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@lucid.com>.
   6  *      Permission granted for noncommercial use and redistribution.
   7  *
   8  *      To compile:
   9  *
  10  *      flex newspeak.l
  11  *      cc -O -o newspeak lex.yy.c
  12  *      rm lex.yy.c
  13  *
  14  *      This must be compiled with "flex", not normal "lex".  Lex has
  15  *      some builtin, unchangable limits which this program exceeds.
  16  *      This has been verified to work with flex version 2.3.7, and
  17  *      may not work with flex version 2.3.6.
  18  *
  19  *      You can get 'flex' by anonymous FTP to prep.ai.mit.edu, or
  20  *      anywhere else you get GNU software.
  21  *
  22  *      Run this like so:  "newspeak < some-text-file | more"
  23  *
  24  *      There are some design notes near the end.  Suggestions and
  25  *      improvements to this code are more than welcome.
  26  *      Hail Big Brother!
  27  *---------------------------------------------------------------------------
  28  */
  29 #include <stdio.h>
  30
  31 unsigned int fcounter = 0;
  32
  33 #define Y0      yytext[0]
  34 #define Y1      yytext[1]
  35 #define Y2      yytext[2]
  36 #define Y3      yytext[3]
  37 #define C       caseify
  38 #define C1      caseify1
  39
  40 #define CAP(string)     capstring(0, (string),0)
  41 #define WCAP(string)    capstring(Y0,(string),1)
  42
  43 #define COMP(string1,string2)   compstring(0, (string1),(string2),0)
  44 #define WCOMP(string1,string2)  compstring(Y0,(string1),(string2),1)
  45
  46 #define DUMP()                  printf("%s",yytext)
  47 %}
  48
  49 %e 8000
  50 %p 9000
  51 %n 2000
  52 %k 3000
  53 %a 5000
  54 %o 4000
  55
  56 W       (([     ]?\n[   ]*)|[   ]|[\"'`])
  57 ES      [.,;:!?/]
  58 EW      ({W}|{ES}|"'")
  59
  60 YOUNG   (([Yy]oung|[Ss]mall|[Ll]ittle){W})
  61 DEAD    (([Dd]ead|[Dd]eceased){W})
  62 THE     ([Tt]he{W}?)
  63 ANDOR   ({W}(([Aa]nd)|([Oo]r)){W}?)
  64 COLOR   (red|green|blue|yellow|cyan|magenta|purple|orange|mauve|pink|turquoise|brown|gr[ae]y)
  65
  66 %%
  67
  68 %{
  69         /******************************
  70          *          PLUSwords         *
  71          ******************************/
  72 %}
  73
  74 [Qq]uite{EW}/[A-Za-z][A-Za-z]                   CAP("plus");
  75 [Rr]ather{EW}/[A-Za-z][A-Za-z][A-Za-z]          CAP("plus");
  76 [Kk]ind{EW}of{EW}/[A-Za-z][A-Za-z][A-Za-z]      CAP("plus");
  77 [Kk]inda{EW}/[A-Za-z][A-Za-z][A-Za-z]           CAP("plus");
  78 [Mm]ore{W}than{W}a{W}(little|bit){W}            CAP("plus");
  79 [Pp]ro-                                         CAP("plus");
  80
  81 [Hh]undreds{W}of{W}[Tt]housands                 |
  82 [Hh]undreds{ANDOR}[Tt]housands                  |
  83 [Hh]undreds{W}if{W}not{W}[Tt]housands           |
  84 [Tt]housands                                    |
  85 [Mm]illions                                     CAP("doubleplusmany");
  86
  87 [Dd]ozens                                       CAP("many");
  88 [Hh]undreds                                     CAP("plusmany");
  89
  90 ([Bb]right|[Ll]ight|[Ii]ntense){W}/{COLOR}      CAP("plus");
  91 ([Dd]im|[Ff]aded|[Dd]ark|[Pp]ale){W}/{COLOR}    CAP("plusun");
  92 ([Dd]im|[Ff]aded|[Dd]ark|[Pp]ale)               CAP("unlight");
  93
  94 [Ee]very                                        DUMP();
  95 [Vv]ery{W}                                      |
  96 [Rr]eally{W}                                    |
  97 [Tt]erribly{W}                                  |
  98 [Aa]wesome({W})?                                |
  99 [Aa]wfully{W}                                   CAP("doubleplus");
 100
 101 [Ww]hopping{EW}                                 CAP("plusbig");
 102
 103 "O.K."                                          |
 104 [Aa]ll({W})?[Rr]ight                            |
 105 [Oo][Kk][Aa][Yy]                                CAP("plusgood");
 106 {W}OK/{W}                                       WCAP("plusgood");
 107
 108 ([Tt]oo|[Oo]verly|[Tt]hat){W}[Mm]uch            CAP("plusmuch");
 109
 110 {W}[Bb]ad/{EW}          WCAP("ungood");
 111 {W}[Pp]oor/{EW}         WCAP("ungood");
 112 {W}[Ll]ame/{EW}         WCAP("ungood");
 113 {W}[Pp]itiful/{EW}      WCAP("ungood");
 114 {W}[Nn]asty/{EW}        WCAP("plusungood");
 115 {W}[Hh]orrid/{EW}       WCAP("doubleplus ungood");
 116 {W}[Hh]orrible/{EW}     WCAP("doubleplus ungood");
 117 {W}[Aa]wful/{W}         WCAP("doubleplus ungood");
 118 {W}[Ee]vil/{W}          WCAP("doubleplus ungood");
 119
 120 %{
 121         /******************************
 122          *           Titles           *
 123          ******************************/
 124 %}
 125
 126 {W}[Ss]ir/{EW}          WCAP("citizen");
 127 {W}[Mm]r"."/{EW}        WCAP("brother");
 128 [Mm]ister/{EW}          CAP("brother");
 129 [Mm]adame?              CAP("sister");
 130 {W}[Mm]iss/{EW}         WCAP("sister");
 131 [Mm]a"'"?am/{EW}        CAP("sister");
 132 {W}[Mm]r?s"."/{EW}      WCAP("sister");
 133 Mrs/{EW}                CAP("sister");
 134
 135 {YOUNG}?[Cc]hildren                             CAP("young citizens");
 136 {YOUNG}?[Bb]oys{ANDOR}[Gg]irl/s                 CAP("young citizens");
 137 {YOUNG}?([Kk]id|[Gg]irl|[Bb]oy|[Cc]hild)/{EW}   CAP("young citizen");
 138
 139 [Ff]ellow               CAP("citizen");
 140
 141 [Nn]on{W}?"-"?citizen   CAP("unperson");
 142 [Nn]on{W}?"-"?member    CAP("unperson");
 143 [Cc]riminal/s?          CAP("unperson");
 144 {DEAD}(man|woman)       CAP("unperson");
 145 {DEAD}(men|women)       CAP("unpersons");
 146
 147 [Ii]n{W}[Pp]erson       DUMP();
 148
 149 {W}[Uu]ser              WCOMP("party ","worker");
 150 [Ss]tudent              COMP("party ","worker");
 151 [Cc]itizen/s?{EW}       COMP("party ","worker");
 152 [Pp]erson/s?{EW}        COMP("party ","worker");
 153 [Pp]eople               COMP("party ","workers");
 154
 155 [Ss]enator              |
 156 [Cc]ongressman          |
 157 [Ss]upervisor           |
 158 [Pp]rofessor            printf("Inner Party Member");
 159 [Pp]rof"."/{EW}         printf("Inner Party Member");
 160 [Pp]rof/{EW}            printf("Inner Party Member");
 161
 162 Representative/s?       printf("Inner Party Member");
 163 representatives         printf("Inner Party Members");
 164
 165 [Ww]hite{W}[Cc]ollar            |
 166 [Uu]pper{W}[Cc]lass             COMP("inner ","party");
 167 [Mm]iddle{W}[Cc]lass            CAP("party");
 168 [Bb]lue{W}[Cc]ollar             |
 169 [Ww]orking{W}[Cc]lass           |
 170 [Ll]ower{W}[Cc]lass             CAP("prole");
 171 ([Ff]ool|[Ii]diot)/s?{EW}       CAP("prole");
 172 [Ss]tupidity                    CAP("proleness");
 173
 174 %{
 175         /******************************
 176          *        Organizations       *
 177          ******************************/
 178 %}
 179
 180 [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");
 181 {THE}?[Rr]epublican{W}[Pp]arty          COMP("mini","luv");
 182 {THE}?[Dd]emocratic{W}[Pp]arty          COMP("mini","plenty");
 183
 184 {THE}?Congress                                  printf("MiniPax");
 185 {THE}?[Ss]enate                                 printf("MiniPax");
 186 {THE}?[Hh]ouse{W}[Oo]f{W}[Rr]epresentatives     printf("MiniPax");
 187 {THE}?[Ss]tate{W}[Dd]epartment                  printf("MiniPax");
 188 {THE}?[Ss]tate{W}[Dd]ept"."?                    printf("MiniPax");
 189 {THE}?[Dd]efen[cs]e{W}[Dd]epartment             |
 190 {THE}?[Dd]efen[cs]e{W}[Dd]ept"."?               |
 191 {THE}?[Ww]ar{W}[Dd]epartment                    |
 192 {THE}?[Ww]ar{W}[Dd]ept"."?                      |
 193 {THE}?[Hh]ouse{W}of{W}[Cc]ommons                |
 194 {THE}?Pentagon                                  |
 195 {THE}?[Ff]eds                                   |
 196 {THE}?FCC                                       |
 197 {THE}?D[Oo]D                                    |
 198 {THE}"D."[Oo]".D."                              |
 199 {THE}?[Ss]ecret{W}[Ss]ervice                    COMP("mini","luv");
 200 {THE}?White{W}House                             |
 201 {THE}?Kremlin                                   printf("MiniTrue");
 202 {THE}?(CIA|NSA|FBI|MI"-"?5)/{EW}                printf("MiniTrue");
 203 {THE}?("C.I.A."|"N.S.A."|"F.B.I.")/{EW}         printf("MiniTrue");
 204 {THE}?[Aa]rchive/s?                             COMP("mini","rec");
 205 {THE}?[Ll]ibrary                                COMP("mini","rec");
 206 {THE}?[Ll]ibraries                              COMP("mini","recs");
 207
 208 [Tt]hought{W}[Pp]olice|[Nn]azis?                COMP("think","pol");
 209 [Vv]ice{W}[Ss]quad                              COMP("sex","pol");
 210 PMRC|"P.M.R.C."                                 COMP("sex","pol");
 211
 212 [Oo]fficer                                      CAP("minister");
 213
 214 {THE}?[Dd]epartment{EW}of{EW}.                  |
 215 {THE}?[Dd]ept"."?{EW}of{EW}.                    |
 216 {THE}?[Uu]niversity{EW}of{EW}.                  |
 217 {THE}?[Uu]niv"."?{EW}of{EW}.                    |
 218 {THE}?[Dd]ept"."?{EW}of{EW}.                    |
 219 {THE}?([Ss]ub"-"?)?[Cc]omm?itt?ee{EW}(of|on){EW}.       |
 220 {THE}?[Ss]chool{EW}of{EW}.                      {
 221         if ((yytext[yyleng-1] >= 'A') && (yytext[yyleng-1] <= 'Z'))
 222            /* "the school of Law" --> "MiniLaw" */
 223            printf("Mini%c",yytext[yyleng-1]);
 224         else if ((yytext[yyleng-1] >= 'a') && (yytext[yyleng-1] <= 'z'))
 225            /* "the school of law" --> "MiniLaw" (not "Minilaw") */
 226            printf("Mini%c",yytext[yyleng-1] - ('a' - 'A'));
 227            /* "the school of 5 things" --> "Ministry of 5 things" */
 228         else printf("Ministry of %c",yytext[yyleng-1]);
 229         }
 230
 231 [Dd]epartment                   |
 232 [Uu]niversity                   CAP("ministry");
 233 [Uu]niv"."?/{W}                 CAP("ministry");
 234 [Dd]ept"."?/{W}                 CAP("ministry");
 235 ([Ss]ub"-"?)?[Cc]omm?itt?ee     CAP("ministry");
 236
 237 {THE}[Pp]roject/{EW}            CAP("the Three Year Plan");
 238 [Oo]ur{W}[Pp]roject/{EW}        CAP("our Three Year Plan");
 239 [Bb]udget                       printf("Three Year Plan");
 240 [Pp]roject/{ES}                 printf("Three Year Plan");
 241
 242 {W}({THE}|([aa]{W}))[Pp]roject  printf("%cthe Three Year Plan",Y0);
 243
 244 [A-Za-z]+"'"[Ss]/{W}(law|Law|LAW|book|Book|BOOK|rule|Rule|RULE){EW}     printf("Goldstein's");
 245
 246 %{
 247         /******************************
 248          *           Actions          *
 249          ******************************/
 250 %}
 251
 252 [Ii]n{W}love{EW}        CAP("SexCriming");
 253 [Ll]ove{W}you/{EW}      CAP("love Big Brother");
 254 [Ll]ove{W}me/{EW}       CAP("love Big Brother");
 255
 256 [Cc]loning              |
 257 [Rr]eproduction         |
 258 [Cc]elibacy             |
 259 [Pp]rocreation          COMP("good","sex");
 260
 261 [Cc]elibate             |
 262 [Pp]rocreate            COMP("good","sexwise");
 263
 264 [Tt]elevisions?         |
 265 TVs?                    |
 266 [Tt]"."[Vv]"."s?        |
 267 [Rr]adios?              |
 268 [Nn]ews{W}?[Pp]apers?   |
 269 [Jj]ournalism           |
 270 [Mm]ovies?              |
 271 [Rr]ock{EW}?"-"?(and|"&"|"'"?n"'"?){EW}?"-"?[Rr]oll({W}[Mm]usic)?            |
 272 (([Rr]ock|[Cc]lassical|[Ii]ndustrial|[Pp]op|[Dd]ance|[Rr]ap){W})?[Mm]usic    |
 273 [Tt]unes                |
 274 [Mm]oney                |
 275 [Cc]ash                 |
 276 [Cc]omic{W}[Bb]ooks?    |
 277 ([Ss]tar{W}?)?[Tt]rek   COMP("prole","feed");
 278
 279 [Pp]eace{W}[Mm]ovement          |
 280 [Pp]eace{W}[Pp]rotest           |
 281 [Aa]nti{EW}[Ww]ar               |
 282 ([Pp]assive{W})?[Rr]esistance   |
 283 [Cc]reativity                   |
 284 [Tt]reason                      |
 285 [Rr]esearch             COMP("crime","think");
 286 %{
 287         /******************************
 288          *           Religion         *
 289          ******************************/
 290 %}
 291
 292 [Jj]esus{W}[Cc]hrist            |
 293 [Jj]esus                        |
 294 {THE}?[Bb]uddh?a                |
 295 [Mm]ohamm?ed                    |
 296 [Mm]artin{W}[Ll]uther{W}[Kk]ing |
 297 J\.?\ ?R\.?\ \"?Bob\"?\ Dobbs   printf("doubleplus crimethinker");
 298
 299 ([Jj]esse{W})?[Hh]elms          |
 300 ([RrDd]on(ald)?{W})?[Rr]ea?gan  |
 301 [Gg]eorge{W}[Gg]uscoria         printf("doubleplus goodthinker");
 302
 303 [Jj]ewish                       COMP("crime","thinkwise");
 304 [Jj]ew                          |
 305 [Cc]hristian                    |
 306 [Mm]oslem                       |
 307 [Bb]uddhist                     |
 308 [Aa]thiest                      |
 309 [Aa]gnostic                     COMP("crime","thinker");
 310
 311 [Ff]aith                        COMP("belly","feel");
 312
 313 %{
 314         /******************************
 315          *            Places          *
 316          ******************************/
 317 %}
 318
 319 [Ee]ngland|{THE}?[Uu]nited{W}[Kk]ingdom                 |
 320 ({THE}?[Uu]nited{W}[Ss]tates{W}[Oo]f{W})?[Aa]merica     |
 321 {THE}?[Uu]nited{W}[Ss]tates|USA|"U.S.A."|[Cc]anada      |
 322 [Gg]ermany|[Ii]srael|[Ee]urope                          printf("Oceana");
 323
 324 Iranian|Iraqu?i|Libyan|Russian|African|Egyptian         printf("Eurasian");
 325 Iran|Iraq|Libya|Russia|Africa|Egypt                     |
 326 ([Ss]audi{W})?Arabia|{THE}?Soviet{W}Union               printf("Eurasia");
 327 [Ss]oviet                                               printf("Eurasian");
 328
 329 [Cc]hinese|[Jj]apanese|[Tt]aiwanese                     |
 330 [Pp]hillipino|[Ii]ndian|[Aa]ustralian|[Mm]exican        |
 331 [Nn]icaraguan|[Ss]alvadori?an                           printf("Eastasian");
 332 China|[Jj]apan|[Tt]aiwan|{THE}?[Pp]hillipines|[Ii]ndia  |
 333 [Aa]ustralia|[Mm]exico|[Nn]icaragua|[Ee]l{W}[Ss]alvador printf("Eastasia");
 334
 335 [Kk]uwaiti                      printf("Eurasian");
 336 [Kk]uwait                       printf("The Malabar Front");
 337
 338 %{
 339         /******************************
 340          * Miscelaneous Translations  *
 341          ******************************/
 342 %}
 343
 344 {W}[Ff]aster            WCAP("plus speedful");
 345 {W}[Ss]lower            WCAP("plus unspeedful");
 346 {W}[Ff]ast              WCAP("speedful");
 347 {W}[Ss]low              WCAP("unspeedful");
 348
 349 [Mm]odern               CAP("plusnew");
 350 [Aa]ncient              CAP("plusunnew");
 351 {W}old/{W}              WCAP("plusunnew");
 352
 353 [Hh]appiness            CAP("joyfulness");
 354 [Hh]appy                CAP("joyful");
 355 [Qq]uick                CAP("speedful");
 356 {W}[Ss]peedy            WCAP("speedful");
 357 [Hh]eavy                CAP("weightful");
 358 [Hh]eavill?y            CAP("weightfully");
 359 [Ss]ick(ly)?            CAP("unhealthful");
 360
 361 [Gg]ross                |
 362 [Ss]ickening            |
 363 [Ff]oul                 |
 364 [Pp]utrid               |
 365 [Dd]isgusting           COMP("crime","thinkful");
 366
 367 [Ss]mash                |
 368 [Cc]rush                |
 369 [Oo]bliterate           |
 370 [Aa]nnihilate           |
 371 [Nn]eutralize           |
 372 [Dd]emolish             |
 373 [Dd]estroy              CAP("unbuild");
 374
 375 [Ii]nanimate            CAP("unlifeful");
 376 [Ss]ociety|[Cc]ulture   printf("IngSoc");
 377 [A-Za-z]+isi?m/{EW}     printf("Goldsteinism");
 378 [A-Za-z]+ist/{EW}       printf("Goldsteinist");
 379
 380 {W}[Dd]ead              WCAP("unlifeful");
 381 {W}[Dd]eath             WCAP("unlife");
 382 {W}[Ll]ie               WCAP("untruth");
 383 {W}[Ff]alsehood         WCAP("untruth");
 384 {W}[Mm]istake/{EW}      WCAP("untruth");
 385 {W}[Ww]hisper           WCAP("unshout");
 386 {W}[Pp]roud             WCAP("prideful");
 387
 388 [Ff]alse                CAP("untrue");
 389 [Dd]ark                 CAP("unlight");
 390 [Bb]lack                CAP("unwhite");
 391 [Ff]orbidden            CAP("unallowed");
 392 [Ff]orbid               CAP("unallow");
 393 [Ff]ailure              CAP("unsuccess");
 394 [Ff]ail/{EW}            CAP("unwin");
 395
 396 [Ss]tatistics?/{EW}             CAP("propaganda");
 397 {W}[Aa]n{W}[Aa]nn?ouncement     WCAP("a NewsFlash");
 398 [Aa]nn?ouncement                printf("NewsFlash");
 399 [Ii]nstructions?                printf("B. B. DayOrder");
 400
 401 [Aa]lmost|[Nn]early     CAP("within measurable distance of");
 402 [Ff]unny                CAP("humorful");
 403
 404 [Dd]oom                 CAP("unsave");
 405 [Cc]haos                CAP("unorder");
 406 [Cc]haotic              CAP("unorderful");
 407 [Ee]nslaved             CAP("protected");
 408 [Ee]nslave              CAP("protect");
 409 [Dd]angerous            CAP("unsafewise");
 410 [Dd]anger               CAP("unsafe");
 411 ([Bb]lind{W})?[Oo]bedience              COMP("ing","soc");
 412 \"?[Nn]ew{W}[Ww]orld{W}[Oo]rder\"?      printf("IngSoc");
 413
 414 [Pp]rivacy              |
 415 [Ii]ndividuality        COMP("own","life");
 416
 417 IMHO            printf("for the love of Big Brother");
 418
 419 [Ee]motion(al|s)?       |
 420 [Cc]onviction           |
 421 [Bb]elie(f|ve)          |
 422 [Aa]ccept(ance)?        COMP("belly","feel");
 423
 424 [Dd]emocracy                    |
 425 [Ll]iberty                      |
 426 [Ff]reedom                      |
 427 [Jj]ustice                      |
 428 {THE}?[Aa]merican{W}[Ww]ay      |
 429 [Ss]ubversion                   |
 430 [Pp]assion              COMP("crime","think");
 431
 432 [Oo]bscenity            |
 433 [Pp]ornography          |
 434 [Oo]rgasm               |
 435 [Ee]rotica              COMP("sex","crime");
 436 [Ss]exy                 |
 437 [Oo]bscene              |
 438 [Pp]ornographic         |
 439 [Ee]rotic               COMP("sex","crimeful");
 440
 441 [Cc]ritic/s?{W}         COMP("crime","thinker");
 442
 443 [Ii]nfant{W}[Mm]ortality        COMP("inf","mort");
 444
 445 [Ff]amilies                     |
 446 [Pp]arents                      COMP("family ","units");
 447 [Mm]other{ANDOR}[Ff]ather       |
 448 [Bb]rother{ANDOR}[Ss]ister      COMP("family ","unit");
 449 {W}[Pp]arent/s?{EW}             WCOMP("family ","unit");
 450 [Ff]amily                       COMP("family ","unit");
 451
 452 God/{EW}                                        printf("Big Brother");
 453 [Pp]res(ident|".")({W}([Bb]ush|[Rr]eagan))?     printf("Big Brother");
 454 [Pp]rime{W}[Mm]inister                          printf("Big Brother");
 455
 456 ([Gg][Nn][Uu]{W}([Ee]macs|EMACS){W})?[Gg]eneral{W}[Pp]ublic{W}[Ll]icense        printf("NewSpeak Dictionary");
 457
 458 (questioning|murder|ass?ass?ination)/{ES}       printf("interrogation");
 459
 460 [Ss]keptic/{EW}                 CAP("unperson");
 461 [Ss]illy                        CAP("foolhardy");
 462 {W}[A-Za-z][A-Za-z]?illy        DUMP();
 463 [Ss]outhern|[Ss]outherly        CAP("southwise");
 464 [Nn]orthern|[Nn]ortherly        CAP("northwise");
 465 [Ee]astern|[Ee]easterly         CAP("eastwise");
 466 [Ww]estern|[Ww]esterly          CAP("westwise");
 467 [Pl]leasant                     CAP("goodwise");
 468 [Vv]iolent                      CAP("unpeacewise");
 469 [Vv]iolence                     CAP("unpeaceness");
 470 [Ii]ndifference                 CAP("uncarefulness");
 471 [Ii]ndifferent                  CAP("uncareful");
 472 [Bb]elly                        CAP("abdomen");
 473 [Cc]omic                        CAP("humorwise");
 474 {W}[Uu]nless                    WCAP("lest");
 475 usually                         printf("usualwise");
 476
 477 [Gg]uerillas            COMP("party ","workers");
 478
 479 [Ww]ar/{EW}             CAP("engagement");
 480 [Dd]efen[cs]e/{EW}      CAP("peace");
 481 [Ii]nvasion             CAP("liberation");
 482
 483 %{
 484         /******************************
 485          *     Syllable Rewriting     *
 486          ******************************
 487         isn't ___       is un___
 488         not the ___     the un___
 489         not my ___      my un___
 490         anti___         un___           (etc...)
 491         ___cally        ___wise
 492         ___ally         ___wise
 493         ___lly          ___wise
 494         ___ly           ___wise
 495         ___aic          ___wise
 496         ___lic          ___wise
 497         ___nnic         ___wise
 498         <VOWEL>tric     ___wise
 499         ___ic           ___wise
 500         <VOWEL>ous      ___ful
 501         <CONSONANT>ous  ___eful
 502         ___less         un___ful
 503
 504          */
 505 %}
 506
 507 [Ii]sn"'"t{W}my{W}              CAP("is my un");
 508 [Ii]s{W}not{W}my{W}             CAP("is my un");
 509 [Ii]sn"'"t{W}[Tt]he{W}          CAP("is the un");
 510 [Ii]s{W}not{W}[Tt]he{W}         CAP("is the un");
 511 [Ii]sn"'"t{W}[Ii]n{W}[Tt]he{W}  CAP("is in the un");
 512 [Ii]s{W}not{W}[Ii]n{W}[Tt]he{W} CAP("is in the un");
 513 [Ii]t"'"?s{W}not{W}[Tt]he{W}    CAP("it's the un");
 514 [Ii]sn"'"t{W}                   CAP("is un");
 515 [Ii]s{W}not{W}                  CAP("is un");
 516 [Nn]ot{W}[Tt]he{W}              CAP("the un");
 517 [Nn]ot{W}[Mm]y{W}               CAP("my un");
 518 [Nn]ot{W}[Aa]{W}                CAP("an un");
 519 [Nn]ot{W}have{W}                CAP("has un");
 520 [Nn]ot{W}be{W}                  CAP("be un");
 521 [Nn]ot{W}[Oo]nly/{W}            CAP("unonly"); /* avoid "unonwise" */
 522
 523 [Aa]{W}[Nn]ot{W}        |
 524 [Aa]{W}[Nn]on"-"?       printf("%cn%cun",Y0,Y1); /* "a non_" -> "an un_" */
 525
 526 %{
 527 /*      {W}[Nn]ot{W}    |       */
 528 %}
 529 {W}[Ii]l"-"?/[a-z][a-z]         WCAP("un");
 530 {W}[Aa]nti"-"?                  |
 531 {W}[Nn]on"-"?                   WCAP("un");
 532
 533 robably|ventually|[Oo]bvious|[Bb]asic|{W}[Oo]nly|otally |
 534 [Aa]rctic|holic|{EW}ally|{EW}[Aa]pply|{W}[Tt]opic       DUMP();
 535
 536 {W}([Tt]raf|[Pp]aci|[Ss]peci)fi/c{W}    DUMP();
 537 {W}(ma|tra)gi/c{W}                      DUMP();
 538 {W}(pub|cyc|re|fro|gar)li/c{W}          DUMP();
 539 {W}(eth|cli|to)ni/c{W}                  DUMP();
 540 {W}(E|cle|met|cit)ri/c{W}               DUMP();
 541 {W}(ch|ep|tr?op|t|mus|stat|att)i/c{W}   DUMP();
 542 {W}only/{W}                             DUMP();
 543 {W}[Aa]tlantic                          DUMP();
 544
 545 [ \t\n][drstwDRSTW]ally         printf("%c%cally", Y0, Y1);
 546
 547 [a-z]ically/{W}                 printf("%cwise", Y0);
 548 [a-z]ally/{W}                   printf("%cwise", Y0);
 549 [a-z][a-z]lly/{W}               printf("%c%cwise", Y0,Y1);
 550 [a-z][a-z][a-z]ly/{W}           printf("%c%c%cwise", Y0,Y1,Y2);
 551 [a-z]ical/{W}                   printf("%cwise", Y0);
 552
 553 [a-km-qs-z]aic/{EW}             printf("%cwise",  Y0);  /* not laic, raic */
 554 [a-z]lic/{EW}                   printf("%clwise", Y0);
 555 [a-z]nnic/{EW}                  printf("%cnwise", Y0);
 556 [a-z][aeiou]tric/{EW}           printf("%c%ctwise", Y0, Y1);
 557 [a-z]tric/{EW}                  printf("%cwise", Y0);
 558 [a-z]ic/{EW}                    printf("%cwise",  Y0);
 559 [a-z]lly/{EW}                   printf("%cwise", Y0);
 560 [a-z]ly/{EW}                    printf("%cwise", Y0);
 561
 562 [aeiouy][^aeiouy]ous/{EW}       printf("%c%cful",Y0,Y1);
 563 [^aeiouy]ous/{EW}               printf("%ceful",Y0);
 564 [^e]ous/{EW}                    printf("%cful",Y0);
 565
 566 [A-Za-z]+less/{EW}  {   yytext[yyleng-4] = '\0';
 567                         if (((yytext[1] < 'A') || (yytext[1] > 'Z')) &&
 568                             (yytext[0] >= 'A') && (yytext[0] <= 'Z'))
 569                            yytext[0] = yytext[0] - ('a' - 'A');
 570                         printf("%cn%sful",C('u'), yytext);
 571                     }
 572
 573 ".  "           {       printf("%s",yytext);
 574                         fcounter &= 15;
 575                         if (14 == fcounter++) printf("(fnord)  ");
 576                 }
 577
 578 [.,!?]\"([^\n\".!]+[.!])?\n/[\n\t ]     printf("%c  Hail Big Brother!\"%s",Y0,yytext+1);
 579 \"([.,!?][^\n\".!]+[.!])?\n/[\n\t ]     printf("%c  Hail Big Brother!\"%s",Y1,yytext+2);
 580
 581 .                       printf(yytext);
 582 \n                      printf("\n");
 583
 584 %{
 585 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 586  * Brief description of Orwell's NewSpeak:
 587  * ---------------------------------------
 588  * Each word was made to serve as noun, verb, adjective and adverb.  In
 589  * particular, nouns and verbs were made exactly the same -- hence "think"
 590  * replaced "thought" as a noun (viz. "crimethink" = thought crime, "thinkpol"
 591  * = thought police).  Adjectives were formed by adding "-ful" to a noun/verb;
 592  * adverbs were formed by adding "-wise."  Hence "speedful" = fast,
 593  * "speedwise" = quickly, "unspeedwise" = slowly.
 594  *
 595  * The decision on which word should be negated was made fairly randomly;
 596  * "dark" could be "unlight," or "light" could be "undark".  But in all cases
 597  * the most important objective (aside from ideological restriction) was
 598  * euphony; the words had to be easily pronounceable.
 599  *
 600  * Most verb inflections were made regular; "he thinked," "he goed," "he
 601  * eated"; only the auxiliaries and modals (to be, to have, may, shall, will,
 602  * etc.) were allowed to inflect irregularly.  Plurals were also regularized:
 603  * "mans," "oxes," "childs."  [This isn't implemented here.]
 604  *
 605  * There were three sets of words.  The A vocabulary was for everyday use:
 606  * car, man, red, good, etc.  It was restricted to fairly simple words.
 607  *
 608  * The B vocabulary consisted of political/ideological words with very
 609  * complicated connotations.  All of the B words were compound words --
 610  * bellyfeel (blind emotional acceptance of the ideology), oldthink (the way
 611  * of thought before the Revolution), crimethink, Ingsoc, goodsex (intercourse
 612  * solely for the purpose of making babies and with no physical enjoyment on
 613  * the part of the female), sexcrime (any kind of sex but goodsex, including
 614  * sex for its own sake), goodthink (thinking in accordance with Ingsoc), and
 615  * so on.  These words were also subject to the modifications mentioned
 616  * above--hence "goodthinker," "goodthinkful," "goodthinkwise."
 617  *
 618  * The C vocabulary consisted of scientific and technical words (though there
 619  * was no longer any word for "science," any connotation it might have being
 620  * subsumed into "Ingsoc").
 621  *
 622  * Implementing a translator for all of this would be really complicated --
 623  * I'd need rather extensive lists of the "irregular" words (so they could be
 624  * regularized), as well as lists of politically meaningful words (so they
 625  * could be excised or translated into either "goodthink" or "crimethink," as
 626  * appropriate).  Any kind of sexual topic should become "sexcrime" (it being
 627  * unlikely that any talk of sex these days would fit into "goodsex").
 628  *
 629  * Basically, the reason it's hard is that NewSpeak was intended to *decrease*
 630  * the vocabulary, and subsume complicated ideas into politically correct
 631  * words so that you wouldn't have to understand the ideas anymore; you'd just
 632  * have to emit the right words.  So to properly "translate" anything into
 633  * NewSpeak, you have to cut the vocabulary way down.
 634  * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 635  */
 636 %}
 637
 638 %%
 639 main()
 640 {
 641         yylex();
 642         printf("\nHail Big Brother!\n");
 643 }
 644
 645
 646 capstring(firstchar,string,i)
 647     char firstchar, *string;
 648     int i;
 649 {
 650         if (firstchar != 0) putchar(firstchar);
 651         putchar( (yytext[i] <= 'Z') ? (string[0] - ' ') : string[0]);
 652         printf("%s",string+1);
 653 }
 654
 655 compstring(firstchar,string1,string2,i)
 656     char firstchar, *string1, *string2;
 657     int i;
 658 {
 659         capstring(firstchar,string1,i);
 660         capstring(0,string2,i);
 661 }
 662
 663 _caseify(c,i)
 664     char c;
 665     int i;
 666 {
 667         if (yytext[i] <= 'Z') return (c - ' ');
 668         else return (c);
 669 }
 670
 671 caseify(c)
 672     char c;
 673 {       _caseify(c,0);  }
 674
 675 caseify1(c)
 676     char c;
 677 {       _caseify(c,1);  }