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