| 1 | package Convert::UUlib; |
1 | package Convert::UUlib; |
|
|
2 | |
|
|
3 | no warnings; |
|
|
4 | use strict; |
| 2 | |
5 | |
| 3 | use Carp; |
6 | use Carp; |
| 4 | |
7 | |
| 5 | require Exporter; |
8 | require Exporter; |
| 6 | require DynaLoader; |
9 | require DynaLoader; |
| 7 | |
10 | |
| 8 | $VERSION = 0.31; |
11 | our $VERSION = '1.12'; |
| 9 | |
12 | |
| 10 | @ISA = qw(Exporter DynaLoader); |
13 | our @ISA = qw(Exporter DynaLoader); |
| 11 | |
14 | |
| 12 | @_consts = qw( |
15 | our @_consts = qw( |
| 13 | ACT_COPYING ACT_DECODING ACT_ENCODING ACT_IDLE ACT_SCANNING |
16 | ACT_COPYING ACT_DECODING ACT_ENCODING ACT_IDLE ACT_SCANNING |
| 14 | |
17 | |
| 15 | FILE_DECODED FILE_ERROR FILE_MISPART FILE_NOBEGIN FILE_NODATA |
18 | FILE_DECODED FILE_ERROR FILE_MISPART FILE_NOBEGIN FILE_NODATA |
| 16 | FILE_NOEND FILE_OK FILE_READ FILE_TMPFILE |
19 | FILE_NOEND FILE_OK FILE_READ FILE_TMPFILE |
| 17 | |
20 | |
| 18 | MSG_ERROR MSG_FATAL MSG_MESSAGE MSG_NOTE MSG_PANIC MSG_WARNING |
21 | MSG_ERROR MSG_FATAL MSG_MESSAGE MSG_NOTE MSG_PANIC MSG_WARNING |
| 19 | |
22 | |
|
|
23 | OPT_RBUF OPT_WBUF |
| 20 | OPT_BRACKPOL OPT_DEBUG OPT_DESPERATE OPT_DUMBNESS OPT_ENCEXT |
24 | OPT_BRACKPOL OPT_DEBUG OPT_DESPERATE OPT_DUMBNESS OPT_ENCEXT |
| 21 | OPT_ERRNO OPT_FAST OPT_IGNMODE OPT_IGNREPLY OPT_OVERWRITE OPT_PREAMB |
25 | OPT_ERRNO OPT_FAST OPT_IGNMODE OPT_IGNREPLY OPT_OVERWRITE OPT_PREAMB |
| 22 | OPT_PROGRESS OPT_SAVEPATH OPT_TINYB64 OPT_USETEXT OPT_VERBOSE |
26 | OPT_PROGRESS OPT_SAVEPATH OPT_TINYB64 OPT_USETEXT OPT_VERBOSE |
| 23 | OPT_VERSION OPT_REMOVE OPT_MOREMIME OPT_DOTDOT |
27 | OPT_VERSION OPT_REMOVE OPT_MOREMIME OPT_DOTDOT OPT_AUTOCHECK |
| 24 | |
28 | |
| 25 | RET_CANCEL RET_CONT RET_EXISTS RET_ILLVAL RET_IOERR RET_NODATA |
29 | RET_CANCEL RET_CONT RET_EXISTS RET_ILLVAL RET_IOERR RET_NODATA |
| 26 | RET_NOEND RET_NOMEM RET_OK RET_UNSUP |
30 | RET_NOEND RET_NOMEM RET_OK RET_UNSUP |
| 27 | |
31 | |
| 28 | B64_ENCODED BH_ENCODED PT_ENCODED QP_ENCODED |
32 | B64_ENCODED BH_ENCODED PT_ENCODED QP_ENCODED |
| 29 | XX_ENCODED UU_ENCODED YENC_ENCODED |
33 | XX_ENCODED UU_ENCODED YENC_ENCODED |
| 30 | ); |
34 | ); |
| 31 | |
35 | |
| 32 | @_funcs = qw( |
36 | our @_funcs = qw( |
| 33 | Initialize CleanUp GetOption SetOption strerror SetMsgCallback |
37 | Initialize CleanUp GetOption SetOption strerror SetMsgCallback |
| 34 | SetBusyCallback SetFileCallback SetFNameFilter SetFileNameCallback |
38 | SetBusyCallback SetFileCallback SetFNameFilter SetFileNameCallback |
| 35 | FNameFilter LoadFile GetFileListItem RenameFile DecodeToTemp |
39 | FNameFilter LoadFile GetFileListItem RenameFile DecodeToTemp |
| 36 | RemoveTemp DecodeFile InfoFile Smerge QuickDecode EncodeMulti |
40 | RemoveTemp DecodeFile InfoFile Smerge QuickDecode EncodeMulti |
| 37 | EncodePartial EncodeToStream EncodeToFile E_PrepSingle |
41 | EncodePartial EncodeToStream EncodeToFile E_PrepSingle |
| 38 | E_PrepPartial |
42 | E_PrepPartial |
| 39 | |
43 | |
| 40 | straction strencoding strmsglevel |
44 | straction strencoding strmsglevel |
| 41 | ); |
45 | ); |
| 42 | |
46 | |
| 43 | @EXPORT = @_consts; |
47 | our @EXPORT = @_consts; |
| 44 | @EXPORT_OK = @_funcs; |
48 | our @EXPORT_OK = @_funcs; |
| 45 | %EXPORT_TAGS = (all => [@_consts,@_funcs], constants => \@_consts); |
49 | our %EXPORT_TAGS = (all => [@_consts,@_funcs], constants => \@_consts); |
| 46 | |
50 | |
| 47 | bootstrap Convert::UUlib $VERSION; |
51 | bootstrap Convert::UUlib $VERSION; |
| 48 | |
52 | |
| 49 | Initialize(); |
53 | Initialize(); |
| 50 | |
54 | |
| 51 | # not when < 5.005_6x |
55 | # not when < 5.005_6x |
| 52 | # END { CleanUp() } |
56 | # END { CleanUp() } |
| 53 | |
57 | |
| 54 | for (@_consts) { |
58 | for (@_consts) { |
| 55 | my $constant = constant($_); |
59 | my $constant = constant($_); |
|
|
60 | no strict 'refs'; |
| 56 | *$_ = sub () { $constant }; |
61 | *$_ = sub () { $constant }; |
| 57 | } |
62 | } |
| 58 | |
63 | |
| 59 | # action code -> string mapping |
64 | # action code -> string mapping |
| 60 | sub straction($) { |
65 | sub straction($) { |
| … | |
… | |
| 151 | OPT_PROGRESS retrieve progress information |
156 | OPT_PROGRESS retrieve progress information |
| 152 | OPT_USETEXT handle text messages |
157 | OPT_USETEXT handle text messages |
| 153 | OPT_PREAMB handle Mime preambles/epilogues |
158 | OPT_PREAMB handle Mime preambles/epilogues |
| 154 | OPT_TINYB64 detect short B64 outside of Mime |
159 | OPT_TINYB64 detect short B64 outside of Mime |
| 155 | OPT_ENCEXT extension for single-part encoded files |
160 | OPT_ENCEXT extension for single-part encoded files |
| 156 | OPT_REMOVE remove input files after decoding |
161 | OPT_REMOVE remove input files after decoding (dangerous) |
| 157 | OPT_MOREMIME strict MIME adherence |
162 | OPT_MOREMIME strict MIME adherence |
| 158 | OPT_DOTDOT .. unescaping has not yet been done on input files |
163 | OPT_DOTDOT ".."-unescaping has not yet been done on input files |
|
|
164 | OPT_RBUF set default read I/O buffer size in bytes |
|
|
165 | OPT_WBUF set default write I/O buffer size in bytes |
|
|
166 | OPT_AUTOCHECK automatically check file list after every loadfile |
| 159 | |
167 | |
| 160 | =head2 Result/Error codes |
168 | =head2 Result/Error codes |
| 161 | |
169 | |
| 162 | RET_OK everything went fine |
170 | RET_OK everything went fine |
| 163 | RET_IOERR I/O Error - examine errno |
171 | RET_IOERR I/O Error - examine errno |
| … | |
… | |
| 203 | |
211 | |
| 204 | Initialize is automatically called when the module is loaded and allocates |
212 | Initialize is automatically called when the module is loaded and allocates |
| 205 | quite a small amount of memory for todays machines ;) CleanUp releases that |
213 | quite a small amount of memory for todays machines ;) CleanUp releases that |
| 206 | again. |
214 | again. |
| 207 | |
215 | |
|
|
216 | On my machine, a fairly complete decode with DBI backend needs about 10MB |
|
|
217 | RSS to decode 20000 files. |
|
|
218 | |
| 208 | =over 4 |
219 | =over 4 |
| 209 | |
220 | |
| 210 | =item Initialize |
221 | =item Initialize |
| 211 | |
222 | |
| 212 | Not normally necessary, (re-)initializes the library. |
223 | Not normally necessary, (re-)initializes the library. |
| … | |
… | |
| 254 | |
265 | |
| 255 | =head2 Loading sourcefiles, optionally fuzzy merge and start decoding |
266 | =head2 Loading sourcefiles, optionally fuzzy merge and start decoding |
| 256 | |
267 | |
| 257 | =over 4 |
268 | =over 4 |
| 258 | |
269 | |
| 259 | =item ($retval, $count) = LoadFile $fname, [$id, [$delflag]] |
270 | =item ($retval, $count) = LoadFile $fname, [$id, [$delflag, [$partno]]] |
| 260 | |
271 | |
| 261 | Load the given file and scan it for encoded contents. Optionally tag it |
272 | Load the given file and scan it for encoded contents. Optionally tag it |
| 262 | with the given id, and if C<$delflag> is true, delete the file after it is |
273 | with the given id, and if C<$delflag> is true, delete the file after it |
| 263 | no longer necessary. |
274 | is no longer necessary. If you are certain of the part number, you can |
|
|
275 | specify it as the last argument. |
|
|
276 | |
|
|
277 | A better (usually faster) way of doing this is using the C<SetFNameFilter> |
|
|
278 | functionality. |
| 264 | |
279 | |
| 265 | =item $retval = Smerge $pass |
280 | =item $retval = Smerge $pass |
| 266 | |
281 | |
| 267 | If you are desperate, try to call C<Smerge> with increasing C<$pass> |
282 | If you are desperate, try to call C<Smerge> with increasing C<$pass> |
| 268 | values, beginning at C<0>, to try to merge parts that usually would not |
283 | values, beginning at C<0>, to try to merge parts that usually would not |
| 269 | have been merged. |
284 | have been merged. |
| 270 | |
285 | |
| 271 | Most probably this will result in garbled files, so never do this by |
286 | Most probably this will result in garbled files, so never do this by |
| 272 | default. |
287 | default, except: |
|
|
288 | |
|
|
289 | If the C<OPT_AUTOCHECK> option has been disabled (by default it is |
|
|
290 | enabled) to speed up file loading, then you I<have> to call C<Smerge -1> |
|
|
291 | after loading all files as an additional pre-pass (which is normally done |
|
|
292 | by C<LoadFile>). |
| 273 | |
293 | |
| 274 | =item $item = GetFileListItem $item_number |
294 | =item $item = GetFileListItem $item_number |
| 275 | |
295 | |
| 276 | Return the C<$item> structure for the C<$item_number>'th found file, or |
296 | Return the C<$item> structure for the C<$item_number>'th found file, or |
| 277 | C<undef> of no file with that number exists. |
297 | C<undef> of no file with that number exists. |
| … | |
… | |
| 423 | =head1 LARGE EXAMPLE DECODER |
443 | =head1 LARGE EXAMPLE DECODER |
| 424 | |
444 | |
| 425 | This is the file C<example-decoder> from the distribution, put here |
445 | This is the file C<example-decoder> from the distribution, put here |
| 426 | instead of more thorough documentation. |
446 | instead of more thorough documentation. |
| 427 | |
447 | |
|
|
448 | #!/usr/bin/perl |
|
|
449 | |
| 428 | # decode all the files in the directory uusrc/ and copy |
450 | # decode all the files in the directory uusrc/ and copy |
| 429 | # the resulting files to uudst/ |
451 | # the resulting files to uudst/ |
| 430 | |
452 | |
| 431 | use Convert::UUlib ':all'; |
453 | use Convert::UUlib ':all'; |
| 432 | |
454 | |
| 433 | sub namefilter { |
455 | sub namefilter { |
| 434 | my($path)=@_; |
456 | my ($path) = @_; |
|
|
457 | |
| 435 | $path=~s/^.*[\/\\]//; |
458 | $path=~s/^.*[\/\\]//; |
|
|
459 | |
| 436 | $path; |
460 | $path |
| 437 | } |
461 | } |
| 438 | |
462 | |
| 439 | sub busycb { |
463 | sub busycb { |
| 440 | my ($action, $curfile, $partno, $numparts, $percent, $fsize) = @_; |
464 | my ($action, $curfile, $partno, $numparts, $percent, $fsize) = @_; |
| 441 | $_[0]=straction($action); |
465 | $_[0]=straction($action); |
| 442 | print "busy_callback(", (join ",",@_), ")\n"; |
466 | print "busy_callback(", (join ",",@_), ")\n"; |
| 443 | 0; |
467 | 0 |
| 444 | } |
468 | } |
| 445 | |
469 | |
|
|
470 | SetOption OPT_RBUF, 128*1024; |
|
|
471 | SetOption OPT_WBUF, 1024*1024; |
| 446 | SetOption OPT_IGNMODE, 1; |
472 | SetOption OPT_IGNMODE, 1; |
|
|
473 | SetOption OPT_IGNMODE, 1; |
| 447 | SetOption OPT_VERBOSE, 1; |
474 | SetOption OPT_VERBOSE, 1; |
| 448 | |
475 | |
| 449 | # show the three ways you can set callback functions. I normally |
476 | # show the three ways you can set callback functions. I normally |
| 450 | # prefer the one with the sub inplace. |
477 | # prefer the one with the sub inplace. |
| 451 | SetFNameFilter \&namefilter; |
478 | SetFNameFilter \&namefilter; |
| 452 | |
479 | |
| 453 | SetBusyCallback "busycb", 333; |
480 | SetBusyCallback "busycb", 333; |
| 454 | |
481 | |
| 455 | SetMsgCallback sub { |
482 | SetMsgCallback sub { |
| 456 | my ($msg, $level) = @_; |
483 | my ($msg, $level) = @_; |
| 457 | print uc strmsglevel $_[1], ": $msg\n"; |
484 | print uc strmsglevel $_[1], ": $msg\n"; |
| 458 | }; |
485 | }; |
| 459 | |
486 | |
| 460 | # the following non-trivial FileNameCallback takes care |
487 | # the following non-trivial FileNameCallback takes care |
| 461 | # of some subject lines not detected properly by uulib: |
488 | # of some subject lines not detected properly by uulib: |
| 462 | SetFileNameCallback sub { |
489 | SetFileNameCallback sub { |
| 463 | return unless $_[1]; # skip "Re:"-plies et al. |
490 | return unless $_[1]; # skip "Re:"-plies et al. |
| 464 | local $_ = $_[0]; |
491 | local $_ = $_[0]; |
| 465 | |
492 | |
| 466 | # the following rules are rather effective on some newsgroups, |
493 | # the following rules are rather effective on some newsgroups, |
| 467 | # like alt.binaries.games.anime, where non-mime, uuencoded data |
494 | # like alt.binaries.games.anime, where non-mime, uuencoded data |
| 468 | # is very common |
495 | # is very common |
| 469 | |
496 | |
| 470 | # if we find some *.rar, take it as the filename |
497 | # if we find some *.rar, take it as the filename |
| 471 | return $1 if /(\S{3,}\.(?:[rstuvwxyz]\d\d|rar))\s/i; |
498 | return $1 if /(\S{3,}\.(?:[rstuvwxyz]\d\d|rar))\s/i; |
| 472 | |
499 | |
| 473 | # one common subject format |
500 | # one common subject format |
| 474 | return $1 if /- "(.{2,}?\..+?)" (?:yenc )?\(\d+\/\d+\)/i; |
501 | return $1 if /- "(.{2,}?\..+?)" (?:yenc )?\(\d+\/\d+\)/i; |
| 475 | |
502 | |
| 476 | # - filename.par (04/55) |
503 | # - filename.par (04/55) |
| 477 | return $1 if /- "?(\S{3,}\.\S+?)"? (?:yenc )?\(\d+\/\d+\)/i; |
504 | return $1 if /- "?(\S{3,}\.\S+?)"? (?:yenc )?\(\d+\/\d+\)/i; |
| 478 | |
505 | |
| 479 | # - (xxx) No. 1 sayuri81.jpg 756565 bytes |
506 | # - (xxx) No. 1 sayuri81.jpg 756565 bytes |
| 480 | # - (20 files) No.17 Roseanne.jpg [2/2] |
507 | # - (20 files) No.17 Roseanne.jpg [2/2] |
| 481 | return $1 if /No\.[ 0-9]+ (\S+\....) (?:\d+ bytes )?\[/; |
508 | return $1 if /No\.[ 0-9]+ (\S+\....) (?:\d+ bytes )?\[/; |
| 482 | |
509 | |
|
|
510 | # try to detect some common forms of filenames |
|
|
511 | return $1 if /([a-z0-9_\-+.]{3,}\.[a-z]{3,4}(?:.\d+))/i; |
|
|
512 | |
| 483 | # otherwise just pass what we have |
513 | # otherwise just pass what we have |
| 484 | return (); |
514 | () |
| 485 | }; |
515 | }; |
| 486 | |
516 | |
| 487 | # now read all files in the directory uusrc/* |
517 | # now read all files in the directory uusrc/* |
| 488 | for(<uusrc/*>) { |
518 | for(<uusrc/*>) { |
| 489 | my($retval,$count)=LoadFile ($_, $_, 1); |
519 | my ($retval, $count) = LoadFile ($_, $_, 1); |
| 490 | print "file($_), status(", strerror $retval, ") parts($count)\n"; |
520 | print "file($_), status(", strerror $retval, ") parts($count)\n"; |
| 491 | } |
521 | } |
| 492 | |
522 | |
| 493 | SetOption OPT_SAVEPATH, "uudst/"; |
523 | SetOption OPT_SAVEPATH, "uudst/"; |
| 494 | |
524 | |
| 495 | # now wade through all files and their source parts |
525 | # now wade through all files and their source parts |
| 496 | $i = 0; |
526 | $i = 0; |
| 497 | while ($uu = GetFileListItem($i)) { |
527 | while ($uu = GetFileListItem $i) { |
| 498 | $i++; |
528 | $i++; |
| 499 | print "file nr. $i"; |
529 | print "file nr. $i"; |
| 500 | print " state ", $uu->state; |
530 | print " state ", $uu->state; |
| 501 | print " mode ", $uu->mode; |
531 | print " mode ", $uu->mode; |
| 502 | print " uudet ", strencoding $uu->uudet; |
532 | print " uudet ", strencoding $uu->uudet; |
| 503 | print " size ", $uu->size; |
533 | print " size ", $uu->size; |
| 504 | print " filename ", $uu->filename; |
534 | print " filename ", $uu->filename; |
| 505 | print " subfname ", $uu->subfname; |
535 | print " subfname ", $uu->subfname; |
| 506 | print " mimeid ", $uu->mimeid; |
536 | print " mimeid ", $uu->mimeid; |
| 507 | print " mimetype ", $uu->mimetype; |
537 | print " mimetype ", $uu->mimetype; |
| 508 | print "\n"; |
538 | print "\n"; |
| 509 | |
539 | |
| 510 | # print additional info about all parts |
540 | # print additional info about all parts |
| 511 | for ($uu->parts) { |
541 | for ($uu->parts) { |
| 512 | while (my ($k, $v) = each %$_) { |
542 | while (my ($k, $v) = each %$_) { |
| 513 | print "$k > $v, "; |
543 | print "$k > $v, "; |
| 514 | } |
544 | } |
| 515 | print "\n"; |
545 | print "\n"; |
| 516 | } |
546 | } |
| 517 | |
547 | |
| 518 | $uu->decode_temp; |
548 | print $uu->filename; |
| 519 | print " temporarily decoded to ", $uu->binfile, "\n"; |
549 | |
| 520 | $uu->remove_temp; |
550 | $uu->remove_temp; |
| 521 | |
551 | |
| 522 | print strerror $uu->decode; |
552 | if (my $err = $uu->decode ()) { |
|
|
553 | print ", ", strerror $err, "\n"; |
|
|
554 | } else { |
| 523 | print " saved as uudst/", $uu->filename, "\n"; |
555 | print ", saved as uudst/", $uu->filename, "\n"; |
| 524 | } |
556 | } |
|
|
557 | } |
| 525 | |
558 | |
| 526 | print "cleanup...\n"; |
559 | print "cleanup...\n"; |
| 527 | |
560 | |
| 528 | CleanUp(); |
561 | CleanUp; |
| 529 | |
562 | |
| 530 | =head1 AUTHOR |
563 | =head1 AUTHOR |
| 531 | |
564 | |
| 532 | Marc Lehmann <pcg@goof.com>, the original uulib library was written |
565 | Marc Lehmann <schmorp@schmorp.de>, the original uulib library was written |
| 533 | by Frank Pilhofer <fp@informatik.uni-frankfurt.de>, and later heavily |
566 | by Frank Pilhofer <fp@informatik.uni-frankfurt.de>, and later heavily |
| 534 | bugfixed by Marc Lehmann. |
567 | bugfixed by Marc Lehmann. |
| 535 | |
568 | |
| 536 | =head1 SEE ALSO |
569 | =head1 SEE ALSO |
| 537 | |
570 | |
