vcs.maemo.org Git - maemian/blob - nokia-lintian/lib/Spelling.pm

   1 # -*- perl -*-
   2 # Spelling -- check for common spelling errors
   3
   4 # Copyright (C) 1998 Richard Braakman
   5 #
   6 # This program is free software; you can redistribute it and/or modify
   7 # it under the terms of the GNU General Public License as published by
   8 # the Free Software Foundation; either version 2 of the License, or
   9 # (at your option) any later version.
  10 #
  11 # This program is distributed in the hope that it will be useful,
  12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 # GNU General Public License for more details.
  15 #
  16 # You should have received a copy of the GNU General Public License
  17 # along with this program.  If not, you can find it on the World Wide
  18 # Web at http://www.gnu.org/copyleft/gpl.html, or write to the Free
  19 # Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston,
  20 # MA 02110-1301, USA.
  21
  22 package Spelling;
  23 use strict;
  24 use Tags;
  25
  26 use Exporter;
  27 our @ISA = qw(Exporter);
  28 our @EXPORT = qw(spelling_check spelling_check_picky);
  29
  30 # All spelling errors that have been observed "in the wild" in package
  31 # descriptions are added here, on the grounds that if they occurred once they
  32 # are more likely to occur again.
  33
  34 # Misspellings of "compatibility", "separate", and "similar" are particularly
  35 # common.
  36
  37 # Be careful with corrections that involve punctuation, since the check is a
  38 # bit rough with punctuation.  For example, I had to delete the correction of
  39 # "builtin" to "built-in".
  40
  41 our %CORRECTIONS = qw(
  42                       accesnt accent
  43                       accelleration acceleration
  44                       accessable accessible
  45                       accomodate accommodate
  46                       acess access
  47                       acording according
  48                       additionaly additionally
  49                       adress address
  50                       adresses addresses
  51                       adviced advised
  52                       afecting affecting
  53                       albumns albums
  54                       alegorical allegorical
  55                       algorith algorithm
  56                       allpication application
  57                       altough although
  58                       alows allows
  59                       amoung among
  60                       amout amount
  61                       analysator analyzer
  62                       ang and
  63                       appropiate appropriate
  64                       arraival arrival
  65                       artifical artificial
  66                       artillary artillery
  67                       attemps attempts
  68                       authentification authentication
  69                       automaticly automatically
  70                       automatize automate
  71                       automatized automated
  72                       automatizes automates
  73                       auxilliary auxiliary
  74                       availavility availability
  75                       availble available
  76                       avaliable available
  77                       availiable available
  78                       backgroud background
  79                       baloons balloons
  80                       becomming becoming
  81                       becuase because
  82                       calender calendar
  83                       cariage carriage
  84                       challanges challenges
  85                       changable changeable
  86                       charachters characters
  87                       charcter character
  88                       choosen chosen
  89                       colorfull colorful
  90                       comand command
  91                       commerical commercial
  92                       comminucation communication
  93                       commoditiy commodity
  94                       compability compatibility
  95                       compatability compatibility
  96                       compatable compatible
  97                       compatibiliy compatibility
  98                       compatibilty compatibility
  99                       compleatly completely
 100                       complient compliant
 101                       compres compress
 102                       containes contains
 103                       containts contains
 104                       contence contents
 105                       continous continuous
 106                       contraints constraints
 107                       convertor converter
 108                       convinient convenient
 109                       cryptocraphic cryptographic
 110                       deamon daemon
 111                       debain Debian
 112                       debians Debian\'s
 113                       decompres decompress
 114                       definate definite
 115                       definately definitely
 116                       dependancies dependencies
 117                       dependancy dependency
 118                       dependant dependent
 119                       developement development
 120                       developped developed
 121                       deveolpment development
 122                       devided divided
 123                       dictionnary dictionary
 124                       diplay display
 125                       disapeared disappeared
 126                       dissapears disappears
 127                       documentaion documentation
 128                       docuentation documentation
 129                       documantation documentation
 130                       dont don\'t
 131                       easilly easily
 132                       ecspecially especially
 133                       edditable editable
 134                       editting editing
 135                       eletronic electronic
 136                       enchanced enhanced
 137                       enhaced enhanced
 138                       encorporating incorporating
 139                       enlightnment enlightenment
 140                       enterily entirely
 141                       enviroiment environment
 142                       environement environment
 143                       excellant excellent
 144                       exlcude exclude
 145                       exprimental experimental
 146                       extention extension
 147                       failuer failure
 148                       familar familiar
 149                       fatser faster
 150                       fetaures features
 151                       forse force
 152                       fortan fortran
 153                       framwork framework
 154                       fuction function
 155                       fuctions functions
 156                       functionnality functionality
 157                       functonality functionality
 158                       functionaly functionally
 159                       futhermore furthermore
 160                       generiously generously
 161                       grahical graphical
 162                       grahpical graphical
 163                       grapic graphic
 164                       guage gauge
 165                       halfs halves
 166                       heirarchically hierarchically
 167                       helpfull helpful
 168                       hierachy hierarchy
 169                       hierarchie hierarchy
 170                       howver however
 171                       implemantation implementation
 172                       incomming incoming
 173                       incompatabilities incompatibilities
 174                       indended intended
 175                       indendation indentation
 176                       independant independent
 177                       informatiom information
 178                       initalize initialize
 179                       inofficial unofficial
 180                       integreated integrated
 181                       integrety integrity
 182                       integrey integrity
 183                       intendet intended
 184                       interchangable interchangeable
 185                       intermittant intermittent
 186                       jave java
 187                       langage language
 188                       langauage language
 189                       langugage language
 190                       lauch launch
 191                       lesstiff lesstif
 192                       libaries libraries
 193                       libary library
 194                       licenceing licencing
 195                       loggin login
 196                       logile logfile
 197                       loggging logging
 198                       maintainance maintenance
 199                       maintainence maintenance
 200                       makeing making
 201                       managable manageable
 202                       manoeuvering maneuvering
 203                       mathimatic mathematic
 204                       mathimatics mathematics
 205                       mathimatical mathematical
 206                       ment meant
 207                       modulues modules
 208                       monochromo monochrome
 209                       multidimensionnal multidimensional
 210                       navagating navigating
 211                       nead need
 212                       neccesary necessary
 213                       neccessary necessary
 214                       necesary necessary
 215                       nescessary necessary
 216                       noticable noticeable
 217                       o\'caml OCaml
 218                       optionnal optional
 219                       orientatied orientated
 220                       orientied oriented
 221                       pacakge package
 222                       pachage package
 223                       packacge package
 224                       packege package
 225                       packge package
 226                       pakage package
 227                       particularily particularly
 228                       persistant persistent
 229                       plattform platform
 230                       ploting plotting
 231                       protable portable
 232                       posible possible
 233                       postgressql PostgreSQL
 234                       powerfull powerful
 235                       prefered preferred
 236                       prefferably preferably
 237                       prepaired prepared
 238                       princliple principle
 239                       priorty priority
 240                       proccesors processors
 241                       proces process
 242                       processsing processing
 243                       processessing processing
 244                       progams programs
 245                       programers programmers
 246                       programm program
 247                       programms programs
 248                       promps prompts
 249                       pronnounced pronounced
 250                       prononciation pronunciation
 251                       pronouce pronounce
 252                       protcol protocol
 253                       protocoll protocol
 254                       recieve receive
 255                       recieved received
 256                       redircet redirect
 257                       refence reference
 258                       regulamentations regulations
 259                       remoote remote
 260                       repectively respectively
 261                       replacments replacements
 262                       requiere require
 263                       runnning running
 264                       safly safely
 265                       savable saveable
 266                       searchs searches
 267                       separatly separately
 268                       seperate separate
 269                       seperated separated
 270                       seperately separately
 271                       seperatly separately
 272                       serveral several
 273                       setts sets
 274                       similiar similar
 275                       simliar similar
 276                       speach speech
 277                       speling spelling
 278                       splitted split
 279                       standart standard
 280                       staically statically
 281                       staticly statically
 282                       succesful successful
 283                       succesfully successfully
 284                       suplied supplied
 285                       suport support
 286                       suppport support
 287                       supportin supporting
 288                       synax syntax
 289                       synchonized synchronized
 290                       syncronize synchronize
 291                       syncronizing synchronizing
 292                       syncronus synchronous
 293                       syste system
 294                       sythesis synthesis
 295                       taht that
 296                       throught through
 297                       useable usable
 298                       usefull useful
 299                       usera users
 300                       usetnet Usenet
 301                       utilites utilities
 302                       utillities utilities
 303                       utilties utilities
 304                       utiltity utility
 305                       utitlty utility
 306                       variantions variations
 307                       varient variant
 308                       verson version
 309                       vicefersa vice-versa
 310                       yur your
 311                       wheter whether
 312                       wierd weird
 313                       xwindows X
 314                      );
 315
 316 # The format above doesn't allow spaces.
 317 $CORRECTIONS{'alot'} = 'a lot';
 318
 319 # Picky corrections, applied before lowercasing the word.  These are only
 320 # applied to things known to be entirely English text, such as package
 321 # descriptions, and should not be applied to files that may contain
 322 # configuration fragments or more informal files such as debian/copyright.
 323 our %CORRECTIONS_CASE = qw(
 324                            D-BUS D-Bus
 325                            d-bus D-Bus
 326                            dbus D-Bus
 327                            debian Debian
 328                            english English
 329                            french French
 330                            EMacs Emacs
 331                            Gconf GConf
 332                            gconf GConf
 333                            german German
 334                            Gnome GNOME
 335                            gnome GNOME
 336                            Gnome-VFS GnomeVFS
 337                            Gnome-Vfs GnomeVFS
 338                            GnomeVfs GnomeVFS
 339                            gnome-vfs GnomeVFS
 340                            gnomevfs GnomeVFS
 341                            Gobject GObject
 342                            gobject GObject
 343                            Gstreamer GStreamer
 344                            gstreamer GStreamer
 345                            GTK GTK+
 346                            gtk+ GTK+
 347                            kde KDE
 348                            meta-package metapackage
 349                            MYSQL MySQL
 350                            Mysql MySQL
 351                            mysql MySQL
 352                            linux Linux
 353                            Latex LaTeX
 354                            latex LaTeX
 355                            OCAML OCaml
 356                            Ocaml OCaml
 357                            ocaml OCaml
 358                            OpenLdap OpenLDAP
 359                            Openldap OpenLDAP
 360                            openldap OpenLDAP
 361                            Postgresql PostgreSQL
 362                            postgresql PostgreSQL
 363                            python Python
 364                            russian Russian
 365                            SkoleLinux Skolelinux
 366                            skolelinux Skolelinux
 367                            SLang S-Lang
 368                            S-lang S-Lang
 369                            s-lang S-Lang
 370                            TCL Tcl
 371                            tcl Tcl
 372                            TEX TeX
 373                            Tex TeX
 374                            TeTeX teTeX
 375                            Tetex teTeX
 376                            tetex teTeX
 377                            TK Tk
 378                            tk Tk
 379                            Xemacs XEmacs
 380                            XEMacs XEmacs
 381                            XFCE Xfce
 382                            XFce Xfce
 383                            xfce Xfce
 384                           );
 385
 386 # The format above doesn't allow spaces.
 387 $CORRECTIONS_CASE{'Debian-Edu'} = 'Debian Edu';
 388 $CORRECTIONS_CASE{'debian-edu'} = 'Debian Edu';
 389 $CORRECTIONS_CASE{'TeXLive'} = 'TeX Live';
 390 $CORRECTIONS_CASE{'TeX-Live'} = 'TeX Live';
 391 $CORRECTIONS_CASE{'TeXlive'} = 'TeX Live';
 392 $CORRECTIONS_CASE{'TeX-live'} = 'TeX Live';
 393 $CORRECTIONS_CASE{'texlive'} = 'TeX Live';
 394 $CORRECTIONS_CASE{'tex-live'} = 'TeX Live';
 395
 396 # -----------------------------------
 397
 398 sub _tag {
 399     my @args = grep { defined($_) } @_;
 400     tag(@args);
 401 }
 402
 403 # Check spelling of $text and report the tag $tag if we find anything.
 404 # $filename, if included, is given as the first argument to the tag.  If it's
 405 # not defined, it will be omitted.
 406 sub spelling_check {
 407     my ($tag, $text, $filename) = @_;
 408     return unless $text;
 409
 410     for my $word (split(/\s+/, $text)) {
 411         $word = lc $word;
 412
 413         # Try deleting the non-alphabetic parts from the word.  Treat
 414         # apostrophes specially: only delete them if they occur at the
 415         # beginning or end of the word.
 416         #
 417         # FIXME: Should do something that's aware of Unicode character
 418         # classes rather than only handling ISO 8859-15 characters.
 419         $word =~ s/(^\')|[^\w\xc0-\xd6\xd8-\xf6\xf8-\xff\'-]+|(\'\z)//g;
 420         if (exists $CORRECTIONS{$word}) {
 421             _tag($tag, $filename, $word, $CORRECTIONS{$word});
 422         }
 423     }
 424
 425     # Special case for correcting a multi-word string.
 426     if ($text =~ m,Debian/GNU\s+Linux,) {
 427         _tag($tag, $filename, "Debian/GNU Linux", "Debian GNU/Linux");
 428     }
 429 }
 430
 431 # Check spelling of $text against pickier corrections, such as common
 432 # capitalization mistakes.  This check is separate from spelling_check since
 433 # it isn't appropriate for some files (such as changelog).  Takes $text to
 434 # check spelling in and $tag to report if we find anything.  $filename, if
 435 # included, is given as the first argument to the tag.  If it's not defined,
 436 # it will be omitted.
 437 sub spelling_check_picky {
 438     my ($tag, $text, $filename) = @_;
 439
 440     for my $word (split(/\s+/, $text)) {
 441         $word =~ s/^\(|[).,?!:;]+$//g;
 442         if (exists $CORRECTIONS_CASE{$word}) {
 443             _tag($tag, $filename, $word, $CORRECTIONS_CASE{$word});
 444             next;
 445         }
 446     }
 447     if ($text =~ m,meta\s+package,) {
 448         _tag($tag, $filename, "meta package", "metapackage");
 449     }
 450 }
 451
 452 1;
 453
 454 # Local Variables:
 455 # indent-tabs-mode: nil
 456 # cperl-indent-level: 4
 457 # End:
 458 # vim: syntax=perl sw=4 sts=4 ts=4 et shiftround