shithub: mc

Info  •  Files  •  Log  •  Branches

Fix capture groups with '+'

    Because of the way we generated the group id when compiling,
    and the way it interacted with the duplication of the subnodes
    for '+', we generated two capture groups for. This is because

      '(a)+ became (a)(a)*.

    This change fixes it so that both capture groups, while generated
    the same way, get the same capture id.

--- a/compile.myr

+++ b/compile.myr

@@ -25,7 +25,7 @@

 	`Class	[char, char]

 	/* meta */

-	`Cap	tree#

+	`Cap	[std.size, tree#] /* id, tree */

 	`Bol	/* beginning of line */

 	`Eol	/* end of line */

 ;;

@@ -87,8 +87,6 @@

 /* generates bytecode from an AST */

 const gen = {re, t

-	var m

-

 	match t#

 	|`Alt	(a, b): genalt(re, a, b)

 	|`Cat	(a, b): gen(re, a); gen(re, b)

@@ -105,8 +103,7 @@

 	/* meta */

 	|`Bol:	append(re, `Ibol)

 	|`Eol:	append(re, `Ibol)

-	|`Cap	a:

-		m = re.nmatch++

+	|`Cap	(m, a):

 		append(re, `Ilbra m)

 		gen(re, a)

 		append(re, `Irbra m)

@@ -325,8 +322,8 @@

 		std.put("Class (%c-%c)\n", a, b)

 	/* meta */

-	| `Cap	a:

-		std.put("Cap\n")

+	| `Cap	(m, a):

+		std.put("Cap %i\n", m)

 		dump(re, a, indent + 1)

 	;;

 }

@@ -417,9 +414,9 @@

 	/* lower prec operators */

 	| '|':	-> `None

 	| ')':	-> `None

-	| '*':	-> `Fail (`Badrep)

-	| '+':	-> `Fail (`Badrep)

-	| '?':	-> `Fail (`Badrep)

+	| '*':	-> `Fail `Badrep

+	| '+':	-> `Fail `Badrep

+	| '?':	-> `Fail `Badrep

 	| '[':	-> chrclass(re)

 	| '.':	getc(re); ret = mk(`Class (0, std.Maxcharval))

 	| '^':	getc(re); ret = mk(`Bol)

@@ -427,17 +424,18 @@

 	| '(':	

 		getc(re)

 		match altexpr(re)

-		| `Some s:	ret = mk(`Cap s)

-		| `None:	-> `Fail (`Emptyparen)

+		| `Some s:

+			if matchc(re, ')')

+				-> `Some mk(`Cap (re.nmatch++, s))

+			else

+				-> `Fail `Unbalanced

+			;;

+		| `None:	-> `Fail `Emptyparen

 		;;

-		if !matchc(re, ')')

-			astfree(ret)

-			-> `Fail (`Unbalanced)

-		;;

 	| '\\':

 		getc(re) /* consume the slash */

 		if re.pat.len == 0

-			-> `Fail (`Earlystop)

+			-> `Fail `Earlystop

 		;;

 		-> escaped(re)

 	| c:

@@ -700,7 +698,7 @@

 	| `Class (a, b):	

 	/* meta */

-	| `Cap	a:	astfree(a)

+	| `Cap	(m, a):	astfree(a)

 	;;

 	std.free(t)

 }

--- a/test/data/regex-capture-expected

+++ b/test/data/regex-capture-expected

@@ -4,10 +4,13 @@

 Matched. 2 matches

 match 0: Abcde

 match 1: bcd

+Matched. 2 matches

+match 0: abab

+match 1: b

 Matched. 3 matches

 match 0: Abcde

-match 1: bcd

-match 2: c

+match 1: c

+match 2: bcd

 Matched. 4 matches

 match 0: aaaa

 match 1: a

--- a/test/regex-capture.myr

+++ b/test/regex-capture.myr

@@ -3,6 +3,7 @@

 const main = {

 	testmatch("A(.*)", "Abc")

 	testmatch("A(.*)e", "Abcde")

+	testmatch("(a|b)+", "abab")

 	testmatch("A(b(.*)d)e", "Abcde")

 	testmatch("(a?)(a*)(a?)", "aaaa")

 }