Changeset 5681
- Timestamp:
- 2003-10-20T14:59:00+13:00 (21 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
trunk/gsdl/perllib/plugins/BasPlug.pm
r4873 r5681 43 43 use DateExtract; 44 44 use ghtml; 45 use gsprintf; 45 46 use printusage; 47 46 48 47 49 my $unicode_list = … … 126 128 127 129 my $options = { 'name' => "BasPlug", 128 'desc' => " Base class for all the import plugins.",130 'desc' => "{BasPlug.desc}", 129 131 'inherits' => "No", 130 132 'args' => $arguments }; 133 134 135 sub gsprintf 136 { 137 return &gsprintf::gsprintf(@_); 138 } 131 139 132 140 … … 233 241 # Produce the usage information using the data structure above 234 242 if ($isleafclass) { 235 print STDERR " usage: plugin $pluginname [options]\n\n"; 243 &gsprintf(STDERR, " {common.usage}: plugin $pluginname [{common.options}]\n\n"); 244 # print STDERR " usage: classify $classifiername [options]\n\n"; 236 245 } 237 246 … … 242 251 243 252 if ($isleafclass) { 244 print STDERR " specific options:\n"; 253 &gsprintf(STDERR, " {common.specific_options}:\n"); 254 # print STDERR " specific options:\n"; 245 255 } 246 256 else { 247 print STDERR " general options (from $pluginname):\n"; 257 &gsprintf(STDERR, " {common.general_options}:\n", $pluginname); 258 # print STDERR " general options (from $classifiername):\n"; 248 259 } 249 260 … … 392 403 q^cover_image^, \$self->{'cover_image'}, 393 404 "allow_extra_options")) { 394 395 print STDERR "\nThe $plugin_name plugin uses an incorrect general option (general options are those\n"; 396 print STDERR "available to all plugins). Check your collect.cfg configuration file.\n"; 405 406 &gsprintf(STDERR, "\n{BasPlug.bad_general_option}\n", $plugin_name); 397 407 $self->print_txt_usage(""); # Use default resource bundle 398 408 die "\n"; … … 483 493 484 494 if ($self->is_recursive()) { 485 die "BasPlug::read function must be implemented in sub-class for recursive plugins\n";495 &gsprintf(STDERR, "{BasPlug.read_must_be_implemented}") && die "\n"; 486 496 } 487 497 … … 521 531 if (!length ($text)) { 522 532 my $plugin_name = ref ($self); 523 print $outhandle "$plugin_name: ERROR: $file contains no text\n"if $self->{'verbosity'};533 &gsprintf($outhandle, "$plugin_name: {BasPlug.file_has_no_text}\n", $filename) if $self->{'verbosity'}; 524 534 525 535 my $failhandle = $self->{'failhandle'}; 526 print $failhandle "$file: " . ref($self) . ": file contains no text\n"; 536 &gsprintf($failhandle, "$file: " . ref($self) . ": {BasPlug.empty_file}\n"); 537 # print $failhandle "$file: " . ref($self) . ": file contains no text\n"; 527 538 $self->{'num_not_processed'} ++; 528 539 … … 563 574 my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj) = @_; 564 575 565 die "Basplug::process function must be implemented in sub-class\n"; 576 &gsprintf(STDERR, "BasPlug::process {common.must_be_implemented}\n") && die "\n"; 577 # die "Basplug::process function must be implemented in sub-class\n"; 566 578 567 579 return undef; # never gets here … … 579 591 { 580 592 my $outhandle = $self->{'outhandle'}; 581 print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'}; 593 &gsprintf($outhandle, "{BasPlug.read_denied}\n", $filename) if $self->{'verbosity'}; 594 # print $outhandle "Read permission denied for $filename\n" if $self->{'verbosity'}; 582 595 return; 583 596 } … … 585 598 $$textref = ""; 586 599 587 open (FILE, $filename) || die "BasPlug::read_file could not open $filename for reading ($!)\n"; 600 open (FILE, $filename) || (&gsprintf(STDERR, "BasPlug::read_file {BasPlug.could_not_open_for_reading} ($!)\n", $filename) && die "\n"); 601 # open (FILE, $filename) || die "BasPlug::read_file could not open $filename for reading ($!)\n"; 588 602 589 603 if ($encoding eq "ascii") { … … 621 635 my $plugin_name = ref ($self); 622 636 my $outhandle = $self->{'outhandle'}; 623 print $outhandle "$plugin_name: WARNING: $filename was read using $encoding encoding but "; 624 print $outhandle "appears to be encoded as $extracted_encoding.\n"; 637 &gsprintf($outhandle, "$plugin_name: {BasPlug.wrong_encoding}\n", $filename, $encoding, $extracted_encoding); 638 # print $outhandle "$plugin_name: WARNING: $filename was read using $encoding encoding but "; 639 # print $outhandle "appears to be encoded as $extracted_encoding.\n"; 625 640 } 626 641 } else { … … 640 655 641 656 # read in file 642 open (FILE, $filename) || die "BasPlug::get_language_encoding could not open $filename for reading ($!)\n";657 open (FILE, $filename) || (&gsprintf(STDERR, "BasPlug::get_language_encoding {BasPlug.could_not_open_for_reading} ($!)\n", $filename) && die "\n"); # die "BasPlug::get_language_encoding could not open $filename for reading ($!)\n"; 643 658 undef $/; 644 659 my $text = <FILE>; … … 681 696 if ($self->{'input_encoding'} ne 'auto') { 682 697 if ($self->{'extract_language'} && $self->{'verbosity'}) { 683 print $outhandle "BasPlug: WARNING: language could not be extracted from $filename - "; 684 print $outhandle "defaulting to $self->{'default_language'}\n"; 698 &gsprintf($outhandle, "BasPlug: {BasPlug.could_not_extract_language}\n", $filename, $self->{'default_language'}); 699 # print $outhandle "BasPlug: WARNING: language could not be extracted from $filename - "; 700 # print $outhandle "defaulting to $self->{'default_language'}\n"; 685 701 } 686 702 return ($self->{'default_language'}, $self->{'input_encoding'}); … … 688 704 } else { 689 705 if ($self->{'verbosity'}) { 690 print $outhandle "BASPlug: WARNING: language could not be extracted from $filename - "; 691 print $outhandle "defaulting to $self->{'default_language'}.\n"; 706 &gsprintf($outhandle, "BasPlug: {BasPlug.could_not_extract_language}\n", $filename, $self->{'default_language'}); 707 # print $outhandle "BASPlug: WARNING: language could not be extracted from $filename - "; 708 # print $outhandle "defaulting to $self->{'default_language'}.\n"; 692 709 } 693 710 return ($self->{'default_language'}, $best_encoding); … … 699 716 if (!defined $language) { 700 717 if ($self->{'verbosity'}) { 701 print $outhandle "BasPlug: WARNING: language could not be extracted from $filename - "; 702 print $outhandle "defaulting to $self->{'default_language'}\n"; 718 &gsprintf($outhandle, "BasPlug: {BasPlug.could_not_extract_language}\n", $filename, $self->{'default_language'}); 719 # print $outhandle "BasPlug: WARNING: language could not be extracted from $filename - "; 720 # print $outhandle "defaulting to $self->{'default_language'}\n"; 703 721 } 704 722 $language = $self->{'default_language'}; … … 706 724 if (!defined $encoding) { 707 725 if ($self->{'verbosity'}) { 708 print $outhandle "BasPlug: WARNING: encoding could not be extracted from $filename - "; 709 print $outhandle "defaulting to $self->{'default_encoding'}\n"; 726 &gsprintf($outhandle, "BasPlug: {BasPlug.could_not_extract_encoding}\n", $filename, $self->{'default_encoding'}); 727 # print $outhandle "BasPlug: WARNING: encoding could not be extracted from $filename - "; 728 # print $outhandle "defaulting to $self->{'default_encoding'}\n"; 710 729 } 711 730 $encoding = $self->{'default_encoding'}; … … 715 734 !defined $encodings::encodings->{$encoding}) { 716 735 if ($self->{'verbosity'}) { 717 print $outhandle "BasPlug: WARNING: $filename appears to be encoded in an unsupported encoding ($encoding) - "; 718 print $outhandle "using $self->{'default_encoding'}\n"; 736 &gsprintf($outhandle, "BasPlug: {BasPlug.unsupported_encoding}\n", $filename, $encoding, $self->{'default_encoding'}); 737 # print $outhandle "BasPlug: WARNING: $filename appears to be encoded in an unsupported encoding ($encoding) - "; 738 # print $outhandle "using $self->{'default_encoding'}\n"; 719 739 } 720 740 $encoding = $self->{'default_encoding'}; … … 783 803 my $outhandle = $self->{'outhandle'}; 784 804 785 print $outhandle " extracting email addresses ...\n" 805 # print $outhandle " extracting email addresses ...\n" 806 &gsprintf($outhandle, " {BasPlug.extracting_emails}...\n") 786 807 if ($self->{'verbosity'} > 2); 787 808 … … 794 815 push @email2, $address; 795 816 $doc_obj->add_utf8_metadata ($thissection, "emailAddress", $address); 796 print $outhandle " extracting $address\n" 817 # print $outhandle " extracting $address\n" 818 &gsprintf($outhandle, " {BasPlug.extracting} $address\n") 797 819 if ($self->{'verbosity'} > 3); 798 820 } 799 821 } 800 print $outhandle " done extracting email addresses.\n" 822 # print $outhandle " done extracting email addresses.\n" 823 &gsprintf($outhandle, " {BasPlug.done_email_extract}\n") 801 824 if ($self->{'verbosity'} > 2); 802 803 825 } 804 826 805 827 # extract metadata 806 sub auto_extract_metadata { 807 828 sub auto_extract_metadata { 808 829 809 830 my $self = shift (@_); … … 842 863 my $keyphrases = $list[0]; #first arg is keyphrase list 843 864 my $stems = $list[1]; #second arg is stemmed keyphrase list 844 print STDERR "keyphrases: $keyphrases\n"; 845 print STDERR "stems: $stems\n"; 865 &gsprintf(STDERR, "{BasPlug.keyphrases}: $keyphrases\n"); 866 # print STDERR "keyphrases: $keyphrases\n"; 867 &gsprintf(STDERR, "{BasPlug.stems}: $stems\n"); 868 # print STDERR "stems: $stems\n"; 846 869 $thissection = $doc_obj->get_top_section(); #add metadata to top section 847 870 $doc_obj->add_metadata($thissection, "kea", $keyphrases); … … 903 926 my $outhandle = $self->{'outhandle'}; 904 927 905 print $outhandle " extracting acronyms ...\n" 928 # print $outhandle " extracting acronyms ...\n" 929 &gsprintf($outhandle, " {BasPlug.extracting_acronyms}...\n") 906 930 if ($self->{'verbosity'} > 2); 907 931 … … 916 940 if ($thisAcro eq $acro->to_string()) { 917 941 $seen_before = "true"; 918 print $outhandle " already seen ". $acro->to_string() . "\n" 942 # print $outhandle " already seen ". $acro->to_string() . "\n" 943 &gsprintf($outhandle, " {BasPlug.already_seen} " . $acro->to_string() . "\n") 919 944 if ($self->{'verbosity'} >= 4); 920 } 945 } 921 946 } 922 947 … … 927 952 #do the normal acronym 928 953 $doc_obj->add_utf8_metadata($thissection, "Acronym", $acro->to_string()); 929 print $outhandle " adding ". $acro->to_string() . "\n" 954 # print $outhandle " adding ". $acro->to_string() . "\n" 955 &gsprintf($outhandle, " {BasPlug.adding} " . $acro->to_string() . "\n") 930 956 if ($self->{'verbosity'} > 3); 931 932 } 933 } 934 print $outhandle " done extracting acronyms. \n" 957 } 958 } 959 960 # print $outhandle " done extracting acronyms. \n" 961 &gsprintf($outhandle, " {BasPlug.done_acronym_extract}\n") 935 962 if ($self->{'verbosity'} > 2); 936 963 } … … 941 968 my $outhandle = $self->{'outhandle'}; 942 969 943 print $outhandle " marking up acronyms ...\n" 970 # print $outhandle " marking up acronyms ...\n" 971 &gsprintf($outhandle, " {BasPlug.marking_up_acronyms}...\n") 944 972 if ($self->{'verbosity'} > 2); 945 973 … … 947 975 $text = &acronym::markup_acronyms($text, $self); 948 976 949 print $outhandle " done marking up acronyms. \n" 977 # print $outhandle " done marking up acronyms. \n" 978 &gsprintf($outhandle, " {BasPlug.done_acronym_markup}\n") 950 979 if ($self->{'verbosity'} > 2); 951 980
Note:
See TracChangeset
for help on using the changeset viewer.