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_linemarks (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 "/*"[\t ]*<[\t ]*"private"[\t ]*>" */"  { scanner->private = TRUE; }
  76 "/*"[\t ]*<[\t ]*"public"[\t ]*>" */"   { scanner->private = FALSE; }
  77 "//".*                                  { }
  78
  79 "#define "[a-zA-Z_][a-zA-Z_0-9]*"("     { yyless (yyleng - 1); return FUNCTION_MACRO; }
  80 "#define "[a-zA-Z_][a-zA-Z_0-9]*        { return OBJECT_MACRO; }
  81 "#pragma ".*"\n"                        { /* Ignore pragma. */ }
  82
  83 "# "[0-9]+" ".*"\n"                     { process_linemarks(scanner); }
  84 "#"                                     { }
  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 ELLIPSIS; }
  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 '='; }
 111 "<"                                     { return '<'; }
 112 ">"                                     { return '>'; }
 113 "+="                                    { return ADDEQ; }
 114 "-="                                    { return SUBEQ; }
 115 "*="                                    { return MULEQ; }
 116 "/="                                    { return DIVEQ; }
 117 "%="                                    { return MODEQ; }
 118 "^="                                    { return XOREQ; }
 119 "&="                                    { return ANDEQ; }
 120 "|="                                    { return OREQ; }
 121 "<<"                                    { return SL; }
 122 ">>"                                    { return SR; }
 123 "<<="                                   { return SLEQ; }
 124 ">>="                                   { return SREQ; }
 125 "=="                                    { return EQ; }
 126 "!="                                    { return NOTEQ; }
 127 "<="                                    { return LTEQ; }
 128 ">="                                    { return GTEQ; }
 129 "&&"                                    { return ANDAND; }
 130 "||"                                    { return OROR; }
 131 "++"                                    { return PLUSPLUS; }
 132 "--"                                    { return MINUSMINUS; }
 133 ","                                     { return ','; }
 134 "->"                                    { return ARROW; }
 135
 136 "__asm"                                 { if (!parse_ignored_macro()) REJECT; }
 137 "__asm__"                               { if (!parse_ignored_macro()) REJECT; }
 138 "__attribute__"                         { if (!parse_ignored_macro()) REJECT; }
 139 "__attribute"                           { if (!parse_ignored_macro()) REJECT; }
 140 "__const"                               { return CONST; }
 141 "__extension__"                         { return EXTENSION; }
 142 "__inline"                              { return INLINE; }
 143 "__nonnull"                             { if (!parse_ignored_macro()) REJECT; }
 144 "__signed__"                            { return SIGNED; }
 145 "__restrict"                            { return RESTRICT; }
 146 "__typeof"                              { if (!parse_ignored_macro()) REJECT; }
 147 "_Bool"                                 { return BOOL; }
 148
 149 [a-zA-Z_][a-zA-Z_0-9]*                  { if (scanner->macro_scan) return IDENTIFIER; else REJECT; }
 150
 151 "asm"                                   { if (!parse_ignored_macro()) REJECT; }
 152 "auto"                                  { return AUTO; }
 153 "break"                                 { return BREAK; }
 154 "case"                                  { return CASE; }
 155 "char"                                  { return CHAR; }
 156 "const"                                 { return CONST; }
 157 "continue"                              { return CONTINUE; }
 158 "default"                               { return DEFAULT; }
 159 "do"                                    { return DO; }
 160 "double"                                { return DOUBLE; }
 161 "else"                                  { return ELSE; }
 162 "enum"                                  { return ENUM; }
 163 "extern"                                { return EXTERN; }
 164 "float"                                 { return FLOAT; }
 165 "for"                                   { return FOR; }
 166 "goto"                                  { return GOTO; }
 167 "if"                                    { return IF; }
 168 "inline"                                { return INLINE; }
 169 "int"                                   { return INT; }
 170 "long"                                  { return LONG; }
 171 "register"                              { return REGISTER; }
 172 "restrict"                              { return RESTRICT; }
 173 "return"                                { return RETURN; }
 174 "short"                                 { return SHORT; }
 175 "signed"                                { return SIGNED; }
 176 "sizeof"                                { return SIZEOF; }
 177 "static"                                { return STATIC; }
 178 "struct"                                { return STRUCT; }
 179 "switch"                                { return SWITCH; }
 180 "typedef"                               { return TYPEDEF; }
 181 "union"                                 { return UNION; }
 182 "unsigned"                              { return UNSIGNED; }
 183 "void"                                  { return VOID; }
 184 "volatile"                              { return VOLATILE; }
 185 "while"                                 { return WHILE; }
 186
 187 [a-zA-Z_][a-zA-Z_0-9]*                  { return check_identifier(scanner, yytext); }
 188
 189 "0"[xX][0-9a-fA-F]+{intsuffix}?         { return INTEGER; }
 190 "0"[0-7]+{intsuffix}?                   { return INTEGER; }
 191 [0-9]+{intsuffix}?                      { return INTEGER; }
 192
 193 {fracconst}{exppart}?{floatsuffix}?     { return FLOATING; }
 194 [0-9]+{exppart}{floatsuffix}?           { return FLOATING; }
 195
 196 "'"{chartext}*"'"                       { return CHARACTER; }
 197 "L'"{chartext}*"'"                      { return CHARACTER; }
 198
 199 "\""{stringtext}*"\""                   { return STRING; }
 200 "L\""{stringtext}*"\""                  { return STRING; }
 201
 202 .                                       { if (yytext[0]) fprintf(stderr, "%s:%d: unexpected character `%c'\n", scanner->current_filename, lineno, yytext[0]); }
 203
 204 %%
 205
 206 static int
 207 yywrap (void)
 208 {
 209   return 1;
 210 }
 211
 212
 213 static void
 214 parse_comment (GISourceScanner *scanner)
 215 {
 216   GString *string;
 217   int c1, c2;
 218   GISourceComment *comment;
 219   int comment_lineno;
 220
 221   c1 = input();
 222   c2 = input();
 223
 224   string = g_string_new ("");
 225
 226   comment_lineno = lineno;
 227
 228   while (c2 != EOF && !(c1 == '*' && c2 == '/'))
 229     {
 230       g_string_append_c (string, c1);
 231
 232       if (c1 == '\n')
 233         lineno++;
 234
 235       c1 = c2;
 236       c2 = input();
 237
 238     }
 239
 240   comment = g_slice_new (GISourceComment);
 241   comment->comment = g_string_free (string, FALSE);
 242   comment->line = comment_lineno;
 243   comment->filename = g_strdup(scanner->current_filename);
 244
 245   scanner->comments = g_slist_prepend (scanner->comments,
 246                                        comment);
 247 }
 248
 249 static int
 250 check_identifier (GISourceScanner *scanner,
 251                   const char  *s)
 252 {
 253         /*
 254          * This function checks if `s' is a type name or an
 255          * identifier.
 256          */
 257
 258         if (gi_source_scanner_is_typedef (scanner, s)) {
 259                 return TYPEDEF_NAME;
 260         } else if (strcmp (s, "__builtin_va_list") == 0) {
 261                 return TYPEDEF_NAME;
 262         }
 263
 264         return IDENTIFIER;
 265 }
 266
 267 /*
 268  * # linenum "filename" flags
 269  *  See http://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html
 270  **/
 271
 272 static void
 273 process_linemarks (GISourceScanner *scanner)
 274 {
 275         char filename[1025];
 276         char *compress;
 277         char *real;
 278
 279         sscanf(yytext, "# %d \"%1024[^\"]\"", &lineno, filename);
 280
 281         compress = g_strcompress (filename);
 282         real = g_realpath (filename);
 283         if (real) {
 284                 g_free (scanner->current_filename);
 285                 scanner->current_filename = real;
 286         } else {
 287                 g_free (real);
 288         }
 289         g_free (compress);
 290 }
 291
 292 /*
 293  * This parses a macro which is ignored, such as
 294  * __attribute__((x)) or __asm__ (x)
 295  */
 296 static int
 297 parse_ignored_macro (void)
 298 {
 299         int c;
 300         int nest;
 301
 302         while ((c = input ()) != EOF && isspace (c))
 303                 ;
 304         if (c != '(')
 305                 return FALSE;
 306
 307         nest = 0;
 308         while ((c = input ()) != EOF && (nest > 0 || c != ')')) {
 309                 if (c == '(')
 310                         nest++;
 311                 else if (c == ')')
 312                         nest--;
 313                 else if (c == '"') {
 314                         while ((c = input ()) != EOF && c != '"') {
 315                                 if (c == '\\')
 316                                         c = input ();
 317                         }
 318                 } else if (c == '\'') {
 319                         c = input ();
 320                         if (c == '\\')
 321                                 c = input ();
 322                         else if (c == '\'')
 323                                 return FALSE;
 324                         c = input ();
 325                         if (c != '\'')
 326                                 return FALSE;
 327                 } else if (c == '\n')
 328                         lineno++;
 329         }
 330
 331         return TRUE;
 332 }