[15867] | 1 | package AcronymExtractor;
|
---|
| 2 |
|
---|
| 3 | use acronym;
|
---|
| 4 | use PrintInfo;
|
---|
[15887] | 5 | use strict;
|
---|
[15867] | 6 |
|
---|
| 7 | BEGIN {
|
---|
| 8 | @AcronymExtractor::ISA = ('PrintInfo');
|
---|
| 9 | }
|
---|
| 10 |
|
---|
| 11 | my $arguments = [
|
---|
| 12 | { 'name' => "extract_acronyms",
|
---|
| 13 | 'desc' => "{AcronymExtractor.extract_acronyms}",
|
---|
| 14 | 'type' => "flag",
|
---|
| 15 | 'reqd' => "no" },
|
---|
| 16 | { 'name' => "markup_acronyms",
|
---|
| 17 | 'desc' => "{AcronymExtractor.markup_acronyms}",
|
---|
| 18 | 'type' => "flag",
|
---|
| 19 | 'reqd' => "no" } ];
|
---|
| 20 |
|
---|
| 21 | my $options = { 'name' => "AcronymExtractor",
|
---|
| 22 | 'desc' => "{AcronymExtractor.desc}",
|
---|
| 23 | 'abstract' => "yes",
|
---|
| 24 | 'inherits' => "yes",
|
---|
| 25 | 'args' => $arguments };
|
---|
| 26 |
|
---|
| 27 |
|
---|
| 28 | sub new {
|
---|
| 29 | my ($class) = shift (@_);
|
---|
| 30 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
| 31 | push(@$pluginlist, $class);
|
---|
| 32 |
|
---|
| 33 | if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
|
---|
| 34 | if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
|
---|
| 35 |
|
---|
[15881] | 36 | my $self = new PrintInfo($pluginlist, $inputargs, $hashArgOptLists,1);
|
---|
[15867] | 37 |
|
---|
| 38 | return bless $self, $class;
|
---|
| 39 |
|
---|
| 40 | }
|
---|
| 41 |
|
---|
| 42 |
|
---|
| 43 | # initialise metadata extractors
|
---|
| 44 | sub initialise_acronym_extractor {
|
---|
| 45 | my $self = shift (@_);
|
---|
| 46 |
|
---|
| 47 | if ($self->{'extract_acronyms'} || $self->{'markup_acronyms'}) {
|
---|
| 48 | &acronym::initialise_acronyms();
|
---|
| 49 | }
|
---|
| 50 | }
|
---|
| 51 |
|
---|
| 52 | # finalise metadata extractors
|
---|
| 53 | sub finalise_acronym_extractor {
|
---|
| 54 | my $self = shift (@_);
|
---|
| 55 |
|
---|
| 56 | if ($self->{'extract_acronyms'} || $self->{'markup_acronyms'}) {
|
---|
| 57 | &acronym::finalise_acronyms();
|
---|
| 58 | }
|
---|
| 59 | }
|
---|
| 60 |
|
---|
| 61 | # extract metadata
|
---|
| 62 | sub extract_acronym_metadata {
|
---|
| 63 |
|
---|
| 64 | my $self = shift (@_);
|
---|
| 65 | my ($doc_obj) = @_;
|
---|
| 66 |
|
---|
| 67 |
|
---|
| 68 | if ($self->{'extract_acronyms'}) {
|
---|
| 69 | my $thissection = $doc_obj->get_top_section();
|
---|
| 70 | while (defined $thissection) {
|
---|
| 71 | my $text = $doc_obj->get_text($thissection);
|
---|
| 72 | $self->extract_acronyms (\$text, $doc_obj, $thissection) if $text =~ /./;
|
---|
| 73 | $thissection = $doc_obj->get_next_section ($thissection);
|
---|
| 74 | }
|
---|
| 75 | }
|
---|
| 76 |
|
---|
| 77 | if ($self->{'markup_acronyms'}) {
|
---|
| 78 | my $thissection = $doc_obj->get_top_section();
|
---|
| 79 | while (defined $thissection) {
|
---|
| 80 | my $text = $doc_obj->get_text($thissection);
|
---|
| 81 | $text = $self->markup_acronyms ($text, $doc_obj, $thissection);
|
---|
| 82 | $doc_obj->delete_text($thissection);
|
---|
| 83 | $doc_obj->add_text($thissection, $text);
|
---|
| 84 | $thissection = $doc_obj->get_next_section ($thissection);
|
---|
| 85 | }
|
---|
| 86 | }
|
---|
| 87 |
|
---|
| 88 | }
|
---|
| 89 |
|
---|
| 90 |
|
---|
| 91 |
|
---|
| 92 | # extract acronyms from a section in a document. progress is
|
---|
| 93 | # reported to outhandle based on the verbosity. both the Acronym
|
---|
| 94 | # and the AcronymKWIC metadata items are created.
|
---|
| 95 |
|
---|
| 96 | sub extract_acronyms {
|
---|
| 97 | my $self = shift (@_);
|
---|
| 98 | my ($textref, $doc_obj, $thissection) = @_;
|
---|
| 99 | my $outhandle = $self->{'outhandle'};
|
---|
| 100 |
|
---|
| 101 | # print $outhandle " extracting acronyms ...\n"
|
---|
| 102 | gsprintf($outhandle, " {BasPlug.extracting_acronyms}...\n")
|
---|
| 103 | if ($self->{'verbosity'} > 2);
|
---|
| 104 |
|
---|
| 105 | my $acro_array = &acronym::acronyms($textref);
|
---|
| 106 |
|
---|
| 107 | foreach my $acro (@$acro_array) {
|
---|
| 108 |
|
---|
| 109 | #check that this is the first time ...
|
---|
| 110 | my $seen_before = "false";
|
---|
| 111 | my $previous_data = $doc_obj->get_metadata($thissection, "Acronym");
|
---|
| 112 | foreach my $thisAcro (@$previous_data) {
|
---|
| 113 | if ($thisAcro eq $acro->to_string()) {
|
---|
| 114 | $seen_before = "true";
|
---|
| 115 | if ($self->{'verbosity'} >= 4) {
|
---|
| 116 | gsprintf($outhandle, " {BasPlug.already_seen} " .
|
---|
| 117 | $acro->to_string() . "\n");
|
---|
| 118 | }
|
---|
| 119 | }
|
---|
| 120 | }
|
---|
| 121 |
|
---|
| 122 | if ($seen_before eq "false") {
|
---|
| 123 | #write it to the file ...
|
---|
| 124 | $acro->write_to_file();
|
---|
| 125 |
|
---|
| 126 | #do the normal acronym
|
---|
| 127 | $doc_obj->add_utf8_metadata($thissection, "Acronym", $acro->to_string());
|
---|
| 128 | gsprintf($outhandle, " {BasPlug.adding} ".$acro->to_string()."\n")
|
---|
| 129 | if ($self->{'verbosity'} > 3);
|
---|
| 130 | }
|
---|
| 131 | }
|
---|
| 132 |
|
---|
| 133 | gsprintf($outhandle, " {BasPlug.done_acronym_extract}\n")
|
---|
| 134 | if ($self->{'verbosity'} > 2);
|
---|
| 135 | }
|
---|
| 136 |
|
---|
| 137 | sub markup_acronyms {
|
---|
| 138 | my $self = shift (@_);
|
---|
| 139 | my ($text, $doc_obj, $thissection) = @_;
|
---|
| 140 | my $outhandle = $self->{'outhandle'};
|
---|
| 141 |
|
---|
| 142 | gsprintf($outhandle, " {BasPlug.marking_up_acronyms}...\n")
|
---|
| 143 | if ($self->{'verbosity'} > 2);
|
---|
| 144 |
|
---|
| 145 | #self is passed in to check for verbosity ...
|
---|
| 146 | $text = &acronym::markup_acronyms($text, $self);
|
---|
| 147 |
|
---|
| 148 | gsprintf($outhandle, " {BasPlug.done_acronym_markup}\n")
|
---|
| 149 | if ($self->{'verbosity'} > 2);
|
---|
| 150 |
|
---|
| 151 | return $text;
|
---|
| 152 | }
|
---|
| 153 |
|
---|
| 154 | 1;
|
---|