/usr/share/perl5/Text/PDF/File.pm is in libtext-pdf-perl 0.31-1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 | package Text::PDF::File;
=head1 NAME
Text::PDF::File - Holds the trailers and cross-reference tables for a PDF file
=head1 SYNOPSIS
$p = Text::PDF::File->open("filename.pdf", 1);
$p->new_obj($obj_ref);
$p->free_obj($obj_ref);
$p->append_file;
$p->close_file;
$p->release; # IMPORTANT!
=head1 DESCRIPTION
This class keeps track of the directory aspects of a PDF file. There are two
parts to the directory: the main directory object which is the parent to all
other objects and a chain of cross-reference tables and corresponding trailer
dictionaries starting with the main directory object.
=head1 INSTANCE VARIABLES
Within this class hierarchy, rather than making everything visible via methods,
which would be a lot of work, there are various instance variables which are
accessible via associative array referencing. To distinguish instance variables
from content variables (which may come from the PDF content itself), each such
variable will start with a space.
Variables which do not start with a space directly reflect elements in a PDF
dictionary. In the case of a Text::PDF::File, the elements reflect those in the
trailer dictionary.
Since some variables are not designed for class users to access, variables are
marked in the documentation with (R) to indicate that such an entry should only
be used as read-only information. (P) indicates that the information is private
and not designed for user use at all, but is included in the documentation for
completeness and to ensure that nobody else tries to use it.
=over
=item newroot
This variable allows the user to create a new root entry to occur in the trailer
dictionary which is output when the file is written or appended. If you wish to
over-ride the root element in the dictionary you have, use this entry to indicate
that without losing the current Root entry. Notice that newroot should point to
a PDF level object and not just to a dictionary which does not have object status.
=item INFILE (R)
Contains the filehandle used to read this information into this PDF directory. Is
an IO object.
=item fname (R)
This is the filename which is reflected by INFILE, or the original IO object passed
in.
=item update (R)
This indicates that the read file has been opened for update and that at some
point, $p->appendfile() can be called to update the file with the changes that
have been made to the memory representation.
=item maxobj (R)
Contains the first useable object number above any that have already appeared
in the file so far.
=item outlist (P)
This is a list of Objind which are to be output when the next appendfile or outfile
occurs.
=item firstfree (P)
Contains the first free object in the free object list. Free objects are removed
from the front of the list and added to the end.
=item lastfree (P)
Contains the last free object in the free list. It may be the same as the firstfree
if there is only one free object.
=item objcache (P)
All objects are held in the cache to ensure that a system only has one occurrence of
each object. In effect, the objind class acts as a container type class to hold the
PDF object structure and it would be unfortunate if there were two identical
place-holders floating around a system.
=item epos (P)
The end location of the read-file.
=back
Each trailer dictionary contains a number of private instance variables which
hold the chain together.
=over
=item loc (P)
Contains the location of the start of the cross-reference table preceding the
trailer.
=item xref (P)
Contains an anonymous array of each cross-reference table entry.
=item prev (P)
A reference to the previous table. Note this differs from the Prev entry which
is in PDF which contains the location of the previous cross-reference table.
=back
=head1 METHODS
=cut
use strict;
no strict "refs";
use vars qw($cr $irreg_char $reg_char $ws_char $delim_char %types $VERSION);
# no warnings qw(uninitialized);
use IO::File;
# Now for the basic PDF types
use Text::PDF::Utils;
use Text::PDF::Array;
use Text::PDF::Bool;
use Text::PDF::Dict;
use Text::PDF::Name;
use Text::PDF::Number;
use Text::PDF::Objind;
use Text::PDF::String;
use Text::PDF::Page;
use Text::PDF::Pages;
use Text::PDF::Null;
# VERSION now taken from Text::PDF.pm
#$VERSION = "0.27"; # MJPH 15-MAY-2006 Fix minor bug in Pages.pm
#$VERSION = "0.26"; # MJPH 19-MAY-2005 Get a release out!
#$VERSION = "0.25"; # MJPH 20-JAN-2003 fix realised in read_obj x y R, fix Text::PDF::Pages::add_page
#$VERSION = "0.24"; # MJPH 28-AUG-2002 out_obj may call new_obj
#$VERSION = "0.23"; # MJPH 14-AUG-2002 Fix MANIFEST
#$VERSION = "0.22"; # MJPH 26-JUL-2002 Add Text::PDF::File::copy, tidy up update(), sort out out_trailer
# Fix to not remove string final CRs when reading dictionaries
#$VERSION = "0.21"; # GJ 8-JUN-2002 Tidy up regexps, add Text::PDF::Null
#$VERSION = "0.20"; # MJPH 27-APR-2002 $trailer->{'Size'} becomes max num objects, fix line end problem,
# remove warnings, update release code
#$VERSION = "0.19"; # MJPH 5-FEB-2002 fix hex keys and ASCII85 filter
#$VERSION = "0.18"; # MJPH 1-DEC-2001 add encryption hooks
#$VERSION = "0.17"; # GST 18-JUL-2001 Handle \) in strings and tidy up endobj handling, no uninitialized warnings
#$VERSION = "0.16"; # GST 18-JUL-2001 Major performance tweaks
#$VERSION = "0.15"; # GST 30-MAY-2001 Memory leaks fixed
#$VERSION = "0.14"; # MJPH 2-MAY-2001 More little bug fixes, added read_objnum
#$VERSION = "0.13"; # MJPH 23-MAR-2001 General bug fix release
#$VERSION = "0.12"; # MJPH 29-JUL-2000 Add font subsetting, random page insertion
#$VERSION = "0.11"; # MJPH 18-JUL-2000 Add pdfstamp.plx and more debugging
#$VERSION = "0.10"; # MJPH 27-JUN-2000 Tidy up some bugs - names
#$VERSION = "0.09"; # MJPH 31-MAR-2000 Copy trailer dictionary properly
#$VERSION = "0.08"; # MJPH 07-FEB-2000 Add null element
#$VERSION = "0.07"; # MJPH 01-DEC-1999 Debug for pdfbklt
#$VERSION = "0.06"; # MJPH 11-SEP-1999 Sort out unixisms
#$VERSION = "0.05"; # MJPH 9-SEP-1999 Add ship_out
#$VERSION = "0.04"; # MJPH 14-JUL-1999 Correct paths for tarball release
#$VERSION = "0.03"; # MJPH 14-JUL-1999 Correct paths for tarball release
#$VERSION = "0.02"; # MJPH 30-JUN-1999 Transfer from old library
BEGIN
{
my ($t, $type);
$ws_char = '[ \t\r\n\f\0]';
$delim_char = '[][<>{}()/%]';
$reg_char = '[^][<>{}()/% \t\r\n\f\0]';
$irreg_char = '[][<>{}()/% \t\r\n\f\0]';
$cr = "$ws_char*(?:\015|\012|(?:\015\012))";
%types = (
'Page' => 'Text::PDF::Page',
'Pages' => 'Text::PDF::Pages'
);
foreach $type (keys %types)
{
$t = $types{$type};
$t =~ s|::|/|og;
require "$t.pm";
}
}
=head2 Text::PDF::File->new
Creates a new, empty file object which can act as the host to other PDF objects.
Since there is no file associated with this object, it is assumed that the
object is created in readiness for creating a new PDF file.
=cut
sub new
{
my ($class, $root) = @_;
my ($self) = $class->_new;
unless ($root)
{
$root = PDFDict();
$root->{'Type'} = PDFName("Catalog");
}
$self->new_obj($root);
$self->{'Root'} = $root;
$self;
}
=head2 $p = Text::PDF::File->open($filename, $update)
Opens the file and reads all the trailers and cross reference tables to build
a complete directory of objects.
$update specifies whether this file is being opened for updating and editing,
or simply to be read.
$filename may be an IO object
=cut
sub open
{
my ($class, $fname, $update) = @_;
my ($self, $buf, $xpos, $end, $tdict, $k);
my ($fh);
$self = $class->_new;
if (ref $fname)
{
$self->{' INFILE'} = $fname;
$fh = $fname;
}
else
{
$fh = IO::File->new(($update ? "+" : "") . "<$fname") || return undef;
$self->{' INFILE'} = $fh;
}
binmode $fh;
if ($update)
{
$self->{' update'} = 1;
$self->{' OUTFILE'} = $fh;
$self->{' fname'} = $fname;
}
$fh->read($buf, 255);
if ($buf !~ m/^\%pdf\-1\.(\d)\s*$cr/moi)
{ die "$fname not a PDF file version 1.x"; }
else
{ $self->{' Version'} = $1; }
$fh->seek(0, 2); # go to end of file
$end = $fh->tell();
$self->{' epos'} = $end;
if (!$fh->seek(($end > 1024 ? $end - 1024 : 0, 0)))
{ die "Seek failed when reading PDF file $fname"; }
$fh->read($buf, 1024);
if ($buf !~ m/startxref$cr([0-9]+)$cr\%\%eof.*?$/oi)
{ die "Malformed PDF file $fname"; }
$xpos = $1;
$tdict = $self->readxrtr($xpos, $self);
foreach $k (keys %{$tdict})
{ $self->{$k} = $tdict->{$k}; }
return $self;
}
=head2 $p->release()
Releases ALL of the memory used by the PDF document and all of its component
objects. After calling this method, do B<NOT> expect to have anything left in
the C<Text::PDF::File> object (so if you need to save, be sure to do it before
calling this method).
B<NOTE>, that it is important that you call this method on any
C<Text::PDF::File> object when you wish to destruct it and free up its memory.
Internally, PDF files have an enormous number of cross-references and this
causes circular references within the internal data structures. Calling
'C<release()>' forces a brute-force cleanup of the data structures, freeing up
all of the memory. Once you've called this method, though, don't expect to be
able to do anything else with the C<Text::PDF::File> object; it'll have B<no>
internal state whatsoever.
B<Developer note:> As part of the brute-force cleanup done here, this method
will throw a warning message whenever unexpected key values are found within
the C<Text::PDF::File> object. This is done to help ensure that any unexpected
and unfreed values are brought to your attention so that you can bug us to keep
the module updated properly; otherwise the potential for memory leaks due to
dangling circular references will exist.
=cut
sub release
{
my ($self, $force) = @_;
my (@tofree);
# first, close the input file if it is still open
close($self->{' INFILE'});
# delete stuff that we know we can, here
if ($force)
{
foreach my $key (keys %{$self})
{
push(@tofree,$self->{$key});
$self->{$key}=undef;
delete($self->{$key});
}
}
else
{ @tofree = map { delete $self->{$_} } keys %{$self}; }
while (my $item = shift @tofree)
{
my $ref = ref($item);
if (UNIVERSAL::can($item, 'release'))
{ $item->release($force); }
elsif ($ref eq 'ARRAY')
{ push( @tofree, @{$item} ); }
elsif (UNIVERSAL::isa($ref, 'HASH'))
{ release($item, $force); }
}
# check that everything has gone - it better had!
foreach my $key (keys %{$self})
{ warn ref($self) . " still has '$key' key left after release.\n"; }
}
=head2 $p->append_file()
Appends the objects for output to the read file and then appends the appropriate tale.
=cut
sub append_file
{
my ($self) = @_;
my ($tdict, $fh, $t);
return undef unless ($self->{' update'});
$fh = $self->{' INFILE'};
if ($self->{' version'} > $self->{' Version'})
{
$fh->seek(0,0);
$fh->print("%PDF-1.$self->{' version'}\n");
}
$tdict = PDFDict();
$tdict->{'Prev'} = PDFNum($self->{' loc'});
$tdict->{'Info'} = $self->{'Info'};
if (defined $self->{' newroot'})
{ $tdict->{'Root'} = $self->{' newroot'}; }
else
{ $tdict->{'Root'} = $self->{'Root'}; }
$tdict->{'Size'} = $self->{'Size'};
# added v0.09
foreach $t (grep ($_ !~ m/^[\s\-]/o, keys %$self))
{ $tdict->{$t} = $self->{$t} unless defined $tdict->{$t}; }
$fh->seek($self->{' epos'}, 0);
$self->out_trailer($tdict, $self->{' update'});
close($self->{' OUTFILE'});
}
=head2 $p->out_file($fname)
Writes a PDF file to a file of the given filename based on the current list of
objects to be output. It creates the trailer dictionary based on information
in $self.
$fname may be an IO object;
=cut
sub out_file
{
my ($self, $fname) = @_;
$self->create_file($fname);
$self->close_file;
$self;
}
=head2 $p->create_file($fname)
Creates a new output file (no check is made of an existing open file) of
the given filename or IO object. Note, make sure that $p->{' version'} is set
correctly before calling this function.
=cut
sub create_file
{
my ($self, $fname) = @_;
my ($fh);
$self->{' fname'} = $fname;
if (ref $fname)
{ $fh = $fname; }
else
{
$fh = IO::File->new(">$fname") || die "Unable to open $fname for writing";
binmode $fh;
}
$self->{' OUTFILE'} = $fh;
$fh->print('%PDF-1.' . ($self->{' version'} || '2') . "\n");
$fh->print("%�쏢\n"); # and some binary stuff in a comment
$self;
}
=head2 $p->close_file
Closes up the open file for output by outputting the trailer etc.
=cut
sub close_file
{
my ($self) = @_;
my ($fh, $tdict, $t);
$tdict = PDFDict();
$tdict->{'Info'} = $self->{'Info'} if defined $self->{'Info'};
$tdict->{'Root'} = (defined $self->{' newroot'} and $self->{' newroot'} ne "") ? $self->{' newroot'} : $self->{'Root'};
# remove all freed objects from the outlist, AND the outlist_cache if not updating
# NO! Don't do that thing! In fact, let out_trailer do the opposite!
$tdict->{'Size'} = $self->{'Size'} || PDFNum(1);
$tdict->{'Prev'} = PDFNum($self->{' loc'}) if ($self->{' loc'});
if ($self->{' update'})
{
foreach $t (grep ($_ !~ m/^[\s\-]/o, keys %$self))
{ $tdict->{$t} = $self->{$t} unless defined $tdict->{$t}; }
$fh = $self->{' INFILE'};
$fh->seek($self->{' epos'}, 0);
}
$self->out_trailer($tdict, $self->{' update'});
close($self->{' OUTFILE'});
MacPerl::SetFileInfo("CARO", "TEXT", $self->{' fname'})
if ($^O eq "MacOS" && !ref($self->{' fname'}));
$self;
}
=head2 ($value, $str) = $p->readval($str, %opts)
Reads a PDF value from the current position in the file. If $str is too short
then read some more from the current location in the file until the whole object
is read. This is a recursive call which may slurp in a whole big stream (unprocessed).
Returns the recursive data structure read and also the current $str that has been
read from the file.
=cut
sub readval
{
my ($self, $str, %opts) = @_;
my ($fh) = $self->{' INFILE'};
my ($res, $key, $value, $k);
$str = update($fh, $str);
if ($str =~ m/^<</so)
{
$str = substr ($str, 2);
$str = update($fh, $str);
$res = PDFDict();
while ($str !~ m/^>>/o)
{
if ($str =~ s|^/($reg_char+)||o)
{
$k = Text::PDF::Name::name_to_string ($1, $self);
($value, $str) = $self->readval($str, %opts);
$res->{$k} = $value;
}
$str = update($fh, $str); # thanks gareth.jones@stud.man.ac.uk
}
$str =~ s/^>>//o;
$str = update($fh, $str);
# streams can't be followed by a lone carriage-return.
if (($str =~ s/^stream(?:(?:\015\012)|\012)//o)
&& ($res->{'Length'}->val != 0)) # stream
{
$k = $res->{'Length'}->val;
$res->{' streamsrc'} = $fh;
$res->{' streamloc'} = $fh->tell - length($str);
unless ($opts{'nostreams'})
{
if ($k > length($str))
{
$value = $str;
$k -= length($str);
read ($fh, $str, $k + 11); # slurp the whole stream!
} else
{ $value = ''; }
$value .= substr($str, 0, $k);
$res->{' stream'} = $value;
$res->{' nofilt'} = 1;
$str = update($fh, $str);
$str =~ s/^endstream//o;
}
}
if (defined $res->{'Type'} && defined $types{$res->{'Type'}->val})
{
bless $res, $types{$res->{'Type'}->val};
$res->init($self);
}
# gdj: FIXME: if any of the ws chars were crs, then the whole
# string might not have been read.
} elsif ($str =~ m/^([0-9]+)$ws_char+([0-9]+)$ws_char+R/so) # objind
{
$k = $1;
$value = $2;
$str =~ s/^([0-9]+)$ws_char+([0-9]+)$ws_char+R//so;
unless ($res = $self->test_obj($k, $value))
{
$res = Text::PDF::Objind->new();
$res->{' objnum'} = $k;
$res->{' objgen'} = $value;
$res->{' realised'} = 0;
$res->{' parent'} = $self;
$self->add_obj($res, $k, $value);
}
# gdj: FIXME: if any of the ws chars were crs, then the whole
# string might not have been read.
} elsif ($str =~ m/^([0-9]+)$ws_char+([0-9]+)$ws_char+obj/so) # object data
{
my ($obj);
$k = $1;
$value = $2;
$str =~ s/^([0-9]+)$ws_char+([0-9]+)$ws_char+obj//so;
($obj, $str) = $self->readval($str, %opts, 'objnum' => $k, 'objgen' => $value);
if ($res = $self->test_obj($k, $value))
{ $res->merge($obj); }
else
{
$res = $obj;
$self->add_obj($res, $k, $value);
$res->{' realised'} = 1;
}
$str = update($fh, $str); # thanks to kundrat@kundrat.sk
$str =~ s/^endobj//o;
} elsif ($str =~ m|^/($reg_char+)|so) # name
{
# " <- Fix colourization
$value = $1;
$str =~ s|^/($reg_char+)||so;
$res = Text::PDF::Name->from_pdf($value, $self);
} elsif ($str =~ m/^\(/o) # literal string
{
$str =~ s/^\(//o;
# We now need to find an unbalanced, unescaped right-paren.
# This can't be done with regexps.
my ($value) = "";
# The current level of nesting, when this reaches 0 we have finished.
my ($nested) = 1;
while (1) {
# Remove everything up to the first (possibly escaped) paren.
$str =~ /^((?:[^\\()]|\\[^()])*)(.*)/so;
$value .= $1;
$str = $2;
if ($str =~ /^(\\[()])/o) {
# An escaped paren. This would be tricky to do with
# the regexp above (it's very difficult to be certain
# that all cases are covered so I think it's better to
# deal with them explicitly).
$str = substr ($str, 2);
$value = $value . $1;
} elsif ($str =~ /^\)/o) {
# Right paren
$nested--;
$str = substr ($str, 1);
if ($nested == 0)
{ last; }
$value = $value . ')';
} elsif ($str =~ /^\(/o) {
# Left paren
$nested++;
$str = substr ($str, 1);
$value = $value . '(';
} else {
# No parens, we must read more. We don't use update
# because we don't want to remove whitespace or
# comments.
$fh->read($str, 255, length($str)) or die "Unterminated string.";
}
}
$res = Text::PDF::String->from_pdf($value);
} elsif ($str =~ m/^</o) # hex-string
{
$str =~ s/^<//o;
$fh->read($str, 255, length($str)) while (0 > index( $str, '>' ));
($value, $str) = ($str =~ /^(.*?)>(.*?)$/so);
$res = Text::PDF::String->from_pdf("<" . $value . ">");
} elsif ($str =~ m/^\[/o) # array
{
$str =~ s/^\[//o;
$str = update($fh, $str);
$res = PDFArray();
while ($str !~ m/^\]/o)
{
($value, $str) = $self->readval($str, %opts);
$res->add_elements($value);
$str = update($fh, $str);
}
$str =~ s/^\]//o;
} elsif ($str =~ m/^(true|false)$irreg_char/o) # boolean
{
$value = $1;
$str =~ s/^(?:true|false)//o;
$res = Text::PDF::Bool->from_pdf($value);
} elsif ($str =~ m/^([+-.0-9]+)$irreg_char/o) # number
{
$value = $1;
$str =~ s/^([+-.0-9]+)//o;
$res = Text::PDF::Number->from_pdf($value);
} elsif ($str =~ m/^null$irreg_char/o)
{
$str =~ s/^null//o;
$res = Text::PDF::Null->new;
} else
{
die "Can't parse `$str' near " . ($fh->tell()) . " length " . length($str) . ".";
}
$str =~ s/^$ws_char*//os;
return ($res, $str);
}
=head2 $ref = $p->read_obj($objind, %opts)
Given an indirect object reference, locate it and read the object returning
the read in object.
=cut
sub read_obj
{
my ($self, $objind, %opts) = @_;
my ($loc, $res, $str, $oldloc);
# return ($objind) if $self->{' objects'}{$objind->uid};
$res = $self->read_objnum($objind->{' objnum'}, $objind->{' objgen'}, %opts) || return undef;
$objind->merge($res) unless ($objind eq $res);
return $objind;
}
=head2 $ref = $p->read_objnum($num, $gen, %opts)
Returns a fully read object of given number and generation in this file
=cut
sub read_objnum
{
my ($self, $num, $gen, %opts) = @_;
my ($res, $loc, $str, $oldloc);
$loc = $self->locate_obj($num, $gen) || return undef;
$oldloc = $self->{' INFILE'}->tell;
$self->{' INFILE'}->seek($loc, 0);
($res, $str) = $self->readval('', %opts, 'objnum' => $num, 'objgen' => $gen);
$self->{' INFILE'}->seek($oldloc, 0);
$res;
}
=head2 $objind = $p->new_obj($obj)
Creates a new, free object reference based on free space in the cross reference chain.
If nothing free then thinks up a new number. If $obj then turns that object into this
new object rather than returning a new object.
=cut
sub new_obj
{
my ($self, $base) = @_;
my ($res);
my ($tdict, $i, $ni, $ng);
return $base if ($base->is_obj($self));
if (defined $self->{' free'} and scalar @{$self->{' free'}} > 0)
{
$res = shift(@{$self->{' free'}});
if (defined $base)
{
my ($num, $gen) = @{$self->{' objects'}{$res->uid}};
$self->remove_obj($res);
$self->add_obj($base, $num, $gen);
return $self->out_obj($base);
}
else
{
$self->{' objects'}{$res->uid}[2] = 0;
return $res;
}
}
$tdict = $self;
while (defined $tdict)
{
$i = $tdict->{' xref'}{defined($i)?$i:''}[0];
while (defined $i and $i != 0)
{
($ni, $ng) = @{$tdict->{' xref'}{$i}};
if (!defined $self->locate_obj($i, $ng))
{
if (defined $base)
{
$self->add_obj($base, $i, $ng);
return $base;
}
else
{
$res = $self->test_obj($i, $ng)
|| $self->add_obj(Text::PDF::Objind->new(), $i, $ng);
$tdict->{' xref'}{$i}[0] = $tdict->{' xref'}{$i}[0];
$self->out_obj($res);
return $res;
}
}
$i = $ni;
}
$tdict = $tdict->{' prev'};
}
$i = $self->{' maxobj'}++;
if (defined $base)
{
$self->add_obj($base, $i, 0);
$self->out_obj($base);
return $base;
}
else
{
$res = $self->add_obj(Text::PDF::Objind->new(), $i, 0);
$self->out_obj($res);
return $res;
}
}
=head2 $p->out_obj($objind)
Indicates that the given object reference should appear in the output xref
table whether with data or freed.
=cut
sub out_obj
{
my ($self, $obj) = @_;
# don't add objects that aren't real objects!
if (!defined $self->{' objects'}{$obj->uid})
{ return $self->new_obj($obj); }
# This is why we've been keeping the outlist CACHE around; to speed
# up this method by orders of magnitude (it saves up from having to
# grep the full outlist each time through as we'll just do a lookup
# in the hash) (which is super-fast).
elsif (!exists $self->{' outlist_cache'}{$obj->uid})
{
push( @{$self->{' outlist'}}, $obj );
$self->{' outlist_cache'}{$obj->uid}++;
}
$obj;
}
=head2 $p->free_obj($objind)
Marks an object reference for output as being freed.
=cut
sub free_obj
{
my ($self, $obj) = @_;
push(@{$self->{' free'}}, $obj);
$self->{' objects'}{$obj->uid}[2] = 1;
$self->out_obj($obj);
}
=head2 $p->remove_obj($objind)
Removes the object from all places where we might remember it
=cut
sub remove_obj
{
my ($self, $objind) = @_;
# who says it has to be fast
delete $self->{' objects'}{$objind->uid};
delete $self->{' outlist_cache'}{$objind->uid};
delete $self->{' printed_cache'}{$objind};
@{$self->{' outlist'}} = grep($_ ne $objind, @{$self->{' outlist'}});
@{$self->{' printed'}} = grep($_ ne $objind, @{$self->{' printed'}});
$self->{' objcache'}{$objind->{' objnum'}, $objind->{' objgen'}} = undef
if ($self->{' objcache'}{$objind->{' objnum'}, $objind->{' objgen'}} eq $objind);
$self;
}
=head2 $p->ship_out(@objects)
Ships the given objects (or all objects for output if @objects is empty) to
the currently open output file (assuming there is one). Freed objects are not
shipped, and once an object is shipped it is switched such that this file
becomes its source and it will not be shipped again unless out_obj is called
again. Notice that a shipped out object can be re-output or even freed, but
that it will not cause the data already output to be changed.
=cut
sub ship_out
{
my ($self, @objs) = @_;
my ($n, $fh, $objind, $i, $j);
my ($objnum, $objgen);
return unless defined($fh = $self->{' OUTFILE'});
seek($fh, 0, 2); # go to the end of the file
@objs = @{$self->{' outlist'}} unless (scalar @objs > 0);
foreach $objind (@objs)
{
next unless $objind->is_obj($self);
$j = -1;
for ($i = 0; $i < scalar @{$self->{' outlist'}}; $i++)
{
if ($self->{' outlist'}[$i] eq $objind)
{
$j = $i;
last;
}
}
next if ($j < 0);
splice(@{$self->{' outlist'}}, $j, 1);
delete $self->{' outlist_cache'}{$objind->uid};
next if grep {$_ eq $objind} @{$self->{' free'}};
$self->{' locs'}{$objind->uid} = $fh->tell;
($objnum, $objgen) = @{$self->{' objects'}{$objind->uid}}[0..1];
$fh->printf("%d %d obj\n", $objnum, $objgen);
$objind->outobjdeep($fh, $self, 'objnum' => $objnum, 'objgen' => $objgen);
$fh->print("\nendobj\n");
# Note that we've output this obj, not forgetting to update the cache
# of whats printed.
unless (exists $self->{' printed_cache'}{$objind})
{
push( @{$self->{' printed'}}, $objind );
$self->{' printed_cache'}{$objind}++;
}
}
$self;
}
=head2 $p->copy($outpdf, \&filter)
Iterates over every object in the file reading the object, calling filter with the object
and outputting the result. if filter is not defined, then just copies input to output.
=cut
sub copy
{
my ($self, $out, $filt) = @_;
my ($tdict, $i, $nl, $ng, $nt, $res, $obj, $minl, $mini, $ming);
foreach $i (grep (!m/^[\s\-]/o, keys %{$self}))
{ $out->{$i} = $self->{$i} unless defined $out->{$i}; }
$tdict = $self;
while (defined $tdict)
{
foreach $i (sort {$a <=> $b} keys %{$tdict->{' xref'}})
{
($nl, $ng, $nt) = @{$tdict->{' xref'}{$i}};
next unless $nt eq 'n';
if ($nl < $minl || $mini == 0)
{
$mini = $i;
$ming = $ng;
$minl = $nl;
}
unless ($obj = $self->test_obj($i, $ng))
{
$obj = Text::PDF::Objind->new();
$obj->{' objnum'} = $i;
$obj->{' objgen'} = $ng;
$self->add_obj($obj, $i, $ng);
$obj->{' parent'} = $self;
$obj->{' realised'} = 0;
}
$obj->realise;
$res = defined $filt ? &{$filt}($obj) : $obj;
$out->new_obj($res) unless (!$res || $res->is_obj($out));
}
$tdict = $tdict->{' prev'};
}
# test for linearized and remove it from output
$obj = $self->test_obj($mini, $ming);
if ($obj->isa('Text::PDF::Dict') && $obj->{'Linearized'})
{ $out->free_obj($obj); }
$self;
}
=head1 PRIVATE METHODS & FUNCTIONS
The following methods and functions are considered private to this class. This
does not mean you cannot use them if you have a need, just that they aren't really
designed for users of this class.
=head2 $offset = $p->locate_obj($num, $gen)
Returns a file offset to the object asked for by following the chain of cross
reference tables until it finds the one you want.
=cut
sub locate_obj
{
my ($self, $num, $gen) = @_;
my ($tdict, $ref);
$tdict = $self;
while (defined $tdict)
{
if (ref $tdict->{' xref'}{$num})
{
$ref = $tdict->{' xref'}{$num};
if ($ref->[1] == $gen)
{
return $ref->[0] if ($ref->[2] eq "n");
return undef; # if $ref->[2] eq "f"
}
}
$tdict = $tdict->{' prev'}
}
return undef;
}
=head2 update($fh, $str)
Keeps reading $fh for more data to ensure that $str has at least a line full
for C<readval> to work on. At this point we also take the opportunity to ignore
comments.
=cut
sub update
{
my ($fh, $str) = @_;
$str =~ s/^$ws_char*//o;
while ($str !~ m/$cr/o && !$fh->eof)
{
$fh->read($str, 255, length($str));
$str =~ s/^$ws_char*//so;
while ($str =~ m/^\%/o)
{
$fh->read($str, 255, length($str)) while ($str !~ m/$cr/o && !$fh->eof);
$str =~ s/^\%(.*)$cr$ws_char*//so;
}
}
return $str;
}
=head2 $objind = $p->test_obj($num, $gen)
Tests the cache to see whether an object reference (which may or may not have
been getobj()ed) has been cached. Returns it if it has.
=cut
sub test_obj
{ $_[0]->{' objcache'}{$_[1], $_[2]}; }
=head2 $p->add_obj($objind)
Adds the given object to the internal object cache.
=cut
sub add_obj
{
my ($self, $obj, $num, $gen) = @_;
$self->{' objcache'}{$num, $gen} = $obj;
$self->{' objects'}{$obj->uid} = [$num, $gen];
return $obj;
}
=head2 $tdict = $p->readxrtr($xpos)
Recursive function which reads each of the cross-reference and trailer tables
in turn until there are no more.
Returns a dictionary corresponding to the trailer chain. Each trailer also
includes the corresponding cross-reference table.
The structure of the xref private element in a trailer dictionary is of an
anonymous hash of cross reference elements by object number. Each element
consists of an array of 3 elements corresponding to the three elements read
in [location, generation number, free or used]. See the PDF Specification
for details.
=cut
sub readxrtr
{
my ($self, $xpos) = @_;
my ($tdict, $xlist, $buf, $xmin, $xnum, $fh, $xdiff);
$fh = $self->{' INFILE'};
$fh->seek($xpos, 0);
$fh->read($buf, 22);
if ($buf !~ m/^xref$cr/oi)
{ die "Malformed xref in PDF file $self->{' fname'}"; }
$buf =~ s/^xref$cr//oi;
$xlist = {};
while ($buf =~ m/^([0-9]+)$ws_char+([0-9]+)$cr(.*?)$/so)
{
$xmin = $1;
$xnum = $2;
$buf = $3;
$xdiff = length($buf);
$fh->read($buf, 20 * $xnum - $xdiff + 15, $xdiff);
while ($xnum-- > 0 && $buf =~ s/^0*([0-9]*)$ws_char+0*([0-9]+)$ws_char+([nf])$cr//o)
{ $xlist->{$xmin++} = [$1, $2, $3]; }
}
if ($buf !~ /^trailer$cr/oi)
{ die "Malformed trailer in PDF file $self->{' fname'} at " . ($fh->tell - length($buf)); }
$buf =~ s/^trailer$cr//oi;
($tdict, $buf) = $self->readval($buf);
$tdict->{' loc'} = $xpos;
$tdict->{' xref'} = $xlist;
$self->{' maxobj'} = $xmin if $xmin > $self->{' maxobj'};
$tdict->{' prev'} = $self->readxrtr($tdict->{'Prev'}->val)
if (defined $tdict->{'Prev'} && $tdict->{'Prev'}->val != 0);
return $tdict;
}
=head2 $p->out_trailer($tdict)
Outputs the body and trailer for a PDF file by outputting all the objects in
the ' outlist' and then outputting a xref table for those objects and any
freed ones. It then outputs the trailing dictionary and the trailer code.
=cut
sub out_trailer
{
my ($self, $tdict, $update) = @_;
my ($objind, $j, $i, $iend, @xreflist, $first, $k, $xref, $tloc, @freelist);
my (%locs, $size);
my ($fh) = $self->{' OUTFILE'};
while (@{$self->{' outlist'}})
{ $self->ship_out; }
# foreach $objind (@{$self->{' outlist'}})
# {
# next if ($self->{' objects'}{$objind->uid}[2]);
# $locs{$objind->uid} = $fh->tell;
# $fh->printf("%d %d obj\n", @{$self->{' objects'}{$objind->uid}}[0..1]);
# $objind->outobjdeep($fh, $self);
# $fh->print("\nendobj\n");
# }
# $size = @{$self->{' printed'}} + @{$self->{' free'}};
# $tdict->{'Size'} = PDFNum($tdict->{'Size'}->val + $size);
# PDFSpec 1.3 says for /Size: (Required) Total number of entries in the file�s
# cross-reference table, including the original table and all updates. Which
# is what the previous two lines implement.
# But this seems to make Acrobat croak on saving so we try the following from
# basil.duval@epfl.ch
$tdict->{'Size'} = PDFNum($self->{' maxobj'});
$tloc = $fh->tell;
$fh->print("xref\n");
@xreflist = sort {$self->{' objects'}{$a->uid}[0] <=>
$self->{' objects'}{$b->uid}[0]}
(@{$self->{' printed'}}, @{$self->{' free'}});
unless ($update)
{
$i = 1;
for ($j = 0; $j < @xreflist; $j++)
{
my (@inserts);
$k = $xreflist[$j];
while ($i < $self->{' objects'}{$k->uid}[0])
{
my ($n) = Text::PDF::Objind->new();
$self->add_obj($n, $i, 0);
$self->free_obj($n);
push(@inserts, $n);
$i++;
}
splice(@xreflist, $j, 0, @inserts);
$j += @inserts;
$i++;
}
}
@freelist = sort {$self->{' objects'}{$a->uid}[0] <=>
$self->{' objects'}{$b->uid}[0]} @{$self->{' free'}};
$j = 0; $first = -1; $k = 0;
for ($i = 0; $i <= $#xreflist + 1; $i++)
{
# if ($i == 0)
# {
# $first = $i; $j = $xreflist[0]->{' objnum'};
# $fh->printf("0 1\n%010d 65535 f \n", $ff);
# }
if ($i > $#xreflist || $self->{' objects'}{$xreflist[$i]->uid}[0] != $j + 1)
{
$fh->print(($first == -1 ? "0 " : "$self->{' objects'}{$xreflist[$first]->uid}[0] ") . ($i - $first) . "\n");
if ($first == -1)
{
$fh->printf("%010d 65535 f \n", defined $freelist[$k] ? $self->{' objects'}{$freelist[$k]->uid}[0] : 0);
$first = 0;
}
for ($j = $first; $j < $i; $j++)
{
$xref = $xreflist[$j];
if (defined $freelist[$k] && defined $xref && "$freelist[$k]" eq "$xref")
{
$k++;
$fh->print(pack("A10AA5A4",
sprintf("%010d", (defined $freelist[$k] ?
$self->{' objects'}{$freelist[$k]->uid}[0] : 0)), " ",
sprintf("%05d", $self->{' objects'}{$xref->uid}[1] + 1),
" f \n"));
} else
{
$fh->print(pack("A10AA5A4", sprintf("%010d", $self->{' locs'}{$xref->uid}), " ",
sprintf("%05d", $self->{' objects'}{$xref->uid}[1]),
" n \n"));
}
}
$first = $i;
$j = $self->{' objects'}{$xreflist[$i]->uid}[0] if ($i < scalar @xreflist);
} else
{ $j++; }
}
$fh->print("trailer\n");
$tdict->outobjdeep($fh, $self);
$fh->print("\nstartxref\n$tloc\n" . '%%EOF' . "\n");
}
=head2 Text::PDF::File->_new
Creates a very empty PDF file object (used by new and open)
=cut
sub _new
{
my ($class) = @_;
my ($self) = {};
bless $self, $class;
$self->{' outlist'} = [];
$self->{' outlist_cache'} = {}; # A cache of whats in the 'outlist'
$self->{' maxobj'} = 1;
$self->{' objcache'} = {};
$self->{' objects'} = {};
$self;
}
1;
=head1 AUTHOR
Martin Hosken Martin_Hosken@sil.org
Copyright Martin Hosken 1999 and onwards
No warranty or expression of effectiveness, least of all regarding anyone's
safety, is implied in this software or documentation.
=head2 Licensing
This Perl Text::PDF module is licensed under the Perl Artistic License.
|