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  *      Permissiom granted for noncommercial use and redistribution.
  17  *      To compile:
  18  *
  19  *      flex newspeak.l
  20  *      cc -O -o newspeak lex.yy.c
  21  *      rm lex.yy.c
  22  *
  23  *      This must be compiled with "flex", not normal "lex".  Lex has
  24  *      some builtin, unchangable limits which this program exceeds.
  25  *      This has been verified to work with flex version 2.3.7, and
  26  *      may not work with flex version 2.3.6.
  27  *
  28  *      You can get 'flex' by anonymous FTP to prep.ai.mit.edu, or
  29  *      anywhere else you get GNU software.
  30  *
  31  *      Run this like so:  "newspeak < some-text-file | more"
  32  *
  33  *      There are some design notes near the end.  Suggestions and
  34  *      improvements to this code are more than welcome.
  35  *      Hail Big Brother!
  36  *---------------------------------------------------------------------------
  37  */
  38 #include <stdio.h>
  39
  40 unsigned int fcounter = 0;
  41
  42 #define Y0      yytext[0]
  43 #define Y1      yytext[1]
  44 #define Y2      yytext[2]
  45 #define Y3      yytext[3]
  46 #define C       caseify
  47 #define C1      caseify1
  48
  49 #define CAP(string)     capstring(0, (string),0)
  50 #define WCAP(string)    capstring(Y0,(string),1)
  51
  52 #define COMP(string1,string2)   compstring(0, (string1),(string2),0)
  53 #define WCOMP(string1,string2)  compstring(Y0,(string1),(string2),1)
  54
  55 #define DUMP()                  printf("%s",yytext)
  56 %}
  57
  58 %e 8000
  59 %p 9000
  60 %n 2000
  61 %k 3000
  62 %a 5000
  63 %o 4000
  64
  65 W       (([     ]?\n[   ]*)|[   ]|[\"'`])
  66 ES      [.,;:!?/]
  67 EW      ({W}|{ES}|"'")
  68
  69 YOUNG   (([Yy]oung|[Ss]mall|[Ll]ittle){W})
  70 DEAD    (([Dd]ead|[Dd]eceased){W})
  71 THE     ([Tt]he{W}?)
  72 ANDOR   ({W}(([Aa]nd)|([Oo]r)){W}?)
  73 COLOR   (red|green|blue|yellow|cyan|magenta|purple|orange|mauve|pink|turquoise|brown|gr[ae]y)
  74
  75 %%
  76
  77 %{
  78         /******************************
  79          *          PLUSwords         *
  80          ******************************/
  81 %}
  82
  83 [Qq]uite{EW}/[A-Za-z][A-Za-z]                   CAP("plus");
  84 [Rr]ather{EW}/[A-Za-z][A-Za-z][A-Za-z]          CAP("plus");
  85 [Kk]ind{EW}of{EW}/[A-Za-z][A-Za-z][A-Za-z]      CAP("plus");
  86 [Kk]inda{EW}/[A-Za-z][A-Za-z][A-Za-z]           CAP("plus");
  87 [Mm]ore{W}than{W}a{W}(little|bit){W}            CAP("plus");
  88 [Pp]ro-                                         CAP("plus");
  89
  90 [Hh]undreds{W}of{W}[Tt]housands                 |
  91 [Hh]undreds{ANDOR}[Tt]housands                  |
  92 [Hh]undreds{W}if{W}not{W}[Tt]housands           |
  93 [Tt]housands                                    |
  94 [Mm]illions                                     CAP("doubleplusmany");
  95
  96 [Dd]ozens                                       CAP("many");
  97 [Hh]undreds                                     CAP("plusmany");
  98
  99 ([Bb]right|[Ll]ight|[Ii]ntense){W}/{COLOR}      CAP("plus");
 100 ([Dd]im|[Ff]aded|[Dd]ark|[Pp]ale){W}/{COLOR}    CAP("plusun");
 101 ([Dd]im|[Ff]aded|[Dd]ark|[Pp]ale)               CAP("unlight");
 102
 103 [Ee]very                                        DUMP();
 104 [Vv]ery{W}                                      |
 105 [Rr]eally{W}                                    |
 106 [Tt]erribly{W}                                  |
 107 [Aa]wesome({W})?                                |
 108 [Aa]wfully{W}                                   CAP("doubleplus");
 109
 110 [Ww]hopping{EW}                                 CAP("plusbig");
 111
 112 "O.K."                                          |
 113 [Aa]ll({W})?[Rr]ight                            |
 114 [Oo][Kk][Aa][Yy]                                CAP("plusgood");
 115 {W}OK/{W}                                       WCAP("plusgood");
 116
 117 ([Tt]oo|[Oo]verly|[Tt]hat){W}[Mm]uch            CAP("plusmuch");
 118
 119 {W}[Bb]ad/{EW}          WCAP("ungood");
 120 {W}[Pp]oor/{EW}         WCAP("ungood");
 121 {W}[Ll]ame/{EW}         WCAP("ungood");
 122 {W}[Pp]itiful/{EW}      WCAP("ungood");
 123 {W}[Nn]asty/{EW}        WCAP("plusungood");
 124 {W}[Hh]orrid/{EW}       WCAP("doubleplus ungood");
 125 {W}[Hh]orrible/{EW}     WCAP("doubleplus ungood");
 126 {W}[Aa]wful/{W}         WCAP("doubleplus ungood");
 127 {W}[Ee]vil/{W}          WCAP("doubleplus ungood");
 128
 129 %{
 130         /******************************
 131          *           Titles           *
 132          ******************************/
 133 %}
 134
 135 {W}[Ss]ir/{EW}          WCAP("citizen");
 136 {W}[Mm]r"."/{EW}        WCAP("brother");
 137 [Mm]ister/{EW}          CAP("brother");
 138 [Mm]adame?              CAP("sister");
 139 {W}[Mm]iss/{EW}         WCAP("sister");
 140 [Mm]a"'"?am/{EW}        CAP("sister");
 141 {W}[Mm]r?s"."/{EW}      WCAP("sister");
 142 Mrs/{EW}                CAP("sister");
 143
 144 {YOUNG}?[Cc]hildren                             CAP("young citizens");
 145 {YOUNG}?[Bb]oys{ANDOR}[Gg]irl/s                 CAP("young citizens");
 146 {YOUNG}?([Kk]id|[Gg]irl|[Bb]oy|[Cc]hild)/{EW}   CAP("young citizen");
 147
 148 [Ff]ellow               CAP("citizen");
 149
 150 [Nn]on{W}?"-"?citizen   CAP("unperson");
 151 [Nn]on{W}?"-"?member    CAP("unperson");
 152 [Cc]riminal/s?          CAP("unperson");
 153 {DEAD}(man|woman)       CAP("unperson");
 154 {DEAD}(men|women)       CAP("unpersons");
 155
 156 [Ii]n{W}[Pp]erson       DUMP();
 157
 158 {W}[Uu]ser              WCOMP("party ","worker");
 159 [Ss]tudent              COMP("party ","worker");
 160 [Cc]itizen/s?{EW}       COMP("party ","worker");
 161 [Pp]erson/s?{EW}        COMP("party ","worker");
 162 [Pp]eople               COMP("party ","workers");
 163
 164 [Ss]enator              |
 165 [Cc]ongressman          |
 166 [Ss]upervisor           |
 167 [Pp]rofessor            printf("Inner Party Member");
 168 [Pp]rof"."/{EW}         printf("Inner Party Member");
 169 [Pp]rof/{EW}            printf("Inner Party Member");
 170
 171 Representative/s?       printf("Inner Party Member");
 172 representatives         printf("Inner Party Members");
 173
 174 [Ww]hite{W}[Cc]ollar            |
 175 [Uu]pper{W}[Cc]lass             COMP("inner ","party");
 176 [Mm]iddle{W}[Cc]lass            CAP("party");
 177 [Bb]lue{W}[Cc]ollar             |
 178 [Ww]orking{W}[Cc]lass           |
 179 [Ll]ower{W}[Cc]lass             CAP("prole");
 180 ([Ff]ool|[Ii]diot)/s?{EW}       CAP("prole");
 181 [Ss]tupidity                    CAP("proleness");
 182
 183 %{
 184         /******************************
 185          *        Organizations       *
 186          ******************************/
 187 %}
 188
 189 [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");
 190 {THE}?[Rr]epublican{W}[Pp]arty          COMP("mini","luv");
 191 {THE}?[Dd]emocratic{W}[Pp]arty          COMP("mini","plenty");
 192
 193 {THE}?Congress                                  printf("MiniPax");
 194 {THE}?[Ss]enate                                 printf("MiniPax");
 195 {THE}?[Hh]ouse{W}[Oo]f{W}[Rr]epresentatives     printf("MiniPax");
 196 {THE}?[Ss]tate{W}[Dd]epartment                  printf("MiniPax");
 197 {THE}?[Ss]tate{W}[Dd]ept"."?                    printf("MiniPax");
 198 {THE}?[Dd]efen[cs]e{W}[Dd]epartment             |
 199 {THE}?[Dd]efen[cs]e{W}[Dd]ept"."?               |
 200 {THE}?[Ww]ar{W}[Dd]epartment                    |
 201 {THE}?[Ww]ar{W}[Dd]ept"."?                      |
 202 {THE}?[Hh]ouse{W}of{W}[Cc]ommons                |
 203 {THE}?Pentagon                                  |
 204 {THE}?[Ff]eds                                   |
 205 {THE}?FCC                                       |
 206 {THE}?D[Oo]D                                    |
 207 {THE}"D."[Oo]".D."                              |
 208 {THE}?[Ss]ecret{W}[Ss]ervice                    COMP("mini","luv");
 209 {THE}?White{W}House                             |
 210 {THE}?Kremlin                                   printf("MiniTrue");
 211 {THE}?(CIA|NSA|FBI|MI"-"?5)/{EW}                printf("MiniTrue");
 212 {THE}?("C.I.A."|"N.S.A."|"F.B.I.")/{EW}         printf("MiniTrue");
 213 {THE}?[Aa]rchive/s?                             COMP("mini","rec");
 214 {THE}?[Ll]ibrary                                COMP("mini","rec");
 215 {THE}?[Ll]ibraries                              COMP("mini","recs");
 216
 217 [Tt]hought{W}[Pp]olice|[Nn]azis?                COMP("think","pol");
 218 [Vv]ice{W}[Ss]quad                              COMP("sex","pol");
 219 PMRC|"P.M.R.C."                                 COMP("sex","pol");
 220
 221 [Oo]fficer                                      CAP("minister");
 222
 223 {THE}?[Dd]epartment{EW}of{EW}.                  |
 224 {THE}?[Dd]ept"."?{EW}of{EW}.                    |
 225 {THE}?[Uu]niversity{EW}of{EW}.                  |
 226 {THE}?[Uu]niv"."?{EW}of{EW}.                    |
 227 {THE}?[Dd]ept"."?{EW}of{EW}.                    |
 228 {THE}?([Ss]ub"-"?)?[Cc]omm?itt?ee{EW}(of|on){EW}.       |
 229 {THE}?[Ss]chool{EW}of{EW}.                      {
 230         if ((yytext[yyleng-1] >= 'A') && (yytext[yyleng-1] <= 'Z'))
 231            /* "the school of Law" --> "MiniLaw" */
 232            printf("Mini%c",yytext[yyleng-1]);
 233         else if ((yytext[yyleng-1] >= 'a') && (yytext[yyleng-1] <= 'z'))
 234            /* "the school of law" --> "MiniLaw" (not "Minilaw") */
 235            printf("Mini%c",yytext[yyleng-1] - ('a' - 'A'));
 236            /* "the school of 5 things" --> "Ministry of 5 things" */
 237         else printf("Ministry of %c",yytext[yyleng-1]);
 238         }
 239
 240 [Dd]epartment                   |
 241 [Uu]niversity                   CAP("ministry");
 242 [Uu]niv"."?/{W}                 CAP("ministry");
 243 [Dd]ept"."?/{W}                 CAP("ministry");
 244 ([Ss]ub"-"?)?[Cc]omm?itt?ee     CAP("ministry");
 245
 246 {THE}[Pp]roject/{EW}            CAP("the Three Year Plan");
 247 [Oo]ur{W}[Pp]roject/{EW}        CAP("our Three Year Plan");
 248 [Bb]udget                       printf("Three Year Plan");
 249 [Pp]roject/{ES}                 printf("Three Year Plan");
 250
 251 {W}({THE}|([aa]{W}))[Pp]roject  printf("%cthe Three Year Plan",Y0);
 252
 253 [A-Za-z]+"'"[Ss]/{W}(law|Law|LAW|book|Book|BOOK|rule|Rule|RULE){EW}     printf("Goldstein's");
 254
 255 %{
 256         /******************************
 257          *           Actions          *
 258          ******************************/
 259 %}
 260
 261 [Ii]n{W}love{EW}        CAP("SexCriming");
 262 [Ll]ove{W}you/{EW}      CAP("love Big Brother");
 263 [Ll]ove{W}me/{EW}       CAP("love Big Brother");
 264
 265 [Cc]loning              |
 266 [Rr]eproduction         |
 267 [Cc]elibacy             |
 268 [Pp]rocreation          COMP("good","sex");
 269
 270 [Cc]elibate             |
 271 [Pp]rocreate            COMP("good","sexwise");
 272
 273 [Tt]elevisions?         |
 274 TVs?                    |
 275 [Tt]"."[Vv]"."s?        |
 276 [Rr]adios?              |
 277 [Nn]ews{W}?[Pp]apers?   |
 278 [Jj]ournalism           |
 279 [Mm]ovies?              |
 280 [Rr]ock{EW}?"-"?(and|"&"|"'"?n"'"?){EW}?"-"?[Rr]oll({W}[Mm]usic)?            |
 281 (([Rr]ock|[Cc]lassical|[Ii]ndustrial|[Pp]op|[Dd]ance|[Rr]ap){W})?[Mm]usic    |
 282 [Tt]unes                |
 283 [Mm]oney                |
 284 [Cc]ash                 |
 285 [Cc]omic{W}[Bb]ooks?    |
 286 ([Ss]tar{W}?)?[Tt]rek   COMP("prole","feed");
 287
 288 [Pp]eace{W}[Mm]ovement          |
 289 [Pp]eace{W}[Pp]rotest           |
 290 [Aa]nti{EW}[Ww]ar               |
 291 ([Pp]assive{W})?[Rr]esistance   |
 292 [Cc]reativity                   |
 293 [Tt]reason                      |
 294 [Rr]esearch             COMP("crime","think");
 295 %{
 296         /******************************
 297          *           Religion         *
 298          ******************************/
 299 %}
 300
 301 [Jj]esus{W}[Cc]hrist            |
 302 [Jj]esus                        |
 303 {THE}?[Bb]uddh?a                |
 304 [Mm]ohamm?ed                    |
 305 [Mm]artin{W}[Ll]uther{W}[Kk]ing |
 306 J\.?\ ?R\.?\ \"?Bob\"?\ Dobbs   printf("doubleplus crimethinker");
 307
 308 ([Jj]esse{W})?[Hh]elms          |
 309 ([RrDd]on(ald)?{W})?[Rr]ea?gan  |
 310 [Gg]eorge{W}[Gg]uscoria         printf("doubleplus goodthinker");
 311
 312 [Jj]ewish                       COMP("crime","thinkwise");
 313 [Jj]ew                          |
 314 [Cc]hristian                    |
 315 [Mm]oslem                       |
 316 [Bb]uddhist                     |
 317 [Aa]thiest                      |
 318 [Aa]gnostic                     COMP("crime","thinker");
 319
 320 [Ff]aith                        COMP("belly","feel");
 321
 322 %{
 323         /******************************
 324          *            Places          *
 325          ******************************/
 326 %}
 327
 328 [Ee]ngland|{THE}?[Uu]nited{W}[Kk]ingdom                 |
 329 ({THE}?[Uu]nited{W}[Ss]tates{W}[Oo]f{W})?[Aa]merica     |
 330 {THE}?[Uu]nited{W}[Ss]tates|USA|"U.S.A."|[Cc]anada      |
 331 [Gg]ermany|[Ii]srael|[Ee]urope                          printf("Oceana");
 332
 333 Iranian|Iraqu?i|Libyan|Russian|African|Egyptian         printf("Eurasian");
 334 Iran|Iraq|Libya|Russia|Africa|Egypt                     |
 335 ([Ss]audi{W})?Arabia|{THE}?Soviet{W}Union               printf("Eurasia");
 336 [Ss]oviet                                               printf("Eurasian");
 337
 338 [Cc]hinese|[Jj]apanese|[Tt]aiwanese                     |
 339 [Pp]hillipino|[Ii]ndian|[Aa]ustralian|[Mm]exican        |
 340 [Nn]icaraguan|[Ss]alvadori?an                           printf("Eastasian");
 341 China|[Jj]apan|[Tt]aiwan|{THE}?[Pp]hillipines|[Ii]ndia  |
 342 [Aa]ustralia|[Mm]exico|[Nn]icaragua|[Ee]l{W}[Ss]alvador printf("Eastasia");
 343
 344 [Kk]uwaiti                      printf("Eurasian");
 345 [Kk]uwait                       printf("The Malabar Front");
 346
 347 %{
 348         /******************************
 349          * Miscelaneous Translations  *
 350          ******************************/
 351 %}
 352
 353 {W}[Ff]aster            WCAP("plus speedful");
 354 {W}[Ss]lower            WCAP("plus unspeedful");
 355 {W}[Ff]ast              WCAP("speedful");
 356 {W}[Ss]low              WCAP("unspeedful");
 357
 358 [Mm]odern               CAP("plusnew");
 359 [Aa]ncient              CAP("plusunnew");
 360 {W}old/{W}              WCAP("plusunnew");
 361
 362 [Hh]appiness            CAP("joyfulness");
 363 [Hh]appy                CAP("joyful");
 364 [Qq]uick                CAP("speedful");
 365 {W}[Ss]peedy            WCAP("speedful");
 366 [Hh]eavy                CAP("weightful");
 367 [Hh]eavill?y            CAP("weightfully");
 368 [Ss]ick(ly)?            CAP("unhealthful");
 369
 370 [Gg]ross                |
 371 [Ss]ickening            |
 372 [Ff]oul                 |
 373 [Pp]utrid               |
 374 [Dd]isgusting           COMP("crime","thinkful");
 375
 376 [Ss]mash                |
 377 [Cc]rush                |
 378 [Oo]bliterate           |
 379 [Aa]nnihilate           |
 380 [Nn]eutralize           |
 381 [Dd]emolish             |
 382 [Dd]estroy              CAP("unbuild");
 383
 384 [Ii]nanimate            CAP("unlifeful");
 385 [Ss]ociety|[Cc]ulture   printf("IngSoc");
 386 [A-Za-z]+isi?m/{EW}     printf("Goldsteinism");
 387 [A-Za-z]+ist/{EW}       printf("Goldsteinist");
 388
 389 {W}[Dd]ead              WCAP("unlifeful");
 390 {W}[Dd]eath             WCAP("unlife");
 391 {W}[Ll]ie               WCAP("untruth");
 392 {W}[Ff]alsehood         WCAP("untruth");
 393 {W}[Mm]istake/{EW}      WCAP("untruth");
 394 {W}[Ww]hisper           WCAP("unshout");
 395 {W}[Pp]roud             WCAP("prideful");
 396
 397 [Ff]alse                CAP("untrue");
 398 [Dd]ark                 CAP("unlight");
 399 [Bb]lack                CAP("unwhite");
 400 [Ff]orbidden            CAP("unallowed");
 401 [Ff]orbid               CAP("unallow");
 402 [Ff]ailure              CAP("unsuccess");
 403 [Ff]ail/{EW}            CAP("unwin");
 404
 405 [Ss]tatistics?/{EW}             CAP("propaganda");
 406 {W}[Aa]n{W}[Aa]nn?ouncement     WCAP("a NewsFlash");
 407 [Aa]nn?ouncement                printf("NewsFlash");
 408 [Ii]nstructions?                printf("B. B. DayOrder");
 409
 410 [Aa]lmost|[Nn]early     CAP("within measurable distance of");
 411 [Ff]unny                CAP("humorful");
 412
 413 [Dd]oom                 CAP("unsave");
 414 [Cc]haos                CAP("unorder");
 415 [Cc]haotic              CAP("unorderful");
 416 [Ee]nslaved             CAP("protected");
 417 [Ee]nslave              CAP("protect");
 418 [Dd]angerous            CAP("unsafewise");
 419 [Dd]anger               CAP("unsafe");
 420 ([Bb]lind{W})?[Oo]bedience              COMP("ing","soc");
 421 \"?[Nn]ew{W}[Ww]orld{W}[Oo]rder\"?      printf("IngSoc");
 422
 423 [Pp]rivacy              |
 424 [Ii]ndividuality        COMP("own","life");
 425
 426 IMHO            printf("for the love of Big Brother");
 427
 428 [Ee]motion(al|s)?       |
 429 [Cc]onviction           |
 430 [Bb]elie(f|ve)          |
 431 [Aa]ccept(ance)?        COMP("belly","feel");
 432
 433 [Dd]emocracy                    |
 434 [Ll]iberty                      |
 435 [Ff]reedom                      |
 436 [Jj]ustice                      |
 437 {THE}?[Aa]merican{W}[Ww]ay      |
 438 [Ss]ubversion                   |
 439 [Pp]assion              COMP("crime","think");
 440
 441 [Oo]bscenity            |
 442 [Pp]ornography          |
 443 [Oo]rgasm               |
 444 [Ee]rotica              COMP("sex","crime");
 445 [Ss]exy                 |
 446 [Oo]bscene              |
 447 [Pp]ornographic         |
 448 [Ee]rotic               COMP("sex","crimeful");
 449
 450 [Cc]ritic/s?{W}         COMP("crime","thinker");
 451
 452 [Ii]nfant{W}[Mm]ortality        COMP("inf","mort");
 453
 454 [Ff]amilies                     |
 455 [Pp]arents                      COMP("family ","units");
 456 [Mm]other{ANDOR}[Ff]ather       |
 457 [Bb]rother{ANDOR}[Ss]ister      COMP("family ","unit");
 458 {W}[Pp]arent/s?{EW}             WCOMP("family ","unit");
 459 [Ff]amily                       COMP("family ","unit");
 460
 461 God/{EW}                                        printf("Big Brother");
 462 [Pp]res(ident|".")({W}([Bb]ush|[Rr]eagan))?     printf("Big Brother");
 463 [Pp]rime{W}[Mm]inister                          printf("Big Brother");
 464
 465 ([Gg][Nn][Uu]{W}([Ee]macs|EMACS){W})?[Gg]eneral{W}[Pp]ublic{W}[Ll]icense        printf("NewSpeak Dictionary");
 466
 467 (questioning|murder|ass?ass?ination)/{ES}       printf("interrogation");
 468
 469 [Ss]keptic/{EW}                 CAP("unperson");
 470 [Ss]illy                        CAP("foolhardy");
 471 {W}[A-Za-z][A-Za-z]?illy        DUMP();
 472 [Ss]outhern|[Ss]outherly        CAP("southwise");
 473 [Nn]orthern|[Nn]ortherly        CAP("northwise");
 474 [Ee]astern|[Ee]easterly         CAP("eastwise");
 475 [Ww]estern|[Ww]esterly          CAP("westwise");
 476 [Pl]leasant                     CAP("goodwise");
 477 [Vv]iolent                      CAP("unpeacewise");
 478 [Vv]iolence                     CAP("unpeaceness");
 479 [Ii]ndifference                 CAP("uncarefulness");
 480 [Ii]ndifferent                  CAP("uncareful");
 481 [Bb]elly                        CAP("abdomen");
 482 [Cc]omic                        CAP("humorwise");
 483 {W}[Uu]nless                    WCAP("lest");
 484 usually                         printf("usualwise");
 485
 486 [Gg]uerillas            COMP("party ","workers");
 487
 488 [Ww]ar/{EW}             CAP("engagement");
 489 [Dd]efen[cs]e/{EW}      CAP("peace");
 490 [Ii]nvasion             CAP("liberation");
 491
 492 %{
 493         /******************************
 494          *     Syllable Rewriting     *
 495          ******************************
 496         isn't ___       is un___
 497         not the ___     the un___
 498         not my ___      my un___
 499         anti___         un___           (etc...)
 500         ___cally        ___wise
 501         ___ally         ___wise
 502         ___lly          ___wise
 503         ___ly           ___wise
 504         ___aic          ___wise
 505         ___lic          ___wise
 506         ___nnic         ___wise
 507         <VOWEL>tric     ___wise
 508         ___ic           ___wise
 509         <VOWEL>ous      ___ful
 510         <CONSONANT>ous  ___eful
 511         ___less         un___ful
 512
 513          */
 514 %}
 515
 516 [Ii]sn"'"t{W}my{W}              CAP("is my un");
 517 [Ii]s{W}not{W}my{W}             CAP("is my un");
 518 [Ii]sn"'"t{W}[Tt]he{W}          CAP("is the un");
 519 [Ii]s{W}not{W}[Tt]he{W}         CAP("is the un");
 520 [Ii]sn"'"t{W}[Ii]n{W}[Tt]he{W}  CAP("is in the un");
 521 [Ii]s{W}not{W}[Ii]n{W}[Tt]he{W} CAP("is in the un");
 522 [Ii]t"'"?s{W}not{W}[Tt]he{W}    CAP("it's the un");
 523 [Ii]sn"'"t{W}                   CAP("is un");
 524 [Ii]s{W}not{W}                  CAP("is un");
 525 [Nn]ot{W}[Tt]he{W}              CAP("the un");
 526 [Nn]ot{W}[Mm]y{W}               CAP("my un");
 527 [Nn]ot{W}[Aa]{W}                CAP("an un");
 528 [Nn]ot{W}have{W}                CAP("has un");
 529 [Nn]ot{W}be{W}                  CAP("be un");
 530 [Nn]ot{W}[Oo]nly/{W}            CAP("unonly"); /* avoid "unonwise" */
 531
 532 [Aa]{W}[Nn]ot{W}        |
 533 [Aa]{W}[Nn]on"-"?       printf("%cn%cun",Y0,Y1); /* "a non_" -> "an un_" */
 534
 535 %{
 536 /*      {W}[Nn]ot{W}    |       */
 537 %}
 538 {W}[Ii]l"-"?/[a-z][a-z]         WCAP("un");
 539 {W}[Aa]nti"-"?                  |
 540 {W}[Nn]on"-"?                   WCAP("un");
 541
 542 robably|ventually|[Oo]bvious|[Bb]asic|{W}[Oo]nly|otally |
 543 [Aa]rctic|holic|{EW}ally|{EW}[Aa]pply|{W}[Tt]opic       DUMP();
 544
 545 {W}([Tt]raf|[Pp]aci|[Ss]peci)fi/c{W}    DUMP();
 546 {W}(ma|tra)gi/c{W}                      DUMP();
 547 {W}(pub|cyc|re|fro|gar)li/c{W}          DUMP();
 548 {W}(eth|cli|to)ni/c{W}                  DUMP();
 549 {W}(E|cle|met|cit)ri/c{W}               DUMP();
 550 {W}(ch|ep|tr?op|t|mus|stat|att)i/c{W}   DUMP();
 551 {W}only/{W}                             DUMP();
 552 {W}[Aa]tlantic                          DUMP();
 553
 554 [ \t\n][drstwDRSTW]ally         printf("%c%cally", Y0, Y1);
 555
 556 [a-z]ically/{W}                 printf("%cwise", Y0);
 557 [a-z]ally/{W}                   printf("%cwise", Y0);
 558 [a-z][a-z]lly/{W}               printf("%c%cwise", Y0,Y1);
 559 [a-z][a-z][a-z]ly/{W}           printf("%c%c%cwise", Y0,Y1,Y2);
 560 [a-z]ical/{W}                   printf("%cwise", Y0);
 561
 562 [a-km-qs-z]aic/{EW}             printf("%cwise",  Y0);  /* not laic, raic */
 563 [a-z]lic/{EW}                   printf("%clwise", Y0);
 564 [a-z]nnic/{EW}                  printf("%cnwise", Y0);
 565 [a-z][aeiou]tric/{EW}           printf("%c%ctwise", Y0, Y1);
 566 [a-z]tric/{EW}                  printf("%cwise", Y0);
 567 [a-z]ic/{EW}                    printf("%cwise",  Y0);
 568 [a-z]lly/{EW}                   printf("%cwise", Y0);
 569 [a-z]ly/{EW}                    printf("%cwise", Y0);
 570
 571 [aeiouy][^aeiouy]ous/{EW}       printf("%c%cful",Y0,Y1);
 572 [^aeiouy]ous/{EW}               printf("%ceful",Y0);
 573 [^e]ous/{EW}                    printf("%cful",Y0);
 574
 575 [A-Za-z]+less/{EW}  {   yytext[yyleng-4] = '\0';
 576                         if (((yytext[1] < 'A') || (yytext[1] > 'Z')) &&
 577                             (yytext[0] >= 'A') && (yytext[0] <= 'Z'))
 578                            yytext[0] = yytext[0] - ('a' - 'A');
 579                         printf("%cn%sful",C('u'), yytext);
 580                     }
 581
 582 ".  "           {       printf("%s",yytext);
 583                         fcounter &= 15;
 584                         if (14 == fcounter++) printf("(fnord)  ");
 585                 }
 586
 587 [.,!?]\"([^\n\".!]+[.!])?\n/[\n\t ]     printf("%c  Hail Big Brother!\"%s",Y0,yytext+2);
 588 \"([.,!?][^\n\".!]+[.!])?\n/[\n\t ]     printf("%c  Hail Big Brother!\"%s",Y1,yytext+2);
 589
 590 .                       printf(yytext);
 591 \n                      printf("\n");
 592
 593 %{
 594 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 595  * Brief description of Orwell's NewSpeak:
 596  * ---------------------------------------
 597  * Each word was made to serve as noun, verb, adjective and adverb.  In
 598  * particular, nouns and verbs were made exactly the same -- hence "think"
 599  * replaced "thought" as a noun (viz. "crimethink" = thought crime, "thinkpol"
 600  * = thought police).  Adjectives were formed by adding "-ful" to a noun/verb;
 601  * adverbs were formed by adding "-wise."  Hence "speedful" = fast,
 602  * "speedwise" = quickly, "unspeedwise" = slowly.
 603  *
 604  * The decision on which word should be negated was made fairly randomly;
 605  * "dark" could be "unlight," or "light" could be "undark".  But in all cases
 606  * the most important objective (aside from ideological restriction) was
 607  * euphony; the words had to be easily pronounceable.
 608  *
 609  * Most verb inflections were made regular; "he thinked," "he goed," "he
 610  * eated"; only the auxiliaries and modals (to be, to have, may, shall, will,
 611  * etc.) were allowed to inflect irregularly.  Plurals were also regularized:
 612  * "mans," "oxes," "childs."  [This isn't implemented here.]
 613  *
 614  * There were three sets of words.  The A vocabulary was for everyday use:
 615  * car, man, red, good, etc.  It was restricted to fairly simple words.
 616  *
 617  * The B vocabulary consisted of political/ideological words with very
 618  * complicated connotations.  All of the B words were compound words --
 619  * bellyfeel (blind emotional acceptance of the ideology), oldthink (the way
 620  * of thought before the Revolution), crimethink, Ingsoc, goodsex (intercourse
 621  * solely for the purpose of making babies and with no physical enjoyment on
 622  * the part of the female), sexcrime (any kind of sex but goodsex, including
 623  * sex for its own sake), goodthink (thinking in accordance with Ingsoc), and
 624  * so on.  These words were also subject to the modifications mentioned
 625  * above--hence "goodthinker," "goodthinkful," "goodthinkwise."
 626  *
 627  * The C vocabulary consisted of scientific and technical words (though there
 628  * was no longer any word for "science," any connotation it might have being
 629  * subsumed into "Ingsoc").
 630  *
 631  * Implementing a translator for all of this would be really complicated --
 632  * I'd need rather extensive lists of the "irregular" words (so they could be
 633  * regularized), as well as lists of politically meaningful words (so they
 634  * could be excised or translated into either "goodthink" or "crimethink," as
 635  * appropriate).  Any kind of sexual topic should become "sexcrime" (it being
 636  * unlikely that any talk of sex these days would fit into "goodsex").
 637  *
 638  * Basically, the reason it's hard is that NewSpeak was intended to *decrease*
 639  * the vocabulary, and subsume complicated ideas into politically correct
 640  * words so that you wouldn't have to understand the ideas anymore; you'd just
 641  * have to emit the right words.  So to properly "translate" anything into
 642  * NewSpeak, you have to cut the vocabulary way down.
 643  * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 644  */
 645 %}
 646
 647 %%
 648 main()
 649 {
 650         yylex();
 651         printf("\nHail Big Brother!\n");
 652 }
 653
 654
 655 capstring(firstchar,string,i)
 656     char firstchar, *string;
 657     int i;
 658 {
 659         if (firstchar != 0) putchar(firstchar);
 660         putchar( (yytext[i] <= 'Z') ? (string[0] - ' ') : string[0]);
 661         printf("%s",string+1);
 662 }
 663
 664 compstring(firstchar,string1,string2,i)
 665     char firstchar, *string1, *string2;
 666     int i;
 667 {
 668         capstring(firstchar,string1,i);
 669         capstring(0,string2,i);
 670 }
 671
 672 _caseify(c,i)
 673     char c;
 674     int i;
 675 {
 676         if (yytext[i] <= 'Z') return (c - ' ');
 677         else return (c);
 678 }
 679
 680 caseify(c)
 681     char c;
 682 {       _caseify(c,0);  }
 683
 684 caseify1(c)
 685     char c;
 686 {       _caseify(c,1);  }