giscanner/scannerlexer.l

   1 /* -*- Mode: C -*-
   2  * GObject introspection: C lexer
   3  *
   4  * Copyright (c) 1997 Sandro Sigala  <ssigala@globalnet.it>
   5  * Copyright (c) 2007-2008 Jürg Billeter  <j@bitron.ch>
   6  *
   7  * All rights reserved.
   8  *
   9  * Redistribution and use in source and binary forms, with or without
  10  * modification, are permitted provided that the following conditions
  11  * are met:
  12  * 1. Redistributions of source code must retain the above copyright
  13  *    notice, this list of conditions and the following disclaimer.
  14  * 2. Redistributions in binary form must reproduce the above copyright
  15  *    notice, this list of conditions and the following disclaimer in the
  16  *    documentation and/or other materials provided with the distribution.
  17  *
  18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
  19  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  20  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  21  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
  22  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  23  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  24  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  25  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  26  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  27  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  28  */
  29
  30 %{
  31 #include <ctype.h>
  32 #include <stdio.h>
  33
  34 #include <glib.h>
  35 #include "sourcescanner.h"
  36 #include "scannerparser.h"
  37 #include "grealpath.h"
  38
  39 int lineno;
  40
  41 extern int yylex (GISourceScanner *scanner);
  42 #define YY_DECL int yylex (GISourceScanner *scanner)
  43 static int yywrap (void);
  44 static void parse_comment (GISourceScanner *scanner);
  45 static void process_directive (GISourceScanner *scanner);
  46 static int check_identifier (GISourceScanner *scanner, const char *);
  47 static int parse_ignored_macro (void);
  48 %}
  49
  50 intsuffix                               ([uU][lL]?[lL]?)|([lL][lL]?[uU]?)
  51 fracconst                               ([0-9]*\.[0-9]+)|([0-9]+\.)
  52 exppart                                 [eE][-+]?[0-9]+
  53 floatsuffix                             [fFlL]
  54 chartext                                ([^\\\'])|(\\.)
  55 stringtext                              ([^\\\"])|(\\.)
  56
  57 %%
  58
  59 "\n"                                    { ++lineno; } /* " */
  60 "\\\n"                                  { ++lineno; }
  61 [\t\f\v\r ]+                            { /* Ignore whitespace. */ }
  62
  63 "/*"                                    { parse_comment(scanner); }
  64 "//".*                                  { }
  65
  66 "#define "[a-zA-Z_][a-zA-Z_0-9]*"("     { yyless (yyleng - 1); return FUNCTION_MACRO; }
  67 "#define "[a-zA-Z_][a-zA-Z_0-9]*        { return OBJECT_MACRO; }
  68
  69 "#"                                     { process_directive(scanner); }
  70
  71 "{"                                     { return '{'; }
  72 "<%"                                    { return '{'; }
  73 "}"                                     { return '}'; }
  74 "%>"                                    { return '}'; }
  75 "["                                     { return '['; }
  76 "<:"                                    { return '['; }
  77 "]"                                     { return ']'; }
  78 ":>"                                    { return ']'; }
  79 "("                                     { return '('; }
  80 ")"                                     { return ')'; }
  81 ";"                                     { return ';'; }
  82 ":"                                     { return ':'; }
  83 "..."                                   { return ELLIPSIS; }
  84 "?"                                     { return '?'; }
  85 "."                                     { return '.'; }
  86 "+"                                     { return '+'; }
  87 "-"                                     { return '-'; }
  88 "*"                                     { return '*'; }
  89 "/"                                     { return '/'; }
  90 "%"                                     { return '%'; }
  91 "^"                                     { return '^'; }
  92 "&"                                     { return '&'; }
  93 "|"                                     { return '|'; }
  94 "~"                                     { return '~'; }
  95 "!"                                     { return '!'; }
  96 "="                                     { return '='; }
  97 "<"                                     { return '<'; }
  98 ">"                                     { return '>'; }
  99 "+="                                    { return ADDEQ; }
 100 "-="                                    { return SUBEQ; }
 101 "*="                                    { return MULEQ; }
 102 "/="                                    { return DIVEQ; }
 103 "%="                                    { return MODEQ; }
 104 "^="                                    { return XOREQ; }
 105 "&="                                    { return ANDEQ; }
 106 "|="                                    { return OREQ; }
 107 "<<"                                    { return SL; }
 108 ">>"                                    { return SR; }
 109 "<<="                                   { return SLEQ; }
 110 ">>="                                   { return SREQ; }
 111 "=="                                    { return EQ; }
 112 "!="                                    { return NOTEQ; }
 113 "<="                                    { return LTEQ; }
 114 ">="                                    { return GTEQ; }
 115 "&&"                                    { return ANDAND; }
 116 "||"                                    { return OROR; }
 117 "++"                                    { return PLUSPLUS; }
 118 "--"                                    { return MINUSMINUS; }
 119 ","                                     { return ','; }
 120 "->"                                    { return ARROW; }
 121
 122 "__asm"                                 { if (!parse_ignored_macro()) REJECT; }
 123 "__asm__"                               { if (!parse_ignored_macro()) REJECT; }
 124 "__attribute__"                         { if (!parse_ignored_macro()) REJECT; }
 125 "__attribute"                           { if (!parse_ignored_macro()) REJECT; }
 126 "__const"                               { return CONST; }
 127 "__extension__"                         { return EXTENSION; }
 128 "__inline"                              { return INLINE; }
 129 "__nonnull"                             { if (!parse_ignored_macro()) REJECT; }
 130 "__signed__"                            { return SIGNED; }
 131 "__restrict"                            { return RESTRICT; }
 132 "__typeof"                              { if (!parse_ignored_macro()) REJECT; }
 133 "_Bool"                                 { return BOOL; }
 134
 135 [a-zA-Z_][a-zA-Z_0-9]*                  { if (scanner->macro_scan) return IDENTIFIER; else REJECT; }
 136
 137 "asm"                                   { if (!parse_ignored_macro()) REJECT; }
 138 "auto"                                  { return AUTO; }
 139 "break"                                 { return BREAK; }
 140 "case"                                  { return CASE; }
 141 "char"                                  { return CHAR; }
 142 "const"                                 { return CONST; }
 143 "continue"                              { return CONTINUE; }
 144 "default"                               { return DEFAULT; }
 145 "do"                                    { return DO; }
 146 "double"                                { return DOUBLE; }
 147 "else"                                  { return ELSE; }
 148 "enum"                                  { return ENUM; }
 149 "extern"                                { return EXTERN; }
 150 "float"                                 { return FLOAT; }
 151 "for"                                   { return FOR; }
 152 "goto"                                  { return GOTO; }
 153 "if"                                    { return IF; }
 154 "inline"                                { return INLINE; }
 155 "int"                                   { return INT; }
 156 "long"                                  { return LONG; }
 157 "register"                              { return REGISTER; }
 158 "restrict"                              { return RESTRICT; }
 159 "return"                                { return RETURN; }
 160 "short"                                 { return SHORT; }
 161 "signed"                                { return SIGNED; }
 162 "sizeof"                                { return SIZEOF; }
 163 "static"                                { return STATIC; }
 164 "struct"                                { return STRUCT; }
 165 "switch"                                { return SWITCH; }
 166 "typedef"                               { return TYPEDEF; }
 167 "union"                                 { return UNION; }
 168 "unsigned"                              { return UNSIGNED; }
 169 "void"                                  { return VOID; }
 170 "volatile"                              { return VOLATILE; }
 171 "while"                                 { return WHILE; }
 172
 173 [a-zA-Z_][a-zA-Z_0-9]*                  { return check_identifier(scanner, yytext); }
 174
 175 "0"[xX][0-9a-fA-F]+{intsuffix}?         { return INTEGER; }
 176 "0"[0-7]+{intsuffix}?                   { return INTEGER; }
 177 [0-9]+{intsuffix}?                      { return INTEGER; }
 178
 179 {fracconst}{exppart}?{floatsuffix}?     { return FLOATING; }
 180 [0-9]+{exppart}{floatsuffix}?           { return FLOATING; }
 181
 182 "'"{chartext}*"'"                       { return CHARACTER; }
 183 "L'"{chartext}*"'"                      { return CHARACTER; }
 184
 185 "\""{stringtext}*"\""                   { return STRING; }
 186 "L\""{stringtext}*"\""                  { return STRING; }
 187
 188 .                                       { fprintf(stderr, "%s:%d: unexpected character `%c'\n", scanner->current_filename, lineno, yytext[0]); }
 189
 190 %%
 191
 192 static int
 193 yywrap (void)
 194 {
 195   return 1;
 196 }
 197
 198
 199 static void
 200 parse_gtkdoc (GISourceScanner *scanner,
 201               gchar           *symbol,
 202               int             *c1,
 203               int             *c2)
 204 {
 205   gboolean isline = FALSE;
 206   GString *line_buf;
 207   char *line;
 208   gchar **parts;
 209   GISourceDirective *directive;
 210   char *name,*value;
 211   GSList *directives;
 212   GSList *options = NULL;
 213   char *rname;
 214   int n_parts;
 215
 216   line_buf = g_string_new ("");
 217
 218   do
 219     {
 220       *c1 = *c2;
 221       if (*c1 == '\n')
 222         {
 223           isline = TRUE;
 224           break;
 225         }
 226       g_string_append_c (line_buf, *c1);
 227       *c2 = input();
 228     } while (*c2 != EOF && !(*c1 == '*' && *c2 == '/'));
 229
 230   if (!isline)
 231     {
 232       g_string_free (line_buf, TRUE);
 233       return;
 234     }
 235
 236   line = g_string_free (line_buf, FALSE);
 237
 238   /* Ignore lines that don't have a : - this is a hack but avoids
 239    * trying to parse too many things as annotations
 240    */
 241   if (!strchr (line, ':'))
 242     {
 243       g_free (line);
 244       return;
 245     }
 246
 247   parts = g_strsplit (line, ":", 3);
 248   n_parts = g_strv_length (parts);
 249
 250   if (g_ascii_strcasecmp (parts[0], "eprecated") == 0)
 251     {
 252       if (n_parts == 3)
 253         options = g_slist_prepend (options, g_strdup_printf ("%s: %s", parts[1], parts[2]));
 254       else if (n_parts == 2)
 255         options = g_slist_prepend (options, g_strdup (parts[1]));
 256       else
 257         options = g_slist_prepend (options, g_strdup (""));
 258       name = parts[0];
 259       value = NULL;
 260     }
 261   else if (n_parts >= 2)
 262     {
 263       name = parts[0];
 264
 265       if (n_parts == 3)
 266         {
 267           char *ptr = g_strdup (parts[1]);
 268           char *start;
 269           char *end;
 270
 271           options = NULL;
 272           start = strchr (ptr, '(');
 273           while (start != NULL)
 274             {
 275               end = strchr (start, ')');
 276               if (end)
 277                 {
 278                   options = g_slist_prepend (options, g_strndup (start+1, end-(start+1)));
 279                   start = strchr (end+1, '(');
 280                 }
 281               else
 282                 {
 283                   break;
 284                 }
 285             }
 286           g_free (ptr);
 287           value = parts[2];
 288         }
 289       else
 290         value = parts[1];
 291     }
 292   else /* parts == 1 */
 293     {
 294       name = parts[0];
 295       value = NULL;
 296     }
 297
 298   /*
 299    * Special cases for global annotations.
 300    * Context-sensitive parsing would probably be the right way to go.
 301    */
 302   if (g_ascii_strncasecmp ("eturn", name, 5) == 0)
 303     rname = "return";
 304   else if (g_ascii_strncasecmp ("eprecated", name, 9) == 0)
 305     rname = "deprecated";
 306   else
 307     rname = name;
 308
 309   directive = gi_source_directive_new (rname, value, options);
 310   directives = g_hash_table_lookup (scanner->directives_map, symbol);
 311   directives = g_slist_prepend (directives, directive);
 312   g_hash_table_replace (scanner->directives_map,
 313                         g_strdup (symbol), directives);
 314
 315   g_strfreev (parts);
 316   g_free (line);
 317 }
 318
 319
 320 static void
 321 parse_comment (GISourceScanner *scanner)
 322 {
 323   GString *symbol = NULL;
 324   gboolean startofline = FALSE, have_symbol = FALSE, start1 = FALSE, start_symbol = FALSE;
 325   int c1, c2;
 326
 327   c1 = input();
 328   c2 = input();
 329
 330   while (c2 != EOF && !(c1 == '*' && c2 == '/'))
 331     {
 332       if (c1 == ':')
 333         have_symbol = TRUE;
 334       else if (c1 == '\n')
 335          start1 = TRUE;
 336       else if (c1 == '*' && start1)
 337          start_symbol = TRUE;
 338       else if (!have_symbol && start_symbol)
 339         {
 340           if (!symbol)
 341             symbol = g_string_new ("");
 342           if (c1 != ' ')
 343             g_string_append_c (symbol, c1);
 344         }
 345
 346       if (c1 == '\n')
 347         {
 348           ++lineno;
 349           startofline = TRUE;
 350         }
 351
 352       c1 = c2;
 353       c2 = input();
 354
 355       if ((c1 != '*' && c1 != ' '))
 356           startofline = FALSE;
 357
 358       if (startofline && (c1 == ' ') && (c2 == '@' || (c2 == 'r') || (c2 == 'R') || (c2 == 'D')))
 359         {
 360            c1 = c2;
 361            c2 = input();
 362            if (symbol)
 363              parse_gtkdoc (scanner, symbol->str, &c1, &c2);
 364         }
 365     }
 366
 367   if (symbol)
 368     g_string_free (symbol, TRUE);
 369
 370 }
 371
 372 static int
 373 check_identifier (GISourceScanner *scanner,
 374                   const char  *s)
 375 {
 376         /*
 377          * This function checks if `s' is a type name or an
 378          * identifier.
 379          */
 380
 381         if (gi_source_scanner_is_typedef (scanner, s)) {
 382                 return TYPEDEF_NAME;
 383         } else if (strcmp (s, "__builtin_va_list") == 0) {
 384                 return TYPEDEF_NAME;
 385         }
 386
 387         return IDENTIFIER;
 388 }
 389
 390 static void
 391 process_directive (GISourceScanner *scanner)
 392 {
 393         /* extract current filename from #line directives */
 394         GString *filename_builder;
 395         gboolean in_string, found_filename;
 396
 397         lineno = 0;
 398         found_filename = FALSE;
 399         in_string = FALSE;
 400         filename_builder = g_string_new ("");
 401
 402         int c = input ();
 403         while (c != EOF && c != '\n') {
 404                 if (!in_string) {
 405                         if (c == '\"') {
 406                                 in_string = TRUE;
 407                                 found_filename = TRUE;
 408                         } else if (c >= '0' && c <= '9') {
 409                                 if (!found_filename) {
 410                                         lineno = lineno * 10 + (c - '0');
 411                                 }
 412                         }
 413                 } else {
 414                         if (c == '\"') {
 415                                 in_string = FALSE;
 416                         } else if (c == '\\') {
 417                                 g_string_append_c (filename_builder, c);
 418                                 c = input ();
 419                                 g_string_append_c (filename_builder, c);
 420                         } else {
 421                                 g_string_append_c (filename_builder, c);
 422                         }
 423                 }
 424                 c = input ();
 425         }
 426
 427         if (filename_builder->len > 0) {
 428                 char *filename = g_strcompress (filename_builder->str);
 429                 if (g_realpath (filename))
 430                   {
 431                     g_free (scanner->current_filename);
 432                     scanner->current_filename = g_realpath (filename);
 433                     g_assert (scanner->current_filename);
 434                     g_free(filename);
 435                   }
 436         }
 437
 438         g_string_free (filename_builder, TRUE);
 439 }
 440
 441 /*
 442  * This parses a macro which is ignored, such as
 443  * __attribute__((x)) or __asm__ (x)
 444  */
 445 static int
 446 parse_ignored_macro (void)
 447 {
 448         int c;
 449         int nest;
 450
 451         while ((c = input ()) != EOF && isspace (c))
 452                 ;
 453         if (c != '(')
 454                 return FALSE;
 455
 456         nest = 0;
 457         while ((c = input ()) != EOF && (nest > 0 || c != ')')) {
 458                 if (c == '(')
 459                         nest++;
 460                 else if (c == ')')
 461                         nest--;
 462                 else if (c == '"') {
 463                         while ((c = input ()) != EOF && c != '"') {
 464                                 if (c == '\\')
 465                                         c = input ();
 466                         }
 467                 } else if (c == '\'') {
 468                         c = input ();
 469                         if (c == '\\')
 470                                 c = input ();
 471                         else if (c == '\'')
 472                                 return FALSE;
 473                         c = input ();
 474                         if (c != '\'')
 475                                 return FALSE;
 476                 } else if (c == '\n')
 477                         lineno++;
 478         }
 479
 480         return TRUE;
 481 }