| 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 |
| 24 | |
28 | |
| … | |
… | |
| 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($) { |
| … | |
… | |
| 154 | OPT_TINYB64 detect short B64 outside of Mime |
159 | OPT_TINYB64 detect short B64 outside of Mime |
| 155 | OPT_ENCEXT extension for single-part encoded files |
160 | OPT_ENCEXT extension for single-part encoded files |
| 156 | OPT_REMOVE remove input files after decoding (dangerous) |
161 | OPT_REMOVE remove input files after decoding (dangerous) |
| 157 | OPT_MOREMIME strict MIME adherence |
162 | OPT_MOREMIME strict MIME adherence |
| 158 | OPT_DOTDOT ".."-unescaping has not yet been done on input files |
163 | OPT_DOTDOT ".."-unescaping has not yet been done on input files |
|
|
164 | OPT_RBUF set default read I/O buffer size in bytes *EXPERIMENTAL* |
|
|
165 | OPT_WBUF set default write I/O buffer size in bytes *EXPERIMENTAL* |
| 159 | |
166 | |
| 160 | =head2 Result/Error codes |
167 | =head2 Result/Error codes |
| 161 | |
168 | |
| 162 | RET_OK everything went fine |
169 | RET_OK everything went fine |
| 163 | RET_IOERR I/O Error - examine errno |
170 | RET_IOERR I/O Error - examine errno |
| … | |
… | |
| 257 | |
264 | |
| 258 | =head2 Loading sourcefiles, optionally fuzzy merge and start decoding |
265 | =head2 Loading sourcefiles, optionally fuzzy merge and start decoding |
| 259 | |
266 | |
| 260 | =over 4 |
267 | =over 4 |
| 261 | |
268 | |
| 262 | =item ($retval, $count) = LoadFile $fname, [$id, [$delflag]] |
269 | =item ($retval, $count) = LoadFile $fname, [$id, [$delflag, [$partno]]] |
| 263 | |
270 | |
| 264 | Load the given file and scan it for encoded contents. Optionally tag it |
271 | Load the given file and scan it for encoded contents. Optionally tag it |
| 265 | with the given id, and if C<$delflag> is true, delete the file after it is |
272 | with the given id, and if C<$delflag> is true, delete the file after it |
| 266 | no longer necessary. |
273 | is no longer necessary. If you are certain of the part number, you can |
|
|
274 | specify it as the last argument. |
|
|
275 | |
|
|
276 | A better (usually faster) way of doing this is using the C<SetFNameFilter> |
|
|
277 | functionality. |
| 267 | |
278 | |
| 268 | =item $retval = Smerge $pass |
279 | =item $retval = Smerge $pass |
| 269 | |
280 | |
| 270 | If you are desperate, try to call C<Smerge> with increasing C<$pass> |
281 | If you are desperate, try to call C<Smerge> with increasing C<$pass> |
| 271 | values, beginning at C<0>, to try to merge parts that usually would not |
282 | values, beginning at C<0>, to try to merge parts that usually would not |
| … | |
… | |
| 426 | =head1 LARGE EXAMPLE DECODER |
437 | =head1 LARGE EXAMPLE DECODER |
| 427 | |
438 | |
| 428 | This is the file C<example-decoder> from the distribution, put here |
439 | This is the file C<example-decoder> from the distribution, put here |
| 429 | instead of more thorough documentation. |
440 | instead of more thorough documentation. |
| 430 | |
441 | |
|
|
442 | #!/usr/bin/perl |
|
|
443 | |
| 431 | # decode all the files in the directory uusrc/ and copy |
444 | # decode all the files in the directory uusrc/ and copy |
| 432 | # the resulting files to uudst/ |
445 | # the resulting files to uudst/ |
| 433 | |
446 | |
| 434 | use Convert::UUlib ':all'; |
447 | use Convert::UUlib ':all'; |
| 435 | |
448 | |
| 436 | sub namefilter { |
449 | sub namefilter { |
| 437 | my($path)=@_; |
450 | my ($path) = @_; |
|
|
451 | |
| 438 | $path=~s/^.*[\/\\]//; |
452 | $path=~s/^.*[\/\\]//; |
|
|
453 | |
| 439 | $path; |
454 | $path |
| 440 | } |
455 | } |
| 441 | |
456 | |
| 442 | sub busycb { |
457 | sub busycb { |
| 443 | my ($action, $curfile, $partno, $numparts, $percent, $fsize) = @_; |
458 | my ($action, $curfile, $partno, $numparts, $percent, $fsize) = @_; |
| 444 | $_[0]=straction($action); |
459 | $_[0]=straction($action); |
| 445 | print "busy_callback(", (join ",",@_), ")\n"; |
460 | print "busy_callback(", (join ",",@_), ")\n"; |
| 446 | 0; |
461 | 0 |
| 447 | } |
462 | } |
| 448 | |
463 | |
|
|
464 | SetOption OPT_RBUF, 128*1024; |
|
|
465 | SetOption OPT_WBUF, 1024*1024; |
| 449 | SetOption OPT_IGNMODE, 1; |
466 | SetOption OPT_IGNMODE, 1; |
|
|
467 | SetOption OPT_IGNMODE, 1; |
| 450 | SetOption OPT_VERBOSE, 1; |
468 | SetOption OPT_VERBOSE, 1; |
| 451 | |
469 | |
| 452 | # show the three ways you can set callback functions. I normally |
470 | # show the three ways you can set callback functions. I normally |
| 453 | # prefer the one with the sub inplace. |
471 | # prefer the one with the sub inplace. |
| 454 | SetFNameFilter \&namefilter; |
472 | SetFNameFilter \&namefilter; |
| 455 | |
473 | |
| 456 | SetBusyCallback "busycb", 333; |
474 | SetBusyCallback "busycb", 333; |
| 457 | |
475 | |
| 458 | SetMsgCallback sub { |
476 | SetMsgCallback sub { |
| 459 | my ($msg, $level) = @_; |
477 | my ($msg, $level) = @_; |
| 460 | print uc strmsglevel $_[1], ": $msg\n"; |
478 | print uc strmsglevel $_[1], ": $msg\n"; |
| 461 | }; |
479 | }; |
| 462 | |
480 | |
| 463 | # the following non-trivial FileNameCallback takes care |
481 | # the following non-trivial FileNameCallback takes care |
| 464 | # of some subject lines not detected properly by uulib: |
482 | # of some subject lines not detected properly by uulib: |
| 465 | SetFileNameCallback sub { |
483 | SetFileNameCallback sub { |
| 466 | return unless $_[1]; # skip "Re:"-plies et al. |
484 | return unless $_[1]; # skip "Re:"-plies et al. |
| 467 | local $_ = $_[0]; |
485 | local $_ = $_[0]; |
| 468 | |
486 | |
|
|
487 | return $1 if /(\S+\s+IMG_\d+.jpg)/i; |
|
|
488 | |
| 469 | # the following rules are rather effective on some newsgroups, |
489 | # the following rules are rather effective on some newsgroups, |
| 470 | # like alt.binaries.games.anime, where non-mime, uuencoded data |
490 | # like alt.binaries.games.anime, where non-mime, uuencoded data |
| 471 | # is very common |
491 | # is very common |
| 472 | |
492 | |
| 473 | # if we find some *.rar, take it as the filename |
493 | # if we find some *.rar, take it as the filename |
| 474 | return $1 if /(\S{3,}\.(?:[rstuvwxyz]\d\d|rar))\s/i; |
494 | return $1 if /(\S{3,}\.(?:[rstuvwxyz]\d\d|rar))\s/i; |
| 475 | |
495 | |
| 476 | # one common subject format |
496 | # one common subject format |
| 477 | return $1 if /- "(.{2,}?\..+?)" (?:yenc )?\(\d+\/\d+\)/i; |
497 | return $1 if /- "(.{2,}?\..+?)" (?:yenc )?\(\d+\/\d+\)/i; |
| 478 | |
498 | |
| 479 | # - filename.par (04/55) |
499 | # - filename.par (04/55) |
| 480 | return $1 if /- "?(\S{3,}\.\S+?)"? (?:yenc )?\(\d+\/\d+\)/i; |
500 | return $1 if /- "?(\S{3,}\.\S+?)"? (?:yenc )?\(\d+\/\d+\)/i; |
| 481 | |
501 | |
| 482 | # - (xxx) No. 1 sayuri81.jpg 756565 bytes |
502 | # - (xxx) No. 1 sayuri81.jpg 756565 bytes |
| 483 | # - (20 files) No.17 Roseanne.jpg [2/2] |
503 | # - (20 files) No.17 Roseanne.jpg [2/2] |
| 484 | return $1 if /No\.[ 0-9]+ (\S+\....) (?:\d+ bytes )?\[/; |
504 | return $1 if /No\.[ 0-9]+ (\S+\....) (?:\d+ bytes )?\[/; |
| 485 | |
505 | |
|
|
506 | # try to detect some common forms of filenames |
|
|
507 | return $1 if /([a-z0-9_\-+.]{3,}\.[a-z]{3,4}(?:.\d+))/i; |
|
|
508 | |
| 486 | # otherwise just pass what we have |
509 | # otherwise just pass what we have |
| 487 | return (); |
510 | () |
| 488 | }; |
511 | }; |
| 489 | |
512 | |
| 490 | # now read all files in the directory uusrc/* |
513 | # now read all files in the directory uusrc/* |
| 491 | for(<uusrc/*>) { |
514 | for(<uusrc/*>) { |
| 492 | my($retval,$count)=LoadFile ($_, $_, 1); |
515 | my ($retval, $count) = LoadFile ($_, $_, 1); |
| 493 | print "file($_), status(", strerror $retval, ") parts($count)\n"; |
516 | print "file($_), status(", strerror $retval, ") parts($count)\n"; |
| 494 | } |
517 | } |
| 495 | |
518 | |
| 496 | SetOption OPT_SAVEPATH, "uudst/"; |
519 | SetOption OPT_SAVEPATH, "uudst/"; |
| 497 | |
520 | |
| 498 | # now wade through all files and their source parts |
521 | # now wade through all files and their source parts |
| 499 | $i = 0; |
522 | $i = 0; |
| 500 | while ($uu = GetFileListItem($i)) { |
523 | while ($uu = GetFileListItem $i) { |
| 501 | $i++; |
524 | $i++; |
| 502 | print "file nr. $i"; |
525 | print "file nr. $i"; |
| 503 | print " state ", $uu->state; |
526 | print " state ", $uu->state; |
| 504 | print " mode ", $uu->mode; |
527 | print " mode ", $uu->mode; |
| 505 | print " uudet ", strencoding $uu->uudet; |
528 | print " uudet ", strencoding $uu->uudet; |
| 506 | print " size ", $uu->size; |
529 | print " size ", $uu->size; |
| 507 | print " filename ", $uu->filename; |
530 | print " filename ", $uu->filename; |
| 508 | print " subfname ", $uu->subfname; |
531 | print " subfname ", $uu->subfname; |
| 509 | print " mimeid ", $uu->mimeid; |
532 | print " mimeid ", $uu->mimeid; |
| 510 | print " mimetype ", $uu->mimetype; |
533 | print " mimetype ", $uu->mimetype; |
| 511 | print "\n"; |
534 | print "\n"; |
| 512 | |
535 | |
| 513 | # print additional info about all parts |
536 | # print additional info about all parts |
| 514 | for ($uu->parts) { |
537 | for ($uu->parts) { |
| 515 | while (my ($k, $v) = each %$_) { |
538 | while (my ($k, $v) = each %$_) { |
| 516 | print "$k > $v, "; |
539 | print "$k > $v, "; |
| 517 | } |
540 | } |
| 518 | print "\n"; |
541 | print "\n"; |
| 519 | } |
542 | } |
| 520 | |
543 | |
| 521 | $uu->decode_temp; |
544 | print $uu->filename; |
| 522 | print " temporarily decoded to ", $uu->binfile, "\n"; |
545 | |
| 523 | $uu->remove_temp; |
546 | $uu->remove_temp; |
| 524 | |
547 | |
| 525 | print strerror $uu->decode; |
548 | if (my $err = $uu->decode ()) { |
|
|
549 | print ", ", strerror $err, "\n"; |
|
|
550 | } else { |
| 526 | print " saved as uudst/", $uu->filename, "\n"; |
551 | print ", saved as uudst/", $uu->filename, "\n"; |
| 527 | } |
552 | } |
|
|
553 | } |
| 528 | |
554 | |
| 529 | print "cleanup...\n"; |
555 | print "cleanup...\n"; |
| 530 | |
556 | |
| 531 | CleanUp(); |
557 | CleanUp; |
| 532 | |
558 | |
| 533 | =head1 AUTHOR |
559 | =head1 AUTHOR |
| 534 | |
560 | |
| 535 | Marc Lehmann <pcg@goof.com>, the original uulib library was written |
561 | Marc Lehmann <schmorp@schmorp.de>, the original uulib library was written |
| 536 | by Frank Pilhofer <fp@informatik.uni-frankfurt.de>, and later heavily |
562 | by Frank Pilhofer <fp@informatik.uni-frankfurt.de>, and later heavily |
| 537 | bugfixed by Marc Lehmann. |
563 | bugfixed by Marc Lehmann. |
| 538 | |
564 | |
| 539 | =head1 SEE ALSO |
565 | =head1 SEE ALSO |
| 540 | |
566 | |
