libguile/read.c

   1 /* Copyright (C) 1995,1996,1997,1999,2000,2001,2003, 2004 Free Software
   2  * Foundation, Inc.
   3  *
   4  * This library is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU Lesser General Public
   6  * License as published by the Free Software Foundation; either
   7  * version 2.1 of the License, or (at your option) any later version.
   8  *
   9  * This library is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * Lesser General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU Lesser General Public
  15  * License along with this library; if not, write to the Free Software
  16  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  17  */
  18
  19
  20 \f
  21
  22 #include <stdio.h>
  23 #include "libguile/_scm.h"
  24 #include "libguile/chars.h"
  25 #include "libguile/eval.h"
  26 #include "libguile/unif.h"
  27 #include "libguile/keywords.h"
  28 #include "libguile/alist.h"
  29 #include "libguile/srcprop.h"
  30 #include "libguile/hashtab.h"
  31 #include "libguile/hash.h"
  32 #include "libguile/ports.h"
  33 #include "libguile/root.h"
  34 #include "libguile/strings.h"
  35 #include "libguile/strports.h"
  36 #include "libguile/vectors.h"
  37 #include "libguile/validate.h"
  38
  39 #include "libguile/read.h"
  40
  41 \f
  42
  43 SCM_GLOBAL_SYMBOL (scm_sym_dot, ".");
  44 SCM_SYMBOL (scm_keyword_prefix, "prefix");
  45
  46 scm_t_option scm_read_opts[] = {
  47   { SCM_OPTION_BOOLEAN, "copy", 0,
  48     "Copy source code expressions." },
  49   { SCM_OPTION_BOOLEAN, "positions", 0,
  50     "Record positions of source code expressions." },
  51   { SCM_OPTION_BOOLEAN, "case-insensitive", 0,
  52     "Convert symbols to lower case."},
  53   { SCM_OPTION_SCM, "keywords", SCM_UNPACK (SCM_BOOL_F),
  54     "Style of keyword recognition: #f or 'prefix."}
  55 #if SCM_ENABLE_ELISP
  56   ,
  57   { SCM_OPTION_BOOLEAN, "elisp-vectors", 0,
  58     "Support Elisp vector syntax, namely `[...]'."},
  59   { SCM_OPTION_BOOLEAN, "elisp-strings", 0,
  60     "Support `\\(' and `\\)' in strings."}
  61 #endif
  62 };
  63
  64 /*
  65   Give meaningful error messages for errors
  66
  67   We use the format
  68
  69   FILE:LINE:COL: MESSAGE
  70   This happened in ....
  71
  72   This is not standard GNU format, but the test-suite likes the real
  73   message to be in front.
  74
  75  */
  76
  77
  78 static void
  79 scm_input_error(char const * function,
  80                 SCM port, const char * message, SCM arg)
  81 {
  82   char *fn = SCM_STRINGP (SCM_FILENAME(port))
  83     ? SCM_STRING_CHARS(SCM_FILENAME(port))
  84     : "#<unknown port>";
  85
  86   SCM string_port =  scm_open_output_string ();
  87   SCM string = SCM_EOL;
  88   scm_simple_format (string_port,
  89                      scm_makfrom0str ("~A:~S:~S: ~A"),
  90                      scm_list_4 (scm_makfrom0str (fn),
  91                                  scm_int2num (SCM_LINUM (port) + 1),
  92                                  scm_int2num (SCM_COL (port) + 1),
  93                                  scm_makfrom0str (message)));
  94
  95
  96   string = scm_get_output_string (string_port);
  97   scm_close_output_port (string_port);
  98   scm_error_scm (scm_str2symbol ("read-error"),
  99                  scm_makfrom0str (function),
 100                  string,
 101                  arg,
 102                  SCM_BOOL_F);
 103 }
 104
 105
 106 SCM_DEFINE (scm_read_options, "read-options-interface", 0, 1, 0,
 107             (SCM setting),
 108             "Option interface for the read options. Instead of using\n"
 109             "this procedure directly, use the procedures @code{read-enable},\n"
 110             "@code{read-disable}, @code{read-set!} and @code{read-options}.")
 111 #define FUNC_NAME s_scm_read_options
 112 {
 113   SCM ans = scm_options (setting,
 114                          scm_read_opts,
 115                          SCM_N_READ_OPTIONS,
 116                          FUNC_NAME);
 117   if (SCM_COPY_SOURCE_P)
 118     SCM_RECORD_POSITIONS_P = 1;
 119   return ans;
 120 }
 121 #undef FUNC_NAME
 122
 123 /* An association list mapping extra hash characters to procedures.  */
 124 static SCM *scm_read_hash_procedures;
 125
 126 SCM_DEFINE (scm_read, "read", 0, 1, 0,
 127             (SCM port),
 128             "Read an s-expression from the input port @var{port}, or from\n"
 129             "the current input port if @var{port} is not specified.\n"
 130             "Any whitespace before the next token is discarded.")
 131 #define FUNC_NAME s_scm_read
 132 {
 133   int c;
 134   SCM tok_buf, copy;
 135
 136   if (SCM_UNBNDP (port))
 137     port = scm_cur_inp;
 138   SCM_VALIDATE_OPINPORT (1, port);
 139
 140   c = scm_flush_ws (port, (char *) NULL);
 141   if (EOF == c)
 142     return SCM_EOF_VAL;
 143   scm_ungetc (c, port);
 144
 145   tok_buf = scm_allocate_string (30);
 146   return scm_lreadr (&tok_buf, port, &copy);
 147 }
 148 #undef FUNC_NAME
 149
 150
 151
 152 char *
 153 scm_grow_tok_buf (SCM *tok_buf)
 154 {
 155   size_t oldlen = SCM_STRING_LENGTH (*tok_buf);
 156   SCM newstr = scm_allocate_string (2 * oldlen);
 157   size_t i;
 158
 159   for (i = 0; i != oldlen; ++i)
 160     SCM_STRING_CHARS (newstr) [i] = SCM_STRING_CHARS (*tok_buf) [i];
 161
 162   *tok_buf = newstr;
 163   return SCM_STRING_CHARS (newstr);
 164 }
 165
 166
 167
 168 int
 169 scm_flush_ws (SCM port, const char *eoferr)
 170 {
 171   register int c;
 172   while (1)
 173     switch (c = scm_getc (port))
 174       {
 175       case EOF:
 176       goteof:
 177         if (eoferr)
 178           {
 179             scm_input_error (eoferr,
 180                              port,
 181                              "end of file",
 182                              SCM_EOL);
 183           }
 184         return c;
 185       case ';':
 186       lp:
 187         switch (c = scm_getc (port))
 188           {
 189           case EOF:
 190             goto goteof;
 191           default:
 192             goto lp;
 193           case SCM_LINE_INCREMENTORS:
 194             break;
 195           }
 196         break;
 197       case SCM_LINE_INCREMENTORS:
 198       case SCM_SINGLE_SPACES:
 199       case '\t':
 200         break;
 201       default:
 202         return c;
 203       }
 204 }
 205
 206
 207
 208 int
 209 scm_casei_streq (char *s1, char *s2)
 210 {
 211   while (*s1 && *s2)
 212     if (scm_downcase((int)*s1) != scm_downcase((int)*s2))
 213       return 0;
 214     else
 215       {
 216         ++s1;
 217         ++s2;
 218       }
 219   return !(*s1 || *s2);
 220 }
 221
 222
 223 /* recsexpr is used when recording expressions
 224  * constructed by read:sharp.
 225  */
 226 static SCM
 227 recsexpr (SCM obj, long line, int column, SCM filename)
 228 {
 229   if (!SCM_CONSP(obj)) {
 230     return obj;
 231   } else {
 232     SCM tmp = obj, copy;
 233     /* If this sexpr is visible in the read:sharp source, we want to
 234        keep that information, so only record non-constant cons cells
 235        which haven't previously been read by the reader. */
 236     if (SCM_FALSEP (scm_whash_lookup (scm_source_whash, obj)))
 237       {
 238         if (SCM_COPY_SOURCE_P)
 239           {
 240             copy = scm_cons (recsexpr (SCM_CAR (obj), line, column, filename),
 241                              SCM_UNDEFINED);
 242             while ((tmp = SCM_CDR (tmp)) && SCM_CONSP (tmp))
 243               {
 244                 SCM_SETCDR (copy, scm_cons (recsexpr (SCM_CAR (tmp),
 245                                                       line,
 246                                                       column,
 247                                                       filename),
 248                                             SCM_UNDEFINED));
 249                 copy = SCM_CDR (copy);
 250               }
 251             SCM_SETCDR (copy, tmp);
 252           }
 253         else
 254           {
 255             recsexpr (SCM_CAR (obj), line, column, filename);
 256             while ((tmp = SCM_CDR (tmp)) && SCM_CONSP (tmp))
 257               recsexpr (SCM_CAR (tmp), line, column, filename);
 258             copy = SCM_UNDEFINED;
 259           }
 260         scm_whash_insert (scm_source_whash,
 261                           obj,
 262                           scm_make_srcprops (line,
 263                                              column,
 264                                              filename,
 265                                              copy,
 266                                              SCM_EOL));
 267       }
 268     return obj;
 269   }
 270 }
 271
 272 /* Consume an SCSH-style block comment.  Assume that we've already
 273    read the initial `#!', and eat characters until we get a
 274    newline/exclamation-point/sharp-sign/newline sequence.
 275
 276    A carriage return is also reocgnized as a newline. */
 277
 278 static void
 279 skip_scsh_block_comment (SCM port)
 280 #define FUNC_NAME "skip_scsh_block_comment"
 281 {
 282   int state = 0;
 283
 284   for (;;)
 285     {
 286       int c = scm_getc (port);
 287
 288       if (c == EOF)
 289         SCM_MISC_ERROR ("unterminated `#! ... !#' comment", SCM_EOL);
 290
 291       if (state == 1 && c == '!')
 292         state = 2;
 293       else if (state == 2 && c == '#')
 294         state = 3;
 295       else if (state == 3 && (c == '\n' || c == '\r'))
 296         return;
 297       else if (c == '\n' || c == '\r')
 298         state = 1;
 299       else
 300         state = 0;
 301     }
 302 }
 303 #undef FUNC_NAME
 304
 305
 306 static SCM scm_get_hash_procedure(int c);
 307 static SCM scm_i_lreadparen (SCM *, SCM, char *, SCM *, char);
 308
 309 static char s_list[]="list";
 310 static char s_vector[]="vector";
 311
 312 SCM
 313 scm_lreadr (SCM *tok_buf, SCM port, SCM *copy)
 314 #define FUNC_NAME "scm_lreadr"
 315 {
 316   int c;
 317   size_t j;
 318   SCM p;
 319
 320  tryagain:
 321   c = scm_flush_ws (port, s_scm_read);
 322  tryagain_no_flush_ws:
 323   switch (c)
 324     {
 325     case EOF:
 326       return SCM_EOF_VAL;
 327
 328     case '(':
 329       return SCM_RECORD_POSITIONS_P
 330         ? scm_lreadrecparen (tok_buf, port, s_list, copy)
 331         : scm_i_lreadparen (tok_buf, port, s_list, copy, ')');
 332     case ')':
 333       scm_input_error (FUNC_NAME, port,"unexpected \")\"", SCM_EOL);
 334       goto tryagain;
 335
 336 #if SCM_ENABLE_ELISP
 337     case '[':
 338       if (SCM_ELISP_VECTORS_P)
 339         {
 340           p = scm_i_lreadparen (tok_buf, port, s_vector, copy, ']');
 341           return SCM_NULLP (p) ? scm_nullvect : scm_vector (p);
 342         }
 343       goto read_token;
 344 #endif
 345     case '\'':
 346       p = scm_sym_quote;
 347       goto recquote;
 348     case '`':
 349       p = scm_sym_quasiquote;
 350       goto recquote;
 351     case ',':
 352       c = scm_getc (port);
 353       if ('@' == c)
 354         p = scm_sym_uq_splicing;
 355       else
 356         {
 357           scm_ungetc (c, port);
 358           p = scm_sym_unquote;
 359         }
 360     recquote:
 361       p = scm_cons2 (p,
 362                      scm_lreadr (tok_buf, port, copy),
 363                      SCM_EOL);
 364       if (SCM_RECORD_POSITIONS_P)
 365         scm_whash_insert (scm_source_whash,
 366                           p,
 367                           scm_make_srcprops (SCM_LINUM (port),
 368                                              SCM_COL (port) - 1,
 369                                              SCM_FILENAME (port),
 370                                              SCM_COPY_SOURCE_P
 371                                              ? (*copy = scm_cons2 (SCM_CAR (p),
 372                                                                    SCM_CAR (SCM_CDR (p)),
 373                                                                    SCM_EOL))
 374                                              : SCM_UNDEFINED,
 375                                              SCM_EOL));
 376       return p;
 377     case '#':
 378       c = scm_getc (port);
 379
 380       {
 381         /* Check for user-defined hash procedure first, to allow
 382            overriding of builtin hash read syntaxes.  */
 383         SCM sharp = scm_get_hash_procedure (c);
 384         if (!SCM_FALSEP (sharp))
 385           {
 386             int line = SCM_LINUM (port);
 387             int column = SCM_COL (port) - 2;
 388             SCM got;
 389
 390             got = scm_call_2 (sharp, SCM_MAKE_CHAR (c), port);
 391             if (SCM_EQ_P (got, SCM_UNSPECIFIED))
 392               goto handle_sharp;
 393             if (SCM_RECORD_POSITIONS_P)
 394               return *copy = recsexpr (got, line, column,
 395                                        SCM_FILENAME (port));
 396             else
 397               return got;
 398           }
 399       }
 400     handle_sharp:
 401       switch (c)
 402         {
 403         case '(':
 404           p = scm_i_lreadparen (tok_buf, port, s_vector, copy, ')');
 405           return SCM_NULLP (p) ? scm_nullvect : scm_vector (p);
 406
 407         case 't':
 408         case 'T':
 409           return SCM_BOOL_T;
 410         case 'f':
 411         case 'F':
 412           return SCM_BOOL_F;
 413
 414         case 'b':
 415         case 'B':
 416         case 'o':
 417         case 'O':
 418         case 'd':
 419         case 'D':
 420         case 'x':
 421         case 'X':
 422         case 'i':
 423         case 'I':
 424         case 'e':
 425         case 'E':
 426           scm_ungetc (c, port);
 427           c = '#';
 428           goto num;
 429
 430         case '!':
 431           /* start of a shell script.  Parse as a block comment,
 432              terminated by !#, just like SCSH.  */
 433           skip_scsh_block_comment (port);
 434           /* EOF is not an error here */
 435           c = scm_flush_ws (port, (char *)NULL);
 436           goto tryagain_no_flush_ws;
 437
 438 #if SCM_HAVE_ARRAYS
 439         case '*':
 440           j = scm_read_token (c, tok_buf, port, 0);
 441           p = scm_istr2bve (SCM_STRING_CHARS (*tok_buf) + 1, (long) (j - 1));
 442           if (!SCM_FALSEP (p))
 443             return p;
 444           else
 445             goto unkshrp;
 446 #endif
 447
 448         case '{':
 449           j = scm_read_token (c, tok_buf, port, 1);
 450           return scm_mem2symbol (SCM_STRING_CHARS (*tok_buf), j);
 451
 452         case '\\':
 453           c = scm_getc (port);
 454           j = scm_read_token (c, tok_buf, port, 0);
 455           if (j == 1)
 456             return SCM_MAKE_CHAR (c);
 457           if (c >= '0' && c < '8')
 458             {
 459               /* Dirk:FIXME::  This type of character syntax is not R5RS
 460                * compliant.  Further, it should be verified that the constant
 461                * does only consist of octal digits.  Finally, it should be
 462                * checked whether the resulting fixnum is in the range of
 463                * characters.  */
 464               p = scm_i_mem2number (SCM_STRING_CHARS (*tok_buf), j, 8);
 465               if (SCM_INUMP (p))
 466                 return SCM_MAKE_CHAR (SCM_INUM (p));
 467             }
 468           for (c = 0; c < scm_n_charnames; c++)
 469             if (scm_charnames[c]
 470                 && (scm_casei_streq (scm_charnames[c], SCM_STRING_CHARS (*tok_buf))))
 471               return SCM_MAKE_CHAR (scm_charnums[c]);
 472           scm_input_error (FUNC_NAME, port, "unknown # object", SCM_EOL);
 473
 474           /* #:SYMBOL is a syntax for keywords supported in all contexts.  */
 475         case ':':
 476           j = scm_read_token ('-', tok_buf, port, 0);
 477           p = scm_mem2symbol (SCM_STRING_CHARS (*tok_buf), j);
 478           return scm_make_keyword_from_dash_symbol (p);
 479
 480         default:
 481         callshrp:
 482           {
 483             SCM sharp = scm_get_hash_procedure (c);
 484
 485             if (!SCM_FALSEP (sharp))
 486               {
 487                 int line = SCM_LINUM (port);
 488                 int column = SCM_COL (port) - 2;
 489                 SCM got;
 490
 491                 got = scm_call_2 (sharp, SCM_MAKE_CHAR (c), port);
 492                 if (SCM_EQ_P (got, SCM_UNSPECIFIED))
 493                   goto unkshrp;
 494                 if (SCM_RECORD_POSITIONS_P)
 495                   return *copy = recsexpr (got, line, column,
 496                                            SCM_FILENAME (port));
 497                 else
 498                   return got;
 499               }
 500           }
 501         unkshrp:
 502         scm_input_error (FUNC_NAME, port, "Unknown # object: ~S",
 503                      scm_list_1 (SCM_MAKE_CHAR (c)));
 504         }
 505
 506     case '"':
 507       j = 0;
 508       while ('"' != (c = scm_getc (port)))
 509         {
 510           if (c == EOF)
 511             str_eof: scm_input_error (FUNC_NAME, port, "end of file in string constant", SCM_EOL);
 512
 513           while (j + 2 >= SCM_STRING_LENGTH (*tok_buf))
 514             scm_grow_tok_buf (tok_buf);
 515
 516           if (c == '\\')
 517             switch (c = scm_getc (port))
 518               {
 519               case EOF:
 520                 goto str_eof;
 521               case '"':
 522               case '\\':
 523                 break;
 524 #if SCM_ENABLE_ELISP
 525               case '(':
 526               case ')':
 527                 if (SCM_ESCAPED_PARENS_P)
 528                   break;
 529                 goto bad_escaped;
 530 #endif
 531               case '\n':
 532                 continue;
 533               case '0':
 534                 c = '\0';
 535                 break;
 536               case 'f':
 537                 c = '\f';
 538                 break;
 539               case 'n':
 540                 c = '\n';
 541                 break;
 542               case 'r':
 543                 c = '\r';
 544                 break;
 545               case 't':
 546                 c = '\t';
 547                 break;
 548               case 'a':
 549                 c = '\007';
 550                 break;
 551               case 'v':
 552                 c = '\v';
 553                 break;
 554               case 'x':
 555                 {
 556                   int a, b;
 557                   a = scm_getc (port);
 558                   if (a == EOF) goto str_eof;
 559                   b = scm_getc (port);
 560                   if (b == EOF) goto str_eof;
 561                   if      ('0' <= a && a <= '9') a -= '0';
 562                   else if ('A' <= a && a <= 'F') a = a - 'A' + 10;
 563                   else if ('a' <= a && a <= 'f') a = a - 'a' + 10;
 564                   else goto bad_escaped;
 565                   if      ('0' <= b && b <= '9') b -= '0';
 566                   else if ('A' <= b && b <= 'F') b = b - 'A' + 10;
 567                   else if ('a' <= b && b <= 'f') b = b - 'a' + 10;
 568                   else goto bad_escaped;
 569                   c = a * 16 + b;
 570                   break;
 571                 }
 572               default:
 573               bad_escaped:
 574                 scm_input_error(FUNC_NAME, port,
 575                                 "illegal character in escape sequence: ~S",
 576                                 scm_list_1 (SCM_MAKE_CHAR (c)));
 577               }
 578           SCM_STRING_CHARS (*tok_buf)[j] = c;
 579           ++j;
 580         }
 581       if (j == 0)
 582         return scm_nullstr;
 583       SCM_STRING_CHARS (*tok_buf)[j] = 0;
 584       return scm_mem2string (SCM_STRING_CHARS (*tok_buf), j);
 585
 586     case '0': case '1': case '2': case '3': case '4':
 587     case '5': case '6': case '7': case '8': case '9':
 588     case '.':
 589     case '-':
 590     case '+':
 591     num:
 592       j = scm_read_token (c, tok_buf, port, 0);
 593       if (j == 1 && (c == '+' || c == '-'))
 594         /* Shortcut:  Detected symbol '+ or '- */
 595         goto tok;
 596
 597       p = scm_i_mem2number (SCM_STRING_CHARS (*tok_buf), j, 10);
 598       if (!SCM_FALSEP (p))
 599         return p;
 600       if (c == '#')
 601         {
 602           if ((j == 2) && (scm_getc (port) == '('))
 603             {
 604               scm_ungetc ('(', port);
 605               c = SCM_STRING_CHARS (*tok_buf)[1];
 606               goto callshrp;
 607             }
 608           scm_input_error (FUNC_NAME, port, "unknown # object", SCM_EOL);
 609         }
 610       goto tok;
 611
 612     case ':':
 613       if (SCM_EQ_P (SCM_PACK (SCM_KEYWORD_STYLE), scm_keyword_prefix))
 614         {
 615           j = scm_read_token ('-', tok_buf, port, 0);
 616           p = scm_mem2symbol (SCM_STRING_CHARS (*tok_buf), j);
 617           return scm_make_keyword_from_dash_symbol (p);
 618         }
 619       /* fallthrough */
 620     default:
 621 #if SCM_ENABLE_ELISP
 622     read_token:
 623 #endif
 624       j = scm_read_token (c, tok_buf, port, 0);
 625       /* fallthrough */
 626
 627     tok:
 628       return scm_mem2symbol (SCM_STRING_CHARS (*tok_buf), j);
 629     }
 630 }
 631 #undef FUNC_NAME
 632
 633
 634 #ifdef _UNICOS
 635 _Pragma ("noopt");              /* # pragma _CRI noopt */
 636 #endif
 637
 638 size_t
 639 scm_read_token (int ic, SCM *tok_buf, SCM port, int weird)
 640 {
 641   register size_t j;
 642   register int c;
 643   register char *p;
 644
 645   c = (SCM_CASE_INSENSITIVE_P ? scm_downcase(ic) : ic);
 646   p = SCM_STRING_CHARS (*tok_buf);
 647
 648   if (weird)
 649     j = 0;
 650   else
 651     {
 652       j = 0;
 653       while (j + 2 >= SCM_STRING_LENGTH (*tok_buf))
 654         p = scm_grow_tok_buf (tok_buf);
 655       p[j] = c;
 656       ++j;
 657     }
 658
 659   while (1)
 660     {
 661       while (j + 2 >= SCM_STRING_LENGTH (*tok_buf))
 662         p = scm_grow_tok_buf (tok_buf);
 663       c = scm_getc (port);
 664       switch (c)
 665         {
 666         case '(':
 667         case ')':
 668 #if SCM_ENABLE_ELISP
 669         case '[':
 670         case ']':
 671 #endif
 672         case '"':
 673         case ';':
 674         case SCM_WHITE_SPACES:
 675         case SCM_LINE_INCREMENTORS:
 676           if (weird
 677 #if SCM_ENABLE_ELISP
 678               || ((!SCM_ELISP_VECTORS_P) && ((c == '[') || (c == ']')))
 679 #endif
 680               )
 681             goto default_case;
 682
 683           scm_ungetc (c, port);
 684         case EOF:
 685         eof_case:
 686           p[j] = 0;
 687           return j;
 688         case '\\':
 689           if (!weird)
 690             goto default_case;
 691           else
 692             {
 693               c = scm_getc (port);
 694               if (c == EOF)
 695                 goto eof_case;
 696               else
 697                 goto default_case;
 698             }
 699         case '}':
 700           if (!weird)
 701             goto default_case;
 702
 703           c = scm_getc (port);
 704           if (c == '#')
 705             {
 706               p[j] = 0;
 707               return j;
 708             }
 709           else
 710             {
 711               scm_ungetc (c, port);
 712               c = '}';
 713               goto default_case;
 714             }
 715
 716         default:
 717         default_case:
 718           {
 719             c = (SCM_CASE_INSENSITIVE_P ? scm_downcase(c) : c);
 720             p[j] = c;
 721             ++j;
 722           }
 723
 724         }
 725     }
 726 }
 727
 728 #ifdef _UNICOS
 729 _Pragma ("opt");                /* # pragma _CRI opt */
 730 #endif
 731
 732 static SCM
 733 scm_i_lreadparen (SCM *tok_buf, SCM port, char *name, SCM *copy, char term_char)
 734 #define FUNC_NAME "scm_i_lreadparen"
 735 {
 736   SCM tmp;
 737   SCM tl;
 738   SCM ans;
 739   int c;
 740
 741   c = scm_flush_ws (port, name);
 742   if (term_char == c)
 743     return SCM_EOL;
 744   scm_ungetc (c, port);
 745   if (SCM_EQ_P (scm_sym_dot, (tmp = scm_lreadr (tok_buf, port, copy))))
 746     {
 747       ans = scm_lreadr (tok_buf, port, copy);
 748     closeit:
 749       if (term_char != (c = scm_flush_ws (port, name)))
 750         scm_input_error (FUNC_NAME, port, "missing close paren", SCM_EOL);
 751       return ans;
 752     }
 753   ans = tl = scm_cons (tmp, SCM_EOL);
 754   while (term_char != (c = scm_flush_ws (port, name)))
 755     {
 756       scm_ungetc (c, port);
 757       if (SCM_EQ_P (scm_sym_dot, (tmp = scm_lreadr (tok_buf, port, copy))))
 758         {
 759           SCM_SETCDR (tl, scm_lreadr (tok_buf, port, copy));
 760           goto closeit;
 761         }
 762       SCM_SETCDR (tl, scm_cons (tmp, SCM_EOL));
 763       tl = SCM_CDR (tl);
 764     }
 765   return ans;
 766 }
 767 #undef FUNC_NAME
 768
 769
 770 SCM
 771 scm_lreadrecparen (SCM *tok_buf, SCM port, char *name, SCM *copy)
 772 #define FUNC_NAME "scm_lreadrecparen"
 773 {
 774   register int c;
 775   register SCM tmp;
 776   register SCM tl, tl2 = SCM_EOL;
 777   SCM ans, ans2 = SCM_EOL;
 778   /* Need to capture line and column numbers here. */
 779   int line = SCM_LINUM (port);
 780   int column = SCM_COL (port) - 1;
 781
 782   c = scm_flush_ws (port, name);
 783   if (')' == c)
 784     return SCM_EOL;
 785   scm_ungetc (c, port);
 786   if (SCM_EQ_P (scm_sym_dot, (tmp = scm_lreadr (tok_buf, port, copy))))
 787     {
 788       ans = scm_lreadr (tok_buf, port, copy);
 789       if (')' != (c = scm_flush_ws (port, name)))
 790         scm_input_error (FUNC_NAME, port, "missing close paren", SCM_EOL);
 791       return ans;
 792     }
 793   /* Build the head of the list structure. */
 794   ans = tl = scm_cons (tmp, SCM_EOL);
 795   if (SCM_COPY_SOURCE_P)
 796     ans2 = tl2 = scm_cons (SCM_CONSP (tmp)
 797                            ? *copy
 798                            : tmp,
 799                            SCM_EOL);
 800   while (')' != (c = scm_flush_ws (port, name)))
 801     {
 802       SCM new_tail;
 803
 804       scm_ungetc (c, port);
 805       if (SCM_EQ_P (scm_sym_dot, (tmp = scm_lreadr (tok_buf, port, copy))))
 806         {
 807           SCM_SETCDR (tl, tmp = scm_lreadr (tok_buf, port, copy));
 808           if (SCM_COPY_SOURCE_P)
 809             SCM_SETCDR (tl2, scm_cons (SCM_CONSP (tmp)
 810                                        ? *copy
 811                                        : tmp,
 812                                        SCM_EOL));
 813           if (')' != (c = scm_flush_ws (port, name)))
 814             scm_input_error (FUNC_NAME, port, "missing close paren", SCM_EOL);
 815           goto exit;
 816         }
 817
 818       new_tail = scm_cons (tmp, SCM_EOL);
 819       SCM_SETCDR (tl, new_tail);
 820       tl = new_tail;
 821
 822       if (SCM_COPY_SOURCE_P)
 823         {
 824           SCM new_tail2 = scm_cons (SCM_CONSP (tmp) ? *copy : tmp, SCM_EOL);
 825           SCM_SETCDR (tl2, new_tail2);
 826           tl2 = new_tail2;
 827         }
 828     }
 829 exit:
 830   scm_whash_insert (scm_source_whash,
 831                     ans,
 832                     scm_make_srcprops (line,
 833                                        column,
 834                                        SCM_FILENAME (port),
 835                                        SCM_COPY_SOURCE_P
 836                                        ? *copy = ans2
 837                                        : SCM_UNDEFINED,
 838                                        SCM_EOL));
 839   return ans;
 840 }
 841 #undef FUNC_NAME
 842
 843
 844 \f
 845
 846 /* Manipulate the read-hash-procedures alist.  This could be written in
 847    Scheme, but maybe it will also be used by C code during initialisation.  */
 848 SCM_DEFINE (scm_read_hash_extend, "read-hash-extend", 2, 0, 0,
 849             (SCM chr, SCM proc),
 850             "Install the procedure @var{proc} for reading expressions\n"
 851             "starting with the character sequence @code{#} and @var{chr}.\n"
 852             "@var{proc} will be called with two arguments:  the character\n"
 853             "@var{chr} and the port to read further data from. The object\n"
 854             "returned will be the return value of @code{read}.")
 855 #define FUNC_NAME s_scm_read_hash_extend
 856 {
 857   SCM this;
 858   SCM prev;
 859
 860   SCM_VALIDATE_CHAR (1, chr);
 861   SCM_ASSERT (SCM_FALSEP (proc)
 862               || SCM_EQ_P (scm_procedure_p (proc), SCM_BOOL_T),
 863               proc, SCM_ARG2, FUNC_NAME);
 864
 865   /* Check if chr is already in the alist.  */
 866   this = *scm_read_hash_procedures;
 867   prev = SCM_BOOL_F;
 868   while (1)
 869     {
 870       if (SCM_NULLP (this))
 871         {
 872           /* not found, so add it to the beginning.  */
 873           if (!SCM_FALSEP (proc))
 874             {
 875               *scm_read_hash_procedures =
 876                 scm_cons (scm_cons (chr, proc), *scm_read_hash_procedures);
 877             }
 878           break;
 879         }
 880       if (SCM_EQ_P (chr, SCM_CAAR (this)))
 881         {
 882           /* already in the alist.  */
 883           if (SCM_FALSEP (proc))
 884             {
 885               /* remove it.  */
 886               if (SCM_FALSEP (prev))
 887                 {
 888                   *scm_read_hash_procedures =
 889                     SCM_CDR (*scm_read_hash_procedures);
 890                 }
 891               else
 892                 scm_set_cdr_x (prev, SCM_CDR (this));
 893             }
 894           else
 895             {
 896               /* replace it.  */
 897               scm_set_cdr_x (SCM_CAR (this), proc);
 898             }
 899           break;
 900         }
 901       prev = this;
 902       this = SCM_CDR (this);
 903     }
 904
 905   return SCM_UNSPECIFIED;
 906 }
 907 #undef FUNC_NAME
 908
 909 /* Recover the read-hash procedure corresponding to char c.  */
 910 static SCM
 911 scm_get_hash_procedure (int c)
 912 {
 913   SCM rest = *scm_read_hash_procedures;
 914
 915   while (1)
 916     {
 917       if (SCM_NULLP (rest))
 918         return SCM_BOOL_F;
 919
 920       if (SCM_CHAR (SCM_CAAR (rest)) == c)
 921         return SCM_CDAR (rest);
 922
 923       rest = SCM_CDR (rest);
 924     }
 925 }
 926
 927 void
 928 scm_init_read ()
 929 {
 930   scm_read_hash_procedures =
 931     SCM_VARIABLE_LOC (scm_c_define ("read-hash-procedures", SCM_EOL));
 932
 933   scm_init_opts (scm_read_options, scm_read_opts, SCM_N_READ_OPTIONS);
 934 #include "libguile/read.x"
 935 }
 936
 937 /*
 938   Local Variables:
 939   c-file-style: "gnu"
 940   End:
 941 */