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 char linebuf[2000];
  41
  42 #undef YY_BUF_SIZE
  43 #define YY_BUF_SIZE 1048576
  44
  45 extern int yylex (GISourceScanner *scanner);
  46 #define YY_DECL int yylex (GISourceScanner *scanner)
  47 static int yywrap (void);
  48 static void parse_comment (GISourceScanner *scanner);
  49 static void process_directive (GISourceScanner *scanner);
  50 static int check_identifier (GISourceScanner *scanner, const char *);
  51 static int parse_ignored_macro (void);
  52 %}
  53
  54 %option nounput
  55
  56 intsuffix                               ([uU][lL]?[lL]?)|([lL][lL]?[uU]?)
  57 fracconst                               ([0-9]*\.[0-9]+)|([0-9]+\.)
  58 exppart                                 [eE][-+]?[0-9]+
  59 floatsuffix                             [fFlL]
  60 chartext                                ([^\\\'])|(\\.)
  61 stringtext                              ([^\\\"])|(\\.)
  62
  63 %%
  64
  65 \n.*                                    { strncpy(linebuf, yytext+1, sizeof(linebuf)); /* save the next line */
  66                                                 linebuf[sizeof(linebuf)-1]='\0';
  67                                                 /* printf("%4d:%s\n",lineno,linebuf); */
  68                                                 yyless(1);      /* give back all but the \n to rescan */
  69                                                 ++lineno;
  70                                         }
  71 "\\\n"                                  { ++lineno; }
  72 [\t\f\v\r ]+                            { /* Ignore whitespace. */ }
  73
  74 "/*"                                    { parse_comment(scanner); }
  75 "//".*                                  { }
  76
  77 "#define "[a-zA-Z_][a-zA-Z_0-9]*"("     { yyless (yyleng - 1); return FUNCTION_MACRO; }
  78 "#define "[a-zA-Z_][a-zA-Z_0-9]*        { return OBJECT_MACRO; }
  79
  80 "#"                                     { process_directive(scanner); }
  81
  82 "{"                                     { return '{'; }
  83 "<%"                                    { return '{'; }
  84 "}"                                     { return '}'; }
  85 "%>"                                    { return '}'; }
  86 "["                                     { return '['; }
  87 "<:"                                    { return '['; }
  88 "]"                                     { return ']'; }
  89 ":>"                                    { return ']'; }
  90 "("                                     { return '('; }
  91 ")"                                     { return ')'; }
  92 ";"                                     { return ';'; }
  93 ":"                                     { return ':'; }
  94 "..."                                   { return ELLIPSIS; }
  95 "?"                                     { return '?'; }
  96 "."                                     { return '.'; }
  97 "+"                                     { return '+'; }
  98 "-"                                     { return '-'; }
  99 "*"                                     { return '*'; }
 100 "/"                                     { return '/'; }
 101 "%"                                     { return '%'; }
 102 "^"                                     { return '^'; }
 103 "&"                                     { return '&'; }
 104 "|"                                     { return '|'; }
 105 "~"                                     { return '~'; }
 106 "!"                                     { return '!'; }
 107 "="                                     { return '='; }
 108 "<"                                     { return '<'; }
 109 ">"                                     { return '>'; }
 110 "+="                                    { return ADDEQ; }
 111 "-="                                    { return SUBEQ; }
 112 "*="                                    { return MULEQ; }
 113 "/="                                    { return DIVEQ; }
 114 "%="                                    { return MODEQ; }
 115 "^="                                    { return XOREQ; }
 116 "&="                                    { return ANDEQ; }
 117 "|="                                    { return OREQ; }
 118 "<<"                                    { return SL; }
 119 ">>"                                    { return SR; }
 120 "<<="                                   { return SLEQ; }
 121 ">>="                                   { return SREQ; }
 122 "=="                                    { return EQ; }
 123 "!="                                    { return NOTEQ; }
 124 "<="                                    { return LTEQ; }
 125 ">="                                    { return GTEQ; }
 126 "&&"                                    { return ANDAND; }
 127 "||"                                    { return OROR; }
 128 "++"                                    { return PLUSPLUS; }
 129 "--"                                    { return MINUSMINUS; }
 130 ","                                     { return ','; }
 131 "->"                                    { return ARROW; }
 132
 133 "__asm"                                 { if (!parse_ignored_macro()) REJECT; }
 134 "__asm__"                               { if (!parse_ignored_macro()) REJECT; }
 135 "__attribute__"                         { if (!parse_ignored_macro()) REJECT; }
 136 "__attribute"                           { if (!parse_ignored_macro()) REJECT; }
 137 "__const"                               { return CONST; }
 138 "__extension__"                         { return EXTENSION; }
 139 "__inline"                              { return INLINE; }
 140 "__nonnull"                             { if (!parse_ignored_macro()) REJECT; }
 141 "__signed__"                            { return SIGNED; }
 142 "__restrict"                            { return RESTRICT; }
 143 "__typeof"                              { if (!parse_ignored_macro()) REJECT; }
 144 "_Bool"                                 { return BOOL; }
 145
 146 [a-zA-Z_][a-zA-Z_0-9]*                  { if (scanner->macro_scan) return IDENTIFIER; else REJECT; }
 147
 148 "asm"                                   { if (!parse_ignored_macro()) REJECT; }
 149 "auto"                                  { return AUTO; }
 150 "break"                                 { return BREAK; }
 151 "case"                                  { return CASE; }
 152 "char"                                  { return CHAR; }
 153 "const"                                 { return CONST; }
 154 "continue"                              { return CONTINUE; }
 155 "default"                               { return DEFAULT; }
 156 "do"                                    { return DO; }
 157 "double"                                { return DOUBLE; }
 158 "else"                                  { return ELSE; }
 159 "enum"                                  { return ENUM; }
 160 "extern"                                { return EXTERN; }
 161 "float"                                 { return FLOAT; }
 162 "for"                                   { return FOR; }
 163 "goto"                                  { return GOTO; }
 164 "if"                                    { return IF; }
 165 "inline"                                { return INLINE; }
 166 "int"                                   { return INT; }
 167 "long"                                  { return LONG; }
 168 "register"                              { return REGISTER; }
 169 "restrict"                              { return RESTRICT; }
 170 "return"                                { return RETURN; }
 171 "short"                                 { return SHORT; }
 172 "signed"                                { return SIGNED; }
 173 "sizeof"                                { return SIZEOF; }
 174 "static"                                { return STATIC; }
 175 "struct"                                { return STRUCT; }
 176 "switch"                                { return SWITCH; }
 177 "typedef"                               { return TYPEDEF; }
 178 "union"                                 { return UNION; }
 179 "unsigned"                              { return UNSIGNED; }
 180 "void"                                  { return VOID; }
 181 "volatile"                              { return VOLATILE; }
 182 "while"                                 { return WHILE; }
 183
 184 [a-zA-Z_][a-zA-Z_0-9]*                  { return check_identifier(scanner, yytext); }
 185
 186 "0"[xX][0-9a-fA-F]+{intsuffix}?         { return INTEGER; }
 187 "0"[0-7]+{intsuffix}?                   { return INTEGER; }
 188 [0-9]+{intsuffix}?                      { return INTEGER; }
 189
 190 {fracconst}{exppart}?{floatsuffix}?     { return FLOATING; }
 191 [0-9]+{exppart}{floatsuffix}?           { return FLOATING; }
 192
 193 "'"{chartext}*"'"                       { return CHARACTER; }
 194 "L'"{chartext}*"'"                      { return CHARACTER; }
 195
 196 "\""{stringtext}*"\""                   { return STRING; }
 197 "L\""{stringtext}*"\""                  { return STRING; }
 198
 199 .                                       { if (yytext[0]) fprintf(stderr, "%s:%d: unexpected character `%c'\n", scanner->current_filename, lineno, yytext[0]); }
 200
 201 %%
 202
 203 static int
 204 yywrap (void)
 205 {
 206   return 1;
 207 }
 208
 209
 210 static void
 211 parse_comment (GISourceScanner *scanner)
 212 {
 213   GString *comment;
 214   int c1, c2;
 215
 216   c1 = input();
 217   c2 = input();
 218
 219   comment = g_string_new ("");
 220
 221   while (c2 != EOF && !(c1 == '*' && c2 == '/'))
 222     {
 223       g_string_append_c (comment, c1);
 224
 225       if (c1 == '\n')
 226         lineno++;
 227
 228       c1 = c2;
 229       c2 = input();
 230
 231     }
 232
 233   scanner->comments = g_slist_prepend (scanner->comments,
 234                                        g_string_free (comment, FALSE));
 235 }
 236
 237 static int
 238 check_identifier (GISourceScanner *scanner,
 239                   const char  *s)
 240 {
 241         /*
 242          * This function checks if `s' is a type name or an
 243          * identifier.
 244          */
 245
 246         if (gi_source_scanner_is_typedef (scanner, s)) {
 247                 return TYPEDEF_NAME;
 248         } else if (strcmp (s, "__builtin_va_list") == 0) {
 249                 return TYPEDEF_NAME;
 250         }
 251
 252         return IDENTIFIER;
 253 }
 254
 255 static void
 256 process_directive (GISourceScanner *scanner)
 257 {
 258         /* extract current filename from #line directives */
 259         GString *filename_builder;
 260         gboolean in_string, found_filename;
 261
 262         lineno = 0;
 263         found_filename = FALSE;
 264         in_string = FALSE;
 265         filename_builder = g_string_new ("");
 266
 267         int c = input ();
 268         while (c != EOF && c != '\n') {
 269                 if (!in_string) {
 270                         if (c == '\"') {
 271                                 in_string = TRUE;
 272                                 found_filename = TRUE;
 273                         } else if (c >= '0' && c <= '9') {
 274                                 if (!found_filename) {
 275                                         lineno = lineno * 10 + (c - '0');
 276                                 }
 277                         }
 278                 } else {
 279                         if (c == '\"') {
 280                                 in_string = FALSE;
 281                         } else if (c == '\\') {
 282                                 g_string_append_c (filename_builder, c);
 283                                 c = input ();
 284                                 g_string_append_c (filename_builder, c);
 285                         } else {
 286                                 g_string_append_c (filename_builder, c);
 287                         }
 288                 }
 289                 c = input ();
 290         }
 291
 292         if (filename_builder->len > 0) {
 293                 char *filename = g_strcompress (filename_builder->str);
 294                 if (g_realpath (filename))
 295                   {
 296                     g_free (scanner->current_filename);
 297                     scanner->current_filename = g_realpath (filename);
 298                     g_assert (scanner->current_filename);
 299                     g_free(filename);
 300                   }
 301         }
 302
 303         g_string_free (filename_builder, TRUE);
 304 }
 305
 306 /*
 307  * This parses a macro which is ignored, such as
 308  * __attribute__((x)) or __asm__ (x)
 309  */
 310 static int
 311 parse_ignored_macro (void)
 312 {
 313         int c;
 314         int nest;
 315
 316         while ((c = input ()) != EOF && isspace (c))
 317                 ;
 318         if (c != '(')
 319                 return FALSE;
 320
 321         nest = 0;
 322         while ((c = input ()) != EOF && (nest > 0 || c != ')')) {
 323                 if (c == '(')
 324                         nest++;
 325                 else if (c == ')')
 326                         nest--;
 327                 else if (c == '"') {
 328                         while ((c = input ()) != EOF && c != '"') {
 329                                 if (c == '\\')
 330                                         c = input ();
 331                         }
 332                 } else if (c == '\'') {
 333                         c = input ();
 334                         if (c == '\\')
 335                                 c = input ();
 336                         else if (c == '\'')
 337                                 return FALSE;
 338                         c = input ();
 339                         if (c != '\'')
 340                                 return FALSE;
 341                 } else if (c == '\n')
 342                         lineno++;
 343         }
 344
 345         return TRUE;
 346 }