usr/src/cmd/geniconvtbl/itm_comp.l

   1 %{
   2 /*
   3  * CDDL HEADER START
   4  *
   5  * The contents of this file are subject to the terms of the
   6  * Common Development and Distribution License, Version 1.0 only
   7  * (the "License").  You may not use this file except in compliance
   8  * with the License.
   9  *
  10  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
  11  * or http://www.opensolaris.org/os/licensing.
  12  * See the License for the specific language governing permissions
  13  * and limitations under the License.
  14  *
  15  * When distributing Covered Code, include this CDDL HEADER in each
  16  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  17  * If applicable, add the following below this CDDL HEADER, with the
  18  * fields enclosed by brackets "[]" replaced with your own identifying
  19  * information: Portions Copyright [yyyy] [name of copyright owner]
  20  *
  21  * CDDL HEADER END
  22  *
  23  * Copyright (c) 1999 by Sun Microsystems, Inc.
  24  * All rights reserved.
  25  */
  26
  27 #pragma ident   "%Z%%M% %I%     %E% SMI"
  28
  29 #include <assert.h>
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <unistd.h>
  33 #include <libintl.h>
  34 #include <string.h>
  35
  36 #include <regexpr.h>
  37
  38 #include "iconv_tm.h"
  39 #include "itmcomp.h"
  40 #include "y.tab.h"
  41
  42 static itm_data_t       *hexadecimal_data(int, char *);
  43 static itm_data_t       *name_data(int, char *);
  44 static void             filename_lineno(void);
  45 static int              at_name_to_token(char *);
  46
  47
  48 %}
  49
  50 %start  norm comment
  51
  52 DECIMAL         ([0-9]+)
  53 OCTAL           (0[0-7][0-7]+)
  54 HEXADECIMAL     (("0x"|"0X")([0-9A-Fa-f])+)
  55 ITMNAME         (([^% \t\n\r])+"%"([^% \t\n\r])+)
  56 ATNAME          "@"([0-9A-Za-z_]+)
  57 NAME            ([A-Za-z_][A-Za-z0-9_]*)
  58 MAPTYPE_NAME    (automatic|dense|index|hash|binary)
  59
  60 %%
  61
  62 [ \t\n]+        ;
  63 "//".*"\n"      ;
  64
  65 ^"#"[ \t]*{DECIMAL}[ \t]*"\"".*"\"".*"\n" {
  66                         filename_lineno();
  67                 }
  68
  69 ^"#".*"\n"      {
  70                         if (NULL == cmd_opt.preprocess) {
  71                                 itm_error(
  72                                 gettext("warning: "
  73                                         "preprocess may be required\n"));
  74                         }
  75                 }
  76
  77 {DECIMAL}       {
  78                         yylval.num = strtoul(yytext, (char **)NULL, 10);
  79                         return (DECIMAL);
  80                 }
  81
  82 {OCTAL}         {       yylval.num = strtoul(yytext, (char **)NULL, 8);
  83                         return (DECIMAL);
  84                 }
  85
  86 {HEXADECIMAL}   {       yylval.data = hexadecimal_data(yyleng - 2, yytext + 2);
  87                         return (HEXADECIMAL);
  88                 }
  89
  90 {ITMNAME}       {       yylval.data = str_to_data(yyleng, yytext);
  91                         return (ITMNAME);
  92                 }
  93
  94 {ATNAME}        {       return at_name_to_token(yytext);
  95                 }
  96
  97 {MAPTYPE_NAME}  {       yylval.num = at_name_to_token(yytext);
  98                         yylval.data = name_data(yyleng, yytext);
  99                         return (MAPTYPE_NAME);
 100
 101                 }
 102
 103 {NAME}          {       yylval.num = at_name_to_token(yytext);
 104                         if (0 != yylval.num) {
 105                                 return (yylval.num);
 106                         } else {
 107                                 yylval.data = name_data(yyleng, yytext);
 108                                 return (NAME);
 109                         }
 110                 }
 111
 112
 113 "{"             {return (CBO);}
 114 "}"             {return (CBC);}
 115 "["             {return (SBO);}
 116 "]"             {return (SBC);}
 117 "("             {return (PO);}
 118 ")"             {return (PC);}
 119 ";"             {return (SC);}
 120 ","             {return (COMMA);}
 121 ":"             {return (COLON);}
 122 "..."           {return (ELLIPSES);}
 123
 124
 125 "="             {return (ASSIGN);}
 126 "||"            {return (LOR);}
 127 "&&"            {return (LAND);}
 128 "|"             {return (OR);}
 129 "^"             {return (XOR);}
 130 "&"             {return (AND);}
 131 "=="            {return (EQ);}
 132 "!="            {return (NE);}
 133 "<"             {return (LT);}
 134 "<="            {return (LE);}
 135 ">"             {return (GT);}
 136 ">="            {return (GE);}
 137 "<<"            {return (SHL);}
 138 ">>"            {return (SHR);}
 139 "+"             {return (PLUS);}
 140 "-"             {return (MINUS);}
 141 "*"             {return (MUL);}
 142 "/"             {return (DIV);}
 143 "%"             {return (MOD);}
 144 "!"             {return (NOT);}
 145 "~"             {return (NEG);}
 146
 147 .               {       itm_error(
 148                                 gettext("Unrecognized token '%1$c' \n"),
 149                                 cmd_opt.my_name, yytext[0]);
 150                         return (0);
 151                 }
 152
 153 %%
 154
 155 /*
 156  * lexinit - starts the Lexical Analyzer off in the right start condition
 157  */
 158 void
 159 lexinit()
 160 {
 161         BEGIN norm;
 162 }
 163
 164 /* does this really need to be here? */
 165 int
 166 yywrap()
 167 {
 168         return (1);
 169 }
 170
 171 void
 172 yyerror(char *s)
 173 {
 174         extern int      yylineno;
 175
 176         itm_error(
 177                 gettext("%1$s: file(%2$s) line(%3$d) last token(%4$s)\n"),
 178                 s, itm_input_file, yylineno, yytext);
 179
 180         exit(ITMC_STATUS_BT);
 181 }
 182
 183 typedef struct {
 184         char    *name;
 185         int     token;
 186 } at_name_token_t;
 187
 188 /*
 189  * NOT: This table must be sorted alphabetically.
 190  */
 191 static at_name_token_t at_table[] = {
 192         "@automatic",   MAPTYPE_AUTO,
 193         "@binary",      MAPTYPE_BINARY,
 194         "@between",     BETWEEN,
 195         "@condition",   CONDITION,
 196         "@default",     ITM_DEFAULT,
 197         "@dense",       MAPTYPE_DENSE,
 198         "@direction",   DIRECTION,
 199         "@discard",     DISCARD,
 200         "@else",        ITM_ELSE,
 201         "@error",       ERROR,
 202         "@escapeseq",   ESCAPESEQ,
 203         "@false",       ITM_FALSE,
 204         "@hash",        MAPTYPE_HASH,
 205         "@identical",   ITM_IDENTICAL,
 206         "@if",          ITM_IF,
 207         "@in",          ITM_IN,
 208         "@index",       MAPTYPE_INDEX,
 209         "@init",        ITM_INIT,
 210         "@input",       ITM_IN,
 211         "@inputsize",   ITM_INSIZE,
 212         "@map",         MAP,
 213         "@maptype",     MAPTYPE,
 214         "@no_change_copy",      ITM_IDENTICAL,
 215         "@nop",         NOP,
 216         "@operation",   OPERATION,
 217         "@out",         ITM_OUT,
 218         "@output",      ITM_OUT,
 219         "@output_byte_length",  RESULTLEN,
 220         "@outputsize",  ITM_OUTSIZE,
 221         "@printchr",    PRINTCHR,
 222         "@printhd",     PRINTHD,
 223         "@printint",    PRINTINT,
 224         "@reset",       RESET,
 225         "@resultlen",   RESULTLEN,
 226         "@return",      RETURN,
 227         "@true",        ITM_TRUE,
 228         "automatic",    MAPTYPE_AUTO,
 229         "between",      BETWEEN,
 230         "binary",       MAPTYPE_BINARY,
 231         "break",        BREAK,
 232         "condition",    CONDITION,
 233         "default",      ITM_DEFAULT,
 234         "dense",        MAPTYPE_DENSE,
 235         "direction",    DIRECTION,
 236         "discard",      DISCARD,
 237         "else",         ITM_ELSE,
 238         "error",        ERROR,
 239         "escapeseq",    ESCAPESEQ,
 240         "false",        ITM_FALSE,
 241         "hash",         MAPTYPE_HASH,
 242         "if",           ITM_IF,
 243         "index",        MAPTYPE_INDEX,
 244         "init",         ITM_INIT,
 245         "input",        ITM_IN,
 246         "inputsize",    ITM_INSIZE,
 247         "map",          MAP,
 248         "maptype",      MAPTYPE,
 249         "no_change_copy",       ITM_IDENTICAL,
 250         "nop",          NOP,
 251         "operation",    OPERATION,
 252         "output",       ITM_OUT,
 253         "output_byte_length",   RESULTLEN,
 254         "outputsize",   ITM_OUTSIZE,
 255         "printchr",     PRINTCHR,
 256         "printhd",      PRINTHD,
 257         "printint",     PRINTINT,
 258         "reset",        RESET,
 259         "return",       RETURN,
 260         "true",         ITM_TRUE,
 261 };
 262
 263 int
 264 at_name_to_token(char *s)
 265 {
 266         int     high;
 267         int     mid;
 268         int     low;
 269         int     result;
 270
 271         TRACE_MESSAGE('l', ("at_name_to_token: %s", s));
 272         for (low = 0, high = (sizeof (at_table) /
 273                                 sizeof (at_name_token_t));
 274             low < high; /* NOP */) {
 275                 mid = (low + high) / 2;
 276                 result = strcmp(s, at_table[mid].name);
 277                 if (result < 0) {
 278                         high = mid;
 279                 } else if (0 < result) {
 280                         low = mid + 1;
 281                 } else { /* 0 == result */
 282                         TRACE_MESSAGE('l', (" %d\n", at_table[mid].token));
 283                         return (at_table[mid].token);
 284                 }
 285         }
 286         TRACE_MESSAGE('l', (" - not found\n"));
 287         return (0);
 288 }
 289
 290 static itm_data_t *
 291 hexadecimal_data(int seqsize, char *seq)
 292 {
 293         itm_data_t      *data;
 294         char            *binary;
 295         int             i, j;
 296         int             val;
 297         int             high;
 298         int             low;
 299         int             size;
 300
 301         /* size is assured to be multiple of 2 */
 302         assert(seqsize != 0);
 303         size = seqsize + 1;
 304         size /= 2;
 305         if (size > MAXSEQUENCE) {
 306                 itm_error(
 307                 gettext(" specified sequence must be less than %$1d\n"),
 308                 MAXSEQUENCE);
 309                 return (NULL);
 310         }
 311         binary = malloc_vital(size);
 312         i = j = 0;
 313         if (seqsize % 2 != 0) {
 314                 low =  *(seq);
 315                 if (('0' <= low) && (low <= '9')) {
 316                         val = (low - '0');
 317                 } else if (('a' <= low) && (low <= 'f')) {
 318                         val = (low - 'a' + 10);
 319                 } else if (('A' <= low) && (low <= 'F')) {
 320                         val = (low - 'A' + 10);
 321                 }
 322                 *(binary + i) = val;
 323                 i++;
 324                 j++;
 325         }
 326         for (/* NOP */; i < size; i++, j += 2) {
 327                 high = *(seq + j);
 328                 low =  *(seq + j + 1);
 329                 if (('0' <= high) && (high <= '9')) {
 330                         val = ((high - '0') << 4);
 331                 } else if (('a' <= high) && (high <= 'f')) {
 332                         val = ((high - 'a' + 10) << 4);
 333                 } else if (('A' <= high) && (high <= 'F')) {
 334                         val = ((high - 'A' + 10) << 4);
 335                 }
 336                 if (('0' <= low) && (low <= '9')) {
 337                         val |= (low - '0');
 338                 } else if (('a' <= low) && (low <= 'f')) {
 339                         val |= (low - 'a' + 10);
 340                 } else if (('A' <= low) && (low <= 'F')) {
 341                         val |= (low - 'A' + 10);
 342                 }
 343                 *(binary + i) = val;
 344         }
 345
 346         data = malloc_vital(sizeof (itm_data_t));
 347
 348         data->size = size;
 349         if (size <= sizeof (data->place)) {
 350                 (void) memmove(&(data->place), binary, size);
 351                 free(binary);
 352         } else {
 353                 data->place.itm_ptr = (itm_place2_t)binary;
 354         }
 355
 356         return (data);
 357 }
 358
 359 static itm_data_t *
 360 name_data(int size, char *seq)
 361 {
 362         itm_data_t *data;
 363
 364
 365         if (size > MAXNAMELENGTH) {
 366                 itm_error(gettext("the length(%d) exceed limitation(%d)"),
 367                 size, MAXNAMELENGTH);
 368                 exit(ITMC_STATUS_BT2);
 369         }
 370         data = malloc_vital(sizeof (itm_data_t));
 371
 372         data->size = size;
 373         if (size <= sizeof (data->place)) {
 374                 (void) memmove(&(data->place), seq, size);
 375         } else {
 376                 data->place.itm_ptr = (itm_place2_t)malloc_vital(size);
 377                 (void) memmove((char *)(data->place.itm_ptr), seq, size);
 378         }
 379
 380         return (data);
 381 }
 382
 383 static void
 384 filename_lineno(void)
 385 {
 386         static char     *re;
 387         static char     restr[] =
 388                         "^#[ \t]*\\([0-9]\\{1,\\}\\)[ \t]*\"\\(.*\\)\".*";
 389         int             match;
 390         static char     *filename;
 391         int             len;
 392         int             lineno;
 393
 394         if (NULL == re) {
 395                 re = compile(restr, NULL, NULL);
 396                 if (NULL == re) {
 397                         itm_error(
 398                                 gettext("REGEXP compile error\n"));
 399                         exit(ITMC_STATUS_BT);
 400                 }
 401         }
 402         match = step(yytext, re);
 403         if (0 != match) {
 404                 lineno = atoi(braslist[0]);
 405                 free(filename);
 406                 len = braelist[1] - braslist[1];
 407                 filename = malloc_vital(len + 1);
 408                 (void) memcpy(filename, braslist[1], len);
 409                 *(filename + len) = '\0';
 410                 itm_input_file = filename;
 411                 yylineno = lineno;
 412         }
 413 }