]> git.saurik.com Git - apt.git/blame_incremental - apt-pkg/contrib/strutl.cc
FileFd: avoid further writing if file failed
[apt.git] / apt-pkg / contrib / strutl.cc
... / ...
CommitLineData
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 <stddef.h>
25#include <stdlib.h>
26#include <time.h>
27#include <string>
28#include <vector>
29#include <ctype.h>
30#include <string.h>
31#include <sstream>
32#include <stdio.h>
33#include <algorithm>
34#include <unistd.h>
35#include <regex.h>
36#include <errno.h>
37#include <stdarg.h>
38#include <iconv.h>
39
40#include <apti18n.h>
41 /*}}}*/
42using namespace std;
43
44// Strip - Remove white space from the front and back of a string /*{{{*/
45// ---------------------------------------------------------------------
46namespace APT {
47 namespace String {
48std::string Strip(const std::string &str)
49{
50 // ensure we have at least one character
51 if (str.empty() == true)
52 return str;
53
54 char const * const s = str.c_str();
55 size_t start = 0;
56 for (; isspace(s[start]) != 0; ++start)
57 ; // find the first not-space
58
59 // string contains only whitespaces
60 if (s[start] == '\0')
61 return "";
62
63 size_t end = str.length() - 1;
64 for (; isspace(s[end]) != 0; --end)
65 ; // find the last not-space
66
67 return str.substr(start, end - start + 1);
68}
69
70bool Endswith(const std::string &s, const std::string &end)
71{
72 if (end.size() > s.size())
73 return false;
74 return (s.substr(s.size() - end.size(), s.size()) == end);
75}
76
77bool Startswith(const std::string &s, const std::string &start)
78{
79 if (start.size() > s.size())
80 return false;
81 return (s.substr(0, start.size()) == start);
82}
83
84}
85}
86 /*}}}*/
87// UTF8ToCodeset - Convert some UTF-8 string for some codeset /*{{{*/
88// ---------------------------------------------------------------------
89/* This is handy to use before display some information for enduser */
90bool UTF8ToCodeset(const char *codeset, const string &orig, string *dest)
91{
92 iconv_t cd;
93 const char *inbuf;
94 char *inptr, *outbuf;
95 size_t insize, bufsize;
96 dest->clear();
97
98 cd = iconv_open(codeset, "UTF-8");
99 if (cd == (iconv_t)(-1)) {
100 // Something went wrong
101 if (errno == EINVAL)
102 _error->Error("conversion from 'UTF-8' to '%s' not available",
103 codeset);
104 else
105 perror("iconv_open");
106
107 return false;
108 }
109
110 insize = bufsize = orig.size();
111 inbuf = orig.data();
112 inptr = (char *)inbuf;
113 outbuf = new char[bufsize];
114 size_t lastError = -1;
115
116 while (insize != 0)
117 {
118 char *outptr = outbuf;
119 size_t outsize = bufsize;
120 size_t const err = iconv(cd, &inptr, &insize, &outptr, &outsize);
121 dest->append(outbuf, outptr - outbuf);
122 if (err == (size_t)(-1))
123 {
124 switch (errno)
125 {
126 case EILSEQ:
127 insize--;
128 inptr++;
129 // replace a series of unknown multibytes with a single "?"
130 if (lastError != insize) {
131 lastError = insize - 1;
132 dest->append("?");
133 }
134 break;
135 case EINVAL:
136 insize = 0;
137 break;
138 case E2BIG:
139 if (outptr == outbuf)
140 {
141 bufsize *= 2;
142 delete[] outbuf;
143 outbuf = new char[bufsize];
144 }
145 break;
146 }
147 }
148 }
149
150 delete[] outbuf;
151
152 iconv_close(cd);
153
154 return true;
155}
156 /*}}}*/
157// strstrip - Remove white space from the front and back of a string /*{{{*/
158// ---------------------------------------------------------------------
159/* This is handy to use when parsing a file. It also removes \n's left
160 over from fgets and company */
161char *_strstrip(char *String)
162{
163 for (;*String != 0 && (*String == ' ' || *String == '\t'); String++);
164
165 if (*String == 0)
166 return String;
167 return _strrstrip(String);
168}
169 /*}}}*/
170// strrstrip - Remove white space from the back of a string /*{{{*/
171// ---------------------------------------------------------------------
172char *_strrstrip(char *String)
173{
174 char *End = String + strlen(String) - 1;
175 for (;End != String - 1 && (*End == ' ' || *End == '\t' || *End == '\n' ||
176 *End == '\r'); End--);
177 End++;
178 *End = 0;
179 return String;
180}
181 /*}}}*/
182// strtabexpand - Converts tabs into 8 spaces /*{{{*/
183// ---------------------------------------------------------------------
184/* */
185char *_strtabexpand(char *String,size_t Len)
186{
187 for (char *I = String; I != I + Len && *I != 0; I++)
188 {
189 if (*I != '\t')
190 continue;
191 if (I + 8 > String + Len)
192 {
193 *I = 0;
194 return String;
195 }
196
197 /* Assume the start of the string is 0 and find the next 8 char
198 division */
199 int Len;
200 if (String == I)
201 Len = 1;
202 else
203 Len = 8 - ((String - I) % 8);
204 Len -= 2;
205 if (Len <= 0)
206 {
207 *I = ' ';
208 continue;
209 }
210
211 memmove(I + Len,I + 1,strlen(I) + 1);
212 for (char *J = I; J + Len != I; *I = ' ', I++);
213 }
214 return String;
215}
216 /*}}}*/
217// ParseQuoteWord - Parse a single word out of a string /*{{{*/
218// ---------------------------------------------------------------------
219/* This grabs a single word, converts any % escaped characters to their
220 proper values and advances the pointer. Double quotes are understood
221 and striped out as well. This is for URI/URL parsing. It also can
222 understand [] brackets.*/
223bool ParseQuoteWord(const char *&String,string &Res)
224{
225 // Skip leading whitespace
226 const char *C = String;
227 for (;*C != 0 && *C == ' '; C++);
228 if (*C == 0)
229 return false;
230
231 // Jump to the next word
232 for (;*C != 0 && isspace(*C) == 0; C++)
233 {
234 if (*C == '"')
235 {
236 C = strchr(C + 1, '"');
237 if (C == NULL)
238 return false;
239 }
240 if (*C == '[')
241 {
242 C = strchr(C + 1, ']');
243 if (C == NULL)
244 return false;
245 }
246 }
247
248 // Now de-quote characters
249 char Buffer[1024];
250 char Tmp[3];
251 const char *Start = String;
252 char *I;
253 for (I = Buffer; I < Buffer + sizeof(Buffer) && Start != C; I++)
254 {
255 if (*Start == '%' && Start + 2 < C &&
256 isxdigit(Start[1]) && isxdigit(Start[2]))
257 {
258 Tmp[0] = Start[1];
259 Tmp[1] = Start[2];
260 Tmp[2] = 0;
261 *I = (char)strtol(Tmp,0,16);
262 Start += 3;
263 continue;
264 }
265 if (*Start != '"')
266 *I = *Start;
267 else
268 I--;
269 Start++;
270 }
271 *I = 0;
272 Res = Buffer;
273
274 // Skip ending white space
275 for (;*C != 0 && isspace(*C) != 0; C++);
276 String = C;
277 return true;
278}
279 /*}}}*/
280// ParseCWord - Parses a string like a C "" expression /*{{{*/
281// ---------------------------------------------------------------------
282/* This expects a series of space separated strings enclosed in ""'s.
283 It concatenates the ""'s into a single string. */
284bool ParseCWord(const char *&String,string &Res)
285{
286 // Skip leading whitespace
287 const char *C = String;
288 for (;*C != 0 && *C == ' '; C++);
289 if (*C == 0)
290 return false;
291
292 char Buffer[1024];
293 char *Buf = Buffer;
294 if (strlen(String) >= sizeof(Buffer))
295 return false;
296
297 for (; *C != 0; C++)
298 {
299 if (*C == '"')
300 {
301 for (C++; *C != 0 && *C != '"'; C++)
302 *Buf++ = *C;
303
304 if (*C == 0)
305 return false;
306
307 continue;
308 }
309
310 if (C != String && isspace(*C) != 0 && isspace(C[-1]) != 0)
311 continue;
312 if (isspace(*C) == 0)
313 return false;
314 *Buf++ = ' ';
315 }
316 *Buf = 0;
317 Res = Buffer;
318 String = C;
319 return true;
320}
321 /*}}}*/
322// QuoteString - Convert a string into quoted from /*{{{*/
323// ---------------------------------------------------------------------
324/* */
325string QuoteString(const string &Str, const char *Bad)
326{
327 std::stringstream Res;
328 for (string::const_iterator I = Str.begin(); I != Str.end(); ++I)
329 {
330 if (strchr(Bad,*I) != 0 || isprint(*I) == 0 ||
331 *I == 0x25 || // percent '%' char
332 *I <= 0x20 || *I >= 0x7F) // control chars
333 {
334 ioprintf(Res, "%%%02hhx", *I);
335 }
336 else
337 Res << *I;
338 }
339 return Res.str();
340}
341 /*}}}*/
342// DeQuoteString - Convert a string from quoted from /*{{{*/
343// ---------------------------------------------------------------------
344/* This undoes QuoteString */
345string DeQuoteString(const string &Str)
346{
347 return DeQuoteString(Str.begin(),Str.end());
348}
349string DeQuoteString(string::const_iterator const &begin,
350 string::const_iterator const &end)
351{
352 string Res;
353 for (string::const_iterator I = begin; I != end; ++I)
354 {
355 if (*I == '%' && I + 2 < end &&
356 isxdigit(I[1]) && isxdigit(I[2]))
357 {
358 char Tmp[3];
359 Tmp[0] = I[1];
360 Tmp[1] = I[2];
361 Tmp[2] = 0;
362 Res += (char)strtol(Tmp,0,16);
363 I += 2;
364 continue;
365 }
366 else
367 Res += *I;
368 }
369 return Res;
370}
371
372 /*}}}*/
373// SizeToStr - Convert a long into a human readable size /*{{{*/
374// ---------------------------------------------------------------------
375/* A max of 4 digits are shown before conversion to the next highest unit.
376 The max length of the string will be 5 chars unless the size is > 10
377 YottaBytes (E24) */
378string SizeToStr(double Size)
379{
380 double ASize;
381 if (Size >= 0)
382 ASize = Size;
383 else
384 ASize = -1*Size;
385
386 /* bytes, KiloBytes, MegaBytes, GigaBytes, TeraBytes, PetaBytes,
387 ExaBytes, ZettaBytes, YottaBytes */
388 char Ext[] = {'\0','k','M','G','T','P','E','Z','Y'};
389 int I = 0;
390 while (I <= 8)
391 {
392 if (ASize < 100 && I != 0)
393 {
394 std::string S;
395 strprintf(S, "%'.1f %c", ASize, Ext[I]);
396 return S;
397 }
398
399 if (ASize < 10000)
400 {
401 std::string S;
402 strprintf(S, "%'.0f %c", ASize, Ext[I]);
403 return S;
404 }
405 ASize /= 1000.0;
406 I++;
407 }
408 return "";
409}
410 /*}}}*/
411// TimeToStr - Convert the time into a string /*{{{*/
412// ---------------------------------------------------------------------
413/* Converts a number of seconds to a hms format */
414string TimeToStr(unsigned long Sec)
415{
416 std::string S;
417 if (Sec > 60*60*24)
418 {
419 //TRANSLATOR: d means days, h means hours, min means minutes, s means seconds
420 strprintf(S,_("%lid %lih %limin %lis"),Sec/60/60/24,(Sec/60/60) % 24,(Sec/60) % 60,Sec % 60);
421 }
422 else if (Sec > 60*60)
423 {
424 //TRANSLATOR: h means hours, min means minutes, s means seconds
425 strprintf(S,_("%lih %limin %lis"),Sec/60/60,(Sec/60) % 60,Sec % 60);
426 }
427 else if (Sec > 60)
428 {
429 //TRANSLATOR: min means minutes, s means seconds
430 strprintf(S,_("%limin %lis"),Sec/60,Sec % 60);
431 }
432 else
433 {
434 //TRANSLATOR: s means seconds
435 strprintf(S,_("%lis"),Sec);
436 }
437 return S;
438}
439 /*}}}*/
440// SubstVar - Substitute a string for another string /*{{{*/
441// ---------------------------------------------------------------------
442/* This replaces all occurrences of Subst with Contents in Str. */
443string SubstVar(const string &Str,const string &Subst,const string &Contents)
444{
445 if (Subst.empty() == true)
446 return Str;
447
448 string::size_type Pos = 0;
449 string::size_type OldPos = 0;
450 string Temp;
451
452 while (OldPos < Str.length() &&
453 (Pos = Str.find(Subst,OldPos)) != string::npos)
454 {
455 if (OldPos != Pos)
456 Temp.append(Str, OldPos, Pos - OldPos);
457 if (Contents.empty() == false)
458 Temp.append(Contents);
459 OldPos = Pos + Subst.length();
460 }
461
462 if (OldPos == 0)
463 return Str;
464
465 if (OldPos >= Str.length())
466 return Temp;
467
468 Temp.append(Str, OldPos, string::npos);
469 return Temp;
470}
471string SubstVar(string Str,const struct SubstVar *Vars)
472{
473 for (; Vars->Subst != 0; Vars++)
474 Str = SubstVar(Str,Vars->Subst,*Vars->Contents);
475 return Str;
476}
477 /*}}}*/
478// OutputInDepth - return a string with separator multiplied with depth /*{{{*/
479// ---------------------------------------------------------------------
480/* Returns a string with the supplied separator depth + 1 times in it */
481std::string OutputInDepth(const unsigned long Depth, const char* Separator)
482{
483 std::string output = "";
484 for(unsigned long d=Depth+1; d > 0; d--)
485 output.append(Separator);
486 return output;
487}
488 /*}}}*/
489// URItoFileName - Convert the uri into a unique file name /*{{{*/
490// ---------------------------------------------------------------------
491/* This converts a URI into a safe filename. It quotes all unsafe characters
492 and converts / to _ and removes the scheme identifier. The resulting
493 file name should be unique and never occur again for a different file */
494string URItoFileName(const string &URI)
495{
496 // Nuke 'sensitive' items
497 ::URI U(URI);
498 U.User.clear();
499 U.Password.clear();
500 U.Access.clear();
501
502 // "\x00-\x20{}|\\\\^\\[\\]<>\"\x7F-\xFF";
503 string NewURI = QuoteString(U,"\\|{}[]<>\"^~_=!@#$%^&*");
504 replace(NewURI.begin(),NewURI.end(),'/','_');
505 return NewURI;
506}
507 /*}}}*/
508// Base64Encode - Base64 Encoding routine for short strings /*{{{*/
509// ---------------------------------------------------------------------
510/* This routine performs a base64 transformation on a string. It was ripped
511 from wget and then patched and bug fixed.
512
513 This spec can be found in rfc2045 */
514string Base64Encode(const string &S)
515{
516 // Conversion table.
517 static char tbl[64] = {'A','B','C','D','E','F','G','H',
518 'I','J','K','L','M','N','O','P',
519 'Q','R','S','T','U','V','W','X',
520 'Y','Z','a','b','c','d','e','f',
521 'g','h','i','j','k','l','m','n',
522 'o','p','q','r','s','t','u','v',
523 'w','x','y','z','0','1','2','3',
524 '4','5','6','7','8','9','+','/'};
525
526 // Pre-allocate some space
527 string Final;
528 Final.reserve((4*S.length() + 2)/3 + 2);
529
530 /* Transform the 3x8 bits to 4x6 bits, as required by
531 base64. */
532 for (string::const_iterator I = S.begin(); I < S.end(); I += 3)
533 {
534 char Bits[3] = {0,0,0};
535 Bits[0] = I[0];
536 if (I + 1 < S.end())
537 Bits[1] = I[1];
538 if (I + 2 < S.end())
539 Bits[2] = I[2];
540
541 Final += tbl[Bits[0] >> 2];
542 Final += tbl[((Bits[0] & 3) << 4) + (Bits[1] >> 4)];
543
544 if (I + 1 >= S.end())
545 break;
546
547 Final += tbl[((Bits[1] & 0xf) << 2) + (Bits[2] >> 6)];
548
549 if (I + 2 >= S.end())
550 break;
551
552 Final += tbl[Bits[2] & 0x3f];
553 }
554
555 /* Apply the padding elements, this tells how many bytes the remote
556 end should discard */
557 if (S.length() % 3 == 2)
558 Final += '=';
559 if (S.length() % 3 == 1)
560 Final += "==";
561
562 return Final;
563}
564 /*}}}*/
565// stringcmp - Arbitrary string compare /*{{{*/
566// ---------------------------------------------------------------------
567/* This safely compares two non-null terminated strings of arbitrary
568 length */
569int stringcmp(const char *A,const char *AEnd,const char *B,const char *BEnd)
570{
571 for (; A != AEnd && B != BEnd; A++, B++)
572 if (*A != *B)
573 break;
574
575 if (A == AEnd && B == BEnd)
576 return 0;
577 if (A == AEnd)
578 return 1;
579 if (B == BEnd)
580 return -1;
581 if (*A < *B)
582 return -1;
583 return 1;
584}
585
586#if __GNUC__ >= 3
587int stringcmp(string::const_iterator A,string::const_iterator AEnd,
588 const char *B,const char *BEnd)
589{
590 for (; A != AEnd && B != BEnd; A++, B++)
591 if (*A != *B)
592 break;
593
594 if (A == AEnd && B == BEnd)
595 return 0;
596 if (A == AEnd)
597 return 1;
598 if (B == BEnd)
599 return -1;
600 if (*A < *B)
601 return -1;
602 return 1;
603}
604int stringcmp(string::const_iterator A,string::const_iterator AEnd,
605 string::const_iterator B,string::const_iterator BEnd)
606{
607 for (; A != AEnd && B != BEnd; A++, B++)
608 if (*A != *B)
609 break;
610
611 if (A == AEnd && B == BEnd)
612 return 0;
613 if (A == AEnd)
614 return 1;
615 if (B == BEnd)
616 return -1;
617 if (*A < *B)
618 return -1;
619 return 1;
620}
621#endif
622 /*}}}*/
623// stringcasecmp - Arbitrary case insensitive string compare /*{{{*/
624// ---------------------------------------------------------------------
625/* */
626int stringcasecmp(const char *A,const char *AEnd,const char *B,const char *BEnd)
627{
628 for (; A != AEnd && B != BEnd; A++, B++)
629 if (tolower_ascii(*A) != tolower_ascii(*B))
630 break;
631
632 if (A == AEnd && B == BEnd)
633 return 0;
634 if (A == AEnd)
635 return 1;
636 if (B == BEnd)
637 return -1;
638 if (tolower_ascii(*A) < tolower_ascii(*B))
639 return -1;
640 return 1;
641}
642#if __GNUC__ >= 3
643int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
644 const char *B,const char *BEnd)
645{
646 for (; A != AEnd && B != BEnd; A++, B++)
647 if (tolower_ascii(*A) != tolower_ascii(*B))
648 break;
649
650 if (A == AEnd && B == BEnd)
651 return 0;
652 if (A == AEnd)
653 return 1;
654 if (B == BEnd)
655 return -1;
656 if (tolower_ascii(*A) < tolower_ascii(*B))
657 return -1;
658 return 1;
659}
660int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
661 string::const_iterator B,string::const_iterator BEnd)
662{
663 for (; A != AEnd && B != BEnd; A++, B++)
664 if (tolower_ascii(*A) != tolower_ascii(*B))
665 break;
666
667 if (A == AEnd && B == BEnd)
668 return 0;
669 if (A == AEnd)
670 return 1;
671 if (B == BEnd)
672 return -1;
673 if (tolower_ascii(*A) < tolower_ascii(*B))
674 return -1;
675 return 1;
676}
677#endif
678 /*}}}*/
679// LookupTag - Lookup the value of a tag in a taged string /*{{{*/
680// ---------------------------------------------------------------------
681/* The format is like those used in package files and the method
682 communication system */
683string LookupTag(const string &Message,const char *Tag,const char *Default)
684{
685 // Look for a matching tag.
686 int Length = strlen(Tag);
687 for (string::const_iterator I = Message.begin(); I + Length < Message.end(); ++I)
688 {
689 // Found the tag
690 if (I[Length] == ':' && stringcasecmp(I,I+Length,Tag) == 0)
691 {
692 // Find the end of line and strip the leading/trailing spaces
693 string::const_iterator J;
694 I += Length + 1;
695 for (; isspace_ascii(*I) != 0 && I < Message.end(); ++I);
696 for (J = I; *J != '\n' && J < Message.end(); ++J);
697 for (; J > I && isspace_ascii(J[-1]) != 0; --J);
698
699 return string(I,J);
700 }
701
702 for (; *I != '\n' && I < Message.end(); ++I);
703 }
704
705 // Failed to find a match
706 if (Default == 0)
707 return string();
708 return Default;
709}
710 /*}}}*/
711// StringToBool - Converts a string into a boolean /*{{{*/
712// ---------------------------------------------------------------------
713/* This inspects the string to see if it is true or if it is false and
714 then returns the result. Several varients on true/false are checked. */
715int StringToBool(const string &Text,int Default)
716{
717 char *ParseEnd;
718 int Res = strtol(Text.c_str(),&ParseEnd,0);
719 // ensure that the entire string was converted by strtol to avoid
720 // failures on "apt-cache show -a 0ad" where the "0" is converted
721 const char *TextEnd = Text.c_str()+Text.size();
722 if (ParseEnd == TextEnd && Res >= 0 && Res <= 1)
723 return Res;
724
725 // Check for positives
726 if (strcasecmp(Text.c_str(),"no") == 0 ||
727 strcasecmp(Text.c_str(),"false") == 0 ||
728 strcasecmp(Text.c_str(),"without") == 0 ||
729 strcasecmp(Text.c_str(),"off") == 0 ||
730 strcasecmp(Text.c_str(),"disable") == 0)
731 return 0;
732
733 // Check for negatives
734 if (strcasecmp(Text.c_str(),"yes") == 0 ||
735 strcasecmp(Text.c_str(),"true") == 0 ||
736 strcasecmp(Text.c_str(),"with") == 0 ||
737 strcasecmp(Text.c_str(),"on") == 0 ||
738 strcasecmp(Text.c_str(),"enable") == 0)
739 return 1;
740
741 return Default;
742}
743 /*}}}*/
744// TimeRFC1123 - Convert a time_t into RFC1123 format /*{{{*/
745// ---------------------------------------------------------------------
746/* This converts a time_t into a string time representation that is
747 year 2000 complient and timezone neutral */
748string TimeRFC1123(time_t Date)
749{
750 struct tm Conv;
751 if (gmtime_r(&Date, &Conv) == NULL)
752 return "";
753
754 char Buf[300];
755 const char *Day[] = {"Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
756 const char *Month[] = {"Jan","Feb","Mar","Apr","May","Jun","Jul",
757 "Aug","Sep","Oct","Nov","Dec"};
758
759 snprintf(Buf, sizeof(Buf), "%s, %02i %s %i %02i:%02i:%02i GMT",Day[Conv.tm_wday],
760 Conv.tm_mday,Month[Conv.tm_mon],Conv.tm_year+1900,Conv.tm_hour,
761 Conv.tm_min,Conv.tm_sec);
762 return Buf;
763}
764 /*}}}*/
765// ReadMessages - Read messages from the FD /*{{{*/
766// ---------------------------------------------------------------------
767/* This pulls full messages from the input FD into the message buffer.
768 It assumes that messages will not pause during transit so no
769 fancy buffering is used.
770
771 In particular: this reads blocks from the input until it believes
772 that it's run out of input text. Each block is terminated by a
773 double newline ('\n' followed by '\n').
774 */
775bool ReadMessages(int Fd, vector<string> &List)
776{
777 char Buffer[64000];
778 // Represents any left-over from the previous iteration of the
779 // parse loop. (i.e., if a message is split across the end
780 // of the buffer, it goes here)
781 string PartialMessage;
782
783 do {
784 int const Res = read(Fd, Buffer, sizeof(Buffer));
785 if (Res < 0 && errno == EINTR)
786 continue;
787
788 // process we read from has died
789 if (Res == 0)
790 return false;
791
792 // No data
793#if EAGAIN != EWOULDBLOCK
794 if (Res < 0 && (errno == EAGAIN || errno == EWOULDBLOCK))
795#else
796 if (Res < 0 && errno == EAGAIN)
797#endif
798 return true;
799 if (Res < 0)
800 return false;
801
802 // extract the message(s) from the buffer
803 char const *Start = Buffer;
804 char const * const End = Buffer + Res;
805
806 char const * NL = (char const *) memchr(Start, '\n', End - Start);
807 if (NL == NULL)
808 {
809 // end of buffer: store what we have so far and read new data in
810 PartialMessage.append(Start, End - Start);
811 Start = End;
812 }
813 else
814 ++NL;
815
816 if (PartialMessage.empty() == false && Start < End)
817 {
818 // if we start with a new line, see if the partial message we have ended with one
819 // so that we properly detect records ending between two read() runs
820 // cases are: \n|\n , \r\n|\r\n and \r\n\r|\n
821 // the case \r|\n\r\n is handled by the usual double-newline handling
822 if ((NL - Start) == 1 || ((NL - Start) == 2 && *Start == '\r'))
823 {
824 if (APT::String::Endswith(PartialMessage, "\n") || APT::String::Endswith(PartialMessage, "\r\n\r"))
825 {
826 PartialMessage.erase(PartialMessage.find_last_not_of("\r\n") + 1);
827 List.push_back(PartialMessage);
828 PartialMessage.clear();
829 while (NL < End && (*NL == '\n' || *NL == '\r')) ++NL;
830 Start = NL;
831 }
832 }
833 }
834
835 while (Start < End) {
836 char const * NL2 = (char const *) memchr(NL, '\n', End - NL);
837 if (NL2 == NULL)
838 {
839 // end of buffer: store what we have so far and read new data in
840 PartialMessage.append(Start, End - Start);
841 break;
842 }
843 ++NL2;
844
845 // did we find a double newline?
846 if ((NL2 - NL) == 1 || ((NL2 - NL) == 2 && *NL == '\r'))
847 {
848 PartialMessage.append(Start, NL2 - Start);
849 PartialMessage.erase(PartialMessage.find_last_not_of("\r\n") + 1);
850 List.push_back(PartialMessage);
851 PartialMessage.clear();
852 while (NL2 < End && (*NL2 == '\n' || *NL2 == '\r')) ++NL2;
853 Start = NL2;
854 }
855 NL = NL2;
856 }
857
858 // we have read at least one complete message and nothing left
859 if (PartialMessage.empty() == true)
860 return true;
861
862 if (WaitFd(Fd) == false)
863 return false;
864 } while (true);
865}
866 /*}}}*/
867// MonthConv - Converts a month string into a number /*{{{*/
868// ---------------------------------------------------------------------
869/* This was lifted from the boa webserver which lifted it from 'wn-v1.07'
870 Made it a bit more robust with a few tolower_ascii though. */
871static int MonthConv(char *Month)
872{
873 switch (tolower_ascii(*Month))
874 {
875 case 'a':
876 return tolower_ascii(Month[1]) == 'p'?3:7;
877 case 'd':
878 return 11;
879 case 'f':
880 return 1;
881 case 'j':
882 if (tolower_ascii(Month[1]) == 'a')
883 return 0;
884 return tolower_ascii(Month[2]) == 'n'?5:6;
885 case 'm':
886 return tolower_ascii(Month[2]) == 'r'?2:4;
887 case 'n':
888 return 10;
889 case 'o':
890 return 9;
891 case 's':
892 return 8;
893
894 // Pretend it is January..
895 default:
896 return 0;
897 }
898}
899 /*}}}*/
900// timegm - Internal timegm if the gnu version is not available /*{{{*/
901// ---------------------------------------------------------------------
902/* Converts struct tm to time_t, assuming the data in tm is UTC rather
903 than local timezone (mktime assumes the latter).
904
905 This function is a nonstandard GNU extension that is also present on
906 the BSDs and maybe other systems. For others we follow the advice of
907 the manpage of timegm and use his portable replacement. */
908#ifndef HAVE_TIMEGM
909static time_t timegm(struct tm *t)
910{
911 char *tz = getenv("TZ");
912 setenv("TZ", "", 1);
913 tzset();
914 time_t ret = mktime(t);
915 if (tz)
916 setenv("TZ", tz, 1);
917 else
918 unsetenv("TZ");
919 tzset();
920 return ret;
921}
922#endif
923 /*}}}*/
924// FullDateToTime - Converts a HTTP1.1 full date strings into a time_t /*{{{*/
925// ---------------------------------------------------------------------
926/* tries to parses a full date as specified in RFC2616 Section 3.3.1
927 with one exception: All timezones (%Z) are accepted but the protocol
928 says that it MUST be GMT, but this one is equal to UTC which we will
929 encounter from time to time (e.g. in Release files) so we accept all
930 here and just assume it is GMT (or UTC) later on */
931bool RFC1123StrToTime(const char* const str,time_t &time)
932{
933 struct tm Tm;
934 setlocale (LC_ALL,"C");
935 bool const invalid =
936 // Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
937 (strptime(str, "%a, %d %b %Y %H:%M:%S %Z", &Tm) == NULL &&
938 // Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
939 strptime(str, "%A, %d-%b-%y %H:%M:%S %Z", &Tm) == NULL &&
940 // Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
941 strptime(str, "%a %b %d %H:%M:%S %Y", &Tm) == NULL);
942 setlocale (LC_ALL,"");
943 if (invalid == true)
944 return false;
945
946 time = timegm(&Tm);
947 return true;
948}
949 /*}}}*/
950// FTPMDTMStrToTime - Converts a ftp modification date into a time_t /*{{{*/
951// ---------------------------------------------------------------------
952/* */
953bool FTPMDTMStrToTime(const char* const str,time_t &time)
954{
955 struct tm Tm;
956 // MDTM includes no whitespaces but recommend and ignored by strptime
957 if (strptime(str, "%Y %m %d %H %M %S", &Tm) == NULL)
958 return false;
959
960 time = timegm(&Tm);
961 return true;
962}
963 /*}}}*/
964// StrToTime - Converts a string into a time_t /*{{{*/
965// ---------------------------------------------------------------------
966/* This handles all 3 popular time formats including RFC 1123, RFC 1036
967 and the C library asctime format. It requires the GNU library function
968 'timegm' to convert a struct tm in UTC to a time_t. For some bizzar
969 reason the C library does not provide any such function :< This also
970 handles the weird, but unambiguous FTP time format*/
971bool StrToTime(const string &Val,time_t &Result)
972{
973 struct tm Tm;
974 char Month[10];
975
976 // Skip the day of the week
977 const char *I = strchr(Val.c_str(), ' ');
978
979 // Handle RFC 1123 time
980 Month[0] = 0;
981 if (sscanf(I," %2d %3s %4d %2d:%2d:%2d GMT",&Tm.tm_mday,Month,&Tm.tm_year,
982 &Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) != 6)
983 {
984 // Handle RFC 1036 time
985 if (sscanf(I," %2d-%3s-%3d %2d:%2d:%2d GMT",&Tm.tm_mday,Month,
986 &Tm.tm_year,&Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) == 6)
987 Tm.tm_year += 1900;
988 else
989 {
990 // asctime format
991 if (sscanf(I," %3s %2d %2d:%2d:%2d %4d",Month,&Tm.tm_mday,
992 &Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec,&Tm.tm_year) != 6)
993 {
994 // 'ftp' time
995 if (sscanf(Val.c_str(),"%4d%2d%2d%2d%2d%2d",&Tm.tm_year,&Tm.tm_mon,
996 &Tm.tm_mday,&Tm.tm_hour,&Tm.tm_min,&Tm.tm_sec) != 6)
997 return false;
998 Tm.tm_mon--;
999 }
1000 }
1001 }
1002
1003 Tm.tm_isdst = 0;
1004 if (Month[0] != 0)
1005 Tm.tm_mon = MonthConv(Month);
1006 else
1007 Tm.tm_mon = 0; // we don't have a month, so pick something
1008 Tm.tm_year -= 1900;
1009
1010 // Convert to local time and then to GMT
1011 Result = timegm(&Tm);
1012 return true;
1013}
1014 /*}}}*/
1015// StrToNum - Convert a fixed length string to a number /*{{{*/
1016// ---------------------------------------------------------------------
1017/* This is used in decoding the crazy fixed length string headers in
1018 tar and ar files. */
1019bool StrToNum(const char *Str,unsigned long &Res,unsigned Len,unsigned Base)
1020{
1021 char S[30];
1022 if (Len >= sizeof(S))
1023 return false;
1024 memcpy(S,Str,Len);
1025 S[Len] = 0;
1026
1027 // All spaces is a zero
1028 Res = 0;
1029 unsigned I;
1030 for (I = 0; S[I] == ' '; I++);
1031 if (S[I] == 0)
1032 return true;
1033
1034 char *End;
1035 Res = strtoul(S,&End,Base);
1036 if (End == S)
1037 return false;
1038
1039 return true;
1040}
1041 /*}}}*/
1042// StrToNum - Convert a fixed length string to a number /*{{{*/
1043// ---------------------------------------------------------------------
1044/* This is used in decoding the crazy fixed length string headers in
1045 tar and ar files. */
1046bool StrToNum(const char *Str,unsigned long long &Res,unsigned Len,unsigned Base)
1047{
1048 char S[30];
1049 if (Len >= sizeof(S))
1050 return false;
1051 memcpy(S,Str,Len);
1052 S[Len] = 0;
1053
1054 // All spaces is a zero
1055 Res = 0;
1056 unsigned I;
1057 for (I = 0; S[I] == ' '; I++);
1058 if (S[I] == 0)
1059 return true;
1060
1061 char *End;
1062 Res = strtoull(S,&End,Base);
1063 if (End == S)
1064 return false;
1065
1066 return true;
1067}
1068 /*}}}*/
1069
1070// Base256ToNum - Convert a fixed length binary to a number /*{{{*/
1071// ---------------------------------------------------------------------
1072/* This is used in decoding the 256bit encoded fixed length fields in
1073 tar files */
1074bool Base256ToNum(const char *Str,unsigned long long &Res,unsigned int Len)
1075{
1076 if ((Str[0] & 0x80) == 0)
1077 return false;
1078 else
1079 {
1080 Res = Str[0] & 0x7F;
1081 for(unsigned int i = 1; i < Len; ++i)
1082 Res = (Res<<8) + Str[i];
1083 return true;
1084 }
1085}
1086 /*}}}*/
1087// Base256ToNum - Convert a fixed length binary to a number /*{{{*/
1088// ---------------------------------------------------------------------
1089/* This is used in decoding the 256bit encoded fixed length fields in
1090 tar files */
1091bool Base256ToNum(const char *Str,unsigned long &Res,unsigned int Len)
1092{
1093 unsigned long long Num;
1094 bool rc;
1095
1096 rc = Base256ToNum(Str, Num, Len);
1097 Res = Num;
1098 if (Res != Num)
1099 return false;
1100
1101 return rc;
1102}
1103 /*}}}*/
1104// HexDigit - Convert a hex character into an integer /*{{{*/
1105// ---------------------------------------------------------------------
1106/* Helper for Hex2Num */
1107static int HexDigit(int c)
1108{
1109 if (c >= '0' && c <= '9')
1110 return c - '0';
1111 if (c >= 'a' && c <= 'f')
1112 return c - 'a' + 10;
1113 if (c >= 'A' && c <= 'F')
1114 return c - 'A' + 10;
1115 return -1;
1116}
1117 /*}}}*/
1118// Hex2Num - Convert a long hex number into a buffer /*{{{*/
1119// ---------------------------------------------------------------------
1120/* The length of the buffer must be exactly 1/2 the length of the string. */
1121bool Hex2Num(const string &Str,unsigned char *Num,unsigned int Length)
1122{
1123 return Hex2Num(APT::StringView(Str), Num, Length);
1124}
1125
1126bool Hex2Num(const APT::StringView Str,unsigned char *Num,unsigned int Length)
1127{
1128 if (Str.length() != Length*2)
1129 return false;
1130
1131 // Convert each digit. We store it in the same order as the string
1132 int J = 0;
1133 for (auto I = Str.begin(); I != Str.end();J++, I += 2)
1134 {
1135 int first_half = HexDigit(I[0]);
1136 int second_half;
1137 if (first_half < 0)
1138 return false;
1139
1140 second_half = HexDigit(I[1]);
1141 if (second_half < 0)
1142 return false;
1143 Num[J] = first_half << 4;
1144 Num[J] += second_half;
1145 }
1146
1147 return true;
1148}
1149 /*}}}*/
1150// TokSplitString - Split a string up by a given token /*{{{*/
1151// ---------------------------------------------------------------------
1152/* This is intended to be a faster splitter, it does not use dynamic
1153 memories. Input is changed to insert nulls at each token location. */
1154bool TokSplitString(char Tok,char *Input,char **List,
1155 unsigned long ListMax)
1156{
1157 // Strip any leading spaces
1158 char *Start = Input;
1159 char *Stop = Start + strlen(Start);
1160 for (; *Start != 0 && isspace(*Start) != 0; Start++);
1161
1162 unsigned long Count = 0;
1163 char *Pos = Start;
1164 while (Pos != Stop)
1165 {
1166 // Skip to the next Token
1167 for (; Pos != Stop && *Pos != Tok; Pos++);
1168
1169 // Back remove spaces
1170 char *End = Pos;
1171 for (; End > Start && (End[-1] == Tok || isspace(End[-1]) != 0); End--);
1172 *End = 0;
1173
1174 List[Count++] = Start;
1175 if (Count >= ListMax)
1176 {
1177 List[Count-1] = 0;
1178 return false;
1179 }
1180
1181 // Advance pos
1182 for (; Pos != Stop && (*Pos == Tok || isspace(*Pos) != 0 || *Pos == 0); Pos++);
1183 Start = Pos;
1184 }
1185
1186 List[Count] = 0;
1187 return true;
1188}
1189 /*}}}*/
1190// VectorizeString - Split a string up into a vector of strings /*{{{*/
1191// ---------------------------------------------------------------------
1192/* This can be used to split a given string up into a vector, so the
1193 propose is the same as in the method above and this one is a bit slower
1194 also, but the advantage is that we have an iteratable vector */
1195vector<string> VectorizeString(string const &haystack, char const &split)
1196{
1197 vector<string> exploded;
1198 if (haystack.empty() == true)
1199 return exploded;
1200 string::const_iterator start = haystack.begin();
1201 string::const_iterator end = start;
1202 do {
1203 for (; end != haystack.end() && *end != split; ++end);
1204 exploded.push_back(string(start, end));
1205 start = end + 1;
1206 } while (end != haystack.end() && (++end) != haystack.end());
1207 return exploded;
1208}
1209 /*}}}*/
1210// StringSplit - split a string into a string vector by token /*{{{*/
1211// ---------------------------------------------------------------------
1212/* See header for details.
1213 */
1214vector<string> StringSplit(std::string const &s, std::string const &sep,
1215 unsigned int maxsplit)
1216{
1217 vector<string> split;
1218 size_t start, pos;
1219
1220 // no separator given, this is bogus
1221 if(sep.size() == 0)
1222 return split;
1223
1224 start = pos = 0;
1225 while (pos != string::npos)
1226 {
1227 pos = s.find(sep, start);
1228 split.push_back(s.substr(start, pos-start));
1229
1230 // if maxsplit is reached, the remaining string is the last item
1231 if(split.size() >= maxsplit)
1232 {
1233 split[split.size()-1] = s.substr(start);
1234 break;
1235 }
1236 start = pos+sep.size();
1237 }
1238 return split;
1239}
1240 /*}}}*/
1241// RegexChoice - Simple regex list/list matcher /*{{{*/
1242// ---------------------------------------------------------------------
1243/* */
1244unsigned long RegexChoice(RxChoiceList *Rxs,const char **ListBegin,
1245 const char **ListEnd)
1246{
1247 for (RxChoiceList *R = Rxs; R->Str != 0; R++)
1248 R->Hit = false;
1249
1250 unsigned long Hits = 0;
1251 for (; ListBegin < ListEnd; ++ListBegin)
1252 {
1253 // Check if the name is a regex
1254 const char *I;
1255 bool Regex = true;
1256 for (I = *ListBegin; *I != 0; I++)
1257 if (*I == '.' || *I == '?' || *I == '*' || *I == '|')
1258 break;
1259 if (*I == 0)
1260 Regex = false;
1261
1262 // Compile the regex pattern
1263 regex_t Pattern;
1264 if (Regex == true)
1265 if (regcomp(&Pattern,*ListBegin,REG_EXTENDED | REG_ICASE |
1266 REG_NOSUB) != 0)
1267 Regex = false;
1268
1269 // Search the list
1270 bool Done = false;
1271 for (RxChoiceList *R = Rxs; R->Str != 0; R++)
1272 {
1273 if (R->Str[0] == 0)
1274 continue;
1275
1276 if (strcasecmp(R->Str,*ListBegin) != 0)
1277 {
1278 if (Regex == false)
1279 continue;
1280 if (regexec(&Pattern,R->Str,0,0,0) != 0)
1281 continue;
1282 }
1283 Done = true;
1284
1285 if (R->Hit == false)
1286 Hits++;
1287
1288 R->Hit = true;
1289 }
1290
1291 if (Regex == true)
1292 regfree(&Pattern);
1293
1294 if (Done == false)
1295 _error->Warning(_("Selection %s not found"),*ListBegin);
1296 }
1297
1298 return Hits;
1299}
1300 /*}}}*/
1301// {str,io}printf - C format string outputter to C++ strings/iostreams /*{{{*/
1302// ---------------------------------------------------------------------
1303/* This is used to make the internationalization strings easier to translate
1304 and to allow reordering of parameters */
1305static bool iovprintf(ostream &out, const char *format,
1306 va_list &args, ssize_t &size) {
1307 char *S = (char*)malloc(size);
1308 ssize_t const n = vsnprintf(S, size, format, args);
1309 if (n > -1 && n < size) {
1310 out << S;
1311 free(S);
1312 return true;
1313 } else {
1314 if (n > -1)
1315 size = n + 1;
1316 else
1317 size *= 2;
1318 }
1319 free(S);
1320 return false;
1321}
1322void ioprintf(ostream &out,const char *format,...)
1323{
1324 va_list args;
1325 ssize_t size = 400;
1326 while (true) {
1327 bool ret;
1328 va_start(args,format);
1329 ret = iovprintf(out, format, args, size);
1330 va_end(args);
1331 if (ret == true)
1332 return;
1333 }
1334}
1335void strprintf(string &out,const char *format,...)
1336{
1337 va_list args;
1338 ssize_t size = 400;
1339 std::ostringstream outstr;
1340 while (true) {
1341 bool ret;
1342 va_start(args,format);
1343 ret = iovprintf(outstr, format, args, size);
1344 va_end(args);
1345 if (ret == true)
1346 break;
1347 }
1348 out = outstr.str();
1349}
1350 /*}}}*/
1351// safe_snprintf - Safer snprintf /*{{{*/
1352// ---------------------------------------------------------------------
1353/* This is a snprintf that will never (ever) go past 'End' and returns a
1354 pointer to the end of the new string. The returned string is always null
1355 terminated unless Buffer == end. This is a better alterantive to using
1356 consecutive snprintfs. */
1357char *safe_snprintf(char *Buffer,char *End,const char *Format,...)
1358{
1359 va_list args;
1360 int Did;
1361
1362 if (End <= Buffer)
1363 return End;
1364 va_start(args,Format);
1365 Did = vsnprintf(Buffer,End - Buffer,Format,args);
1366 va_end(args);
1367
1368 if (Did < 0 || Buffer + Did > End)
1369 return End;
1370 return Buffer + Did;
1371}
1372 /*}}}*/
1373// StripEpoch - Remove the version "epoch" from a version string /*{{{*/
1374// ---------------------------------------------------------------------
1375string StripEpoch(const string &VerStr)
1376{
1377 size_t i = VerStr.find(":");
1378 if (i == string::npos)
1379 return VerStr;
1380 return VerStr.substr(i+1);
1381}
1382 /*}}}*/
1383
1384// tolower_ascii - tolower() function that ignores the locale /*{{{*/
1385// ---------------------------------------------------------------------
1386/* This little function is the most called method we have and tries
1387 therefore to do the absolut minimum - and is notable faster than
1388 standard tolower/toupper and as a bonus avoids problems with different
1389 locales - we only operate on ascii chars anyway. */
1390#undef tolower_ascii
1391int tolower_ascii(int const c) APT_CONST APT_COLD;
1392int tolower_ascii(int const c)
1393{
1394 return tolower_ascii_inline(c);
1395}
1396 /*}}}*/
1397
1398// isspace_ascii - isspace() function that ignores the locale /*{{{*/
1399// ---------------------------------------------------------------------
1400/* This little function is one of the most called methods we have and tries
1401 therefore to do the absolut minimum - and is notable faster than
1402 standard isspace() and as a bonus avoids problems with different
1403 locales - we only operate on ascii chars anyway. */
1404#undef isspace_ascii
1405int isspace_ascii(int const c) APT_CONST APT_COLD;
1406int isspace_ascii(int const c)
1407{
1408 return isspace_ascii_inline(c);
1409}
1410 /*}}}*/
1411
1412// CheckDomainList - See if Host is in a , separate list /*{{{*/
1413// ---------------------------------------------------------------------
1414/* The domain list is a comma separate list of domains that are suffix
1415 matched against the argument */
1416bool CheckDomainList(const string &Host,const string &List)
1417{
1418 string::const_iterator Start = List.begin();
1419 for (string::const_iterator Cur = List.begin(); Cur <= List.end(); ++Cur)
1420 {
1421 if (Cur < List.end() && *Cur != ',')
1422 continue;
1423
1424 // Match the end of the string..
1425 if ((Host.size() >= (unsigned)(Cur - Start)) &&
1426 Cur - Start != 0 &&
1427 stringcasecmp(Host.end() - (Cur - Start),Host.end(),Start,Cur) == 0)
1428 return true;
1429
1430 Start = Cur + 1;
1431 }
1432 return false;
1433}
1434 /*}}}*/
1435// strv_length - Return the length of a NULL-terminated string array /*{{{*/
1436// ---------------------------------------------------------------------
1437/* */
1438size_t strv_length(const char **str_array)
1439{
1440 size_t i;
1441 for (i=0; str_array[i] != NULL; i++)
1442 /* nothing */
1443 ;
1444 return i;
1445}
1446 /*}}}*/
1447// DeEscapeString - unescape (\0XX and \xXX) from a string /*{{{*/
1448// ---------------------------------------------------------------------
1449/* */
1450string DeEscapeString(const string &input)
1451{
1452 char tmp[3];
1453 string::const_iterator it;
1454 string output;
1455 for (it = input.begin(); it != input.end(); ++it)
1456 {
1457 // just copy non-escape chars
1458 if (*it != '\\')
1459 {
1460 output += *it;
1461 continue;
1462 }
1463
1464 // deal with double escape
1465 if (*it == '\\' &&
1466 (it + 1 < input.end()) && it[1] == '\\')
1467 {
1468 // copy
1469 output += *it;
1470 // advance iterator one step further
1471 ++it;
1472 continue;
1473 }
1474
1475 // ensure we have a char to read
1476 if (it + 1 == input.end())
1477 continue;
1478
1479 // read it
1480 ++it;
1481 switch (*it)
1482 {
1483 case '0':
1484 if (it + 2 <= input.end()) {
1485 tmp[0] = it[1];
1486 tmp[1] = it[2];
1487 tmp[2] = 0;
1488 output += (char)strtol(tmp, 0, 8);
1489 it += 2;
1490 }
1491 break;
1492 case 'x':
1493 if (it + 2 <= input.end()) {
1494 tmp[0] = it[1];
1495 tmp[1] = it[2];
1496 tmp[2] = 0;
1497 output += (char)strtol(tmp, 0, 16);
1498 it += 2;
1499 }
1500 break;
1501 default:
1502 // FIXME: raise exception here?
1503 break;
1504 }
1505 }
1506 return output;
1507}
1508 /*}}}*/
1509// URI::CopyFrom - Copy from an object /*{{{*/
1510// ---------------------------------------------------------------------
1511/* This parses the URI into all of its components */
1512void URI::CopyFrom(const string &U)
1513{
1514 string::const_iterator I = U.begin();
1515
1516 // Locate the first colon, this separates the scheme
1517 for (; I < U.end() && *I != ':' ; ++I);
1518 string::const_iterator FirstColon = I;
1519
1520 /* Determine if this is a host type URI with a leading double //
1521 and then search for the first single / */
1522 string::const_iterator SingleSlash = I;
1523 if (I + 3 < U.end() && I[1] == '/' && I[2] == '/')
1524 SingleSlash += 3;
1525
1526 /* Find the / indicating the end of the hostname, ignoring /'s in the
1527 square brackets */
1528 bool InBracket = false;
1529 for (; SingleSlash < U.end() && (*SingleSlash != '/' || InBracket == true); ++SingleSlash)
1530 {
1531 if (*SingleSlash == '[')
1532 InBracket = true;
1533 if (InBracket == true && *SingleSlash == ']')
1534 InBracket = false;
1535 }
1536
1537 if (SingleSlash > U.end())
1538 SingleSlash = U.end();
1539
1540 // We can now write the access and path specifiers
1541 Access.assign(U.begin(),FirstColon);
1542 if (SingleSlash != U.end())
1543 Path.assign(SingleSlash,U.end());
1544 if (Path.empty() == true)
1545 Path = "/";
1546
1547 // Now we attempt to locate a user:pass@host fragment
1548 if (FirstColon + 2 <= U.end() && FirstColon[1] == '/' && FirstColon[2] == '/')
1549 FirstColon += 3;
1550 else
1551 FirstColon += 1;
1552 if (FirstColon >= U.end())
1553 return;
1554
1555 if (FirstColon > SingleSlash)
1556 FirstColon = SingleSlash;
1557
1558 // Find the colon...
1559 I = FirstColon + 1;
1560 if (I > SingleSlash)
1561 I = SingleSlash;
1562 for (; I < SingleSlash && *I != ':'; ++I);
1563 string::const_iterator SecondColon = I;
1564
1565 // Search for the @ after the colon
1566 for (; I < SingleSlash && *I != '@'; ++I);
1567 string::const_iterator At = I;
1568
1569 // Now write the host and user/pass
1570 if (At == SingleSlash)
1571 {
1572 if (FirstColon < SingleSlash)
1573 Host.assign(FirstColon,SingleSlash);
1574 }
1575 else
1576 {
1577 Host.assign(At+1,SingleSlash);
1578 // username and password must be encoded (RFC 3986)
1579 User.assign(DeQuoteString(FirstColon,SecondColon));
1580 if (SecondColon < At)
1581 Password.assign(DeQuoteString(SecondColon+1,At));
1582 }
1583
1584 // Now we parse the RFC 2732 [] hostnames.
1585 unsigned long PortEnd = 0;
1586 InBracket = false;
1587 for (unsigned I = 0; I != Host.length();)
1588 {
1589 if (Host[I] == '[')
1590 {
1591 InBracket = true;
1592 Host.erase(I,1);
1593 continue;
1594 }
1595
1596 if (InBracket == true && Host[I] == ']')
1597 {
1598 InBracket = false;
1599 Host.erase(I,1);
1600 PortEnd = I;
1601 continue;
1602 }
1603 I++;
1604 }
1605
1606 // Tsk, weird.
1607 if (InBracket == true)
1608 {
1609 Host.clear();
1610 return;
1611 }
1612
1613 // Now we parse off a port number from the hostname
1614 Port = 0;
1615 string::size_type Pos = Host.rfind(':');
1616 if (Pos == string::npos || Pos < PortEnd)
1617 return;
1618
1619 Port = atoi(string(Host,Pos+1).c_str());
1620 Host.assign(Host,0,Pos);
1621}
1622 /*}}}*/
1623// URI::operator string - Convert the URI to a string /*{{{*/
1624// ---------------------------------------------------------------------
1625/* */
1626URI::operator string()
1627{
1628 std::stringstream Res;
1629
1630 if (Access.empty() == false)
1631 Res << Access << ':';
1632
1633 if (Host.empty() == false)
1634 {
1635 if (Access.empty() == false)
1636 Res << "//";
1637
1638 if (User.empty() == false)
1639 {
1640 // FIXME: Technically userinfo is permitted even less
1641 // characters than these, but this is not conveniently
1642 // expressed with a blacklist.
1643 Res << QuoteString(User, ":/?#[]@");
1644 if (Password.empty() == false)
1645 Res << ":" << QuoteString(Password, ":/?#[]@");
1646 Res << "@";
1647 }
1648
1649 // Add RFC 2732 escaping characters
1650 if (Access.empty() == false && Host.find_first_of("/:") != string::npos)
1651 Res << '[' << Host << ']';
1652 else
1653 Res << Host;
1654
1655 if (Port != 0)
1656 Res << ':' << Port;
1657 }
1658
1659 if (Path.empty() == false)
1660 {
1661 if (Path[0] != '/')
1662 Res << "/" << Path;
1663 else
1664 Res << Path;
1665 }
1666
1667 return Res.str();
1668}
1669 /*}}}*/
1670// URI::SiteOnly - Return the schema and site for the URI /*{{{*/
1671string URI::SiteOnly(const string &URI)
1672{
1673 ::URI U(URI);
1674 U.User.clear();
1675 U.Password.clear();
1676 U.Path.clear();
1677 return U;
1678}
1679 /*}}}*/
1680// URI::ArchiveOnly - Return the schema, site and cleaned path for the URI /*{{{*/
1681string URI::ArchiveOnly(const string &URI)
1682{
1683 ::URI U(URI);
1684 U.User.clear();
1685 U.Password.clear();
1686 if (U.Path.empty() == false && U.Path[U.Path.length() - 1] == '/')
1687 U.Path.erase(U.Path.length() - 1);
1688 return U;
1689}
1690 /*}}}*/
1691// URI::NoUserPassword - Return the schema, site and path for the URI /*{{{*/
1692string URI::NoUserPassword(const string &URI)
1693{
1694 ::URI U(URI);
1695 U.User.clear();
1696 U.Password.clear();
1697 return U;
1698}
1699 /*}}}*/