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