[vala] gen-introspect: Merge some lexer changes taken from g-i

[Rico Tzschichholz <ricotz src gnome org>]

commit db1ab49336bee30258561682f780fdeefa598f58
Author: Rico Tzschichholz <ricotz ubuntu com>
Date:   Tue Dec 2 14:14:17 2014 +0100

    gen-introspect: Merge some lexer changes taken from g-i

 gobject-introspection/scannerlexer.l |  190 ++++++++++++++++++++++++----------
 1 files changed, 137 insertions(+), 53 deletions(-)
---
diff --git a/gobject-introspection/scannerlexer.l b/gobject-introspection/scannerlexer.l
index d3d24a1..0b7b1bb 100644
--- a/gobject-introspection/scannerlexer.l
+++ b/gobject-introspection/scannerlexer.l
@@ -1,5 +1,5 @@
 /* -*- Mode: C -*-
-/* GObject introspection: C lexer
+ * GObject introspection: C lexer
  *
  * Copyright (c) 1997 Sandro Sigala  <ssigala globalnet it>
  * Copyright (c) 2007-2008 Jürg Billeter  <j bitron ch>
@@ -37,6 +37,7 @@
 #include "grealpath.h"
 
 int lineno;
+char linebuf[2000];
 
 #undef YY_BUF_SIZE
 #define YY_BUF_SIZE 1048576
@@ -45,30 +46,42 @@ extern int yylex (GIGenerator *igenerator);
 #define YY_DECL int yylex (GIGenerator *igenerator)
 static int yywrap (void);
 static void parse_comment (GIGenerator *igenerator);
-static void process_directive (GIGenerator *igenerator);
+static void process_linemarks (GIGenerator *igenerator, gboolean has_line);
 static int check_identifier (GIGenerator *igenerator, const char *);
+static int parse_ignored_macro (void);
+static void print_error (GIGenerator *igenerator);
 %}
 
-intsuffix                              ([uU][lL]?)|([lL][uU]?)
+%option nounput
+
+intsuffix                              ([uU][lL]?[lL]?)|([lL][lL]?[uU]?)
 fracconst                              ([0-9]*\.[0-9]+)|([0-9]+\.)
 exppart                                        [eE][-+]?[0-9]+
 floatsuffix                            [fFlL]
-chartext                               ([^\'])|(\\.) 
-stringtext                             ([^\"])|(\\.)
+chartext                               ([^\\\'])|(\\.)
+stringtext                             ([^\\\"])|(\\.)
 
 %%
 
-"\n"                                   { ++lineno; } /* " */
+\n.*                                   { strncpy(linebuf, yytext+1, sizeof(linebuf)); /* save the next line 
*/
+                                               linebuf[sizeof(linebuf)-1]='\0';
+                                               /* printf("%4d:%s\n",lineno,linebuf); */
+                                               yyless(1);      /* give back all but the \n to rescan */
+                                               ++lineno;
+                                       }
+"\\\n"                                 { ++lineno; }
+
 [\t\f\v\r ]+                           { /* Ignore whitespace. */ }
 
 "/*"                                   { parse_comment(igenerator); }
-"//".*                                 { }
+"//".*                                 { /* Ignore C++ style comments. */ }
 
 "#define "[a-zA-Z_][a-zA-Z_0-9]*"("    { yyless (yyleng - 1); return FUNCTION_MACRO; }
 "#define "[a-zA-Z_][a-zA-Z_0-9]*       { return OBJECT_MACRO; }
 
-"#"                                    { process_directive(igenerator); }
-
+"# "[0-9]+" ".*"\n"                    { process_linemarks(igenerator, FALSE); }
+"#line "[0-9]+" ".*"\n"                        { process_linemarks(igenerator, TRUE); }
+"#"                                    { }
 "{"                                    { return '{'; }
 "<%"                                   { return '{'; }
 "}"                                    { return '}'; }
@@ -120,10 +133,40 @@ stringtext                                ([^\"])|(\\.)
 ","                                    { return ','; }
 "->"                                   { return ARROW; }
 
-[a-zA-Z_][a-zA-Z_0-9]*                 { if (igenerator->macro_scan) return IDENTIFIER; else REJECT; }
+"__asm"[\t\f\v\r ]+"volatile"                  { if (!parse_ignored_macro()) REJECT; }
+"__asm__"[\t\f\v\r ]+"volatile"                { if (!parse_ignored_macro()) REJECT; }
+"__asm__"[\t\f\v\r ]+"__volatile__"    { if (!parse_ignored_macro()) REJECT; }
+"__asm"                                { if (!parse_ignored_macro()) REJECT; }
+"__asm__"                              { if (!parse_ignored_macro()) REJECT; }
+"__attribute__"                        { if (!parse_ignored_macro()) REJECT; }
+"__attribute"                          { if (!parse_ignored_macro()) REJECT; }
+"__const"                               { return CONST; }
+"__extension__"                         { /* Ignore */ }
+"__inline__"                            { return INLINE; }
+"__inline"                             { return INLINE; }
+"__nonnull"                            { if (!parse_ignored_macro()) REJECT; }
+"_Noreturn"                            { /* Ignore */ }
+"__signed__"                           { return SIGNED; }
+"__restrict"                           { return RESTRICT; }
+"__restrict__"                         { return RESTRICT; }
+"__typeof"                             { if (!parse_ignored_macro()) REJECT; }
+"__volatile"                           { return VOLATILE; }
+"__volatile__"                         { return VOLATILE; }
+"_Bool"                                        { return BOOL; }
+"typedef char __static_assert_t".*"\n" { /* Ignore */ }
+"__cdecl"                              { /* Ignore */ }
+"__declspec(deprecated(".*"))"         { /* Ignore */ }
+"__declspec"[\t ]*"("[a-z\t ]+")"      { /* Ignore */ }
+"__stdcall"                            { /* ignore */ }
+"__w64"                                        { /* ignore */ }
+"__int64"                              { return INT; }
+
+
+[a-zA-Z_][a-zA-Z_0-9]*                 { if (igenerator->macro_scan) return check_identifier(igenerator, 
yytext); else REJECT; }
+
+"asm"                                  { if (!parse_ignored_macro()) REJECT; }
 
 "auto"                                 { return AUTO; }
-"_Bool"                                        { return BOOL; }
 "break"                                        { return BREAK; }
 "case"                                 { return CASE; }
 "char"                                 { return CHAR; }
@@ -141,6 +184,10 @@ stringtext                         ([^\"])|(\\.)
 "if"                                   { return IF; }
 "inline"                               { return INLINE; }
 "int"                                  { return INT; }
+"__uint128_t"                          { return INT; }
+"__int128_t"                           { return INT; }
+"__uint128"                            { return INT; }
+"__int128"                             { return INT; }
 "long"                                 { return LONG; }
 "register"                             { return REGISTER; }
 "restrict"                             { return RESTRICT; }
@@ -173,11 +220,12 @@ stringtext                                ([^\"])|(\\.)
 "\""{stringtext}*"\""                  { return STRING; }
 "L\""{stringtext}*"\""                 { return STRING; }
 
-.                                      { fprintf(stderr, "%s:%d: unexpected character `%c'\n", 
igenerator->current_filename, lineno, yytext[0]); }
+.                                      { print_error(igenerator); }
 
 %%
 
-static int yywrap (void)
+static int
+yywrap (void)
 {
   return 1;
 }
@@ -254,7 +302,9 @@ static void parse_comment (GIGenerator *igenerator)
     }
 }
 
-static int check_identifier (GIGenerator *igenerator, const char *s)
+static int
+check_identifier (GIGenerator *igenerator,
+                 const char *s)
 {
        /*
         * This function checks if `s' is a type name or an
@@ -270,49 +320,83 @@ static int check_identifier (GIGenerator *igenerator, const char *s)
        return IDENTIFIER;
 }
 
-static void process_directive (GIGenerator *igenerator)
+/*
+ * # linenum "filename" flags
+ *  See http://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html
+ **/
+
+static void
+process_linemarks (GIGenerator *igenerator, gboolean has_line)
+{
+       char escaped_filename[1025];
+       char *filename;
+       char *real;
+
+       if (has_line)
+               sscanf(yytext, "#line %d \"%1024[^\"]\"", &lineno, escaped_filename);
+       else
+               sscanf(yytext, "# %d \"%1024[^\"]\"", &lineno, escaped_filename);
+
+       filename = g_strcompress (escaped_filename);
+
+        real = g_realpath (filename);
+        if (real)
+          {
+            g_free (filename);
+            filename = real;
+          }
+
+        if (igenerator->current_filename)
+          g_free (igenerator->current_filename);
+       igenerator->current_filename = filename;
+}
+
+/*
+ * This parses a macro which is ignored, such as
+ * __attribute__((x)) or __asm__ (x)
+ */
+static int
+parse_ignored_macro (void)
 {
-       /* extract current filename from #line directives */
-       GString *filename_builder;
-       gboolean in_string, found_filename;
-
-       lineno = 0;
-       found_filename = FALSE;
-       in_string = FALSE;
-       filename_builder = g_string_new ("");
-
-       int c = input ();
-       while (c != EOF && c != '\n') {
-               if (!in_string) {
-                       if (c == '\"') {
-                               in_string = TRUE;
-                               found_filename = TRUE;
-                       } else if (c >= '0' && c <= '9') {
-                               if (!found_filename) {
-                                       lineno = lineno * 10 + (c - '0');
-                               }
+       int c;
+       int nest;
+
+       while ((c = input ()) != EOF && isspace (c))
+               ;
+       if (c != '(')
+               return FALSE;
+
+       nest = 0;
+       while ((c = input ()) != EOF && (nest > 0 || c != ')')) {
+               if (c == '(')
+                       nest++;
+               else if (c == ')')
+                       nest--;
+               else if (c == '"') {
+                       while ((c = input ()) != EOF && c != '"') {
+                               if (c == '\\')
+                                       c = input ();
                        }
-               } else {
-                       if (c == '\"') {
-                               in_string = FALSE;
-                       } else if (c == '\\') {
-                               g_string_append_c (filename_builder, c);
+               } else if (c == '\'') {
+                       c = input ();
+                       if (c == '\\')
                                c = input ();
-                               g_string_append_c (filename_builder, c);
-                       } else {
-                               g_string_append_c (filename_builder, c);
-                       }
-               }
-               c = input ();
+                       else if (c == '\'')
+                               return FALSE;
+                       c = input ();
+                       if (c != '\'')
+                               return FALSE;
+               } else if (c == '\n')
+                       lineno++;
        }
 
-       if (filename_builder->len > 0) {
-               char *filename = g_strcompress (filename_builder->str);
-               g_free (igenerator->current_filename);
-               igenerator->current_filename = g_realpath(filename);
-               g_free(filename);
-       }
-
-       g_string_free (filename_builder, TRUE);
+       return TRUE;
 }
 
+static void
+print_error (GIGenerator *igenerator)
+{
+  if (yytext[0]) {
+    fprintf(stderr, "%s:%d: unexpected character `%c'\n", igenerator->current_filename, lineno, yytext[0]);
+  }
+}