shithub: libmujs

--- a/jsi.h

+++ b/jsi.h

@@ -12,6 +12,7 @@

 #include <math.h>

 #include <float.h>

+typedef struct js_Regexp js_Regexp;

 typedef struct js_Value js_Value;

 typedef struct js_Object js_Object;

 typedef struct js_Ast js_Ast;

@@ -38,13 +39,13 @@

 void js_newfunction(js_State *J, js_Function *function, js_Environment *scope);

 void js_newscript(js_State *J, js_Function *function);

-void *js_toregexp(js_State *J, int idx, int *flags);

+js_Regexp *js_toregexp(js_State *J, int idx);

 void js_dup(js_State *J);

 void js_rot2(js_State *J);

 void js_rot3(js_State *J);

-int js_RegExp_prototype_exec(js_State *J, int idx, const char *text);

+int js_RegExp_prototype_exec(js_State *J, js_Regexp *re, const char *text);

 /* Exception handling */

--- a/jsregexp.c

+++ b/jsregexp.c

@@ -6,89 +6,111 @@

 #include <regex.h>

-int js_RegExp_prototype_exec(js_State *J, int idx, const char *text)

+void js_newregexp(js_State *J, const char *pattern, int flags)

+	char msg[256];

 	js_Object *obj;

-	int flags, opts, last;

 	regex_t *prog;

+	int opts, status;

+	obj = jsV_newobject(J, JS_CREGEXP, J->RegExp_prototype);

+	opts = REG_EXTENDED;

+	if (flags & JS_REGEXP_I) opts |= REG_ICASE;

+	if (flags & JS_REGEXP_M) opts |= REG_NEWLINE;

+	prog = malloc(sizeof (regex_t));

+	status = regcomp(prog, pattern, opts);

+	if (status) {

+		free(prog);

+		regerror(status, prog, msg, sizeof msg);

+		js_syntaxerror(J, "%s", msg);

+	}

+	obj->u.r.prog = prog;

+	obj->u.r.source = pattern;

+	obj->u.r.flags = flags;

+	obj->u.r.last = 0;

+	js_pushobject(J, obj);

+}

+int js_RegExp_prototype_exec(js_State *J, js_Regexp *re, const char *text)

+{

 	regmatch_t m[10];

+	int opts;

 	int i;

-	prog = js_toregexp(J, idx, &flags);

-	obj = js_toobject(J, idx);

 	opts = 0;

-	if (flags & JS_REGEXP_G) {

-		last = obj->u.r.last;

-		if (last < 0 || last > strlen(text)) {

-			obj->u.r.last = 0;

+	if (re->flags & JS_REGEXP_G) {

+		if (re->last < 0 || re->last > strlen(text)) {

+			re->last = 0;

 			js_pushnull(J);

 			return 1;

-		if (last > 0) {

-			text += last;

+		if (re->last > 0) {

+			text += re->last;

 			opts |= REG_NOTBOL;

-	if (!regexec(prog, text, nelem(m), m, opts)) {

+	if (!regexec(re->prog, text, nelem(m), m, opts)) {

 		js_newarray(J);

 		for (i = 0; i < nelem(m) && m[i].rm_so >= 0; ++i) {

 			js_pushlstring(J, text + m[i].rm_so, m[i].rm_eo - m[i].rm_so);

 			js_setindex(J, -2, i);

-		if (flags & JS_REGEXP_G)

-			obj->u.r.last = last + m[0].rm_eo;

+		if (re->flags & JS_REGEXP_G)

+			re->last = re->last + m[0].rm_eo;

 		return 1;

-	if (flags & JS_REGEXP_G)

-		obj->u.r.last = 0;

+	if (re->flags & JS_REGEXP_G)

+		re->last = 0;

 	js_pushnull(J);

 	return 1;

-void js_newregexp(js_State *J, const char *pattern, int flags)

+static int Rp_test(js_State *J, int argc)

-	char msg[256];

-	js_Object *obj;

-	regex_t *prog;

-	int opts, status;

+	js_Regexp *re;

+	const char *text;

+	regmatch_t m[10];

+	int opts;

-	obj = jsV_newobject(J, JS_CREGEXP, J->RegExp_prototype);

+	re = js_toregexp(J, 0);

+	text = js_tostring(J, 1);

-	opts = REG_EXTENDED;

-	if (flags & JS_REGEXP_I) opts |= REG_ICASE;

-	if (flags & JS_REGEXP_M) opts |= REG_NEWLINE;

+	opts = 0;

+	if (re->flags & JS_REGEXP_G) {

+		if (re->last < 0 || re->last > strlen(text)) {

+			re->last = 0;

+			js_pushboolean(J, 0);

+			return 1;

+		}

+		if (re->last > 0) {

+			text += re->last;

+			opts |= REG_NOTBOL;

+		}

+	}

-	prog = malloc(sizeof (regex_t));

-	status = regcomp(prog, pattern, opts);

-	if (status) {

-		free(prog);

-		regerror(status, prog, msg, sizeof msg);

-		js_syntaxerror(J, "%s", msg);

+	if (!regexec(re->prog, text, nelem(m), m, opts)) {

+		if (re->flags & JS_REGEXP_G)

+			re->last = re->last + m[0].rm_eo;

+		js_pushboolean(J, 1);

+		return 1;

-	obj->u.r.prog = prog;

-	obj->u.r.flags = flags;

-	obj->u.r.last = 0;

-	js_pushobject(J, obj);

+	if (re->flags & JS_REGEXP_G)

+		re->last = 0;

-	js_pushstring(J, pattern);

-	js_defproperty(J, -2, "source", JS_READONLY | JS_DONTENUM | JS_DONTDELETE);

-	js_pushboolean(J, flags & JS_REGEXP_G);

-	js_defproperty(J, -2, "global", JS_READONLY | JS_DONTENUM | JS_DONTDELETE);

-	js_pushboolean(J, flags & JS_REGEXP_I);

-	js_defproperty(J, -2, "ignoreCase", JS_READONLY | JS_DONTENUM | JS_DONTDELETE);

-	js_pushboolean(J, flags & JS_REGEXP_M);

-	js_defproperty(J, -2, "multiline", JS_READONLY | JS_DONTENUM | JS_DONTDELETE);

-	// TODO: lastIndex

+	js_pushboolean(J, 0);

+	return 1;

 static int jsB_new_RegExp(js_State *J, int argc)

+	js_Regexp *old;

 	const char *pattern;

 	int flags;

@@ -95,10 +117,9 @@

 	if (js_isregexp(J, 1)) {

 		if (argc > 1)

 			js_typeerror(J, "cannot supply flags when creating one RegExp from another");

-		js_toregexp(J, 1, &flags);

-		js_getproperty(J, 1, "source");

-		pattern = js_tostring(J, -1);

-		js_pop(J, 1);

+		old = js_toregexp(J, 1);

+		pattern = old->source;

+		flags = old->flags;

 	} else if (js_isundefined(J, 1)) {

 		pattern = "";

 		flags = 0;

@@ -138,26 +159,18 @@

 static int Rp_toString(js_State *J, int argc)

-	const char *source;

-	int flags;

+	js_Regexp *re;

 	char *out;

-	js_Object *self = js_toobject(J, 0);

-	if (self->type != JS_CREGEXP)

-		js_typeerror(J, "not a regexp");

+	re = js_toregexp(J, 0);

-	flags = self->u.r.flags;

-	js_getproperty(J, 0, "source");

-	source = js_tostring(J, -1);

-	out = malloc(strlen(source) + 6); /* extra space for //gim */

+	out = malloc(strlen(re->source) + 6); /* extra space for //gim */

 	strcpy(out, "/");

-	strcat(out, source);

+	strcat(out, re->source);

 	strcat(out, "/");

-	if (flags & JS_REGEXP_G) strcat(out, "g");

-	if (flags & JS_REGEXP_I) strcat(out, "i");

-	if (flags & JS_REGEXP_M) strcat(out, "m");

+	if (re->flags & JS_REGEXP_G) strcat(out, "g");

+	if (re->flags & JS_REGEXP_I) strcat(out, "i");

+	if (re->flags & JS_REGEXP_M) strcat(out, "m");

 	if (js_try(J)) {

 		free(out);

@@ -172,47 +185,7 @@

 static int Rp_exec(js_State *J, int argc)

-	return js_RegExp_prototype_exec(J, 0, js_tostring(J, 1));

-}

-static int Rp_test(js_State *J, int argc)

-{

-	int flags, opts, last;

-	js_Object *obj;

-	regmatch_t m[10];

-	regex_t *prog;

-	const char *text;

-	prog = js_toregexp(J, 0, &flags);

-	obj = js_toobject(J, 0);

-	text = js_tostring(J, 1);

-	opts = 0;

-	if (flags & JS_REGEXP_G) {

-		last = obj->u.r.last;

-		if (last < 0 || last > strlen(text)) {

-			obj->u.r.last = 0;

-			js_pushboolean(J, 0);

-			return 1;

-		}

-		if (last > 0) {

-			text += last;

-			opts |= REG_NOTBOL;

-		}

-	}

-	if (!regexec(prog, text, nelem(m), m, opts)) {

-		if (flags & JS_REGEXP_G)

-			obj->u.r.last = last + m[0].rm_eo;

-		js_pushboolean(J, 1);

-		return 1;

-	}

-	if (flags & JS_REGEXP_G)

-		obj->u.r.last = 0;

-	js_pushboolean(J, 0);

-	return 1;

+	return js_RegExp_prototype_exec(J, js_toregexp(J, 0), js_tostring(J, 1));

 void jsB_initregexp(js_State *J)

--- a/jsrun.c

+++ b/jsrun.c

@@ -207,11 +207,11 @@

 	return jsV_toprimitive(J, stackidx(J, idx), hint);

-void *js_toregexp(js_State *J, int idx, int *flags)

+js_Regexp *js_toregexp(js_State *J, int idx)

 	const js_Value *v = stackidx(J, idx);

 	if (v->type == JS_TOBJECT && v->u.object->type == JS_CREGEXP)

-		return *flags = v->u.object->u.r.flags, v->u.object->u.r.prog;

+		return &v->u.object->u.r;

 	js_typeerror(J, "not a regexp");

--- a/jsstring.c

+++ b/jsstring.c

@@ -318,10 +318,9 @@

 static int Sp_match(js_State *J, int argc)

+	js_Regexp *re;

+	regmatch_t m[10];

 	const char *text;

-	regmatch_t m;

-	regex_t *prog;

-	int flags;

 	unsigned int len, a, b, c, e;

 	text = js_tostring(J, 0);

@@ -333,9 +332,9 @@

 	else

 		js_newregexp(J, js_tostring(J, 1), 0);

-	prog = js_toregexp(J, -1, &flags);

-	if (!(flags & JS_REGEXP_G))

-		return js_RegExp_prototype_exec(J, -1, text);

+	re = js_toregexp(J, -1);

+	if (!(re->flags & JS_REGEXP_G))

+		return js_RegExp_prototype_exec(J, re, text);

 	js_newarray(J);

@@ -343,11 +342,11 @@

 	len = 0;

 	a = 0;

 	while (a <= e) {

-		if (regexec(prog, text + a, 1, &m, a > 0 ? REG_NOTBOL : 0))

+		if (regexec(re->prog, text + a, nelem(m), m, a > 0 ? REG_NOTBOL : 0))

 			break;

-		b = a + m.rm_so;

-		c = a + m.rm_eo;

+		b = a + m[0].rm_so;

+		c = a + m[0].rm_eo;

 		js_pushlstring(J, text + b, c - b);

 		js_setindex(J, -2, len++);

@@ -362,10 +361,9 @@

 static int Sp_search(js_State *J, int argc)

+	js_Regexp *re;

+	regmatch_t m[10];

 	const char *text;

-	regmatch_t m;

-	regex_t *prog;

-	int flags;

 	text = js_tostring(J, 0);

@@ -376,10 +374,10 @@

 	else

 		js_newregexp(J, js_tostring(J, 1), 0);

-	prog = js_toregexp(J, -1, &flags);

+	re = js_toregexp(J, -1);

-	if (!regexec(prog, text, 1, &m, 0))

-		js_pushnumber(J, m.rm_so); // TODO: convert to utf-8 index offset

+	if (!regexec(re->prog, text, nelem(m), m, 0))

+		js_pushnumber(J, m[0].rm_so); // TODO: convert to utf-8 index offset

 	else

 		js_pushnumber(J, -1);

@@ -388,17 +386,16 @@

 static int Sp_replace_regexp(js_State *J, int argc)

-	const char *source, *s, *r;

+	js_Regexp *re;

 	regmatch_t m[10];

-	regex_t *prog;

-	int flags;

+	const char *source, *s, *r;

 	struct sbuffer *sb = NULL;

 	int n, x;

 	source = js_tostring(J, 0);

-	prog = js_toregexp(J, 1, &flags);

+	re = js_toregexp(J, 1);

-	if (regexec(prog, source, nelem(m), m, 0)) {

+	if (regexec(re->prog, source, nelem(m), m, 0)) {

 		js_copy(J, 0);

 		return 1;

@@ -453,7 +450,7 @@

-	if (flags & JS_REGEXP_G) {

+	if (re->flags & JS_REGEXP_G) {

 		source = source + m[0].rm_eo;

 		if (n == 0) {

 			if (*source)

@@ -461,7 +458,7 @@

 			else

 				goto end;

-		if (!regexec(prog, source, nelem(m), m, REG_NOTBOL))

+		if (!regexec(re->prog, source, nelem(m), m, REG_NOTBOL))

 			goto loop;

@@ -548,15 +545,15 @@

 static int Sp_split_regexp(js_State *J, int argc)

-	const char *str = js_tostring(J, 0);

-	unsigned int limit = !js_isundefined(J, 2) ? js_touint32(J, 2) : 1 << 30;

+	js_Regexp *re;

 	regmatch_t m[10];

-	regex_t *prog;

-	int flags;

-	unsigned int len, k, e;

+	const char *str;

+	unsigned int limit, len, k, e;

 	unsigned int p, a, b, c;

-	prog = js_toregexp(J, 1, &flags);

+	str = js_tostring(J, 0);

+	re = js_toregexp(J, 1);

+	limit = !js_isundefined(J, 2) ? js_touint32(J, 2) : 1 << 30;

 	js_newarray(J);

 	len = 0;

@@ -565,7 +562,7 @@

 	/* splitting the empty string */

 	if (e == 0) {

-		if (regexec(prog, str, nelem(m), m, 0)) {

+		if (regexec(re->prog, str, nelem(m), m, 0)) {

 			if (len == limit) return 1;

 			js_pushliteral(J, "");

 			js_setindex(J, -2, 0);

@@ -575,7 +572,7 @@

 	p = a = 0;

 	while (a < e) {

-		if (regexec(prog, str + a, nelem(m), m, a > 0 ? REG_NOTBOL : 0))

+		if (regexec(re->prog, str + a, nelem(m), m, a > 0 ? REG_NOTBOL : 0))

 			break; /* no match */

 		b = a + m[0].rm_so;

--- a/jsvalue.h

+++ b/jsvalue.h

@@ -44,6 +44,14 @@

 	} u;

};

+struct js_Regexp

+{

+	void *prog;

+	const char *source;

+	unsigned short flags;

+	unsigned short last;

+};

 struct js_Object

 	js_Class type;

@@ -65,11 +73,7 @@

 			js_CFunction function;

 			js_CFunction constructor;

 		} c;

-		struct {

-			void *prog;

-			unsigned short flags;

-			unsigned short last;

-		} r;

+		js_Regexp r;

 		struct {

 			js_Object *target;

 			js_Iterator *head;