newspeak.l

   1 %{      -*- Mode:Fundamental -*-
   2 /*
   3  *      newspeak.l, version 1.1.
   4  *      Lex filter to transform plain English into NewSpeak.
   5  *      Copyright (c) 1991 Jamie Zawinski <jwz@jwz.org>.
   6  *
   7  *      Permission to use, copy, modify, distribute, and sell this
   8  *      software and its documentation for any purpose is hereby granted
   9  *      without fee, provided that the above copyright notice appear in
  10  *      all copies and that both that copyright notice and this
  11  *      permission notice appear in supporting documentation.  No
  12  *      representations are made about the suitability of this software
  13  *      for any purpose.  It is provided "as is" without express or
  14  *      implied warranty.
  15  *
  16  *      To compile:
  17  *
  18  *      flex newspeak.l
  19  *      cc -O -o newspeak lex.yy.c
  20  *      rm lex.yy.c
  21  *
  22  *      This must be compiled with "flex", not normal "lex".  Lex has
  23  *      some builtin, unchangeable limits which this program exceeds.
  24  *      This has been verified to work with flex version 2.3.7, and
  25  *      may not work with flex version 2.3.6.
  26  *
  27  *      You can get 'flex' by anonymous FTP to prep.ai.mit.edu, or
  28  *      anywhere else you get GNU software.
  29  *
  30  *      Run this like so:  "newspeak < some-text-file | more"
  31  *
  32  *      There are some design notes near the end.  Suggestions and
  33  *      improvements to this code are more than welcome.
  34  *      Hail Big Brother!
  35  *---------------------------------------------------------------------------
  36  */
  37 #include <stdio.h>
  38
  39 unsigned int fcounter = 0;
  40
  41 #define Y0      yytext[0]
  42 #define Y1      yytext[1]
  43 #define Y2      yytext[2]
  44 #define Y3      yytext[3]
  45 #define C       caseify
  46
  47 #define CAP(string)     capstring(0, (string),0)
  48 #define WCAP(string)    capstring(Y0,(string),1)
  49
  50 #define COMP(string1,string2)   compstring(0, (string1),(string2),0)
  51 #define WCOMP(string1,string2)  compstring(Y0,(string1),(string2),1)
  52
  53 #define DUMP()                  printf("%s",yytext)
  54
  55 void capstring(char, char*, int);
  56 void compstring(char, char*, char*, int);
  57 char _caseify(char, int);
  58 char caseify(char);
  59
  60 %}
  61
  62 %e 8000
  63 %p 9000
  64 %n 2000
  65 %k 3000
  66 %a 5000
  67 %o 4000
  68
  69 W       (([     ]?\n[   ]*)|[   ]|[\"'`])
  70 ES      [.,;:!?/]
  71 EW      ({W}|{ES}|"'")
  72
  73 YOUNG   (([Yy]oung|[Ss]mall|[Ll]ittle){W})
  74 DEAD    (([Dd]ead|[Dd]eceased){W})
  75 THE     ([Tt]he{W}?)
  76 ANDOR   ({W}(([Aa]nd)|([Oo]r)){W}?)
  77 COLOR   (red|green|blue|yellow|cyan|magenta|purple|orange|mauve|pink|turquoise|brown|gr[ae]y)
  78
  79 %%
  80
  81 %{
  82         /******************************
  83          *          PLUSwords         *
  84          ******************************/
  85 %}
  86
  87 [Qq]uite{EW}/[A-Za-z][A-Za-z]                   CAP("plus");
  88 [Rr]ather{EW}/[A-Za-z][A-Za-z][A-Za-z]          CAP("plus");
  89 [Kk]ind{EW}of{EW}/[A-Za-z][A-Za-z][A-Za-z]      CAP("plus");
  90 [Kk]inda{EW}/[A-Za-z][A-Za-z][A-Za-z]           CAP("plus");
  91 [Mm]ore{W}than{W}a{W}(little|bit){W}            CAP("plus");
  92 [Pp]ro-                                         CAP("plus");
  93
  94 [Hh]undreds{W}of{W}[Tt]housands                 |
  95 [Hh]undreds{ANDOR}[Tt]housands                  |
  96 [Hh]undreds{W}if{W}not{W}[Tt]housands           |
  97 [Tt]housands                                    |
  98 [Mm]illions                                     CAP("doubleplusmany");
  99
 100 [Dd]ozens                                       CAP("many");
 101 [Hh]undreds                                     CAP("plusmany");
 102
 103 ([Bb]right|[Ll]ight|[Ii]ntense){W}/{COLOR}      CAP("plus");
 104 ([Dd]im|[Ff]aded|[Dd]ark|[Pp]ale){W}/{COLOR}    CAP("plusun");
 105 ([Dd]im|[Ff]aded|[Dd]ark|[Pp]ale)               CAP("unlight");
 106
 107 [Ee]very                                        DUMP();
 108 [Vv]ery{W}                                      |
 109 [Rr]eally{W}                                    |
 110 [Tt]erribly{W}                                  |
 111 [Aa]wesome({W})?                                |
 112 [Aa]wfully{W}                                   CAP("doubleplus");
 113
 114 [Ww]hopping{EW}                                 CAP("plusbig");
 115
 116 "O.K."                                          |
 117 [Aa]ll({W})?[Rr]ight                            |
 118 [Oo][Kk][Aa][Yy]                                CAP("plusgood");
 119 {W}OK/{W}                                       WCAP("plusgood");
 120
 121 ([Tt]oo|[Oo]verly|[Tt]hat){W}[Mm]uch            CAP("plusmuch");
 122
 123 {W}[Bb]ad/{EW}          WCAP("ungood");
 124 {W}[Pp]oor/{EW}         WCAP("ungood");
 125 {W}[Ll]ame/{EW}         WCAP("ungood");
 126 {W}[Pp]itiful/{EW}      WCAP("ungood");
 127 {W}[Nn]asty/{EW}        WCAP("plusungood");
 128 {W}[Hh]orrid/{EW}       WCAP("doubleplus ungood");
 129 {W}[Hh]orrible/{EW}     WCAP("doubleplus ungood");
 130 {W}[Aa]wful/{W}         WCAP("doubleplus ungood");
 131 {W}[Ee]vil/{W}          WCAP("doubleplus ungood");
 132
 133 %{
 134         /******************************
 135          *           Titles           *
 136          ******************************/
 137 %}
 138
 139 {W}[Ss]ir/{EW}          WCAP("citizen");
 140 {W}[Mm]r"."/{EW}        WCAP("brother");
 141 [Mm]ister/{EW}          CAP("brother");
 142 [Mm]adame?              CAP("sister");
 143 {W}[Mm]iss/{EW}         WCAP("sister");
 144 [Mm]a"'"?am/{EW}        CAP("sister");
 145 {W}[Mm]r?s"."/{EW}      WCAP("sister");
 146 Mrs/{EW}                CAP("sister");
 147
 148 {YOUNG}?[Cc]hildren                             CAP("young citizens");
 149 {YOUNG}?[Bb]oys{ANDOR}[Gg]irl/s                 CAP("young citizens");
 150 {YOUNG}?([Kk]id|[Gg]irl|[Bb]oy|[Cc]hild)/{EW}   CAP("young citizen");
 151
 152 [Ff]ellow               CAP("citizen");
 153
 154 [Nn]on{W}?"-"?citizen   CAP("unperson");
 155 [Nn]on{W}?"-"?member    CAP("unperson");
 156 [Cc]riminal/s?          CAP("unperson");
 157 {DEAD}(man|woman)       CAP("unperson");
 158 {DEAD}(men|women)       CAP("unpersons");
 159
 160 [Ii]n{W}[Pp]erson       DUMP();
 161
 162 {W}[Uu]ser              WCOMP("party ","worker");
 163 [Ss]tudent              COMP("party ","worker");
 164 [Cc]itizen/s?{EW}       COMP("party ","worker");
 165 [Pp]erson/s?{EW}        COMP("party ","worker");
 166 [Pp]eople               COMP("party ","workers");
 167
 168 [Ss]enator              |
 169 [Cc]ongressman          |
 170 [Ss]upervisor           |
 171 [Pp]rofessor            printf("Inner Party Member");
 172 [Pp]rof"."/{EW}         printf("Inner Party Member");
 173 [Pp]rof/{EW}            printf("Inner Party Member");
 174
 175 Representative/s?       printf("Inner Party Member");
 176 representatives         printf("Inner Party Members");
 177
 178 [Ww]hite{W}[Cc]ollar            |
 179 [Uu]pper{W}[Cc]lass             COMP("inner ","party");
 180 [Mm]iddle{W}[Cc]lass            CAP("party");
 181 [Bb]lue{W}[Cc]ollar             |
 182 [Ww]orking{W}[Cc]lass           |
 183 [Ll]ower{W}[Cc]lass             CAP("prole");
 184 ([Ff]ool|[Ii]diot)/s?{EW}       CAP("prole");
 185 [Ss]tupidity                    CAP("proleness");
 186
 187 %{
 188         /******************************
 189          *        Organizations       *
 190          ******************************/
 191 %}
 192
 193 [Aa]?{W}([Ww]hite{W}[Hh]ouse|[Gg]ovt\.?|[Gg]overnment){W}([Ss]ource|[Oo]fficial|[Ss]pokes(man|woman|person))    CAP("an Inner Party Member");
 194 {THE}?[Rr]epublican{W}[Pp]arty          COMP("mini","luv");
 195 {THE}?[Dd]emocratic{W}[Pp]arty          COMP("mini","plenty");
 196
 197 {THE}?Congress                                  printf("MiniPax");
 198 {THE}?[Ss]enate                                 printf("MiniPax");
 199 {THE}?[Hh]ouse{W}[Oo]f{W}[Rr]epresentatives     printf("MiniPax");
 200 {THE}?[Ss]tate{W}[Dd]epartment                  printf("MiniPax");
 201 {THE}?[Ss]tate{W}[Dd]ept"."?                    printf("MiniPax");
 202 {THE}?[Dd]efen[cs]e{W}[Dd]epartment             |
 203 {THE}?[Dd]efen[cs]e{W}[Dd]ept"."?               |
 204 {THE}?[Ww]ar{W}[Dd]epartment                    |
 205 {THE}?[Ww]ar{W}[Dd]ept"."?                      |
 206 {THE}?[Hh]ouse{W}of{W}[Cc]ommons                |
 207 {THE}?Pentagon                                  |
 208 {THE}?[Ff]eds                                   |
 209 {THE}?FCC                                       |
 210 {THE}?D[Oo]D                                    |
 211 {THE}"D."[Oo]".D."                              |
 212 {THE}?[Ss]ecret{W}[Ss]ervice                    COMP("mini","luv");
 213 {THE}?White{W}House                             |
 214 {THE}?Kremlin                                   printf("MiniTrue");
 215 {THE}?(CIA|NSA|FBI|MI"-"?5)/{EW}                printf("MiniTrue");
 216 {THE}?("C.I.A."|"N.S.A."|"F.B.I.")/{EW}         printf("MiniTrue");
 217 {THE}?[Aa]rchive/s?                             COMP("mini","rec");
 218 {THE}?[Ll]ibrary                                COMP("mini","rec");
 219 {THE}?[Ll]ibraries                              COMP("mini","recs");
 220
 221 [Tt]hought{W}[Pp]olice|[Nn]azis?                COMP("think","pol");
 222 [Vv]ice{W}[Ss]quad                              COMP("sex","pol");
 223 PMRC|"P.M.R.C."                                 COMP("sex","pol");
 224
 225 [Oo]fficer                                      CAP("minister");
 226
 227 {THE}?[Dd]epartment{EW}of{EW}.                  |
 228 {THE}?[Dd]ept"."?{EW}of{EW}.                    |
 229 {THE}?[Uu]niversity{EW}of{EW}.                  |
 230 {THE}?[Uu]niv"."?{EW}of{EW}.                    |
 231 {THE}?[Dd]ept"."?{EW}of{EW}.                    |
 232 {THE}?([Ss]ub"-"?)?[Cc]omm?itt?ee{EW}(of|on){EW}.       |
 233 {THE}?[Ss]chool{EW}of{EW}.                      {
 234         if ((yytext[yyleng-1] >= 'A') && (yytext[yyleng-1] <= 'Z'))
 235            /* "the school of Law" --> "MiniLaw" */
 236            printf("Mini%c",yytext[yyleng-1]);
 237         else if ((yytext[yyleng-1] >= 'a') && (yytext[yyleng-1] <= 'z'))
 238            /* "the school of law" --> "MiniLaw" (not "Minilaw") */
 239            printf("Mini%c",yytext[yyleng-1] - ('a' - 'A'));
 240            /* "the school of 5 things" --> "Ministry of 5 things" */
 241         else printf("Ministry of %c",yytext[yyleng-1]);
 242         }
 243
 244 [Dd]epartment                   |
 245 [Uu]niversity                   CAP("ministry");
 246 [Uu]niv"."?/{W}                 CAP("ministry");
 247 [Dd]ept"."?/{W}                 CAP("ministry");
 248 ([Ss]ub"-"?)?[Cc]omm?itt?ee     CAP("ministry");
 249
 250 {THE}[Pp]roject/{EW}            CAP("the Three Year Plan");
 251 [Oo]ur{W}[Pp]roject/{EW}        CAP("our Three Year Plan");
 252 [Bb]udget                       printf("Three Year Plan");
 253 [Pp]roject/{ES}                 printf("Three Year Plan");
 254
 255 {W}({THE}|([aa]{W}))[Pp]roject  printf("%cthe Three Year Plan",Y0);
 256
 257 [A-Za-z]+"'"[Ss]/{W}(law|Law|LAW|book|Book|BOOK|rule|Rule|RULE){EW}     printf("Goldstein's");
 258
 259 %{
 260         /******************************
 261          *           Actions          *
 262          ******************************/
 263 %}
 264
 265 [Ii]n{W}love{EW}        CAP("SexCriming");
 266 [Ll]ove{W}you/{EW}      CAP("love Big Brother");
 267 [Ll]ove{W}me/{EW}       CAP("love Big Brother");
 268
 269 [Cc]loning              |
 270 [Rr]eproduction         |
 271 [Cc]elibacy             |
 272 [Pp]rocreation          COMP("good","sex");
 273
 274 [Cc]elibate             |
 275 [Pp]rocreate            COMP("good","sexwise");
 276
 277 [Tt]elevisions?         |
 278 TVs?                    |
 279 [Tt]"."[Vv]"."s?        |
 280 [Rr]adios?              |
 281 [Nn]ews{W}?[Pp]apers?   |
 282 [Jj]ournalism           |
 283 [Mm]ovies?              |
 284 [Rr]ock{EW}?"-"?(and|"&"|"'"?n"'"?){EW}?"-"?[Rr]oll({W}[Mm]usic)?            |
 285 (([Rr]ock|[Cc]lassical|[Ii]ndustrial|[Pp]op|[Dd]ance|[Rr]ap){W})?[Mm]usic    |
 286 [Tt]unes                |
 287 [Mm]oney                |
 288 [Cc]ash                 |
 289 [Cc]omic{W}[Bb]ooks?    |
 290 ([Ss]tar{W}?)?[Tt]rek   COMP("prole","feed");
 291
 292 [Pp]eace{W}[Mm]ovement          |
 293 [Pp]eace{W}[Pp]rotest           |
 294 [Aa]nti{EW}[Ww]ar               |
 295 ([Pp]assive{W})?[Rr]esistance   |
 296 [Cc]reativity                   |
 297 [Tt]reason                      |
 298 [Rr]esearch             COMP("crime","think");
 299 %{
 300         /******************************
 301          *           Religion         *
 302          ******************************/
 303 %}
 304
 305 [Jj]esus{W}[Cc]hrist            |
 306 [Jj]esus                        |
 307 {THE}?[Bb]uddh?a                |
 308 [Mm]ohamm?ed                    |
 309 [Mm]artin{W}[Ll]uther{W}[Kk]ing |
 310 J\.?\ ?R\.?\ \"?Bob\"?\ Dobbs   printf("doubleplus crimethinker");
 311
 312 ([Jj]esse{W})?[Hh]elms          |
 313 ([RrDd]on(ald)?{W})?[Rr]ea?gan  |
 314 [Gg]eorge{W}[Gg]uscoria         printf("doubleplus goodthinker");
 315
 316 [Jj]ewish                       COMP("crime","thinkwise");
 317 [Jj]ew                          |
 318 [Cc]hristian                    |
 319 [Mm]oslem                       |
 320 [Bb]uddhist                     |
 321 [Aa]thiest                      |
 322 [Aa]gnostic                     COMP("crime","thinker");
 323
 324 [Ff]aith                        COMP("belly","feel");
 325
 326 %{
 327         /******************************
 328          *            Places          *
 329          ******************************/
 330 %}
 331
 332 [Ee]ngland|{THE}?[Uu]nited{W}[Kk]ingdom                 |
 333 ({THE}?[Uu]nited{W}[Ss]tates{W}[Oo]f{W})?[Aa]merica     |
 334 {THE}?[Uu]nited{W}[Ss]tates|USA|"U.S.A."|[Cc]anada      |
 335 [Gg]ermany|[Ii]srael|[Ee]urope                          printf("Oceana");
 336
 337 Iranian|Iraqu?i|Libyan|Russian|African|Egyptian         printf("Eurasian");
 338 Iran|Iraq|Libya|Russia|Africa|Egypt                     |
 339 ([Ss]audi{W})?Arabia|{THE}?Soviet{W}Union               printf("Eurasia");
 340 [Ss]oviet                                               printf("Eurasian");
 341
 342 [Cc]hinese|[Jj]apanese|[Tt]aiwanese                     |
 343 [Pp]hillipino|[Ii]ndian|[Aa]ustralian|[Mm]exican        |
 344 [Nn]icaraguan|[Ss]alvadori?an                           printf("Eastasian");
 345 China|[Jj]apan|[Tt]aiwan|{THE}?[Pp]hillipines|[Ii]ndia  |
 346 [Aa]ustralia|[Mm]exico|[Nn]icaragua|[Ee]l{W}[Ss]alvador printf("Eastasia");
 347
 348 [Kk]uwaiti                      printf("Eurasian");
 349 [Kk]uwait                       printf("The Malabar Front");
 350
 351 %{
 352         /******************************
 353          * Miscellaneous Translations *
 354          ******************************/
 355 %}
 356
 357 {W}[Ff]aster            WCAP("plus speedful");
 358 {W}[Ss]lower            WCAP("plus unspeedful");
 359 {W}[Ff]ast              WCAP("speedful");
 360 {W}[Ss]low              WCAP("unspeedful");
 361
 362 [Mm]odern               CAP("plusnew");
 363 [Aa]ncient              CAP("plusunnew");
 364 {W}old/{W}              WCAP("plusunnew");
 365
 366 [Hh]appiness            CAP("joyfulness");
 367 [Hh]appy                CAP("joyful");
 368 [Qq]uick                CAP("speedful");
 369 {W}[Ss]peedy            WCAP("speedful");
 370 [Hh]eavy                CAP("weightful");
 371 [Hh]eavill?y            CAP("weightfully");
 372 [Ss]ick(ly)?            CAP("unhealthful");
 373
 374 [Gg]ross                |
 375 [Ss]ickening            |
 376 [Ff]oul                 |
 377 [Pp]utrid               |
 378 [Dd]isgusting           COMP("crime","thinkful");
 379
 380 [Ss]mash                |
 381 [Cc]rush                |
 382 [Oo]bliterate           |
 383 [Aa]nnihilate           |
 384 [Nn]eutralize           |
 385 [Dd]emolish             |
 386 [Dd]estroy              CAP("unbuild");
 387
 388 [Ii]nanimate            CAP("unlifeful");
 389 [Ss]ociety|[Cc]ulture   printf("IngSoc");
 390 [A-Za-z]+isi?m/{EW}     printf("Goldsteinism");
 391 [A-Za-z]+ist/{EW}       printf("Goldsteinist");
 392
 393 {W}[Dd]ead              WCAP("unlifeful");
 394 {W}[Dd]eath             WCAP("unlife");
 395 {W}[Ll]ie               WCAP("untruth");
 396 {W}[Ff]alsehood         WCAP("untruth");
 397 {W}[Mm]istake/{EW}      WCAP("untruth");
 398 {W}[Ww]hisper           WCAP("unshout");
 399 {W}[Pp]roud             WCAP("prideful");
 400
 401 [Ff]alse                CAP("untrue");
 402 [Dd]ark                 CAP("unlight");
 403 [Bb]lack                CAP("unwhite");
 404 [Ff]orbidden            CAP("unallowed");
 405 [Ff]orbid               CAP("unallow");
 406 [Ff]ailure              CAP("unsuccess");
 407 [Ff]ail/{EW}            CAP("unwin");
 408
 409 [Ss]tatistics?/{EW}             CAP("propaganda");
 410 {W}[Aa]n{W}[Aa]nn?ouncement     WCAP("a NewsFlash");
 411 [Aa]nn?ouncement                printf("NewsFlash");
 412 [Ii]nstructions?                printf("B. B. DayOrder");
 413
 414 [Aa]lmost|[Nn]early     CAP("within measurable distance of");
 415 [Ff]unny                CAP("humorful");
 416
 417 [Dd]oom                 CAP("unsave");
 418 [Cc]haos                CAP("unorder");
 419 [Cc]haotic              CAP("unorderful");
 420 [Ee]nslaved             CAP("protected");
 421 [Ee]nslave              CAP("protect");
 422 [Dd]angerous            CAP("unsafewise");
 423 [Dd]anger               CAP("unsafe");
 424 ([Bb]lind{W})?[Oo]bedience              COMP("ing","soc");
 425 \"?[Nn]ew{W}[Ww]orld{W}[Oo]rder\"?      printf("IngSoc");
 426
 427 [Pp]rivacy              |
 428 [Ii]ndividuality        COMP("own","life");
 429
 430 IMHO            printf("for the love of Big Brother");
 431
 432 [Ee]motion(al|s)?       |
 433 [Cc]onviction           |
 434 [Bb]elie(f|ve)          |
 435 [Aa]ccept(ance)?        COMP("belly","feel");
 436
 437 [Dd]emocracy                    |
 438 [Ll]iberty                      |
 439 [Ff]reedom                      |
 440 [Jj]ustice                      |
 441 {THE}?[Aa]merican{W}[Ww]ay      |
 442 [Ss]ubversion                   |
 443 [Pp]assion              COMP("crime","think");
 444
 445 [Oo]bscenity            |
 446 [Pp]ornography          |
 447 [Oo]rgasm               |
 448 [Ee]rotica              COMP("sex","crime");
 449 [Ss]exy                 |
 450 [Oo]bscene              |
 451 [Pp]ornographic         |
 452 [Ee]rotic               COMP("sex","crimeful");
 453
 454 [Cc]ritic/s?{W}         COMP("crime","thinker");
 455
 456 [Ii]nfant{W}[Mm]ortality        COMP("inf","mort");
 457
 458 [Ff]amilies                     |
 459 [Pp]arents                      COMP("family ","units");
 460 [Mm]other{ANDOR}[Ff]ather       |
 461 [Bb]rother{ANDOR}[Ss]ister      COMP("family ","unit");
 462 {W}[Pp]arent/s?{EW}             WCOMP("family ","unit");
 463 [Ff]amily                       COMP("family ","unit");
 464
 465 God/{EW}                                        printf("Big Brother");
 466 [Pp]res(ident|".")({W}([Bb]ush|[Rr]eagan))?     printf("Big Brother");
 467 [Pp]rime{W}[Mm]inister                          printf("Big Brother");
 468
 469 ([Gg][Nn][Uu]{W}([Ee]macs|EMACS){W})?[Gg]eneral{W}[Pp]ublic{W}[Ll]icense        printf("NewSpeak Dictionary");
 470
 471 (questioning|murder|ass?ass?ination)/{ES}       printf("interrogation");
 472
 473 [Ss]keptic/{EW}                 CAP("unperson");
 474 [Ss]illy                        CAP("foolhardy");
 475 {W}[A-Za-z][A-Za-z]?illy        DUMP();
 476 [Ss]outhern|[Ss]outherly        CAP("southwise");
 477 [Nn]orthern|[Nn]ortherly        CAP("northwise");
 478 [Ee]astern|[Ee]easterly         CAP("eastwise");
 479 [Ww]estern|[Ww]esterly          CAP("westwise");
 480 [Pl]leasant                     CAP("goodwise");
 481 [Vv]iolent                      CAP("unpeacewise");
 482 [Vv]iolence                     CAP("unpeaceness");
 483 [Ii]ndifference                 CAP("uncarefulness");
 484 [Ii]ndifferent                  CAP("uncareful");
 485 [Bb]elly                        CAP("abdomen");
 486 [Cc]omic                        CAP("humorwise");
 487 {W}[Uu]nless                    WCAP("lest");
 488 usually                         printf("usualwise");
 489
 490 [Gg]uerillas            COMP("party ","workers");
 491
 492 [Ww]ar/{EW}             CAP("engagement");
 493 [Dd]efen[cs]e/{EW}      CAP("peace");
 494 [Ii]nvasion             CAP("liberation");
 495
 496 %{
 497         /******************************
 498          *     Syllable Rewriting     *
 499          ******************************
 500         isn't ___       is un___
 501         not the ___     the un___
 502         not my ___      my un___
 503         anti___         un___           (etc...)
 504         ___cally        ___wise
 505         ___ally         ___wise
 506         ___lly          ___wise
 507         ___ly           ___wise
 508         ___aic          ___wise
 509         ___lic          ___wise
 510         ___nnic         ___wise
 511         <VOWEL>tric     ___wise
 512         ___ic           ___wise
 513         <VOWEL>ous      ___ful
 514         <CONSONANT>ous  ___eful
 515         ___less         un___ful
 516
 517          */
 518 %}
 519
 520 [Ii]sn"'"t{W}my{W}              CAP("is my un");
 521 [Ii]s{W}not{W}my{W}             CAP("is my un");
 522 [Ii]sn"'"t{W}[Tt]he{W}          CAP("is the un");
 523 [Ii]s{W}not{W}[Tt]he{W}         CAP("is the un");
 524 [Ii]sn"'"t{W}[Ii]n{W}[Tt]he{W}  CAP("is in the un");
 525 [Ii]s{W}not{W}[Ii]n{W}[Tt]he{W} CAP("is in the un");
 526 [Ii]t"'"?s{W}not{W}[Tt]he{W}    CAP("it's the un");
 527 [Ii]sn"'"t{W}                   CAP("is un");
 528 [Ii]s{W}not{W}                  CAP("is un");
 529 [Nn]ot{W}[Tt]he{W}              CAP("the un");
 530 [Nn]ot{W}[Mm]y{W}               CAP("my un");
 531 [Nn]ot{W}[Aa]{W}                CAP("an un");
 532 [Nn]ot{W}have{W}                CAP("has un");
 533 [Nn]ot{W}be{W}                  CAP("be un");
 534 [Nn]ot{W}[Oo]nly/{W}            CAP("unonly"); /* avoid "unonwise" */
 535
 536 [Aa]{W}[Nn]ot{W}        |
 537 [Aa]{W}[Nn]on"-"?       printf("%cn%cun",Y0,Y1); /* "a non_" -> "an un_" */
 538
 539 %{
 540 /*      {W}[Nn]ot{W}    |       */
 541 %}
 542 {W}[Ii]l"-"?/[a-z][a-z]         WCAP("un");
 543 {W}[Aa]nti"-"?                  |
 544 {W}[Nn]on"-"?                   WCAP("un");
 545
 546 robably|ventually|[Oo]bvious|[Bb]asic|{W}[Oo]nly|otally |
 547 [Aa]rctic|holic|{EW}ally|{EW}[Aa]pply|{W}[Tt]opic       DUMP();
 548
 549 {W}([Tt]raf|[Pp]aci|[Ss]peci)fi/c{W}    DUMP();
 550 {W}(ma|tra)gi/c{W}                      DUMP();
 551 {W}(pub|cyc|re|fro|gar)li/c{W}          DUMP();
 552 {W}(eth|cli|to)ni/c{W}                  DUMP();
 553 {W}(E|cle|met|cit)ri/c{W}               DUMP();
 554 {W}(ch|ep|tr?op|t|mus|stat|att)i/c{W}   DUMP();
 555 {W}only/{W}                             DUMP();
 556 {W}[Aa]tlantic                          DUMP();
 557
 558 [ \t\n][drstwDRSTW]ally         printf("%c%cally", Y0, Y1);
 559
 560 [a-z]ically/{W}                 printf("%cwise", Y0);
 561 [a-z]ally/{W}                   printf("%cwise", Y0);
 562 [a-z][a-z]lly/{W}               printf("%c%cwise", Y0,Y1);
 563 [a-z][a-z][a-z]ly/{W}           printf("%c%c%cwise", Y0,Y1,Y2);
 564 [a-z]ical/{W}                   printf("%cwise", Y0);
 565
 566 [a-km-qs-z]aic/{EW}             printf("%cwise",  Y0);  /* not laic, raic */
 567 [a-z]lic/{EW}                   printf("%clwise", Y0);
 568 [a-z]nnic/{EW}                  printf("%cnwise", Y0);
 569 [a-z][aeiou]tric/{EW}           printf("%c%ctwise", Y0, Y1);
 570 [a-z]tric/{EW}                  printf("%cwise", Y0);
 571 [a-z]ic/{EW}                    printf("%cwise",  Y0);
 572 [a-z]lly/{EW}                   printf("%cwise", Y0);
 573 [a-z]ly/{EW}                    printf("%cwise", Y0);
 574
 575 [aeiouy][^aeiouy]ous/{EW}       printf("%c%cful",Y0,Y1);
 576 [^aeiouy]ous/{EW}               printf("%ceful",Y0);
 577 [^e]ous/{EW}                    printf("%cful",Y0);
 578
 579 [A-Za-z]+less/{EW}  {   yytext[yyleng-4] = '\0';
 580                         if (((yytext[1] < 'A') || (yytext[1] > 'Z')) &&
 581                             (yytext[0] >= 'A') && (yytext[0] <= 'Z'))
 582                            yytext[0] = yytext[0] - ('a' - 'A');
 583                         printf("%cn%sful",C('u'), yytext);
 584                     }
 585
 586 ".  "           {       printf("%s",yytext);
 587                         fcounter &= 15;
 588                         if (14 == fcounter++) printf("(fnord)  ");
 589                 }
 590
 591 [.,!?]\"([^\n\".!]+[.!])?\n/[\n\t ]     printf("%c  Hail Big Brother!\"%s",Y0,yytext+2);
 592 \"([.,!?][^\n\".!]+[.!])?\n/[\n\t ]     printf("%c  Hail Big Brother!\"%s",Y1,yytext+2);
 593
 594 .                       printf("%s", yytext);
 595 \n                      printf("\n");
 596
 597 %{
 598 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 599  * Brief description of Orwell's NewSpeak:
 600  * ---------------------------------------
 601  * Each word was made to serve as noun, verb, adjective and adverb.  In
 602  * particular, nouns and verbs were made exactly the same -- hence "think"
 603  * replaced "thought" as a noun (viz. "crimethink" = thought crime, "thinkpol"
 604  * = thought police).  Adjectives were formed by adding "-ful" to a noun/verb;
 605  * adverbs were formed by adding "-wise."  Hence "speedful" = fast,
 606  * "speedwise" = quickly, "unspeedwise" = slowly.
 607  *
 608  * The decision on which word should be negated was made fairly randomly;
 609  * "dark" could be "unlight," or "light" could be "undark".  But in all cases
 610  * the most important objective (aside from ideological restriction) was
 611  * euphony; the words had to be easily pronounceable.
 612  *
 613  * Most verb inflections were made regular; "he thinked," "he goed," "he
 614  * eated"; only the auxiliaries and modals (to be, to have, may, shall, will,
 615  * etc.) were allowed to inflect irregularly.  Plurals were also regularized:
 616  * "mans," "oxes," "childs."  [This isn't implemented here.]
 617  *
 618  * There were three sets of words.  The A vocabulary was for everyday use:
 619  * car, man, red, good, etc.  It was restricted to fairly simple words.
 620  *
 621  * The B vocabulary consisted of political/ideological words with very
 622  * complicated connotations.  All of the B words were compound words --
 623  * bellyfeel (blind emotional acceptance of the ideology), oldthink (the way
 624  * of thought before the Revolution), crimethink, Ingsoc, goodsex (intercourse
 625  * solely for the purpose of making babies and with no physical enjoyment on
 626  * the part of the female), sexcrime (any kind of sex but goodsex, including
 627  * sex for its own sake), goodthink (thinking in accordance with Ingsoc), and
 628  * so on.  These words were also subject to the modifications mentioned
 629  * above--hence "goodthinker," "goodthinkful," "goodthinkwise."
 630  *
 631  * The C vocabulary consisted of scientific and technical words (though there
 632  * was no longer any word for "science," any connotation it might have being
 633  * subsumed into "Ingsoc").
 634  *
 635  * Implementing a translator for all of this would be really complicated --
 636  * I'd need rather extensive lists of the "irregular" words (so they could be
 637  * regularized), as well as lists of politically meaningful words (so they
 638  * could be excised or translated into either "goodthink" or "crimethink," as
 639  * appropriate).  Any kind of sexual topic should become "sexcrime" (it being
 640  * unlikely that any talk of sex these days would fit into "goodsex").
 641  *
 642  * Basically, the reason it's hard is that NewSpeak was intended to *decrease*
 643  * the vocabulary, and subsume complicated ideas into politically correct
 644  * words so that you wouldn't have to understand the ideas anymore; you'd just
 645  * have to emit the right words.  So to properly "translate" anything into
 646  * NewSpeak, you have to cut the vocabulary way down.
 647  * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 648  */
 649 %}
 650
 651 %%
 652 int main()
 653 {
 654         yylex();
 655         printf("\nHail Big Brother!\n");
 656 }
 657
 658
 659 void capstring(firstchar,string,i)
 660     char firstchar, *string;
 661     int i;
 662 {
 663         if (firstchar != 0) putchar(firstchar);
 664         putchar( (yytext[i] <= 'Z') ? (string[0] - ' ') : string[0]);
 665         printf("%s",string+1);
 666 }
 667
 668 void compstring(firstchar,string1,string2,i)
 669     char firstchar, *string1, *string2;
 670     int i;
 671 {
 672         capstring(firstchar,string1,i);
 673         capstring(0,string2,i);
 674 }
 675
 676 char _caseify(c,i)
 677     char c;
 678     int i;
 679 {
 680         if (yytext[i] <= 'Z') return (c - ' ');
 681         else return (c);
 682 }
 683
 684 char caseify(c)
 685     char c;
 686 {
 687         return _caseify(c,0);
 688 }