vcs.maemo.org Git - maemian/blob - 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                       abandonning abandoning
  43                       abigious ambiguous
  44                       abitrate arbitrate
  45                       abov above
  46                       absolut absolute
  47                       accelleration acceleration
  48                       accesing accessing
  49                       accesnt accent
  50                       accessable accessible
  51                       accessable accessible
  52                       accidentaly accidentally
  53                       accidentually accidentally
  54                       accomodate accommodate
  55                       accomodate accommodate
  56                       accomodates accommodates
  57                       accout account
  58                       acess access
  59                       acording according
  60                       acumulating accumulating
  61                       addional additional
  62                       additionaly additionally
  63                       adress address
  64                       adresses addresses
  65                       adviced advised
  66                       afecting affecting
  67                       albumns albums
  68                       alegorical allegorical
  69                       algorith algorithm
  70                       algorithmical algorithmically
  71                       allpication application
  72                       alows allows
  73                       als also
  74                       altough although
  75                       ambigious ambiguous
  76                       amoung among
  77                       amout amount
  78                       analysator analyzer
  79                       ang and
  80                       annoucement announcement
  81                       appearence appearance
  82                       appropiate appropriate
  83                       appropriatly appropriately
  84                       aquired acquired
  85                       aquired acquired
  86                       architechture architecture
  87                       arguement argument
  88                       arguements arguments
  89                       aritmetic arithmetic
  90                       arne't aren't
  91                       arraival arrival
  92                       artifical artificial
  93                       artillary artillery
  94                       assigment assignment
  95                       assigments assignments
  96                       assistent assistant
  97                       asuming assuming
  98                       atomatically automatically
  99                       attemps attempts
 100                       attruibutes attributes
 101                       authentification authentication
 102                       automaticly automatically
 103                       automatize automate
 104                       automatized automated
 105                       automatizes automates
 106                       auxilliary auxiliary
 107                       avaiable available
 108                       availabled available
 109                       availablity availability
 110                       availale available
 111                       availavility availability
 112                       availble available
 113                       availble available
 114                       availiable available
 115                       avaliable available
 116                       avaliable available
 117                       backgroud background
 118                       baloons balloons
 119                       batery battery
 120                       becomming becoming
 121                       becuase because
 122                       begining beginning
 123                       calender calendar
 124                       cancelation cancellation
 125                       cancelation cancellation
 126                       capabilites capabilities
 127                       capatibilities capabilities
 128                       cariage carriage
 129                       challanges challenges
 130                       changable changeable
 131                       charachters characters
 132                       charcter character
 133                       choosen chosen
 134                       colorfull colorful
 135                       comand command
 136                       comit commit
 137                       commerical commercial
 138                       comminucation communication
 139                       commiting committing
 140                       committ commit
 141                       commoditiy commodity
 142                       compability compatibility
 143                       compatability compatibility
 144                       compatable compatible
 145                       compatibiliy compatibility
 146                       compatibilty compatibility
 147                       compleatly completely
 148                       complient compliant
 149                       compres compress
 150                       compresion compression
 151                       connectinos connections
 152                       consistancy consistency
 153                       containes contains
 154                       containts contains
 155                       contence contents
 156                       continous continuous
 157                       continueing continuing
 158                       contraints constraints
 159                       convertor converter
 160                       convinient convenient
 161                       corected corrected
 162                       cryptocraphic cryptographic
 163                       deamon daemon
 164                       debain Debian
 165                       debians Debian's
 166                       decompres decompress
 167                       definate definite
 168                       definately definitely
 169                       delemiter delimiter
 170                       dependancies dependencies
 171                       dependancy dependency
 172                       dependant dependent
 173                       detabase database
 174                       developement development
 175                       developement development
 176                       developped developed
 177                       deveolpment development
 178                       devided divided
 179                       dictionnary dictionary
 180                       diplay display
 181                       disapeared disappeared
 182                       dispertion dispersion
 183                       dissapears disappears
 184                       docuentation documentation
 185                       documantation documentation
 186                       documentaion documentation
 187                       dont don't
 188                       easilly easily
 189                       ecspecially especially
 190                       edditable editable
 191                       editting editing
 192                       eletronic electronic
 193                       enchanced enhanced
 194                       encorporating incorporating
 195                       endianness endianess
 196                       enhaced enhanced
 197                       enlightnment enlightenment
 198                       enterily entirely
 199                       enviroiment environment
 200                       enviroment environment
 201                       environement environment
 202                       excecutable executable
 203                       exceded exceeded
 204                       excellant excellent
 205                       exlcude exclude
 206                       explicitely explicitly
 207                       expresion expression
 208                       exprimental experimental
 209                       extention extension
 210                       failuer failure
 211                       familar familiar
 212                       fatser faster
 213                       fetaures features
 214                       forse force
 215                       fortan fortran
 216                       forwardig forwarding
 217                       framwork framework
 218                       fuction function
 219                       fuctions functions
 220                       functionaly functionally
 221                       functionnality functionality
 222                       functonality functionality
 223                       futhermore furthermore
 224                       generiously generously
 225                       grahical graphical
 226                       grahpical graphical
 227                       grapic graphic
 228                       guage gauge
 229                       halfs halves
 230                       heirarchically hierarchically
 231                       helpfull helpful
 232                       hierachy hierarchy
 233                       hierarchie hierarchy
 234                       howver however
 235                       implemantation implementation
 236                       incomming incoming
 237                       incompatabilities incompatibilities
 238                       incompatable incompatible
 239                       inconsistant inconsistent
 240                       indendation indentation
 241                       indended intended
 242                       independant independent
 243                       informatiom information
 244                       infromation information
 245                       initalize initialize
 246                       initators initiators
 247                       initializiation initialization
 248                       inofficial unofficial
 249                       integreated integrated
 250                       integrety integrity
 251                       integrey integrity
 252                       intendet intended
 253                       interchangable interchangeable
 254                       intermittant intermittent
 255                       interupted interrupted
 256                       jave java
 257                       langage language
 258                       langauage language
 259                       langugage language
 260                       lauch launch
 261                       lenght length
 262                       lesstiff lesstif
 263                       libaries libraries
 264                       libary library
 265                       libraris libraries
 266                       licenceing licencing
 267                       loggging logging
 268                       loggin login
 269                       logile logfile
 270                       machinary machinery
 271                       maintainance maintenance
 272                       maintainence maintenance
 273                       makeing making
 274                       managable manageable
 275                       manoeuvering maneuvering
 276                       mathimatical mathematical
 277                       mathimatic mathematic
 278                       mathimatics mathematics
 279                       ment meant
 280                       messsages messages
 281                       microprocesspr microprocessor
 282                       milliseonds milliseconds
 283                       miscelleneous miscellaneous
 284                       misformed malformed
 285                       mispelled misspelled
 286                       mmnemonic mnemonic
 287                       modulues modules
 288                       monochorome monochrome
 289                       monochromo monochrome
 290                       monocrome monochrome
 291                       mroe more
 292                       multidimensionnal multidimensional
 293                       navagating navigating
 294                       nead need
 295                       neccesary necessary
 296                       neccessary necessary
 297                       neccessary necessary
 298                       necesary necessary
 299                       negotation negotiation
 300                       nescessary necessary
 301                       nessessary necessary
 302                       noticable noticeable
 303                       notications notifications
 304                       o'caml OCaml
 305                       omitt omit
 306                       ommitted omitted
 307                       optionnal optional
 308                       optmizations optimizations
 309                       orientatied orientated
 310                       orientied oriented
 311                       overaall overall
 312                       pacakge package
 313                       pachage package
 314                       packacge package
 315                       packege package
 316                       packge package
 317                       pakage package
 318                       paramameters parameters
 319                       parameterize parametrize
 320                       paramter parameter
 321                       paramters parameters
 322                       particularily particularly
 323                       pased passed
 324                       peprocessor preprocessor
 325                       perfoming performing
 326                       permissons permissions
 327                       persistant persistent
 328                       plattform platform
 329                       ploting plotting
 330                       posible possible
 331                       postgressql PostgreSQL
 332                       powerfull powerful
 333                       preceeded preceded
 334                       preceeding preceding
 335                       precission precision
 336                       prefered preferred
 337                       prefferably preferably
 338                       prepaired prepared
 339                       primative primitive
 340                       princliple principle
 341                       priorty priority
 342                       proccesors processors
 343                       proces process
 344                       processessing processing
 345                       processpr processor
 346                       processsing processing
 347                       progams programs
 348                       programers programmers
 349                       programm program
 350                       programms programs
 351                       promps prompts
 352                       pronnounced pronounced
 353                       prononciation pronunciation
 354                       pronouce pronounce
 355                       pronunce pronounce
 356                       propery property
 357                       prosess process
 358                       protable portable
 359                       protcol protocol
 360                       protecion protection
 361                       protocoll protocol
 362                       psychadelic psychedelic
 363                       quering querying
 364                       recieved received
 365                       recieved received
 366                       recieve receive
 367                       recieve receive
 368                       reciever receiver
 369                       recognizeable recognizable
 370                       recommanded recommended
 371                       redircet redirect
 372                       redirectrion redirection
 373                       reenabled re-enabled
 374                       reenable re-enable
 375                       reencode re-encode
 376                       refence reference
 377                       registerd registered
 378                       registraration registration
 379                       regulamentations regulations
 380                       remoote remote
 381                       removeable removable
 382                       repectively respectively
 383                       replacments replacements
 384                       requiere require
 385                       requred required
 386                       resizeable resizable
 387                       ressize resize
 388                       ressource resource
 389                       retransmited retransmitted
 390                       runnning running
 391                       safly safely
 392                       savable saveable
 393                       searchs searches
 394                       secund second
 395                       separatly separately
 396                       sepcify specify
 397                       seperated separated
 398                       seperated separated
 399                       seperately separately
 400                       seperate separate
 401                       seperate separate
 402                       seperatly separately
 403                       seperator separator
 404                       sequencial sequential
 405                       serveral several
 406                       setts sets
 407                       similiar similar
 408                       simliar similar
 409                       speach speech
 410                       speciefied specified
 411                       specifed specified
 412                       specificaton specification
 413                       specifing specifying
 414                       speficied specified
 415                       speling spelling
 416                       splitted split
 417                       staically statically
 418                       standart standard
 419                       staticly statically
 420                       subdirectoires subdirectories
 421                       succesfully successfully
 422                       succesful successful
 423                       superceded superseded
 424                       superflous superfluous
 425                       superseeded superseded
 426                       suplied supplied
 427                       suport support
 428                       suppored supported
 429                       supportin supporting
 430                       suppoted supported
 431                       suppported supported
 432                       suppport support
 433                       suspicously suspiciously
 434                       synax syntax
 435                       synchonized synchronized
 436                       syncronize synchronize
 437                       syncronize synchronize
 438                       syncronizing synchronizing
 439                       syncronus synchronous
 440                       syste system
 441                       sythesis synthesis
 442                       taht that
 443                       throught through
 444                       transfering transferring
 445                       trasmission transmission
 446                       treshold threshold
 447                       trigerring triggering
 448                       unexecpted unexpected
 449                       unfortunatelly unfortunately
 450                       unknonw unknown
 451                       unuseful useless
 452                       useable usable
 453                       usefull useful
 454                       usera users
 455                       usetnet Usenet
 456                       utilites utilities
 457                       utillities utilities
 458                       utilties utilities
 459                       utiltity utility
 460                       utitlty utility
 461                       variantions variations
 462                       varient variant
 463                       verbse verbose
 464                       verisons versions
 465                       verison version
 466                       verson version
 467                       vicefersa vice-versa
 468                       wheter whether
 469                       wierd weird
 470                       xwindows X
 471                       yur your
 472                      );
 473
 474 # The format above doesn't allow spaces.
 475 $CORRECTIONS{'alot'} = 'a lot';
 476
 477 # Picky corrections, applied before lowercasing the word.  These are only
 478 # applied to things known to be entirely English text, such as package
 479 # descriptions, and should not be applied to files that may contain
 480 # configuration fragments or more informal files such as debian/copyright.
 481 our %CORRECTIONS_CASE = qw(
 482                            apache Apache
 483                            api API
 484                            Api API
 485                            D-BUS D-Bus
 486                            d-bus D-Bus
 487                            dbus D-Bus
 488                            debian Debian
 489                            english English
 490                            french French
 491                            EMacs Emacs
 492                            Gconf GConf
 493                            gconf GConf
 494                            german German
 495                            Gnome GNOME
 496                            gnome GNOME
 497                            Gnome-VFS GnomeVFS
 498                            Gnome-Vfs GnomeVFS
 499                            GnomeVfs GnomeVFS
 500                            gnome-vfs GnomeVFS
 501                            gnomevfs GnomeVFS
 502                            gnu GNU
 503                            Gnu GNU
 504                            Gobject GObject
 505                            gobject GObject
 506                            Gstreamer GStreamer
 507                            gstreamer GStreamer
 508                            GTK GTK+
 509                            gtk+ GTK+
 510                            Http HTTP
 511                            kde KDE
 512                            meta-package metapackage
 513                            MYSQL MySQL
 514                            Mysql MySQL
 515                            mysql MySQL
 516                            linux Linux
 517                            Latex LaTeX
 518                            latex LaTeX
 519                            OCAML OCaml
 520                            Ocaml OCaml
 521                            ocaml OCaml
 522                            OpenLdap OpenLDAP
 523                            Openldap OpenLDAP
 524                            openldap OpenLDAP
 525                            Postgresql PostgreSQL
 526                            postgresql PostgreSQL
 527                            python Python
 528                            russian Russian
 529                            SkoleLinux Skolelinux
 530                            skolelinux Skolelinux
 531                            SLang S-Lang
 532                            S-lang S-Lang
 533                            s-lang S-Lang
 534                            spanish Spanish
 535                            subversion Subversion
 536                            TCL Tcl
 537                            tcl Tcl
 538                            TEX TeX
 539                            Tex TeX
 540                            TeTeX teTeX
 541                            Tetex teTeX
 542                            tetex teTeX
 543                            TK Tk
 544                            tk Tk
 545                            Xemacs XEmacs
 546                            XEMacs XEmacs
 547                            XFCE Xfce
 548                            XFce Xfce
 549                            xfce Xfce
 550                           );
 551
 552 # The format above doesn't allow spaces.
 553 $CORRECTIONS_CASE{'Debian-Edu'} = 'Debian Edu';
 554 $CORRECTIONS_CASE{'debian-edu'} = 'Debian Edu';
 555 $CORRECTIONS_CASE{'TeXLive'} = 'TeX Live';
 556 $CORRECTIONS_CASE{'TeX-Live'} = 'TeX Live';
 557 $CORRECTIONS_CASE{'TeXlive'} = 'TeX Live';
 558 $CORRECTIONS_CASE{'TeX-live'} = 'TeX Live';
 559 $CORRECTIONS_CASE{'texlive'} = 'TeX Live';
 560 $CORRECTIONS_CASE{'tex-live'} = 'TeX Live';
 561
 562 # -----------------------------------
 563
 564 sub _tag {
 565     my @args = grep { defined($_) } @_;
 566     tag(@args);
 567 }
 568
 569 # Check spelling of $text and report the tag $tag if we find anything.
 570 # $filename, if included, is given as the first argument to the tag.  If it's
 571 # not defined, it will be omitted.
 572 sub spelling_check {
 573     my ($tag, $text, $filename) = @_;
 574     return unless $text;
 575
 576     $text = lc $text;
 577     $text =~ s/[.,;:?!()[\]]//g;
 578
 579     for my $word (split(/\s+/, $text)) {
 580         if (exists $CORRECTIONS{$word}) {
 581             _tag($tag, $filename, $word, $CORRECTIONS{$word});
 582         }
 583     }
 584
 585     # Special case for correcting a multi-word string.
 586     if ($text =~ m,debian/gnu\s+linux,) {
 587         _tag($tag, $filename, "Debian/GNU Linux", "Debian GNU/Linux");
 588     }
 589 }
 590
 591 # Check spelling of $text against pickier corrections, such as common
 592 # capitalization mistakes.  This check is separate from spelling_check since
 593 # it isn't appropriate for some files (such as changelog).  Takes $text to
 594 # check spelling in and $tag to report if we find anything.  $filename, if
 595 # included, is given as the first argument to the tag.  If it's not defined,
 596 # it will be omitted.
 597 sub spelling_check_picky {
 598     my ($tag, $text, $filename) = @_;
 599
 600     # Check this first in case it's contained in square brackets and
 601     # removed below.
 602     if ($text =~ m,meta\s+package,) {
 603         _tag($tag, $filename, "meta package", "metapackage");
 604     }
 605
 606     # Exclude text enclosed in square brackets as it could be a package list
 607     # or similar which may legitimately contain lower-cased versions of
 608     # the words.
 609     $text =~ s/\[.+?\]//sg;
 610     for my $word (split(/\s+/, $text)) {
 611         $word =~ s/^\(|[).,?!:;]+$//g;
 612         if (exists $CORRECTIONS_CASE{$word}) {
 613             _tag($tag, $filename, $word, $CORRECTIONS_CASE{$word});
 614             next;
 615         }
 616     }
 617 }
 618
 619 1;
 620
 621 # Local Variables:
 622 # indent-tabs-mode: nil
 623 # cperl-indent-level: 4
 624 # End:
 625 # vim: syntax=perl sw=4 sts=4 ts=4 et shiftround