his portable replacement. */
840 #ifndef HAVE_TIMEGM
841 static time_t timegm(struct tm *t)
842 {
843 char *tz = getenv("TZ");
844 setenv("TZ", "", 1);
845 tzset();
846 time_t ret = mktime(t);
847 if (tz)
848 setenv("TZ", tz, 1);
849 else
850 unsetenv("TZ");
851 tzset();
852 return ret;
853 }
854 #endif
855 /*}}}*/
856 // FullDateToTime - Converts a HTTP1.1 full date strings into a time_t /*{{{*/
857 // ---------------------------------------------------------------------
858 /* tries to parses a full date as specified in RFC2616 Section 3.3.1
859 with one exception: All timezones (%Z) are accepted but the protocol
860 says that it MUST be GMT, but this one is equal to UTC which we will
861 encounter from time to time (e.g. in Release files) so we accept all
862 here and just assume it is GMT (or UTC) later on */
863 bool RFC1123StrToTime(const char* const str,time_t &time)
864 {
865 struct tm Tm;
866 setlocale (LC_ALL,"C");
867 bool const invalid =
868 // Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
869 (strptime(str, "%a, %d %b %Y %H:%M:%S %Z", &Tm) == NULL &&
870 // Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
871 strptime(str, "%A, %d-%b-%y %H:%M:%S %Z", &Tm) == NULL &&
872 // Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
873 strptime(str, "%a %b %d %H:%M:%S %Y", &Tm) == NULL);
874 setlocale (LC_ALL,"");
875 if (invalid == true)
876 return false;
877
878 time = timegm(&Tm);
879 return true;
880 }
881 /*}}}*/
882 // FTPMDTMStrToTime - Converts a ftp modification date into a time_t /*{{{*/
883 // ---------------------------------------------------------------------
884 /* */
885 bool FTPMDTMStrToTime(const char* const str,time_t &time)
886 {
887 struct tm Tm;
888 // MDTM includes no whitespaces but recommend and ignored by strptime
889 if (strptime(str, "%Y %m %d %H %M %S", &Tm) == NULL)
890 return false;
891
892 time = timegm(&Tm);
893 return true;
894 }
895 /*}}}*/
896 // StrToTime - Converts a string into a time_t /*{{{*/
897 // ---------------------------------------------------------------------
898 /* This handles all 3 populare time formats including RFC 1123, RFC 1036
899 and the C library asctime format. It requires the GNU library function
900 'timegm' to convert a struct tm in UTC to a time_t. For some bizzar
901 reason the C library does not provide any such function :< This also
902 handles the weird, but unambiguous FTP time format*/
903 bool StrToTime(const string &Val,time_t &Result)
904 {
905 struct tm Tm;
906 char Month[10];
907 const char *I = Val.c_str();
908
909 // Skip the day of the week
910 for (;*I != 0 && *I != ' '; I++);
911
912 // Handle RFC 1123 time
913 Month[0] = 0;
914 if (sscanf(I," %d %3s %d %d:%d:%d GMT",&Tm.tm_mday,Month,&Tm.tm_year,
915 &Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) != 6)
916 {
917 // Handle RFC 1036 time
918 if (sscanf(I," %d-%3s-%d %d:%d:%d GMT",&Tm.tm_mday,Month,
919 &Tm.tm_year,&Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) == 6)
920 Tm.tm_year += 1900;
921 else
922 {
923 // asctime format
924 if (sscanf(I," %3s %d %d:%d:%d %d",Month,&Tm.tm_mday,
925 &Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec,&Tm.tm_year) != 6)
926 {
927 // 'ftp' time
928 if (sscanf(Val.c_str(),"%4d%2d%2d%2d%2d%2d",&Tm.tm_year,&Tm.tm_mon,
929 &Tm.tm_mday,&Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) != 6)
930 return false;
931 Tm.tm_mon--;
932 }
933 }
934 }
935
936 Tm.tm_isdst = 0;
937 if (Month[0] != 0)
938 Tm.tm_mon = MonthConv(Month);
939 Tm.tm_year -= 1900;
940
941 // Convert to local time and then to GMT
942 Result = timegm(&Tm);
943 return true;
944 }
945 /*}}}*/
946 // StrToNum - Convert a fixed length string to a number /*{{{*/
947 // ---------------------------------------------------------------------
948 /* This is used in decoding the crazy fixed length string headers in
949 tar and ar files. */
950 bool StrToNum(const char *Str,unsigned long &Res,unsigned Len,unsigned Base)
951 {
952 char S[30];
953 if (Len >= sizeof(S))
954 return false;
955 memcpy(S,Str,Len);
956 S[Len] = 0;
957
958 // All spaces is a zero
959 Res = 0;
960 unsigned I;
961 for (I = 0; S[I] == ' '; I++);
962 if (S[I] == 0)
963 return true;
964
965 char *End;
966 Res = strtoul(S,&End,Base);
967 if (End == S)
968 return false;
969
970 return true;
971 }
972 /*}}}*/
973 // Base256ToNum - Convert a fixed length binary to a number /*{{{*/
974 // ---------------------------------------------------------------------
975 /* This is used in decoding the 256bit encoded fixed length fields in
976 tar files */
977 bool Base256ToNum(const char *Str,unsigned long &Res,unsigned int Len)
978 {
979 if ((Str[0] & 0x80) == 0)
980 return false;
981 else
982 {
983 Res = Str[0] & 0x7F;
984 for(unsigned int i = 1; i < Len; ++i)
985 Res = (Res<<8) + Str[i];
986 return true;
987 }
988 }
989 /*}}}*/
990 // HexDigit - Convert a hex character into an integer /*{{{*/
991 // ---------------------------------------------------------------------
992 /* Helper for Hex2Num */
993 static int HexDigit(int c)
994 {
995 if (c >= '0' && c <= '9')
996 return c - '0';
997 if (c >= 'a' && c <= 'f')
998 return c - 'a' + 10;
999 if (c >= 'A' && c <= 'F')
1000 return c - 'A' + 10;
1001 return 0;
1002 }
1003 /*}}}*/
1004 // Hex2Num - Convert a long hex number into a buffer /*{{{*/
1005 // ---------------------------------------------------------------------
1006 /* The length of the buffer must be exactly 1/2 the length of the string. */
1007 bool Hex2Num(const string &Str,unsigned char *Num,unsigned int Length)
1008 {
1009 if (Str.length() != Length*2)
1010 return false;
1011
1012 // Convert each digit. We store it in the same order as the string
1013 int J = 0;
1014 for (string::const_iterator I = Str.begin(); I != Str.end();J++, I += 2)
1015 {
1016 if (isxdigit(*I) == 0 || isxdigit(I[1]) == 0)
1017 return false;
1018
1019 Num[J] = HexDigit(I[0]) << 4;
1020 Num[J] += HexDigit(I[1]);
1021 }
1022
1023 return true;
1024 }
1025 /*}}}*/
1026 // TokSplitString - Split a string up by a given token /*{{{*/
1027 // ---------------------------------------------------------------------
1028 /* This is intended to be a faster splitter, it does not use dynamic
1029 memories. Input is changed to insert nulls at each token location. */
1030 bool TokSplitString(char Tok,char *Input,char **List,
1031 unsigned long ListMax)
1032 {
1033 // Strip any leading spaces
1034 char *Start = Input;
1035 char *Stop = Start + strlen(Start);
1036 for (; *Start != 0 && isspace(*Start) != 0; Start++);
1037
1038 unsigned long Count = 0;
1039 char *Pos = Start;
1040 while (Pos != Stop)
1041 {
1042 // Skip to the next Token
1043 for (; Pos != Stop && *Pos != Tok; Pos++);
1044
1045 // Back remove spaces
1046 char *End = Pos;
1047 for (; End > Start && (End[-1] == Tok || isspace(End[-1]) != 0); End--);
1048 *End = 0;
1049
1050 List[Count++] = Start;
1051 if (Count >= ListMax)
1052 {
1053 List[Count-1] = 0;
1054 return false;
1055 }
1056
1057 // Advance pos
1058 for (; Pos != Stop && (*Pos == Tok || isspace(*Pos) != 0 || *Pos == 0); Pos++);
1059 Start = Pos;
1060 }
1061
1062 List[Count] = 0;
1063 return true;
1064 }
1065 /*}}}*/
1066 // VectorizeString - Split a string up into a vector of strings /*{{{*/
1067 // ---------------------------------------------------------------------
1068 /* This can be used to split a given string up into a vector, so the
1069 propose is the same as in the method above and this one is a bit slower
1070 also, but the advantage is that we have an iteratable vector */
1071 vector<string> VectorizeString(string const &haystack, char const &split)
1072 {
1073 string::const_iterator start = haystack.begin();
1074 string::const_iterator end = start;
1075 vector<string> exploded;
1076 do {
1077 for (; end != haystack.end() && *end != split; ++end);
1078 exploded.push_back(string(start, end));
1079 start = end + 1;
1080 } while (end != haystack.end() && (++end) != haystack.end());
1081 return exploded;
1082 }
1083 /*}}}*/
1084 // RegexChoice - Simple regex list/list matcher /*{{{*/
1085 // ---------------------------------------------------------------------
1086 /* */
1087 unsigned long RegexChoice(RxChoiceList *Rxs,const char **ListBegin,
1088 const char **ListEnd)
1089 {
1090 for (RxChoiceList *R = Rxs; R->Str != 0; R++)
1091 R->Hit = false;
1092
1093 unsigned long Hits = 0;
1094 for (; ListBegin != ListEnd; ListBegin++)
1095 {
1096 // Check if the name is a regex
1097 const char *I;
1098 bool Regex = true;
1099 for (I = *ListBegin; *I != 0; I++)
1100 if (*I == '.' || *I == '?' || *I == '*' || *I == '|')
1101 break;
1102 if (*I == 0)
1103 Regex = false;
1104
1105 // Compile the regex pattern
1106 regex_t Pattern;
1107 if (Regex == true)
1108 if (regcomp(&Pattern,*ListBegin,REG_EXTENDED | REG_ICASE |
1109 REG_NOSUB) != 0)
1110 Regex = false;
1111
1112 // Search the list
1113 bool Done = false;
1114 for (RxChoiceList *R = Rxs; R->Str != 0; R++)
1115 {
1116 if (R->Str[0] == 0)
1117 continue;
1118
1119 if (strcasecmp(R->Str,*ListBegin) != 0)
1120 {
1121 if (Regex == false)
1122 continue;
1123 if (regexec(&Pattern,R->Str,0,0,0) != 0)
1124 continue;
1125 }
1126 Done = true;
1127
1128 if (R->Hit == false)
1129 Hits++;
1130
1131 R->Hit = true;
1132 }
1133
1134 if (Regex == true)
1135 regfree(&Pattern);
1136
1137 if (Done == false)
1138 _error->Warning(_("Selection %s not found"),*ListBegin);
1139 }
1140
1141 return Hits;
1142 }
1143 /*}}}*/
1144 // ioprintf - C format string outputter to C++ iostreams /*{{{*/
1145 // ---------------------------------------------------------------------
1146 /* This is used to make the internationalization strings easier to translate
1147 and to allow reordering of parameters */
1148 void ioprintf(ostream &out,const char *format,...)
1149 {
1150 va_list args;
1151 va_start(args,format);
1152
1153 // sprintf the description
1154 char S[4096];
1155 vsnprintf(S,sizeof(S),format,args);
1156 out << S;
1157 }
1158 /*}}}*/
1159 // strprintf - C format string outputter to C++ strings /*{{{*/
1160 // ---------------------------------------------------------------------
1161 /* This is used to make the internationalization strings easier to translate
1162 and to allow reordering of parameters */
1163 void strprintf(string &out,const char *format,...)
1164 {
1165 va_list args;
1166 va_start(args,format);
1167
1168 // sprintf the description
1169 char S[4096];
1170 vsnprintf(S,sizeof(S),format,args);
1171 out = string(S);
1172 }
1173 /*}}}*/
1174 // safe_snprintf - Safer snprintf /*{{{*/
1175 // ---------------------------------------------------------------------
1176 /* This is a snprintf that will never (ever) go past 'End' and returns a
1177 pointer to the end of the new string. The returned string is always null
1178 terminated unless Buffer == end. This is a better alterantive to using
1179 consecutive snprintfs. */
1180 char *safe_snprintf(char *Buffer,char *End,const char *Format,...)
1181 {
1182 va_list args;
1183 int Did;
1184
1185 va_start(args,Format);
1186
1187 if (End <= Buffer)
1188 return End;
1189
1190 Did = vsnprintf(Buffer,End - Buffer,Format,args);
1191 if (Did < 0 || Buffer + Did > End)
1192 return End;
1193 return Buffer + Did;
1194 }
1195 /*}}}*/
1196 // StripEpoch - Remove the version "epoch" from a version string /*{{{*/
1197 // ---------------------------------------------------------------------
1198 string StripEpoch(const string &VerStr)
1199 {
1200 size_t i = VerStr.find(":");
1201 if (i == string::npos)
1202 return VerStr;
1203 return VerStr.substr(i+1);
1204 }
1205
1206 // tolower_ascii - tolower() function that ignores the locale /*{{{*/
1207 // ---------------------------------------------------------------------
1208 /* This little function is the most called method we have and tries
1209 therefore to do the absolut minimum - and is noteable faster than
1210 standard tolower/toupper and as a bonus avoids problems with different
1211 locales - we only operate on ascii chars anyway. */
1212 int tolower_ascii(int const c)
1213 {
1214 if (c >= 'A' && c <= 'Z')
1215 return c + 32;
1216 return c;
1217 }
1218 /*}}}*/
1219
1220 // CheckDomainList - See if Host is in a , seperate list /*{{{*/
1221 // ---------------------------------------------------------------------
1222 /* The domain list is a comma seperate list of domains that are suffix
1223 matched against the argument */
1224 bool CheckDomainList(const string &Host,const string &List)
1225 {
1226 string::const_iterator Start = List.begin();
1227 for (string::const_iterator Cur = List.begin(); Cur <= List.end(); Cur++)
1228 {
1229 if (Cur < List.end() && *Cur != ',')
1230 continue;
1231
1232 // Match the end of the string..
1233 if ((Host.size() >= (unsigned)(Cur - Start)) &&
1234 Cur - Start != 0 &&
1235 stringcasecmp(Host.end() - (Cur - Start),Host.end(),Start,Cur) == 0)
1236 return true;
1237
1238 Start = Cur + 1;
1239 }
1240 return false;
1241 }
1242 /*}}}*/
1243
1244 // URI::CopyFrom - Copy from an object /*{{{*/
1245 // ---------------------------------------------------------------------
1246 /* This parses the URI into all of its components */
1247 void URI::CopyFrom(const string &U)
1248 {
1249 string::const_iterator I = U.begin();
1250
1251 // Locate the first colon, this separates the scheme
1252 for (; I < U.end() && *I != ':' ; I++);
1253 string::const_iterator FirstColon = I;
1254
1255 /* Determine if this is a host type URI with a leading double //
1256 and then search for the first single / */
1257 string::const_iterator SingleSlash = I;
1258 if (I + 3 < U.end() && I[1] == '/' && I[2] == '/')
1259 SingleSlash += 3;
1260
1261 /* Find the / indicating the end of the hostname, ignoring /'s in the
1262 square brackets */
1263 bool InBracket = false;
1264 for (; SingleSlash < U.end() && (*SingleSlash != '/' || InBracket == true); SingleSlash++)
1265 {
1266 if (*SingleSlash == '[')
1267 InBracket = true;
1268 if (InBracket == true && *SingleSlash == ']')
1269 InBracket = false;
1270 }
1271
1272 if (SingleSlash > U.end())
1273 SingleSlash = U.end();
1274
1275 // We can now write the access and path specifiers
1276 Access.assign(U.begin(),FirstColon);
1277 if (SingleSlash != U.end())
1278 Path.assign(SingleSlash,U.end());
1279 if (Path.empty() == true)
1280 Path = "/";
1281
1282 // Now we attempt to locate a user:pass@host fragment
1283 if (FirstColon + 2 <= U.end() && FirstColon[1] == '/' && FirstColon[2] == '/')
1284 FirstColon += 3;
1285 else
1286 FirstColon += 1;
1287 if (FirstColon >= U.end())
1288 return;
1289
1290 if (FirstColon > SingleSlash)
1291 FirstColon = SingleSlash;
1292
1293 // Find the colon...
1294 I = FirstColon + 1;
1295 if (I > SingleSlash)
1296 I = SingleSlash;
1297 for (; I < SingleSlash && *I != ':'; I++);
1298 string::const_iterator SecondColon = I;
1299
1300 // Search for the @ after the colon
1301 for (; I < SingleSlash && *I != '@'; I++);
1302 string::const_iterator At = I;
1303
1304 // Now write the host and user/pass
1305 if (At == SingleSlash)
1306 {
1307 if (FirstColon < SingleSlash)
1308 Host.assign(FirstColon,SingleSlash);
1309 }
1310 else
1311 {
1312 Host.assign(At+1,SingleSlash);
1313 // username and password must be encoded (RFC 3986)
1314 User.assign(DeQuoteString(FirstColon,SecondColon));
1315 if (SecondColon < At)
1316 Password.assign(DeQuoteString(SecondColon+1,At));
1317 }
1318
1319 // Now we parse the RFC 2732 [] hostnames.
1320 unsigned long PortEnd = 0;
1321 InBracket = false;
1322 for (unsigned I = 0; I != Host.length();)
1323 {
1324 if (Host[I] == '[')
1325 {
1326 InBracket = true;
1327 Host.erase(I,1);
1328 continue;
1329 }
1330
1331 if (InBracket == true && Host[I] == ']')
1332 {
1333 InBracket = false;
1334 Host.erase(I,1);
1335 PortEnd = I;
1336 continue;
1337 }
1338 I++;
1339 }
1340
1341 // Tsk, weird.
1342 if (InBracket == true)
1343 {
1344 Host.clear();
1345 return;
1346 }
1347
1348 // Now we parse off a port number from the hostname
1349 Port = 0;
1350 string::size_type Pos = Host.rfind(':');
1351 if (Pos == string::npos || Pos < PortEnd)
1352 return;
1353
1354 Port = atoi(string(Host,Pos+1).c_str());
1355 Host.assign(Host,0,Pos);
1356 }
1357 /*}}}*/
1358 // URI::operator string - Convert the URI to a string /*{{{*/
1359 // ---------------------------------------------------------------------
1360 /* */
1361 URI::operator string()
1362 {
1363 string Res;
1364
1365 if (Access.empty() == false)
1366 Res = Access + ':';
1367
1368 if (Host.empty() == false)
1369 {
1370 if (Access.empty() == false)
1371 Res += "//";
1372
1373 if (User.empty() == false)
1374 {
1375 Res += User;
1376 if (Password.empty() == false)
1377 Res += ":" + Password;
1378 Res += "@";
1379 }
1380
1381 // Add RFC 2732 escaping characters
1382 if (Access.empty() == false &&
1383 (Host.find('/') != string::npos || Host.find(':') != string::npos))
1384 Res += '[' + Host + ']';
1385 else
1386 Res += Host;
1387
1388 if (Port != 0)
1389 {
1390 char S[30];
1391 sprintf(S,":%u",Port);
1392 Res += S;
1393 }
1394 }
1395
1396 if (Path.empty() == false)
1397 {
1398 if (Path[0] != '/')
1399 Res += "/" + Path;
1400 else
1401 Res += Path;
1402 }
1403
1404 return Res;
1405 }
1406 /*}}}*/
1407 // URI::SiteOnly - Return the schema and site for the URI /*{{{*/
1408 // ---------------------------------------------------------------------
1409 /* */
1410 string URI::SiteOnly(const string &URI)
1411 {
1412 ::URI U(URI);
1413 U.User.clear();
1414 U.Password.clear();
1415 U.Path.clear();
1416 U.Port = 0;
1417 return U;
1418 }
1419 /*}}}*/
1420 // URI::NoUserPassword - Return the schema, site and path for the URI /*{{{*/
1421 // ---------------------------------------------------------------------
1422 /* */
1423 string URI::NoUserPassword(const string &URI)
1424 {
1425 ::URI U(URI);
1426 U.User.clear();
1427 U.Password.clear();
1428 U.Port = 0;
1429 return U;
1430 }
1431 /*}}}*/