helpcontent2/helpers/update_tree.pl

   1 :
   2 eval 'exec perl -wS $0 ${1+"$@"}'
   3     if 0;
   4 #*************************************************************************
   5 #
   6 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   7 #
   8 # Copyright 2008 by Sun Microsystems, Inc.
   9 #
  10 # OpenOffice.org - a multi-platform office productivity suite
  11 #
  12 # $RCSfile: update_tree.pl,v $
  13 #
  14 # $Revision: 1.13 $
  15 #
  16 # This file is part of OpenOffice.org.
  17 #
  18 # OpenOffice.org is free software: you can redistribute it and/or modify
  19 # it under the terms of the GNU Lesser General Public License version 3
  20 # only, as published by the Free Software Foundation.
  21 #
  22 # OpenOffice.org is distributed in the hope that it will be useful,
  23 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  24 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25 # GNU Lesser General Public License version 3 for more details
  26 # (a copy is included in the LICENSE file that accompanied this code).
  27 #
  28 # You should have received a copy of the GNU Lesser General Public License
  29 # version 3 along with OpenOffice.org.  If not, see
  30 # <http://www.openoffice.org/license.html>
  31 # for a copy of the LGPLv3 License.
  32 #
  33 #*************************************************************************
  34
  35 use Cwd 'abs_path';
  36 use File::Find;
  37 use File::Copy qw/cp mv/;
  38 use File::Basename;
  39 use Benchmark;
  40
  41 my $is_win = "false";
  42
  43 $is_win = "true" , if ( defined $ENV{USE_SHELL} && $ENV{USE_SHELL} eq "4nt" );
  44
  45 $t0 = new Benchmark;
  46 # update the tree files in <platform>/misc/*
  47
  48 $| = 1;
  49
  50 my $prj = $ENV{ENVPRJ};
  51
  52 my $inpath = $ENV{INPATH};
  53 terminate() if ( ! defined $inpath );
  54
  55 my $destpath = $inpath;
  56 my $with_lang = $ENV{WITH_LANG};
  57
  58 if ( defined $ENV{COMMON_OUTDIR} && $ENV{COMMON_OUTDIR} ne "" ) {
  59     $destpath =~ s/$ENV{OUTPATH}/$ENV{COMMON_OUTDIR}/;
  60 }
  61
  62 # Always use / directory separators
  63 $prj =~ s/\\/\//g if defined($prj);
  64 $inpath =~ s/\\/\//g;
  65 $destpath =~ s/\\/\//g;
  66
  67
  68 if ( ! defined $prj ) {
  69 # do someting that works for manual call
  70     ($scriptname = `pwd`) =~ s/\n/\/$0/;
  71     ($tree_src = $scriptname) =~ s/\/update_tree.pl/\/..\/source\/auxiliary/;
  72     ($tree_dest = $scriptname) =~ s/\/update_tree.pl/\/..\/$destpath\/misc/;
  73     ($source_dir = $scriptname) =~ s/\/update_tree.pl/\/..\/source/;
  74     if ( defined $ENV{TRYSDF} && defined $ENV{LOCALIZESDF} )
  75     {
  76         if( defined $ENV{LOCALIZATION_FOUND} && $ENV{LOCALIZATION_FOUND} eq "YES" )
  77         {
  78             $source_dir = $ENV{TRYSDF};
  79         }
  80         else
  81         {
  82             $source_dir = $ENV{LOCALIZESDF};
  83         }
  84         if( $is_win eq "false" ) { $source_dir =~ s/\/auxiliary\/localize.sdf$// ; }
  85         else { $source_dir =~ s/\\auxiliary\\localize.sdf$// ; }
  86
  87     }
  88     #else {die "ERROR: The env variables TRYSDF LOCALIZATION_FOUND LOCALIZESDF not found ... something is wrong!\n";}
  89
  90
  91     $treestrings = "$source_dir/text/shared/tree_strings.xhp";
  92 } else {
  93     $tree_src = "$prj\/source\/auxiliary";
  94     $tree_dest = "$prj\/$destpath\/misc";
  95     $source_dir = "$prj\/source";
  96     $treestrings = "$source_dir/text/shared/tree_strings.xhp";
  97
  98     if ( defined $ENV{TRYSDF} && defined $ENV{LOCALIZESDF} )
  99     {
 100         if( defined $ENV{LOCALIZATION_FOUND} && $ENV{LOCALIZATION_FOUND} eq "YES" )
 101         {
 102             $source_dir = $ENV{TRYSDF};
 103         }
 104         else
 105         {
 106             $source_dir = $ENV{LOCALIZESDF};
 107         }
 108         if( $is_win eq "false" ) { $source_dir =~ s/\/auxiliary\/localize.sdf$// ; }
 109         else { $source_dir =~ s/\\auxiliary\\localize.sdf$// ; }
 110     }
 111     #else {die "ERROR: The env variables TRYSDF LOCALIZATION_FOUND LOCALIZESDF not found ... something is wrong!\n";}
 112
 113 }
 114
 115 # Get the English tree files as master
 116 #-------------------------------
 117 # Update English from xhp
 118 #-------------------------------
 119 &do_english;
 120 #-------------------------------
 121 # Update localizations from sdf
 122 #-------------------------------
 123
 124 @langs = split /\s+/, $with_lang;
 125 &read_loc;
 126 print "################\nUpdating the treefiles for @langs \n";
 127 for $l(@langs) {
 128     if ($l ne "en-US") {
 129         &do_lang($l);
 130     }
 131 }
 132
 133 #-------------------------------
 134 #
 135 $t1 = new Benchmark;
 136 $td = timediff($t1, $t0);
 137 print timestr($td),"\n";
 138
 139 ####################
 140 # SUBS
 141 ####################
 142 sub terminate {
 143     $err = shift;
 144     print "$err\n\n";
 145     $msg = <<"MSG";
 146
 147 update_tree.pl
 148    all languages in WITH_LANG are processed. WITH_LANG=ALL is
 149    not supported in manual calls.
 150
 151    Updates the *.tree files.
 152    At first, the English file is updated based on the English
 153    help topic titles as read from the help files. Then, the
 154    localized tree files are written based on the English tree
 155    file and the localized help topic titles.
 156
 157    Requires a valid SO/OOo environment.
 158 MSG
 159    print "$msg\n";
 160    exit( -1 );
 161    # die "$msg\n";
 162 }
 163
 164 #---------------------------------------------------
 165
 166 sub do_english {
 167     print "Processing en-US\n";
 168     undef %helpsection; undef %node;
 169     &readtreestrings;
 170     &gettreefiles;
 171     &processtreefiles('en-US');
 172 }
 173
 174 #---------------------------------------------------
 175 sub do_lang {
 176     $lng = shift;
 177     print "\n---------------------------------------------------\nProcessing $lng\n";
 178     &processtreefiles($lng);
 179     print "\n";
 180 }
 181
 182 #---------------------------------------------------
 183 sub readtreestrings {
 184     print "Reading tree strings for en-US...";
 185     if (open TREE, $treestrings) {
 186         while (<TREE>) {
 187             chomp;
 188             s/<\/*help:productname>//gis;
 189             if (/help_section/) {
 190                 s/^\s*<.*help_section//;
 191                 s/<\/.*$//;
 192                 ($id = $_) =~ s/^.*id=&quot;(\d+)&quot;.*$/$1/;
 193                 ($title = $_) =~ s/^.*title=&quot;(.*)&quot;.*$/$1/;
 194                 $helpsection{$id} = $title;
 195             }
 196
 197             if (/node id=/) {
 198                 s/^\s*<.*node //;
 199                 s/<\/.*$//;
 200                 ($id = $_) =~ s/^.*id=&quot;(\d+)&quot;.*$/$1/;
 201                 ($title = $_) =~ s/^.*title=&quot;(.*)&quot;.*$/$1/;
 202                 $node{$id} = $title;
 203             }
 204         }
 205         close TREE;
 206     } else {
 207         &terminate("Error opening $treestrings");
 208     }
 209     print "done\n";
 210 }
 211
 212 #------------------------------------
 213 sub gettreefiles {
 214     # Read the tree files from the directory
 215     # this list is also used for all foreign languages
 216     print "Reading tree files...";
 217     if (opendir ENUS, "$tree_src") {
 218         @treeviews = grep /\.tree/, readdir ENUS;
 219         closedir ENUS;
 220     } else {
 221         &terminate("Cannot open directory $tree_src");
 222     }
 223     print "done\n";
 224 }
 225
 226 #------------------------------------
 227 sub processtreefiles {
 228     $lng = shift;
 229     use File::Temp qw/ tempfile /;
 230     use File::Spec;
 231
 232     for $tv(@treeviews) {
 233         print "\nProcessing $tv\n";
 234         @lines = &readtv("$tree_src/$tv");
 235         print "Read ".scalar @lines." lines\n";
 236         for $l(@lines) {
 237             if ($l =~ /topic/) {
 238                 ($id = $l) =~ s/^.*id="([^"]*)".*$/$1/gis;
 239                 ($module = $id) =~ s/^([^\/]*).*$/$1/;
 240                 $id =~ s/^.*?\///;
 241                 $file = "$source_dir/$id";
 242
 243                 if ($lng eq 'en-US') { # english comes from the file
 244                     if (open F,$file) {
 245                         print ".";
 246                         undef $/; $cnt = <F>; close F;
 247                         $cnt =~ s/^.*<title[^>]+id="tit"[^>]*>([^<]*)<\/title>.*$/$1/gis;
 248                         $cnt =~ s/&apos;/\'/gis; $cnt =~ s/&amp;/+/gis;
 249                         $cnt =~ s/&quot;/\'/gis; $cnt =~ s/&/+/gis;
 250                         $l = "<topic id=\"$module/$id\">$cnt</topic>\n";
 251                     } else {
 252                         print "!";
 253                         $l = "<!-- removed $module/$id -->\n";
 254                     }
 255                 } else { # localized comes from the localize sdf
 256                     #print "\nid: $id";
 257                     if (defined($loc_title{$lng}->{$id})) {
 258                         print ".";
 259                         $l = "<topic id=\"$module/$id\">$loc_title{$lng}->{$id}</topic>\n";
 260                     } else {
 261                         print "!";
 262                     }
 263                 }
 264             }
 265
 266             if ($l =~/<node/) {
 267                 ($id = $l) =~ s/^.*id="(\d+)".*$/$1/gis;
 268                 if ($lng eq 'en-US') {
 269                     if (defined($node{$id})) {
 270                         $l =~ s/title="(.*)"/title="$node{$id}"/;
 271                     } else {
 272                         $l =~ s/title="(.*)"/title="NOTFOUND:$id"/;
 273                     }
 274                 } else {
 275                     if (defined($node{$lng}->{$id})) {
 276                         $l =~ s/title="(.*)"/title="$node{$lng}->{$id}"/;
 277                     }
 278                 }
 279             }
 280
 281             if ($l =~/<help_section/) {
 282                 ($id = $l) =~ s/^.*id="(\d+)".*$/$1/gis;
 283                 if ($lng eq 'en-US') {
 284                     if (defined($helpsection{$id})) {
 285                         $l =~ s/title="(.*)"/title="$helpsection{$id}"/;
 286                     } else {
 287                         print "#";
 288                         $l =~ s/title="(.*)"/title="NOTFOUND:$id"/;
 289                     }
 290                 } else {
 291                     if (defined($helpsection{$lng}->{$id})) {
 292                         $l =~ s/title="(.*)"/title="$helpsection{$lng}->{$id}"/;
 293                     }
 294                 }
 295             }
 296         }
 297          if ( ! -d "$tree_dest/$lng" ) {
 298             mkdir "$tree_dest/$lng" or die "\nCouldn't create directory \"$tree_dest/$lng\"";
 299         }
 300         my $treeoutdir = "$tree_dest/$lng";
 301         my $tmpname_template=$tv."_XXXXX";
 302         my ( $treetmpfilehandle, $treetmpfile ) = tempfile($tmpname_template , DIR => File::Spec->tmpdir() );
 303         close $treetmpfilehandle ;
 304         if (open TV, ">$treetmpfile") {
 305             for $line(@lines) {
 306                 $line =~ s/\$\[officename\]/%PRODUCTNAME/g;
 307                 $line =~ s/\$\[officeversion\]/%PRODUCTVERSION/g;
 308                 print TV $line;
 309             }
 310             close TV;
 311             chmod 0664, $treetmpfile or &terminate("Cannot change rights on $treetmpfile");
 312             if( $^O eq 'MSWin32' )
 313             {
 314                 $tree_dest =~ s/\//\\/g ;
 315                 unlink "$tree_dest\\$lng\\$tv" ;
 316                 mv $treetmpfile , "$tree_dest\\$lng\\$tv" or &terminate("Cannot mv $treetmpfile to $tree_dest\\$lng\\$tv" );
 317             }
 318             else
 319             {
 320                 unlink "$tree_dest/$lng/$tv" ;
 321                 my $ret=mv $treetmpfile , "$tree_dest/$lng/$tv$inpath" or &terminate("Cannot write to $tree_dest/$lng/$tv$inpath - Error $!");
 322                 my $ret=mv "$tree_dest/$lng/$tv$inpath" , "$tree_dest/$lng/$tv" or &terminate("Cannot write to $tree_dest/$lng/$tv - Error $!");
 323             }
 324       } else {
 325             &terminate("Cannot write to $tvout");
 326         }
 327     }
 328 }
 329
 330 #------------------------------------
 331 sub readtv {
 332     my $f = shift;
 333     if (open TV, $f) {
 334         $/ = "\n";
 335         my @l = <TV>;
 336         close TV;
 337         return @l;
 338     } else {
 339         &terminate("Error opening $f");
 340     }
 341 }
 342
 343 #------------------------------------
 344 # read entries form localize.sdf files
 345 #------------------------------------
 346 sub read_loc {
 347     print "\n\nReading localized titles...";
 348     $/ = "\n";
 349     my $path = "$source_dir/text";
 350     $path =~ s/\//\\/g , if $is_win eq "true";
 351     print " in $source_dir/text\n";
 352     @files = `find $source_dir/text -name localize.sdf`;
 353     for my $fname (@files) {
 354         $FS = '\t';
 355         print ".";
 356         open(LOCALIZE_SDF, $fname) || die 'Cannot open "localize.sdf".'."$fname";
 357         while (<LOCALIZE_SDF>) {
 358             my $sdf_line = $_;
 359         my ($Fld1,$file,$Fld3,$Fld4,$id,$Fld6,$Fld7,$Fld8,$Fld9,$lang,$text) = split($FS, $sdf_line , 12);
 360             next if ( $Fld1 =~ /^#/);
 361         if ($id eq 'tit') {
 362                 #strip filename
 363                 $file =~ s/.*text\\/text\\/g;
 364                 #convert \ to / in filename
 365                 $file =~ s/\\/\//g;
 366                 #fpe: i46823 - need to encode &s, added encoding
 367                 $text =~ s/&(?!amp;)/&amp;/g;
 368                 # add entry to the hash
 369                 $loc_title{$lang}->{$file} = $text;
 370             }
 371             if ($file =~ /tree_strings.xhp/) {
 372                 #strip filename
 373                 $file =~ s/.*text/text/g;
 374                 #convert \ to / in filename
 375                 $file =~ s/\\/\//g;
 376                 if ($text =~ /^<help_section/) {
 377                     #example: <help_section application="scalc" id="08" title="表計算ドキュメント">
 378                     my ($fld1,$app,$fld3,$id,$fld5,$sec_title) = split('"', $text, 7);
 379                     #fpe: i46823 - need to encode &s, added encoding
 380                     if( defined $sec_title )
 381                     {
 382                         $sec_title =~ s/&(?!amp;)/&amp;/g;
 383                         #unquot \<item ... /\>
 384                         terminate( "\n\nERROR: Bad string in file '$fname' will cause invalid xml tree file \n---\n'$sdf_line'\n---\nPlease remove or replace < = '&lt;' and  > = '&gt;' within the title attribute '$sec_title'\n") , if( $sec_title =~ /[\<\>]/ );
 385                         $helpsection{$lang}->{$id} = $sec_title;
 386                     }
 387                 } elsif ($text =~/<node id=/) {
 388                     # example: <node id="0205" title="Tabelas em documentos de texto">
 389                     # BEWARE: title may contain escaped '"' so only match " not preceded by \
 390                     # using a zero‐width negative look‐behind assertion.
 391                     my ($fld1,$id,$fld3,$node_title,$Fld5) = split(/(?<!\\)"/, $text, 5);
 392                     #fpe: i46823 - need to encode &s, added encoding
 393                     if( defined $node_title )
 394                     {
 395                         $node_title =~ s/&(?!amp;)/&amp;/g;
 396                            terminate( "\n\nERROR: Bad string in '$fname' will cause invalid xml tree file \n---\n'$sdf_line'\n---\nPlease remove or replace < = '&lt;' and  > = '&gt;' within the title attribute '$node_title'\n") , if( $node_title =~ /[\<\>]/ );
 397                     }
 398                     $node{$lang}->{$id} = $node_title;
 399                 }
 400             }
 401         }
 402         close LOCALIZE_SDF;
 403     }
 404     # statistics
 405     $total_elements=0;
 406     foreach $lang (keys %loc_title) {
 407         $no_elements = scalar(keys(%{$loc_title{$lang}}));
 408         push(@langstat, "$lang:\t ".$no_elements." matches\n");
 409         $total_elements += $no_elements;
 410     }
 411     print "\ndone reading a total of ".$total_elements." localized titles for ".scalar(keys(%loc_title))." languages from ".scalar @files ." files\n";
 412     print sort(@langstat);
 413 }
 414