scripts/parse-help-patches.pl

   1 #!/usr/bin/perl
   2
   3 # WARNING!  This script is really ugly!
   4
   5 # to use:
   6 #   1) make a folder, like en-import
   7 #   2) run ../parse-help-patches.pl in it
   8 #   3) voila!
   9 #
  10 # on the server, run:
  11 #   find en-import/ -type f -name \*.txt | ./wiki_import.sh -s 'Top level'
  12
  13 use strict;
  14 use warnings;
  15 use Text::CSV_PP;
  16
  17 my $line = "";
  18 my @lines;
  19 my $lineCount = 0;
  20 my $column;
  21 my $lastColumn = 0;
  22 my $printText = "";
  23
  24 my %classnames = ();
  25 my %xyhash = ();
  26
  27 my $libraryName = "";
  28 my $objectclassName = "";
  29 my $fileName = "";
  30
  31 my $pageName = "";
  32 my $abbreviationName = "";
  33 my $descriptionName = "";
  34 my $categoryName = "";
  35 my $datatypeName = "";
  36
  37 #------------------------------------------------------------------------------#
  38 # TRANSLATION VARIABLES
  39 #------------------------------------------------------------------------------#
  40
  41 # wikipedia terms
  42 my $stub = "";
  43 my $template = "";
  44 my $category = "";
  45 my $infobox = "";
  46
  47 # pdpedia terms
  48 my $objectclass = "";
  49
  50 # page headers
  51 my $inlets = "";
  52 my $outlets = "";
  53 my $arguments = "";
  54 my $messages = "";
  55
  56 # infobox
  57 my $name = "";
  58 my $description = "";
  59 my $abbreviation = "";
  60 my $library = "";
  61 my $author = "";
  62 my $developer = "";
  63 my $releaseVersion = "";
  64 my $releaseDate = "";
  65 my $dependencies = "";
  66 my $license = "";
  67 my $website = "";
  68 my $programmingLanguage = "";
  69 my $platform = "";
  70 my $operatingSystem = "";
  71 my $language = "";
  72 my $dataType = "";
  73 my $distribution = "";
  74 my $status = "";
  75 my $caption = "";
  76
  77 #------------------------------------------------------------------------------#
  78 # Français
  79
  80 # wikipedia terms
  81 $stub = "Ébauche";
  82 $template = "Modèle";
  83 $category = "Catégorie";
  84 $infobox = "Infos";
  85
  86 # pdpedia terms
  87 $objectclass = "Classe d'objets";
  88
  89 # page headers
  90 $inlets = "Entrées";
  91 $outlets = "Sorties";
  92 $arguments = "Arguments";
  93 $messages = "Messages";
  94
  95 # infobox
  96 $name = "Nom";
  97 $description = "Description";
  98 $abbreviation = "Abréviation";
  99 $library = "Bibliothèque";
 100 $author = "Auteur";
 101 $developer = "Développeur";
 102 $releaseVersion = "Version";
 103 $releaseDate = "Date";
 104 $dependencies = "Dépendances";
 105 $license = "Licence";
 106 $website = "Site Web";
 107 $programmingLanguage = "Langage de programmation";
 108 $platform = "Plate-forme";
 109 $operatingSystem = "Système d'exploitation";
 110 $language = "Langue";
 111 $dataType = "Type de données";
 112 $distribution = "Distribution";
 113
 114 $status = "État";
 115 $caption = "Légende";
 116 $language = "Langue";
 117
 118
 119
 120
 121 #------------------------------------------------------------------------------#
 122 # PARSE CSV
 123 #------------------------------------------------------------------------------#
 124 my $csvfile = '/Users/hans/Desktop/TODO/wiki_files_hacked/objectlist.csv';
 125 my $csv = Text::CSV_PP->new();
 126 my %csvhash = ();
 127
 128 open (CSV, "<", $csvfile) or die $!;
 129 my @csvlines = split(/\012\015?|\015\012?/,(join '',<CSV>));
 130 foreach (@csvlines) {
 131          if ($csv->parse($_)) {
 132                   my @columns = $csv->fields();
 133                   $csvhash{ $columns[0] }{ $columns[2] } = "$columns[0],$columns[2],$columns[3],$columns[4],$columns[5],$columns[7]";
 134                   #print("$columns[0],$columns[2] | ");
 135          } else {
 136                   my $err = $csv->error_input;
 137                   print "Failed to parse line: $err";
 138          }
 139 }
 140 close CSV;
 141
 142 #------------------------------------------------------------------------------#
 143 # PARSE HELP FILES
 144 #------------------------------------------------------------------------------#
 145
 146 foreach (`/sw/bin/find /Users/hans/Desktop/TODO/wiki_files_hacked/5.reference/ -type f -name '*.pd'`) {
 147   chop;
 148   $fileName = "";
 149   if( (m|.*/5\.reference/([a-zA-Z0-9_-]+)/(.+)-help\.pd|) ||  (m|.*/5\.reference/([a-zA-Z0-9_-]+)/(.+)\.pd|) ) {
 150 #        print("$1 , $2\t");
 151          if( $1 eq 'zflatspace' ) { $libraryName = "flatspace";}
 152          else {$libraryName = lc($1); }
 153          $objectclassName = $2;
 154          $fileName = $_;
 155   }
 156
 157 #  print "filename: $fileName\n";
 158   if ($fileName) {
 159          $printText = "";            # init container
 160          %xyhash = ();               # init sorting hash
 161
 162          open(HELPPATCH, "$fileName");
 163          undef $/;                                                # $/ defines the "end of record" character
 164          $_ = <HELPPATCH>;                        # read the whole file into the scalar
 165          close HELPPATCH;
 166          $/ = "\n";                                               # Restore for normal behaviour later in script
 167
 168          s| \\||g;                                                # remove Pd-style escaping
 169          s|([^;])\n|$1 |g;                        # remove extra newlines
 170          s|\(http://.*\)\([ \n]\)|[$1]$2|g;
 171
 172          @lines = split(';\n', $_);
 173
 174          foreach (@lines) {
 175                 if (m|^#X text ([0-9]+) ([0-9]+) (.*)|) {
 176                   $xyhash{ $2 }{ $1 } = $3;
 177                   #      print("$lineCount @ $1,$2: $3\n");
 178                 }
 179                 $lineCount++;
 180          }
 181          for ($column = -300; $column < 1501; $column += 300) {
 182                 foreach my $yKey ( sort {$a <=> $b} keys(%xyhash) ) {
 183                   foreach my $xKey ( keys(%{$xyhash{$yKey}}) ) {
 184                          if ( ($xKey > $lastColumn) && ($xKey < $column) ) {
 185                                 $printText .= "$xyhash{$yKey}{$xKey}\n\n";
 186                                 #print("TEST $xKey,$yKey: $xyhash{$yKey}{$xKey}\n");
 187                          }
 188                   }
 189                 }
 190                 $lastColumn = $column;
 191          }
 192
 193          my $myColumns = $csvhash{$libraryName}{$objectclassName};
 194          my @myColumns;
 195          if($myColumns) { @myColumns = split(',', $myColumns); }
 196          if($myColumns[2]) { $abbreviationName = $myColumns[2]; }        if( $classnames{$objectclassName} ) {
 197                 $pageName = "${objectclassName}_(${libraryName})";
 198          } else {
 199                 $pageName = "${objectclassName}";
 200          }
 201
 202
 203          mkdir($libraryName);
 204          if( $classnames{$objectclassName} ) {
 205                 $pageName = "${objectclassName}_(${libraryName})";
 206          } else {
 207                 $pageName = "${objectclassName}";
 208          }
 209          open(OBJECTCLASS, ">$libraryName/${pageName}.txt");
 210          print(OBJECTCLASS "{{Infobox $objectclass\n");
 211          print(OBJECTCLASS "| $name                   = $objectclassName\n");
 212          if($abbreviationName) {
 213                 print(OBJECTCLASS "| $abbreviation           = $abbreviationName\n");}
 214          print(OBJECTCLASS "| $description            = \n");
 215          print(OBJECTCLASS "| $dataType               = \n");
 216          print(OBJECTCLASS "| $library                = [[$libraryName]]\n");
 217          print(OBJECTCLASS "| $author                 = {{$libraryName " . lc(${author}) . "}}\n");
 218          print(OBJECTCLASS "| $license                = {{$libraryName " . lc(${license}) . "}}\n");
 219          print(OBJECTCLASS "| $status                 = {{$libraryName " . lc(${status}) . "}}\n");
 220          print(OBJECTCLASS "| $website                = {{$libraryName " . lc(${website}) . "}}\n");
 221          print(OBJECTCLASS "| $releaseVersion         = {{$libraryName " . lc(${releaseVersion}) . "}}\n");
 222          print(OBJECTCLASS "| $releaseDate            = {{$libraryName " . lc(${releaseDate}) . "}}\n");
 223          print(OBJECTCLASS "| $distribution           = {{$libraryName " . lc(${distribution}) . "}}\n");
 224          print(OBJECTCLASS "| $platform               = [[GNU/Linux]], [[Mac OS X]], [[Windows]]\n");
 225          print(OBJECTCLASS "}}\n\n");
 226 #        print(OBJECTCLASS "\n$printText\n\n\n");
 227          print(OBJECTCLASS "==$inlets==\n\n\n");
 228          print(OBJECTCLASS "==$outlets==\n\n\n");
 229          print(OBJECTCLASS "==$arguments==\n\n\n");
 230 #        print(OBJECTCLASS "==$messages==\n\n\n");
 231          print(OBJECTCLASS "{{${objectclass}-${stub}}} \n\n");
 232          print(OBJECTCLASS "[[$category:$objectclass]]\n");
 233          print(OBJECTCLASS "[[$category:$libraryName]]\n");
 234 #        if($category) {
 235 #               print(OBJECTCLASS "[[$category:$categoryName]]\n");
 236 #        }
 237          print(OBJECTCLASS "\n\n");
 238          print(OBJECTCLASS "[[de:$pageName]]\n");
 239          print(OBJECTCLASS "[[en:$pageName]]\n");
 240          print(OBJECTCLASS "[[es:$pageName]]\n");
 241 #        print(OBJECTCLASS "[[fr:$pageName]]\n");
 242          print(OBJECTCLASS "[[it:$pageName]]\n");
 243          print(OBJECTCLASS "[[ja:$pageName]]\n");
 244          print(OBJECTCLASS "[[nl:$pageName]]\n");
 245          print(OBJECTCLASS "[[pt:$pageName]]\n");
 246          print(OBJECTCLASS "[[ru:$pageName]]\n");
 247          print(OBJECTCLASS "[[tr:$pageName]]\n");
 248          print(OBJECTCLASS "\n\n");
 249          close(OBJECTCLASS);
 250
 251          $classnames{$objectclassName} = 1;
 252   }
 253 }
 254