Port re2c parser from YACC to recursive descent.

svn path=/trunk/yasm/; revision=1718

6 files changed, 288 insertions, 221 deletions

diff --git a/tools/re2c/Makefile.inc b/tools/re2c/Makefile.inc
index 06f47134..abd2db87 100644
--- a/tools/re2c/Makefile.inc
+++ b/tools/re2c/Makefile.inc
@@ -16,7 +16,8 @@ EXTRA_DIST += tools/re2c/code.c
 EXTRA_DIST += tools/re2c/dfa.h
 EXTRA_DIST += tools/re2c/dfa.c
 EXTRA_DIST += tools/re2c/parse.h
-EXTRA_DIST += tools/re2c/re2c-parser.y
+EXTRA_DIST += tools/re2c/parser.h
+EXTRA_DIST += tools/re2c/parser.c
 EXTRA_DIST += tools/re2c/actions.c
 EXTRA_DIST += tools/re2c/scanner.h
 EXTRA_DIST += tools/re2c/scanner.c
@@ -45,8 +46,8 @@ re2c-code.$(OBJEXT): tools/re2c/code.c
 re2c-dfa.$(OBJEXT): tools/re2c/dfa.c
 	$(CC_FOR_BUILD) $(DEFAULT_INCLUDES) $(INCLUDES) -c -o $@ `test -f tools/re2c/dfa.c || echo '$(srcdir)/'`tools/re2c/dfa.c
 
-re2c-parser.$(OBJEXT): re2c-parser.c
-	$(CC_FOR_BUILD) $(DEFAULT_INCLUDES) $(INCLUDES) -c -o $@ `test -f re2c-parser.c || echo '$(srcdir)/'`re2c-parser.c
+re2c-parser.$(OBJEXT): tools/re2c/parser.c
+	$(CC_FOR_BUILD) $(DEFAULT_INCLUDES) $(INCLUDES) -c -o $@ `test -f tools/re2c/parser.c || echo '$(srcdir)/'`tools/re2c/parser.c
 
 re2c-actions.$(OBJEXT): tools/re2c/actions.c
 	$(CC_FOR_BUILD) $(DEFAULT_INCLUDES) $(INCLUDES) -c -o $@ `test -f tools/re2c/actions.c || echo '$(srcdir)/'`tools/re2c/actions.c
@@ -63,41 +64,6 @@ re2c-substr.$(OBJEXT): tools/re2c/substr.c
 re2c-translate.$(OBJEXT): tools/re2c/translate.c
 	$(CC_FOR_BUILD) $(DEFAULT_INCLUDES) $(INCLUDES) -c -o $@ `test -f tools/re2c/translate.c || echo '$(srcdir)/'`tools/re2c/translate.c
 
-re2c-parser.c: tools/re2c/re2c-parser.y
-	$(YACC) $(YFLAGS) $(AM_YFLAGS) `test -f 'tools/re2c/re2c-parser.y' || echo '$(srcdir)/'`tools/re2c/re2c-parser.y
-	if test -f y.tab.h; then \
-	  to=`echo "re2c-parser_H" | sed \
-                -e 'y/abcdefghijklmnopqrstuvwxyz/ABCDEFGHIJKLMNOPQRSTUVWXYZ/' \
-                -e 's/[^ABCDEFGHIJKLMNOPQRSTUVWXYZ]/_/g'`; \
-	  sed "/^#/ s/Y_TAB_H/$$to/g" y.tab.h >re2c-parser.ht; \
-	  rm -f y.tab.h; \
-	  if cmp -s re2c-parser.ht re2c-parser.h; then \
-	    rm -f re2c-parser.ht ;\
-	  else \
-	    mv re2c-parser.ht re2c-parser.h; \
-	  fi; \
-	fi
-	if test -f y.output; then \
-	  mv y.output re2c-parser.output; \
-	fi
-	sed '/^#/ s|y\.tab\.c|re2c-parser.c|' y.tab.c >re2c-parser.ct && mv re2c-parser.ct re2c-parser.c
-	rm -f y.tab.c
-
-re2c-parser.h: re2c-parser.c
-	@if test ! -f $@; then \
-	  rm -f re2c-parser.c; \
-	  $(MAKE) re2c-parser.c; \
-	else :; fi
-
-BUILT_SOURCES += re2c-parser.c
-BUILT_SOURCES += re2c-parser.h
-
-CLEANFILES += re2c-parser.c
-CLEANFILES += re2c-parser.h
-
-EXTRA_DIST += re2c-parser.c
-EXTRA_DIST += re2c-parser.h
-
 EXTRA_DIST += tools/re2c/CHANGELOG
 EXTRA_DIST += tools/re2c/NO_WARRANTY
 EXTRA_DIST += tools/re2c/README
diff --git a/tools/re2c/parser.c b/tools/re2c/parser.c
new file mode 100644
index 00000000..02d5c668
--- /dev/null
+++ b/tools/re2c/parser.c
@@ -0,0 +1,249 @@
+#include <time.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include "tools/re2c/globals.h"
+#include "tools/re2c/parse.h"
+#include "tools/re2c/parser.h"
+
+int yylex(void);
+static RegExp *parse_expr(void);
+static RegExp *parse_diff(void);
+static RegExp *parse_term(void);
+static RegExp *parse_factor(void);
+static RegExp *parse_primary(void);
+
+static unsigned int accept;
+static RegExp *spec;
+static Scanner *in;
+
+static int curtok, peektok;
+yystype yylval;
+static yystype peekval;
+
+#define get_next_token()    (curtok = yylex())
+
+static void
+get_peek_token(void)
+{
+    yystype temp = yylval; /* structure copy */
+    if (peektok != NONE)
+	Scanner_fatal(in, "more than one token of lookahead?");
+    peektok = yylex();
+    peekval = yylval; /* structure copy */
+    yylval = temp;
+}
+
+static void
+yyparse(void)
+{
+    RegExp *re, *look;
+
+    accept = 0;
+    spec = NULL;
+    get_next_token();
+    while (curtok != 0) {
+	switch (curtok) {
+	    case ID:
+		get_peek_token();
+		if (peektok == '=') {
+		    /* ID = expr; */
+		    Symbol *sym = yylval.symbol;
+		    get_next_token(); /* id */
+		    get_next_token(); /* = */
+		    re = parse_expr();
+		    if (curtok != ';')
+			Scanner_fatal(in, "missing `;' after regexp");
+		    get_next_token(); /* ; */
+		    if (sym->re)
+			Scanner_fatal(in, "sym already defined");
+		    sym->re = re;
+		    break;
+		}
+		/*@fallthrough@*/
+	    default:
+		/* rule: expr [/ expr] CODE */
+		re = parse_expr();
+		if (!re)
+		    Scanner_fatal(in, "expression syntax error");
+
+		if (curtok == '/') {
+		    get_next_token(); /* / */
+		    look = parse_expr();
+		} else
+		    look = RegExp_new_NullOp();
+
+		if (curtok != CODE)
+		    Scanner_fatal(in, "missing code after regexp");
+		re = RegExp_new_RuleOp(re, look, yylval.token, accept++);
+		get_next_token(); /* CODE */
+		spec = spec ? mkAlt(spec, re) : re;
+	}
+    }
+}
+
+static RegExp *
+parse_expr(void)
+{
+    RegExp *e, *f;
+    e = parse_diff();
+    while (curtok == '|') {
+	get_next_token(); /* | */
+	f = parse_diff();
+	e = mkAlt(e, f);
+    }
+    return e;
+}
+
+static RegExp *
+parse_diff(void)
+{
+    RegExp *e, *f;
+    e = parse_term();
+    while (curtok == '\\') {
+	get_next_token(); /* \ */
+	f = parse_term();
+	e = mkDiff(e, f);
+	if(!e)
+	    Scanner_fatal(in, "can only difference char sets");
+    }
+    return e;
+}
+
+static RegExp *
+parse_term(void)
+{
+    RegExp *e, *f;
+    e = parse_factor();
+    while ((f = parse_factor())) {
+	e = RegExp_new_CatOp(e, f);
+    }
+    return e;
+}
+
+static RegExp *
+parse_factor(void)
+{
+    RegExp *e;
+    char ch;
+    e = parse_primary();
+    while (curtok == CLOSE || curtok == CLOSESIZE) {
+	switch (curtok) {
+	    case CLOSE:
+		ch = yylval.op;
+		while (get_next_token() == CLOSE) {
+		    if (ch != yylval.op)
+			ch = '*';
+		}
+		switch (ch) {
+		    case '*':
+			e = mkAlt(RegExp_new_CloseOp(e), RegExp_new_NullOp());
+			break;
+		    case '+':
+			e = RegExp_new_CloseOp(e);
+			break;
+		    case '?':
+			e = mkAlt(e, RegExp_new_NullOp());
+			break;
+		}
+		break;
+	    case CLOSESIZE:
+		e = RegExp_new_CloseVOp(e, yylval.extop.minsize,
+					yylval.extop.maxsize);
+		get_next_token();	/* CLOSESIZE */
+		break;
+	    default:
+		Scanner_fatal(in, "parse error");
+		break;
+	}
+    }
+    return e;
+}
+
+static RegExp *
+parse_primary(void)
+{
+    RegExp *e;
+    switch (curtok) {
+	case ID:
+	    if (!yylval.symbol->re)
+		Scanner_fatal(in, "can't find symbol");
+	    e = yylval.symbol->re;
+	    get_next_token();
+	    break;
+	case RANGE:
+	case STRING:
+	    e = yylval.regexp;
+	    get_next_token();
+	    break;
+	case '(':
+	    get_next_token();
+	    e = parse_expr();
+	    if (curtok != ')')
+		Scanner_fatal(in, "missing closing parenthesis");
+	    get_next_token();
+	    break;
+	default:
+	    return NULL;
+    }
+    return e;
+}
+
+int
+yylex(void)
+{
+    if (peektok != NONE) {
+	int tok = peektok;
+	yylval = peekval;
+	peektok = NONE;
+	return tok;
+    }
+    return Scanner_scan(in);
+}
+
+void line_source(FILE *o, unsigned int line)
+{
+    char *	fnamebuf;
+    char *	token;
+
+    if (iFlag)
+	return;
+    fprintf(o, "#line %u \"", line);
+    if( fileName != NULL ) {
+    	fnamebuf = mystrdup( fileName );
+    } else {
+	fnamebuf = mystrdup( "<stdin>" );
+    }
+    token = strtok( fnamebuf, "\\" );
+    for(;;) {
+	fprintf(o, "%s", token);
+	token = strtok( NULL, "\\" );
+	if( token == NULL ) break;
+	fputs("\\\\", o);
+    }
+    fputs("\"\n", o); oline++;
+    free( fnamebuf );
+}
+
+void parse(FILE *i, FILE *o){
+    time_t now;
+
+    time(&now);
+
+    peektok = NONE;
+
+    fputs("/* Generated by re2c 0.9.1-C on ", o);
+    fprintf(o, "%-24s", ctime(&now));
+    fputs(" */\n", o); oline+=2;
+
+    in = Scanner_new(i);
+
+    line_source(o, Scanner_line(in));
+
+    while(Scanner_echo(in, o)){
+	yyparse();
+	if(spec)
+	    genCode(o, spec);
+	line_source(o, Scanner_line(in));
+    }
+}
diff --git a/tools/re2c/parser.h b/tools/re2c/parser.h
new file mode 100644
index 00000000..c433a99d
--- /dev/null
+++ b/tools/re2c/parser.h
@@ -0,0 +1,33 @@
+#ifndef RE2C_PARSER_H
+#define RE2C_PARSER_H
+
+/* Tokens */
+enum yytokentype {
+    CLOSESIZE = 258,
+    CLOSE = 259,
+    ID = 260,
+    CODE = 261,
+    RANGE = 262,
+    STRING = 263,
+    NONE = 264
+};
+
+#define CLOSESIZE 258
+#define CLOSE 259
+#define ID 260
+#define CODE 261
+#define RANGE 262
+#define STRING 263
+#define NONE 264
+
+typedef union {
+    Symbol	*symbol;
+    RegExp	*regexp;
+    Token	*token;
+    char	op;
+    ExtOp	extop;
+} yystype;
+
+extern yystype yylval;
+
+#endif
diff --git a/tools/re2c/re2c-parser.y b/tools/re2c/re2c-parser.y
deleted file mode 100644
index 998cdb05..00000000
--- a/tools/re2c/re2c-parser.y
+++ /dev/null
@@ -1,181 +0,0 @@
-%{
-#include <time.h>
-#include <stdio.h>
-#include <string.h>
-#include <stdlib.h>
-#include "tools/re2c/globals.h"
-#include "tools/re2c/parse.h"
-
-/* Work around bug in Bison 2.1 */
-#define YYPARSE_PARAM	unused
-
-int yylex(void);
-void yyerror(const char*);
-
-static unsigned int accept;
-static RegExp *spec;
-static Scanner *in;
-
-%}
-
-%start	spec
-
-%union {
-    Symbol	*symbol;
-    RegExp	*regexp;
-    Token	*token;
-    char	op;
-    ExtOp	extop;
-}
-
-%token		CLOSESIZE   CLOSE	ID	CODE	RANGE	STRING
-
-%type	<op>		CLOSE
-%type	<op>		close
-%type	<extop>		CLOSESIZE
-%type	<symbol>	ID
-%type	<token>		CODE
-%type	<regexp>	RANGE	STRING
-%type	<regexp>	rule	look	expr	diff	term	factor	primary
-
-%%
-
-spec	:
-		{ accept = 0;
-		  spec = NULL; }
-	|	spec rule
-		{ spec = spec? mkAlt(spec, $2) : $2; }
-	|	spec decl
-	;
-
-decl	:	ID '=' expr ';'
-		{ if($1->re)
-		      Scanner_fatal(in, "sym already defined");
-		  $1->re = $3; }
-	;
-
-rule	:	expr look CODE
-		{ $$ = RegExp_new_RuleOp($1, $2, $3, accept++); }
-	;
-
-look	:
-		{ $$ = RegExp_new_NullOp(); }
-	|	'/' expr
-		{ $$ = $2; }
-	;
-
-expr	:	diff
-		{ $$ = $1; }
-	|	expr '|' diff
-		{ $$ =  mkAlt($1, $3); }
-	;
-
-diff	:	term
-		{ $$ = $1; }
-	|	diff '\\' term
-		{ $$ =  mkDiff($1, $3);
-		  if(!$$)
-		       Scanner_fatal(in, "can only difference char sets");
-		}
-	;
-
-term	:	factor
-		{ $$ = $1; }
-	|	term factor
-		{ $$ = RegExp_new_CatOp($1, $2); }
-	;
-
-factor	:	primary
-		{ $$ = $1; }
-	|	primary close
-		{
-		    switch($2){
-		    case '*':
-			$$ = mkAlt(RegExp_new_CloseOp($1), RegExp_new_NullOp());
-			break;
-		    case '+':
-			$$ = RegExp_new_CloseOp($1);
-			break;
-		    case '?':
-			$$ = mkAlt($1, RegExp_new_NullOp());
-			break;
-		    }
-		}
-	|	primary CLOSESIZE
-		{
-			$$ = RegExp_new_CloseVOp($1, $2.minsize, $2.maxsize);
-		}
-	;
-
-close	:	CLOSE
-		{ $$ = $1; }
-	|	close CLOSE
-		{ $$ = ($1 == $2) ? $1 : '*'; }
-	;
-
-primary	:	ID
-		{ if(!$1->re)
-		      Scanner_fatal(in, "can't find symbol");
-		  $$ = $1->re; }
-	|	RANGE
-		{ $$ = $1; }
-	|	STRING
-		{ $$ = $1; }
-	|	'(' expr ')'
-		{ $$ = $2; }
-	;
-
-%%
-
-void yyerror(const char* s){
-    Scanner_fatal(in, s);
-}
-
-int yylex(){
-    return Scanner_scan(in);
-}
-
-void line_source(FILE *o, unsigned int line)
-{
-    char *	fnamebuf;
-    char *	token;
-
-    if (iFlag)
-	return;
-    fprintf(o, "#line %u \"", line);
-    if( fileName != NULL ) {
-    	fnamebuf = mystrdup( fileName );
-    } else {
-	fnamebuf = mystrdup( "<stdin>" );
-    }
-    token = strtok( fnamebuf, "\\" );
-    for(;;) {
-	fprintf(o, "%s", token);
-	token = strtok( NULL, "\\" );
-	if( token == NULL ) break;
-	fputs("\\\\", o);
-    }
-    fputs("\"\n", o); oline++;
-    free( fnamebuf );
-}
-
-void parse(FILE *i, FILE *o){
-    time_t now;
-
-    time(&now);
-
-    fputs("/* Generated by re2c 0.9.1-C on ", o);
-    fprintf(o, "%-24s", ctime(&now));
-    fputs(" */\n", o); oline+=2;
-
-    in = Scanner_new(i);
-
-    line_source(o, Scanner_line(in));
-
-    while(Scanner_echo(in, o)){
-	yyparse(NULL);
-	if(spec)
-	    genCode(o, spec);
-	line_source(o, Scanner_line(in));
-    }
-}
diff --git a/tools/re2c/scanner.c b/tools/re2c/scanner.c
index fd0ca93e..4640ee4a 100644
--- a/tools/re2c/scanner.c
+++ b/tools/re2c/scanner.c
@@ -6,7 +6,7 @@
 #include "tools/re2c/scanner.h"
 #include "tools/re2c/parse.h"
 #include "tools/re2c/globals.h"
-#include "re2c-parser.h"
+#include "tools/re2c/parser.h"
 
 #ifndef MAX
 #define MAX(a,b) (((a)>(b))?(a):(b))
diff --git a/tools/re2c/scanner.re b/tools/re2c/scanner.re
index 81dcf48b..10af0883 100644
--- a/tools/re2c/scanner.re
+++ b/tools/re2c/scanner.re
@@ -3,7 +3,7 @@
 #include "tools/re2c/scanner.h"
 #include "tools/re2c/parse.h"
 #include "tools/re2c/globals.h"
-#include "re2c-parser.h"
+#include "tools/re2c/parser.h"
 
 #ifndef MAX
 #define MAX(a,b) (((a)>(b))?(a):(b))