| #!/usr/bin/env perl |
| |
| ## see TODOs |
| |
| ### NOTES: |
| ### |
| ### 1. The input for this file has to be UTF8-encoded _without_ SGML-Entites (except the escaped ones -> &amp;, <lt;, >gt;, 'apos;, "quot; |
| ### - XML::LibXML::Reader translates them back to their origin - e.g.: &amp; -> & (because '&' translates to '&'). |
| ### SGML-Entities cannot be processed, because of the first 2 missing lines in the input document (see xces2korap.sh). |
| ### |
| ### 2. Let 'data.xml' only be used/referred by 'sentences.xml', 'paragraphs.xml' and 'structure.xml', since it can contain double blanks (' ' - see comment to '<s/>' in func. |
| ### retr_info) and TODO: also contains not only text inside s-tags (e.g.: <label><hi rend="it bo">Ludwig Pietsch (1824-1911):</hi></label><item><s>Journalist ...) |
| ### |
| |
| use 5.012; |
| |
| use strict; |
| use warnings; |
| use File::Basename; # dirname |
| use IO::Handle qw( ); |
| use IO::Select; |
| |
| use open qw(:std :utf8); # assume utf-8 encoding |
| use Encode qw(encode_utf8 decode_utf8); |
| |
| use Pod::Usage; |
| use Getopt::Long; |
| use XML::CompactTree::XS; |
| use XML::LibXML::Reader; |
| use IO::Compress::Zip qw(zip $ZipError :constants); |
| use IPC::Open2; # open2 |
| |
| my $zip; my $first_write=1; |
| my $outh = *STDOUT; |
| |
| our $VERSION = '0.01'; |
| our $VERSION_MSG = "\ntei2korapxml - v$VERSION\n"; |
| |
| my $_CORR_POS_FOR_EMPTY_S_TAGS = 1; # this should only be deactivated for test purposes (empty s-tags produce an additional blank (look for ' ' in data.xml) |
| my $_CORR_BYTE_POS_FOR_P_TAGS = 1; |
| |
| my $_GEN_TOK_BAS = 0; |
| my $_GEN_TOK_DUMMY = 0; # not needed for drukola or cmc |
| |
| my $_COMPRESSION_METHOD = ZIP_CM_DEFLATE; |
| |
| my $_DEBUG = 0; |
| |
| my $_root_dir = "."; # output base directory |
| my $_header_file = "header.xml"; # filename to save text header in (in Latin1 format -> needed for ... ?) # TODO: why 'Latin1'? |
| my $_data_file = "data.xml"; # filename to save text data in (in UTF-8 format -> needed for ... ?) |
| my $_base_dir = "base"; # directory to store sentence and paragraph file and ... |
| my $_structure_dir = "struct"; |
| my $_structure_file = "structure.xml"; |
| my ($text_id, $text_id_esc); # '$text_id_esc' = escaped version of $text_id - see %ent (needed e.g. for 'Corola-PublishingHouse/Editura "Cronica"/260_a_608') |
| my $dir = ""; # text directory (on top of base) |
| my $dir_crp = ""; # corpus directory (on top of base) |
| my $dir_doc = ""; # document directory (on top of base) |
| my $data = ""; # $data is written to '$_data_file' |
| my %ent=('"', '"', '&','&','<','<','>','>'); # convert '&', '<' and '>' to their corresp. sgml-entities for compat. with KorAP |
| # NOTE: the index still refers to the above char-versions, which are counted as 1 |
| # (search for '&' in data.xml and see corresp. indices in structure.xml, lemma.xml, ana.xml ...) |
| my $header_crp = ""; |
| my $header_doc = ""; |
| my $header_txt = ""; # $header_* are written to '$_header_file' |
| |
| my ($header_fl_crp, $header_fl_doc, $header_fl_txt, $data_fl, $header_prfx, $data_prfx1, $data_prfx2, $data_sfx); |
| $header_prfx = $data_prfx1 = $data_prfx2 = $data_sfx = ""; |
| |
| my ($i, $c); |
| |
| ### needed for sub retr_info |
| my $_XCT_LN=0; # remember line numbers (set to 1 if debugging) |
| my ($reader, $tree_data, $rl, $_IDX, $n, $e, $dl, @oti); |
| # 'dl' stands for 'data length', 'oti' stands for 'open tags index' |
| ($_XCT_LN)?($_IDX=5):($_IDX=4); |
| ### |
| ### needed in func. write_structures |
| my ($ref, $idx, $att_idx); |
| |
| my $buf_in = ""; |
| my $output; |
| |
| ### needed for tokenization |
| my $_tok_file_bas = "tokens.xml"; |
| |
| #my @sentences; # sentence character offsets (start, end, start, end, ...) |
| #my @paragraphs; # paragraph character offsets (start, end, start, end, ...) |
| my @structures; # list of hashes, each hash representing an XML element |
| |
| ## whitespace-related issue (see retr_info()): needed to correct from-offset for some tags |
| my $add_one; |
| my %hws; # hash for indices of whitespace nodes (needed to recorrect from-value) |
| # idea: when closing element, check if it's from-index minus 1 refers to a whitespace node (means: 'from-index - 1' is a key in %hws). |
| # if this is _not_ the case, then the from-value is one to high => correct it by substracting 1 |
| my ($fval, $fval2); |
| $fval=$fval2=0; # init. |
| ## |
| |
| ## needed if $_GEN_TOK_DUMMY is set |
| my $_tok_file_con = "tokens_conservative.xml"; |
| my $_tok_file_agg = "tokens_aggr.xml"; |
| my (@tok_tokens_con, @tok_tokens_agg); |
| my ($m1, $m2, $m3, $m4, $tmp, $p1, $p2, $pr, $offset, $txt); |
| $offset=0; |
| ## |
| |
| # do this to prevent error message 'Global symbol "..." requires explicit package name at ...' |
| my ($_ANNOT_ATT_ANNOT_NAME_RD, $_ANNOT_ATT_LEM_NAME_RD, $_ANNOT_ATT_POS_NAME_WR, $_ANNOT_ATT_MSD_NAME_WR) = ('','','',''); |
| |
| ## needed for handling some inline annotations |
| my $_ANNOT_GEN = 0; # turn processing of inline annotations on/off |
| #my $_ANNOT_DIR_NAME = "connexor"; # name of path/foundry to write information to |
| my $_ANNOT_DIR_NAME = "drukola"; # name of path/foundry to write information to |
| #my $_ANNOT_DIR_NAME = "cmc"; # name of path/foundry to write information to # cmc = computer mediated communication - needed for dck corpus |
| my $_ANNOT_FILE_NAME = "morpho.xml"; # file to write annotations to |
| my $_ANNOT_TAG_NAME_RD = "w"; # from which tag to fetch the below attributes |
| $_ANNOT_ATT_ANNOT_NAME_RD = "ana"; # from which attribute to fetch the msd and pos information |
| #$_ANNOT_ATT_ANNOT_NAME_RD = "type"; # from which attribute to fetch the msd and pos information # must be 'type' for dck corpus |
| # new: comment out, if not needed |
| $_ANNOT_ATT_LEM_NAME_RD = "lemma"; # from which attribute to fetch the lemma information |
| # new: comment out, if not needed |
| $_ANNOT_ATT_POS_NAME_WR = "pos"; # see e.g.: GIT/KorAP-XML-Krill/t/annotation/corpus/doc/0001/mate/morpho.xml |
| # new: comment out, if not needed |
| $_ANNOT_ATT_MSD_NAME_WR = "msd"; # not needed for dck corpus |
| # new: comment out, if not needed |
| my $_ANNOT_ATT_LEM_NAME_WR = "lemma"; # s.a. |
| my (@annot, @oti2); |
| my $inside_annot_tag = -1; |
| ## |
| |
| GetOptions( |
| 'base|b=s' => \$_root_dir, |
| 'help|h' => sub { |
| pod2usage( |
| -verbose => 99, |
| -sections => 'NAME|DESCRIPTION|SYNOPSIS|ARGUMENTS|OPTIONS', |
| -msg => $VERSION_MSG, |
| -output => '-' |
| ) |
| }, |
| 'version|v' => sub { |
| pod2usage( |
| -verbose => 0, |
| -msg => $VERSION_MSG, |
| -output => '-' |
| ) |
| } |
| ); |
| |
| ###### |
| ###### MAIN |
| ###### |
| $_root_dir.='/'; # must always end with a slash at this point |
| $_root_dir=~s/^\.?\///; # remove leading / (only relative paths allowed in IO::Compress::Zip) and ./ |
| |
| my $lc=0; my $tc=0; my ($pfx, $sfx); |
| $header_fl_txt=0; |
| |
| # header_prfx |
| $header_prfx = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"; |
| $header_prfx .= "<?xml-model href=\"header.rng\" type=\"application/xml\" schematypens=\"http://relaxng.org/ns/structure/1.0\"?>\n"; |
| $header_prfx .= "<!DOCTYPE idsCorpus PUBLIC \"-//IDS//DTD IDS-XCES 1.0//EN\" \"http://corpora.ids-mannheim.de/idsxces1/DTD/ids.xcesdoc.dtd\">\n"; |
| |
| $data_prfx1 = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"; |
| $data_prfx1 .= "<?xml-model href=\"text.rng\" type=\"application/xml\" schematypens=\"http://relaxng.org/ns/structure/1.0\"?>\n\n"; |
| $data_prfx1 .= "<raw_text docid=\""; |
| $data_prfx2 .= "\" xmlns=\"http://ids-mannheim.de/ns/KorAP\">\n"; |
| $data_prfx2 .= " <metadata file=\"metadata.xml\" />\n"; |
| $data_prfx2 .= " <text>"; |
| $data_sfx = "</text>\n</raw_text>"; |
| |
| my($chld_out, $chld_in, $pid, $select); |
| if( $_GEN_TOK_BAS ){ |
| startTokenizer(); |
| } |
| |
| while(<STDIN>){ |
| |
| $lc++; # line counter |
| |
| if( $data_fl && m/^(.+)?<\/text>(.+)?$/ ){ # end of text body |
| ### |
| ### write data.xml, sentences.xml, paragraph.xml and structure.xml |
| ### |
| |
| $pfx=$1; $sfx=$2; |
| die "ERROR ($0): main(): input line number $lc: line with closing tag for 'text' contains additional information ... => Aborting\n\tline=$_" |
| if (defined $pfx && $pfx !~ /^\s*$/) || (defined $sfx && $sfx !~ /^\s*$/); |
| |
| if( $dir ne "" ){ |
| |
| ### NOTE: needed for debugging (e.g. whitespace issues) |
| #if( $text_id eq "GOE_AGA.00000" ) |
| #if( $text_id eq "GOE_AGI.04846" ) |
| #{ |
| #open H1, ">/tmp/t123"; |
| #print H1 "$buf_in"; |
| #close H1; |
| #} |
| #### |
| |
| $reader = XML::LibXML::Reader->new(string => "<text>$buf_in</text>", huge => 1); |
| |
| if( $_XCT_LN ){ # _XCT_LINE_NUMBERS is only for debugging |
| #$tree_data = XML::CompactTree::XS::readSubtreeToPerl($reader, XCT_DOCUMENT_ROOT|XCT_IGNORE_WS|XCT_IGNORE_SIGNIFICANT_WS|XCT_IGNORE_COMMENTS|XCT_ATTRIBUTE_ARRAY|XCT_LINE_NUMBERS); |
| # remove 'XCT_IGNORE_WS', because whitespace (in form of one blank) between tags is needed for separation of 'text-nodes' in data.xml |
| # TODO: check out the meanings/differences of 'XCT_IGNORE_WS' and 'XCT_IGNORE_SIGNIFICANT_WS': seems like, 'XCT_IGNORE_SIGNIFICANT_WS' is the one we need |
| $tree_data = XML::CompactTree::XS::readSubtreeToPerl($reader, XCT_DOCUMENT_ROOT|XCT_IGNORE_COMMENTS|XCT_ATTRIBUTE_ARRAY|XCT_LINE_NUMBERS); |
| # $tree_data = XML::CompactTree::XS::readSubtreeToPerl($reader, XCT_DOCUMENT_ROOT|XCT_IGNORE_COMMENTS|XCT_ATTRIBUTE_ARRAY|XCT_IGNORE_WS|XCT_IGNORE_SIGNIFICANT_WS|XCT_LINE_NUMBERS); |
| }else{ |
| $tree_data = XML::CompactTree::XS::readSubtreeToPerl($reader, XCT_DOCUMENT_ROOT|XCT_IGNORE_COMMENTS|XCT_ATTRIBUTE_ARRAY); |
| # $tree_data = XML::CompactTree::XS::readSubtreeToPerl($reader, XCT_DOCUMENT_ROOT|XCT_IGNORE_COMMENTS|XCT_ATTRIBUTE_ARRAY|XCT_IGNORE_WS|XCT_IGNORE_SIGNIFICANT_WS); |
| } |
| |
| @structures=(); @oti=(); |
| |
| if( $_ANNOT_GEN ){ |
| @annot=(); @oti2=() |
| } |
| |
| $dl = 0; |
| |
| $rl = 0; # recursion level |
| |
| # whitespace related issue |
| $add_one = 0; |
| %hws=(); |
| # |
| |
| ### |
| ### recursion |
| retr_info( \$tree_data->[2] ); # parse input data |
| ### |
| ### |
| $rl--; |
| |
| # write data.xml |
| $data=~tr/\n\r/ /; # NOTE: 2 blanks - otherwise offset data would become corrupt |
| $data=encode_utf8($data); |
| |
| # first write it to tokenization pipe to give it some time |
| if( $_GEN_TOK_BAS ){ |
| print $chld_in "$data\n\x03\n"; |
| } |
| |
| print STDERR "DEBUG ($0): main(): Writing (utf8-formatted) xml file $_root_dir$dir/$_data_file\n" if $_DEBUG; |
| if ($first_write){ |
| $first_write=0; |
| ### 1st time: create instance |
| $zip = new IO::Compress::Zip $outh, Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 0, Name => "$_root_dir$dir/$_data_file" |
| or die "ERROR ('$_root_dir$dir/$_data_file'): zip failed: $ZipError\n" |
| }else{ |
| ### close previous stream and open new one |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => "$_root_dir$dir/$_data_file" ) |
| or die "ERROR ('$_root_dir$dir/$_data_file'): zip failed: $ZipError\n" |
| } |
| $data =~ s/(&|<|>)/$ent{$1}/g; |
| $zip->print("$data_prfx1$text_id_esc$data_prfx2$data$data_sfx"); |
| |
| # write structures |
| write_structures() if @structures; |
| |
| # write annotations |
| write_annot() if $_ANNOT_GEN && @annot; |
| |
| # tokenization |
| if( $_GEN_TOK_BAS || $_GEN_TOK_DUMMY ) |
| { |
| write_tok(); |
| if( $_GEN_TOK_DUMMY ){ |
| $offset = 0; @tok_tokens_con=(); @tok_tokens_agg=(); |
| } |
| } |
| |
| $data_fl=0; $data = $dir = $buf_in = ""; # reinit. |
| |
| }else{ # $dir eq "" |
| |
| print STDERR "WARNING ($0): main(): maybe empty textSigle => skipping this text ...\n"; |
| print STDERR "WARNING ($0): main(): text header=$header_txt\n"; |
| print STDERR "WARNING ($0): main(): data=$data\n"; |
| } |
| }elsif( $data_fl ){ # inside text body |
| #print STDERR "inside text body (\$data_fl set)\n"; |
| |
| ### whitespace handling |
| # remove consecutive whitespace at beginning and end (mostly one newline) |
| # to let 'XML::CompactTree::XS' recognize these blanks as 'text-nodes', the option 'XCT_IGNORE_WS' may not be used (see above). |
| s/^\s+//; s/\s+$//; |
| |
| # important!: search for '>[^<][^<]*<' e.g. in 'vim goe_tagged.i5.xml.bz2' => insert a blank before every _first_ tag in a line, that contains text |
| # NOTE: the input text should avoid structures like below (which could also happen to other tags like '<pb ... />' or '<abbr>' etc.): |
| # negative example for '<w>' (see dck corpus in 'DeReKo-2016-II' - H.L. will fix this in 'DeReKo-2017-I'): |
| # <s> |
| # ... |
| # <w id="f1306078.m10.t9" lemma="zu" type="APPRART">zur</w> |
| # <w id="f1306078.m10.t10" lemma="Kenntnis" type="NN">Kenntnis</w> |
| # <w id="f1306078.m10.t11" lemma="nehmen" type="VVINF">nehmen</w> |
| # <w id="f1306078.m10.t12" lemma="," type="$,">,</w> |
| # <w id="f1306078.m10.t13" lemma="dass" type="KOUS">dass</w> |
| # ... |
| # In the above example, the whitespace/blank information between the words is lost. The below perl subst. expression does not recognize, that there may not be a blank before a comma. |
| if(m/<[^>]+>[^<]/){ |
| $tc++; # text counter |
| s/^(.)/ $1/ if $tc>1; # only add a blank, if this is not the first text in the whole text |
| } |
| |
| # add line to buffer |
| $buf_in .= $_; |
| |
| }elsif( $header_fl_txt && m/^(.*<\/idsHeader>)(.+)?$/ ){ # end of text header |
| #print STDERR "end of text header\n"; |
| ### |
| ### write header.xml |
| ### |
| $header_txt .= $1; $header_fl_txt = 0; $sfx=$2; |
| die "ERROR ($0): main(): input line number $lc: line with closing tag for 'idsHeader type=\"text\"' contains additional information ... => Aborting\n\tline=$_" |
| if defined $sfx && $sfx !~ /^\s*$/; |
| if( $dir eq "" ){ |
| print STDERR "WARNING ($0): main(): input line number $lc: empty textSigle in text header => nothing to do ...\ntext header=$header_txt\n"; |
| }else{ |
| print STDERR "DEBUG ($0): Writing file $_root_dir$dir/$_header_file\n" if $_DEBUG; |
| # create zip buffer |
| if ($first_write){ |
| $first_write=0; |
| $zip = new IO::Compress::Zip $outh, Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 0, Name => "$_root_dir$dir/$_header_file" |
| or die "ERROR ('$_root_dir$dir/$_header_file'): zip failed: $ZipError\n" |
| }else{ |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => "$_root_dir$dir/$_header_file" ) |
| or die "ERROR ('$_root_dir$dir/$_header_file'): zip failed: $ZipError\n" |
| } |
| $header_txt=encode_utf8($header_txt); |
| $zip->print("$header_prfx$header_txt"); |
| } |
| }elsif( $header_fl_txt ){ # inside text header |
| #print STDERR "inside text header\n"; |
| if( m/^(.+)?<textSigle>([^<]+)(.+)$/ ){ |
| $pfx=$1; $sfx=$3; $dir=$2; $text_id = $dir; $text_id =~ tr/\//_/; $dir =~ s/("|&|<|>)/$ent{$1}/g; |
| $text_id=encode_utf8($text_id); # otherwise problems with "wide character" where $text_id is printed (see e.g.: http://ahinea.com/en/tech/perl-unicode-struggle.html) |
| # see e.g.: Drukola-Project |
| die "ERROR ($0): main(): input line number $lc: line with textSigle-tag is not in expected format ... => Aborting\n\tline=$_" |
| if (defined $pfx && $pfx !~ /^\s*$/) || $sfx !~ /^<\/textSigle>\s*$/; |
| print STDERR "$0: main(): text_id=".decode_utf8($text_id)."\n"; # some output to see progression |
| $text_id_esc = $text_id; |
| if( $text_id_esc =~ s/("|&|<|>)/$ent{$1}/g ){ |
| $_ =~ s/(<textSigle>)[^<]+(<\/textSigle>)/$1$dir$2/; # do be consistent with escaping, escape also textSigle in Text-Header |
| } |
| $dir =~ tr/\./\//; |
| } |
| $header_txt .= $_; |
| }elsif( $header_fl_doc && m/^(.*<\/idsHeader>)(.+)?$/ ){ # end of document header |
| #print STDERR "end of doc header\n"; |
| $header_doc .= $1; $header_fl_doc = 0; $sfx=$2; |
| die "ERROR ($0): main(): input line number $lc: line with closing tag for 'idsHeader type=\"document\"' contains additional information ... => Aborting\n\tline=$_" |
| if defined $sfx && $sfx !~ /^\s*$/; |
| if( $dir_doc eq "" ){ |
| print STDERR "WARNING ($0): main(): input line number $lc: empty document sigle in document header => nothing to do ...\ndocument header=$header_doc\n"; |
| }else{ |
| print STDERR "DEBUG ($0): Writing file $_root_dir$dir_doc/$_header_file\n" if $_DEBUG; |
| if ($first_write){ |
| $first_write=0; |
| $zip = new IO::Compress::Zip $outh, Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 0, Name => "$_root_dir$dir_doc/$_header_file" |
| or die "ERROR ('$_root_dir$dir_doc/$_header_file'): zip failed: $ZipError\n" |
| }else{ |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => "$_root_dir$dir_doc/$_header_file" ) |
| or die "ERROR ('$_root_dir$dir_doc/$_header_file'): zip failed: $ZipError\n" |
| } |
| $header_doc=encode_utf8($header_doc); |
| $zip->print("$header_prfx$header_doc"); |
| } |
| }elsif( $header_fl_doc ){ # inside document header |
| #print STDERR "inside doc header\n"; |
| if( m/^(.+)?<dokumentSigle>([^<]+)(.+)$/ ){ |
| $pfx=$1; $sfx=$3; $dir_doc=$2; |
| die "ERROR ($0): main(): input line number $lc: line with dokumentSigle-tag is not in expected format ... => Aborting\n\tline=$_" |
| if (defined $pfx && $pfx !~ /^\s*$/) || $sfx !~ /^<\/dokumentSigle>\s*$/; |
| if( $dir_doc =~ s/("|&|<|>)/$ent{$1}/g ){ |
| $_ =~ s/(<dokumentSigle>)[^<]+(<\/dokumentSigle>)/$1$dir_doc$2/; # do be consistent with escaping, escape also textSigle in Document-Header |
| } |
| } |
| $header_doc .= $_; |
| }elsif ( m/^.*(<idsHeader[^>]*type=\"text\".*)$/ ){ # begin of text header |
| #print STDERR "begin of text header\n"; |
| $tc=0; ### reset (needed for whitespace-handling) |
| $header_txt = "$1\n"; $header_fl_txt = 1; |
| }elsif( m/^(.+)?<text(?: [^>]*)?>(.+)?$/ ){ # begin of text body |
| #print STDERR "inside text body\n"; |
| $pfx=$1; $sfx=$2; |
| die "ERROR ($0): main(): input line number $lc: line with opening tag for 'text' contains additional information ... => Aborting\n\tline=$_" |
| if (defined $pfx && $pfx !~ /^\s*$/) || (defined $sfx && $sfx !~ /^\s*$/); |
| $data_fl=1; |
| }elsif ( m/^.*(<idsHeader[^>]*type=\"document\".*)$/ ){ # begin of document header |
| #print STDERR "begin of doc header\n"; |
| $header_doc = "$1\n"; $header_fl_doc = 1; |
| }elsif( $header_fl_crp && m/^(.*<\/idsHeader>)(.+)?$/ ){ # end of corpus header |
| #print STDERR "end of corp header\n"; |
| $header_crp .= $1; $header_fl_crp = 0; $sfx=$2; |
| die "ERROR ($0): main(): input line number $lc: line with closing tag for 'idsHeader type=\"corpus\"' contains additional information ... => Aborting\n\tline=$_" |
| if defined $sfx && $sfx !~ /^\s*$/; |
| if( $dir_crp eq "" ){ |
| print STDERR "WARNING ($0): main(): input line number $lc: empty corpus sigle in corpus header => nothing to do ...\ncorpus header=$header_crp\n"; |
| }else{ |
| print STDERR "DEBUG ($0): Writing file $_root_dir$dir_crp/$_header_file\n" if $_DEBUG; |
| if ($first_write){ |
| $first_write=0; |
| $zip = new IO::Compress::Zip $outh, Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 0, Name => "$_root_dir$dir_crp/$_header_file" |
| or die "ERROR ('$_root_dir$dir_crp/$_header_file'): zip failed: $ZipError\n" |
| }else{ |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => "$_root_dir$dir_crp/$_header_file" ) |
| or die "ERROR ('$_root_dir$dir_crp/$_header_file'): zip failed: $ZipError\n" |
| } |
| $header_crp=encode_utf8($header_crp); |
| $zip->print("$header_prfx$header_crp"); |
| } |
| }elsif( $header_fl_crp ){ # inside corpus header |
| #print STDERR "inside corp header\n"; |
| if( m/^(.+)?<korpusSigle>([^<]+)(.+)$/ ){ |
| $pfx=$1; $sfx=$3; $dir_crp=$2; |
| die "ERROR ($0): main(): input line number $lc: line with korpusSigle-tag is not in expected format ... => Aborting\n\tline=$_" |
| if (defined $pfx && $pfx !~ /^\s*$/) || $sfx !~ /^<\/korpusSigle>\s*$/; |
| if( $dir_crp =~ s/("|&|<|>)/$ent{$1}/g ){ |
| $_ =~ s/(<korpusSigle>)[^<]+(<\/korpusSigle>)/$1$dir_crp$2/; # do be consistent with escaping, escape also textSigle in Corpus-Header |
| } |
| } |
| $header_crp .= $_; |
| }elsif ( m/^.*(<idsHeader[^>]*type=\"corpus\".*)$/ ){ # begin of corpus header |
| #print STDERR "begin of corp header\n"; |
| $header_crp = "$1\n"; $header_fl_crp = 1; |
| } |
| ### |
| |
| }#end: while |
| |
| if( $_GEN_TOK_BAS ){ |
| close($chld_in); |
| close($chld_out); |
| } |
| $zip->close(); |
| |
| ###### |
| ###### SUBS |
| ###### |
| # |
| # $reader = XML::LibXML::Reader->new(IO => STDIN); |
| # $data = XML::CompactTree::XS::readSubtreeToPerl($reader, XCT_DOCUMENT_ROOT|XCT_IGNORE_WS|XCT_IGNORE_SIGNIFICANT_WS|XCT_IGNORE_COMMENTS); |
| # |
| # example: <node a=\"v\"><node1>some <n/> text</node1><node2>more-text</node2></node> |
| # |
| # ( print out values of above example: echo "<node a=\"v\"><node1>some <n/> text</node1><node2>more-text</node2></node>"|perl -e 'use XML::CompactTree::XS; use XML::LibXML::Reader; $reader = XML::LibXML::Reader->new(IO => STDIN);$data = XML::CompactTree::XS::readSubtreeToPerl($reader,XCT_DOCUMENT_ROOT|XCT_IGNORE_WS|XCT_IGNORE_SIGNIFICANT_WS|XCT_IGNORE_COMMENTS|XCT_LINE_NUMBERS);print $data->[2]->[0]->[5]->[1]->[1]' ) |
| # |
| # $data = reference to below array |
| # |
| # [ 0: XML_READER_TYPE_DOCUMENT, |
| # 1: ? |
| # 2: [ 0: [ 0: XML_READER_TYPE_ELEMENT |
| # 1: 'node' |
| # 2: ? |
| # 3: HASH (attributes) |
| # 4: 1 (line number) |
| # 5: [ 0: [ 0: XML_READER_TYPE_ELEMENT |
| # 1: 'node1' |
| # 2: ? |
| # 3: undefined (no attributes) |
| # 4: 1 (line number) |
| # 5: [ 0: [ 0: XML_READER_TYPE_TEXT |
| # 1: 'some ' |
| # ] |
| # 1: [ 0: XML_READER_TYPE_ELEMENT |
| # 1: 'n' |
| # 2: ? |
| # 3: undefined (no attributes) |
| # 4: 1 (line number) |
| # 5: undefined (no child nodes) |
| # ] |
| # 2: [ 0: XML_READER_TYPE_TEXT |
| # 1: ' text' |
| # ] |
| # ] |
| # ] |
| # 1: [ 0: XML_READER_TYPE_ELEMENT |
| # 1: 'node2' |
| # 2: ? |
| # 3: undefined (not attributes) |
| # 4: 1 (line number) |
| # 5: [ 0: [ 0: XML_READER_TYPE_TEXT |
| # 1: 'more-text' |
| # ] |
| # ] |
| # ] |
| # ] |
| # ] |
| # ] |
| # ] |
| # |
| # $data->[0] = 9 (=> type == XML_READER_TYPE_DOCUMENT) |
| # |
| # ref($data->[2]) == ARRAY (with 1 element for 'node') |
| # ref($data->[2]->[0]) == ARRAY (with 6 elements) |
| # |
| # $data->[2]->[0]->[0] == 1 (=> type == XML_READER_TYPE_ELEMENT) |
| # $data->[2]->[0]->[1] == 'node' |
| # ref($data->[2]->[0]->[3]) == HASH (=> ${$data->[2]->[0]->[3]}{a} == 'v') |
| # $data->[2]->[0]->[4] == 1 (line number) |
| # ref($data->[2]->[0]->[5]) == ARRAY (with 2 elements for 'node1' and 'node2') NOTE: child nodes of actual node (see $_IDX) |
| # |
| # ref($data->[2]->[0]->[5]->[0]) == ARRAY (with 6 elements) |
| # $data->[2]->[0]->[5]->[0]->[0] == 1 (=> type == XML_READER_TYPE_ELEMENT) |
| # $data->[2]->[0]->[5]->[0]->[1] == 'node1' |
| # $data->[2]->[0]->[5]->[0]->[3] == undefined (=> no attribute) |
| # $data->[2]->[0]->[5]->[0]->[4] == 1 (line number) |
| # ref($data->[2]->[0]->[5]->[0]->[5]) == ARRAY (with 3 elements for 'some ', '<n/>' and ' text') |
| # |
| # ref($data->[2]->[0]->[5]->[0]->[5]->[0]) == ARRAY (with 2 elements) |
| # $data->[2]->[0]->[5]->[0]->[5]->[0]->[0] == 3 (=> type == XML_READER_TYPE_TEXT) |
| # $data->[2]->[0]->[5]->[0]->[5]->[0]->[1] == 'some ' |
| # |
| # ref($data->[2]->[0]->[5]->[0]->[5]->[1]) == ARRAY (with 5 elements) |
| # $data->[2]->[0]->[5]->[0]->[5]->[1]->[0] == 1 (=> type == XML_READER_TYPE_ELEMENT) |
| # $data->[2]->[0]->[5]->[0]->[5]->[1]->[1] == 'n' |
| # $data->[2]->[0]->[5]->[0]->[5]->[1]->[3] == undefined (=> no attribute) |
| # $data->[2]->[0]->[5]->[0]->[5]->[1]->[4] == 1 (line number) |
| # $data->[2]->[0]->[5]->[0]->[5]->[1]->[5] == undefined (=> no child nodes) |
| # |
| # ref($data->[2]->[0]->[5]->[0]->[5]->[2]) == ARRAY (with 2 elements) |
| # $data->[2]->[0]->[5]->[0]->[5]->[2]->[0] == 3 (=> type == XML_READER_TYPE_TEXT) |
| # $data->[2]->[0]->[5]->[0]->[5]->[2]->[1] == ' text' |
| # |
| # |
| # NOTE: the sub 'retr_info' starts with the array reference '${$_[0]}' (= '\$tree_data->[2]'), which corresponds to '${\$data->[2]}' in the above example. |
| # Hence, the expression '@{${$_[0]}}' corresponds to '@{${\$data->[2]}', '$e' to '${${\$data->[2]}}[0]' (='$data->[2]->[0]') and '$e->[0]' to '${${\$data->[2]}}[0]->[0]' (='$data->[2]->[0]->[0]'. |
| |
| sub retr_info{ |
| |
| $rl++; # rec. level (2 = topmost level inside retr_info() = level for all elements directly under <text>) |
| |
| # $_[0] is a reference to an array reference |
| foreach $e( @{${$_[0]}} ){ # iteration through all array elements |
| |
| ## http://search.cpan.org/~shlomif/XML-LibXML-2.0118/lib/XML/LibXML/Reader.pod#NODE_TYPES |
| if( $e->[0] == XML_READER_TYPE_ELEMENT ){ # element node |
| |
| ### . insert new array (for new tag) into @structures with tag-name and (if present) tag-attributes |
| ### . update @oti (open tags indizes) with @structures highest index (=$#structures); e.g.: @a=(1,2,3) => $#a = 2 |
| ### |
| $n = $e->[1]; # name of tag |
| |
| $inside_annot_tag = $rl if $_ANNOT_GEN && $n eq "$_ANNOT_TAG_NAME_RD"; |
| |
| my @array; |
| push @array, $n; # name of tag |
| push @structures, \@array; |
| |
| push @oti, $#structures; # add highest index of @structures to @oti |
| |
| if( $_ANNOT_GEN && $inside_annot_tag == $rl ){ |
| my @array2; |
| push @array2, $n; |
| push @annot, \@array2; |
| push @oti2, $#annot; |
| } |
| |
| ### handle attributes |
| if( defined $e->[3] ){ # only go on if attributes exist |
| for( $c = 0; $c < @{$e->[3]}; $c += 2 ){ # with 'XCT_ATTRIBUTE_ARRAY', $node->[3] is an array reference of the form [ name1, value1, name2, value2, ....] |
| # of attribute names and corresponding values. NOTE: arrays are faster! (see: http://makepp.sourceforge.net/2.0/perl_performance.html) |
| |
| # '$c' references the 'key' and '$c+1' the 'value' (example: perl -e '@a=(X,Y);push @b,\@a;$ref=$b[$#b];push @{$ref},Z;print @{$ref}' => XYZ) |
| push @{$structures[$#structures]}, ${$e->[3]}[$c], ${$e->[3]}[$c+1]; |
| |
| if( $_ANNOT_GEN && $inside_annot_tag == $rl ){ |
| push @{$annot[$#annot]}, ${$e->[3]}[$c], ${$e->[3]}[$c+1]; |
| } |
| } |
| } |
| ### |
| |
| push @{$structures[$#structures]}, ($dl+$add_one); # from-value |
| |
| if( $_ANNOT_GEN && $inside_annot_tag == $rl ){ |
| push @{$annot[$#annot]}, ($dl+$add_one); |
| } |
| |
| ###### |
| # until here: opening tag-node |
| ###### |
| |
| ### RECURSION |
| if(defined $e->[$_IDX]){ # if $e->[$_IDX] is not defined (=> no array of child nodes, e.g.: <pb id="bka.00001-643-pb643" n="643" TEIform="pb"/>), do no recursion |
| |
| retr_info( \$e->[$_IDX] ); # recursion with array of child nodes |
| |
| $rl--; # return from recursion |
| } |
| ### |
| |
| ###### |
| # from here: closing tag-node |
| ###### |
| #print STDERR "2: dl=$dl\n" if $text_id eq "BIH_BKA.00001"; |
| # |
| { my $ix = pop @oti; |
| my $aix = @{$structures[$ix]}; |
| $fval = ${$structures[$ix]}[$aix-1]; # from-value |
| |
| # whitespace related issue |
| if( $fval>0 && not exists $hws{$fval-1} ){ |
| # previous node was a text-node |
| ${$structures[$ix]}[$aix-1] = $fval-1; # recorrect from-value |
| } |
| # |
| |
| die "ERROR ($0, retr_info()): text_id='$text_id', processing of \@structures: from-value ($fval) is more than 1 greater than to-value ($dl) => please check. aborting ...\n" if ($fval-1) > $dl; |
| ${$structures[$ix]}[$aix-1] = $dl if $fval == $dl+1; # correct from-value |
| |
| push @{$structures[$ix]}, $dl, $rl; # to-value and recursion-level |
| ### use $dl, because the offsets are _between_ the characters (e.g.: word = 'Hallo' => from = 0 (before 'H'), to = 5 (after 'o')) |
| } |
| |
| if ( $_ANNOT_GEN && $inside_annot_tag == $rl ){ |
| my $ix = pop @oti2; |
| my $aix = @{$annot[$ix]}; |
| $fval2 = ${$annot[$ix]}[$aix-1]; # from-value |
| |
| # whitespace related issue |
| if( $fval2>0 && not exists $hws{$fval2-1} ){ |
| # previous node was a text-node |
| ${$annot[$ix]}[$aix-1] = $fval2-1; # recorrect from-value |
| } |
| # |
| |
| die "ERROR ($0, retr_info()): text_id='$text_id', processing of \@annot: from-value ($fval2) is more than 1 greater than to-value ($dl) => please check. aborting ...\n" if ($fval2-1) > $dl; |
| ${$annot[$ix]}[$aix-1] = $dl if $fval2 == $dl+1; # correct from-value |
| |
| push @{$annot[$ix]}, $dl, $rl; # to-value and recursion-level |
| $inside_annot_tag = -1; # reset |
| } |
| |
| # whitespace related issue |
| # clean up |
| delete $hws{$fval-1} if $fval>0 && exists $hws{$fval-1}; |
| delete $hws{$fval2-1} if $_ANNOT_GEN && $fval2>0 && exists $hws{$fval2-1}; |
| # |
| |
| ###### |
| # from here: text nodes |
| ###### |
| }elsif( $e->[0] == XML_READER_TYPE_TEXT || $e->[0] == XML_READER_TYPE_SIGNIFICANT_WHITESPACE ){ # see 'man XML::LibXML::Reader' |
| |
| # set flag, to remember to add 1 to 'from'-$dl (see above), if this text-node was _not_ a whitespace-node |
| # NOTE: this is needed to address e.g. the following situation (see first text in goe-corpus): |
| # '... <head type="main"><s>Campagne in Frankreich</s></head><head type="sub"> <s>1792</s> ...': the from-index of the 2nd head-tag |
| # should be right before '1792' (which was produced by a text-node) and not before the ' ' (which was produced by a whitespace-node) |
| # by the way: '$add_one' is not set (resp. 're'-set) for whitespace-nodes, because they always (incidentally) set '$dl' to the right |
| # from-index of the next node (e.g.: '<w>fu</w> <w>bar</w>' or '<w>fu</w><w> </w><w>bar</w>) - this is actually not the case for |
| # text-nodes (see same previous 'fubar'-example), for which '$dl' points to the from-index of ' ', _IF_ there is alway a ' ' between |
| # two text-nodes. |
| # NOTE: A new problem arises with this approach, _IF_ there is _NOT_ a ' ' between two text-nodes |
| # (see e.g. in romanian corpus 'corola-bucurenci': ... >Anului</w><w lemma="," ana="PUNCTUATION">,</w> <w ...). |
| # In this case, the from-index for the next node would be exactly 1 to high (see in approp. zip-file for above example: |
| # '... <span id="s14" from="101" to="101" l="5"> ...' - the correct span should be: '... <span id="s14" from="100" to="101" l="5"> ...'). |
| # NOTE: it's not possible to solve the above problem here, because the following case would not be covered: |
| # ... <a><b><c><d/></c></b></a> ... (or see in goe-corpus 'GOE/AGI.04846': <div...><p/></div> ...) |
| # Because, no text- and whitespace-nodes are present, the from-values for 'a', 'b' and 'c' can only be corrected, when closing these tags. |
| # But this can be easily done by looking at the actual value of '$add_one' - if it's set, then there was no whitespace-node (and hence no ' ') |
| # before '<a>' and hence the from-values must be corrected. |
| # NOTE: To realize, what's stated in the above note, another information is needed, by which can be checked, if a tag contains text/ws or not |
| # (in the above 'abcd'-example, all tags are empty in the sense of that they don't contain any text/ws). |
| # By saving the recursion level, the 'text containing'-info can be simply checked, by the condition '$last_text_or_ws_node > $rl'. |
| # The condition is true, when there is text below the tag on level $rl. |
| ##$last_text_or_ws_node = $rl; |
| # NOTE: a much simpler solution, which should have the same effect, without following exactly the information-processing, is, to just set the |
| # value of the 'from'-index to the value of the 'to'-index, if "to==from-1" holds and to die with an error, if "(from-1)>to" holds. |
| |
| # whitespace related issue |
| if( $e->[0] == XML_READER_TYPE_SIGNIFICANT_WHITESPACE ){ |
| $add_one=0; |
| $hws{$dl}++; # '++' does not mean a thing here (could be used for consistency checking) |
| }else{ # text-node |
| $add_one=1; |
| } |
| # |
| |
| $data .= $e->[1]; |
| $dl += length($e->[1]); |
| |
| ### needed only for dummy tokenization |
| if( $_GEN_TOK_DUMMY ){ |
| $txt = $e->[1]; |
| if( (substr($txt,0,1) ne ' ' || substr($txt,1,1) ne ' ') ){ # $txt has at least 2 chars, if it is not empty or equal to ' ' |
| |
| ### start: conservative tokenization |
| # '\p{Punct}' is equal to the character class '[-!"#%&'()*,./:;?@[\\\]_{}]' |
| while( $txt =~ /([\p{Punct}]*)([^\p{Punct} \x{9}\n]+(?:([\p{Punct}]+)[^\p{Punct} \x{9}\n]+)*)?([\p{Punct}]*)(?:[ \x{9}\n])?/g ){ |
| |
| $m1=$1; $m2=$2; $m3=$3; $m4=$4; |
| |
| #print STDERR "*** m1=$1, m2="; (defined $m2)?(print STDERR "$2"):(print STDERR "undef");print STDERR ", m3="; (defined $3)?(print STDERR "$3"):(print STDERR "undef"); print STDERR ", m4=$4\n"; |
| if("$m1" ne ""){ # special chars before token |
| $p1=$-[1]; $p2=$+[1]; |
| #print STDERR "A1: ".$m1." -> from $p1 to $p2\n"; |
| if($p2==$p1+1){ |
| if($p1!=0){$tmp=substr($txt,$p1-1,1);$pr=($tmp=~/^[^A-Za-z0-9]/)}else{$pr=0}; |
| if(not $pr){$tmp=substr($txt,$p2,1);$pr=($tmp=~/^[^A-Za-z0-9]/)}; |
| if($pr){push @tok_tokens_con, $p1+$offset; push @tok_tokens_con, $p2+$offset}; # from and to |
| }else{ |
| for($i=0;$i<($p2-$p1);$i++){ |
| #print STDERR "A2: ".substr($m1,$i,1)." -> from $p1 to $p2\n"; |
| push @tok_tokens_con, $p1+$i+$offset; push @tok_tokens_con, $p1+$i+1+$offset; # from and to |
| } |
| } |
| }# "$m1" ne "" |
| #print STDERR "B: "."$m2 -> from ".($-[2]+$offset)." to ".($+[2]+$offset)."\n" if defined $m2; # token (wordform) |
| |
| if(defined $m2){push @tok_tokens_con, $-[2]+$offset; push @tok_tokens_con, $+[2]+$offset}; # from and to |
| |
| if(defined $m3){ |
| $p1=$-[3]; $p2=$+[3]; |
| #print STDERR "C: ".$m3." -> from $p1 to $p2\n"; |
| if($p2==$p1+1){ |
| $tmp=substr($txt,$p2,1);$pr=($tmp=~/^$/);$pr=($tmp=~/^[^A-Za-z0-9]/) if not $pr; # char after match |
| if(not $pr){$tmp=substr($txt,$p1-1,1);$pr=($tmp=~/^[^A-Za-z0-9]/)}; # char before match |
| if($pr){push @tok_tokens_con, $p1+$offset; push @tok_tokens_con, $p2+$offset}; # from and to |
| }else{ # length($m3)>1 => print all chars |
| for($i=0; $i<($p2-$p1); $i++){ |
| #$tmp=substr($m3,$i,1); |
| #print STDERR "C2: $tmp -> from $p1 to $p2\n"; |
| push @tok_tokens_con, $p1+$i+$offset; push @tok_tokens_con, $p1+$i+1+$offset; # from and to |
| } |
| } |
| }# defined $m3 |
| |
| if("$m4" ne ""){ # special chars after token |
| $p1=$-[4]; $p2=$+[4]; |
| #print STDERR "D1: ".$m4." -> from ".($p1+$offset)." to ".($p2+$offset)."\n"; |
| if($p2==$p1+1){ |
| $tmp=substr($txt,$p2,1);$pr=($tmp=~/^$/);$pr=($tmp=~/^[^A-Za-z0-9]/) if not $pr; # char after match |
| if(not $pr){$tmp=substr($txt,$p1-1,1);$pr=($tmp=~/^[^A-Za-z0-9]/)}; # char before match |
| if($pr){push @tok_tokens_con, $p1+$offset; push @tok_tokens_con, $p2+$offset} # from and to |
| }else{ |
| for($i=0;$i<($p2-$p1);$i++){ |
| #print STDERR "D2: ".substr($m4,$i,1)." -> from ".($p1+$i+$offset)." to ".($p1+$i+1+$offset)."\n"; |
| push @tok_tokens_con, $p1+$i+$offset; push @tok_tokens_con, $p1+$i+1+$offset; # from and to |
| } |
| } |
| }# "$m4" ne "" |
| |
| }# end: while |
| ### end: conservative tokenization |
| |
| ### start: aggressive tokenization |
| while( $txt =~ /([^\p{Punct} \x{9}\n]+)(?:([\p{Punct}])|(?:[ \x{9}\n])?)|([\p{Punct}])/g ){ |
| if(defined $1){ |
| push @tok_tokens_agg, $-[1]+$offset; push @tok_tokens_agg, $+[1]+$offset; # from and to |
| if(defined $2){push @tok_tokens_agg, $-[2]+$offset; push @tok_tokens_agg, $+[2]+$offset} # from and to |
| }else{ # defined $3 |
| push @tok_tokens_agg, $-[3]+$offset; push @tok_tokens_agg, $+[3]+$offset # from and to |
| } |
| }# end: while |
| ### end: aggressive tokenization |
| |
| #$offset = $dl+1; |
| $offset = $dl; |
| } |
| } |
| ### |
| # }#fi |
| |
| #print STDERR "1: dl=$dl\n" if $text_id eq "BIH_BKA.00001"; |
| |
| |
| # elsif( $e->[0] == XML_READER_TYPE_ATTRIBUTE ) # attribute node |
| # NOTE: attributes cannot be processed like this |
| |
| ### from here: should not really happen |
| |
| }else{ # not yet handled type |
| |
| die "ERROR ($0): Not yet handled type (\$e->[0]=".$e->[0].") ... => Aborting\n"; |
| |
| } |
| |
| }#end: foreach iteration |
| |
| }#end: sub retr_info |
| |
| |
| sub write_tok { # called from MAIN |
| |
| #print STDERR "$0: write_tok() ...\n"; |
| |
| if( $_GEN_TOK_BAS ) { |
| if( $select->can_read(3600) ){ # wait 60m for external tokenizer |
| $_ = <$chld_out>; |
| my @bounds = split; |
| write_tokenization("$_root_dir$dir/$_base_dir/$_tok_file_bas", $text_id_esc, \@bounds); |
| while($select->can_read(0)) { |
| $_ = <$chld_out>; |
| if (defined $_ && $_ ne '') { |
| print STDERR "WARNING: extra output: $_\n" |
| } else { |
| print STDERR "WARNING: tokenizer seems to have crashed, restarting.\n"; |
| startTokenizer(); |
| } |
| } |
| }else{ |
| $zip->close(); |
| die "ERROR ($0): cannot retrieve token bounds from external tokenizer for text '$text_id' => Aborting ...\n"; |
| } |
| }elsif( $_GEN_TOK_DUMMY ){ |
| write_tokenization("$_root_dir$dir/$_base_dir/$_tok_file_con", $text_id_esc, \@tok_tokens_con); |
| write_tokenization("$_root_dir$dir/$_base_dir/$_tok_file_agg", $text_id_esc, \@tok_tokens_agg); |
| } |
| #print STDERR "$0: write_tok(): DONE\n"; |
| } # end: write_tok |
| |
| sub write_tokenization { |
| my ($fname, $textid_esc, $bounds) = @_; |
| |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => $fname) |
| or die "ERROR ('$fname'): zip failed: $ZipError\n"; |
| $output = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<?xml-model href=\"span.rng\" type=\"application/xml\" schematypens=\"http://relaxng.org/ns/structure/1.0\"?>\n\n<layer docid=\"$text_id_esc\" xmlns=\"http://ids-mannheim.de/ns/KorAP\" version=\"KorAP-0.4\">\n <spanList>\n"; |
| $c=0; for( $i=0; $i < ($#$bounds + 1); $i+=2 ){ |
| $output .= " <span id=\"t_$c\" from=\"".$bounds->[$i]."\" to=\"".$bounds->[$i+1]."\" />\n"; |
| $c++; |
| } |
| $output .= " </spanList>\n</layer>"; |
| $zip->print("$output"); |
| } |
| |
| # write structure file |
| sub write_structures { # called from MAIN |
| |
| #print STDERR "$0: write_structures(): ...\n"; |
| |
| if( $dir eq "" ){ |
| print STDERR "WARNING ($0): write_structures(): empty textSigle => nothing to do ...\n"; |
| return; |
| } |
| |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => "$_root_dir$dir/$_structure_dir/$_structure_file" ) |
| or die "ERROR ('$_root_dir$dir/$_structure_dir/$_structure_file'): zip failed: $ZipError\n"; |
| |
| $output = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<?xml-model href=\"span.rng\" type=\"application/xml\" schematypens=\"http://relaxng.org/ns/structure/1.0\"?>\n\n<layer docid=\"".decode_utf8($text_id_esc)."\" xmlns=\"http://ids-mannheim.de/ns/KorAP\" version=\"KorAP-0.4\">\n <spanList>\n"; |
| |
| $c = 0; |
| |
| foreach $ref (@structures){ |
| |
| (@{$ref}==4)?($idx=1):($idx=@{$ref}-3); # if array '@{$ref}' doesn't contain attributes, then the number of elements in this array is 4 (name, from, to, rec_level), otherwise >4 |
| |
| ### correct last from-value (if the 'second to last' from-value refers to an s-tag, then the last from-value is one to big - see retr_info) |
| if( $#structures == $c && ${$ref}[$idx] == ${$ref}[$idx+1] + 1 ){ |
| ${$ref}[$idx] = ${$ref}[$idx+1]; |
| } |
| ## this consistency check is already done in 'retr_info()' |
| ##elsif( ${$ref}[$idx] > ${$ref}[$idx+1] ){ # consistency check: abort, if this doesn't hold |
| ## #die "ERROR ($0: write_structures(): \$text_id=$text_id, \$c=$c, tag-name=${$ref}[0]): 'from-index=${$ref}[$idx]' > 'to-index=${$ref}[$idx+1]' => please check! aborting ...\n"; |
| ## die "ERROR ($0: write_structures(): \$text_id=$text_id, \$c=$c, tag-name=${$ref}[0]): 'from-index=${$ref}[$idx]' > 'to-index=${$ref}[$idx+1]' => please check! aborting ...\n\n$output"; |
| ##} |
| ### |
| |
| ### at least 'POS' should always be there => remove constraint '$_ANNOT_GEN' |
| #if( $_ANNOT_GEN && ${$ref}[0] ne "$_ANNOT_TAG_NAME_RD" ){ # this is already written in 'write_annot', so don't do it twice |
| if( ${$ref}[0] ne "$_ANNOT_TAG_NAME_RD" ){ # this is already written in 'write_annot', so don't do it twice |
| |
| $output .= " <span id=\"s$c\" from=\"${$ref}[$idx]\" to=\"${$ref}[$idx+1]\" l=\"${$ref}[$idx+2]\">\n"; # l (level): insert information about depth of element in XML-tree (top element = level 1) |
| $output .= " <fs type=\"struct\" xmlns=\"http://www.tei-c.org/ns/1.0\">\n"; |
| $output .= " <f name=\"name\">${$ref}[0]</f>\n"; |
| |
| if ( $idx > 2 ) # attributes |
| { |
| $output .= " <f name=\"attr\">\n <fs type=\"attr\">\n"; |
| |
| for ( $att_idx=1; $att_idx < $idx; $att_idx+=2 ){ |
| ${$ref}[$att_idx+1] =~ s/(&|<|>)/$ent{$1}/g; # see explanation in func. 'write_annot' |
| $output .= " <f name=\"${$ref}[$att_idx]\">${$ref}[$att_idx+1]</f>\n"; # attribute (at index $att_idx) with value (at index $att_idx+1) |
| } |
| |
| $output .= " </fs>\n </f>\n"; |
| } |
| |
| $output .= " </fs>\n </span>\n"; |
| |
| } |
| |
| $c++; |
| |
| }# end: foreach |
| |
| $output .= " </spanList>\n</layer>"; |
| |
| $output=encode_utf8($output); |
| |
| $zip->print("$output"); |
| |
| #print STDERR "$0: write_structures(): DONE\n"; |
| } # end: sub write_structures |
| |
| |
| sub write_annot { # called from MAIN |
| #print STDERR "$0: write_annot(): ...\n"; |
| |
| if( $dir eq "" ){ |
| print STDERR "WARNING ($0): write_annot(): empty textSigle => nothing to do ...\n"; |
| return; |
| } |
| |
| $zip->newStream( Zip64 => 1, TextFlag => 1, Method => $_COMPRESSION_METHOD, Append => 1, Name => "$_root_dir$dir/$_ANNOT_DIR_NAME/$_ANNOT_FILE_NAME" ) |
| or die "ERROR ('$_root_dir$dir/$_ANNOT_DIR_NAME/$_ANNOT_FILE_NAME'): zip failed: $ZipError\n"; |
| |
| $output = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<?xml-model href=\"span.rng\" type=\"application/xml\" schematypens=\"http://relaxng.org/ns/structure/1.0\"?>\n\n<layer docid=\"".decode_utf8($text_id_esc)."\" xmlns=\"http://ids-mannheim.de/ns/KorAP\" version=\"KorAP-0.4\">\n <spanList>\n"; |
| |
| $c = 0; |
| |
| #open H1, ">/tmp/t456" if $text_id eq "GOE_AGA.00000"; |
| foreach $ref (@annot){ |
| |
| #if($text_id eq "GOE_AGA.00000"){ |
| #foreach my $key(@{$ref}){print H1 " $c:$key"}; |
| #print H1 "\n"; |
| #} |
| (@{$ref}==4)?($idx=1):($idx=@{$ref}-3); # if array '@{$ref}' doesn't contain attributes, then the number of elements in this array is 4 (name, from, to, rec_level), otherwise >4 |
| |
| ### correct last from-value (if the 'second to last' from-value refers to a s-tag, then the last from-value is one to big - see retr_info) |
| if( $#annot == $c && ${$ref}[$idx] == ${$ref}[$idx+1] + 1 ){ |
| ${$ref}[$idx] = ${$ref}[$idx+1]; |
| } |
| ### |
| |
| $output .= " <span id=\"s$c\" from=\"${$ref}[$idx]\" to=\"${$ref}[$idx+1]\" l=\"${$ref}[$idx+2]\">\n"; # l (level): insert information about depth of element in XML-tree (top element = level 1) |
| $output .= " <fs type=\"lex\" xmlns=\"http://www.tei-c.org/ns/1.0\">\n"; |
| ##$output .= " <f name=\"${$ref}[0]\">\n"; |
| $output .= " <f name=\"lex\">\n"; |
| |
| if ( $idx > 2 ) # attributes |
| { |
| $output .= " <fs>\n"; |
| |
| for ( $att_idx=1; $att_idx < $idx; $att_idx+=2 ){ |
| |
| ${$ref}[$att_idx+1] =~ s/(&|<|>)/$ent{$1}/g; # e.g.: look for 4th occurence of '…' in '84030_a_82705' in Drukola-Korpus 'corola-bucurenci': |
| # ... <w lemma=">" ana="PUNCTUATION">></w> ... |
| # - the '>' is translated to '>' and hence the result would be '<f name="lemma">></f>' |
| |
| if( "$_ANNOT_ATT_ANNOT_NAME_RD" && ${$ref}[$att_idx] eq "$_ANNOT_ATT_ANNOT_NAME_RD" ){ |
| |
| ${$ref}[$att_idx+1] =~ /^([^ ]+)(?: (.+))?$/; |
| die "ERROR (write_annot()): unexpected format! => Aborting ... (att: ${$ref}[$att_idx+1])\n" if ($_ANNOT_ATT_POS_NAME_WR && not defined $1) || ($_ANNOT_ATT_MSD_NAME_WR && not defined $2); |
| if( "$_ANNOT_ATT_POS_NAME_WR" ){ |
| $output .= " <f name=\"$_ANNOT_ATT_POS_NAME_WR\">"; |
| $output .= "$1" if defined $1; |
| $output .= "</f>\n"; |
| } |
| if( "$_ANNOT_ATT_MSD_NAME_WR" ){ |
| $output .= " <f name=\"$_ANNOT_ATT_MSD_NAME_WR\">"; |
| $output .= "$2" if defined $2; |
| $output .= "</f>\n"; |
| } |
| |
| }elsif( "$_ANNOT_ATT_LEM_NAME_RD" && ${$ref}[$att_idx] eq "$_ANNOT_ATT_LEM_NAME_RD" ){ |
| |
| $output .= " <f name=\"$_ANNOT_ATT_LEM_NAME_WR\">${$ref}[$att_idx+1]</f>\n"; |
| |
| }else{ # all other attributes |
| |
| $output .= " <f name=\"${$ref}[$att_idx]\">${$ref}[$att_idx+1]</f>\n"; # attribute (at index $att_idx) with value (at index $att_idx+1) |
| } |
| |
| } # end:for |
| |
| $output .= " </fs>\n"; |
| } # fi |
| |
| $output .= " </f>\n </fs>\n </span>\n"; |
| |
| $c++; |
| |
| }# end: foreach |
| #close H1 if $text_id eq "GOE_AGA.00000"; |
| |
| $output .= " </spanList>\n</layer>"; |
| |
| $output=encode_utf8($output); |
| |
| $zip->print("$output"); |
| |
| #print STDERR "$0: write_annot(): DONE\n"; |
| |
| } # end: sub write_annot |
| |
| |
| sub startTokenizer { |
| $pid = open2($chld_out, $chld_in, 'java -cp '. join(":", ".", glob(&dirname(__FILE__)."/../target/*.jar"))." de.ids_mannheim.korap.tokenizer.KorAPTokenizerImpl"); |
| $select = IO::Select->new(); |
| $select->add(*$chld_out); |
| } |
| |
| __END__ |
| |
| =pod |
| |
| =encoding utf8 |
| |
| =head1 NAME |
| |
| tei2korapxml - Conversion of TEI P5 based formats to KorAP-XML |
| |
| =head1 SYNOPSIS |
| |
| cat corpus.i5.xml | tei2korapxml > corpus.korapxml.zip |
| |
| =head1 DESCRIPTION |
| |
| C<tei2korapxml> is a script to convert TEI P5 and I5 based documents |
| to the KorAP-XML format. If no specific input is defined, data is |
| read from C<STDIN>. If no specific output is defined, data is written |
| to C<STDOUT>. |
| This program is usually called from inside another script. |
| |
| =head1 INSTALLATION |
| |
| C<tei2korapxml> requires L<libxml2-dev> bindings to build. When |
| these bindings are available, the preferred way to install the script is |
| to use L<cpanm|App::cpanminus>. |
| |
| $ cpanm https://github.com/KorAP/KorAP-XML-TEI.git |
| |
| In case everything went well, the C<tei2korapxml> tool will |
| be available on your command line immediately. |
| Minimum requirement for L<KorAP::XML::TEI> is Perl 5.16. |
| |
| =head1 OPTIONS |
| |
| =over 2 |
| |
| =item B<--base|-b> |
| |
| The base directory for output. Defaults to C<.>. |
| |
| =item B<--help|-h> |
| |
| Print help information. |
| |
| =item B<--version|-v> |
| |
| Print version information. |
| |
| =back |
| |
| =head1 COPYRIGHT AND LICENSE |
| |
| Copyright (C) 2020, L<IDS Mannheim|https://www.ids-mannheim.de/> |
| |
| Author: Peter Harders |
| |
| Contributors: Marc Kupietz, Carsten Schnober, Nils Diewald |
| |
| L<KorAP::XML::TEI> is developed as part of the L<KorAP|https://korap.ids-mannheim.de/> |
| Corpus Analysis Platform at the |
| L<Leibniz Institute for the German Language (IDS)|http://ids-mannheim.de/>, |
| member of the |
| L<Leibniz-Gemeinschaft|http://www.leibniz-gemeinschaft.de/>. |
| |
| This program is free software published under the |
| L<BSD-2 License|https://raw.githubusercontent.com/KorAP/KorAP-XML-TEI/master/LICENSE>. |
| |
| =cut |