From 3837d0364bf4da659cf7c205300caeb14d6c80e9 Mon Sep 17 00:00:00 2001
From: Kyryl Melekhin <k.melekhin@gmail.com>
Date: Sun, 18 Jul 2021 22:33:09 +0000
Subject: [PATCH] drop named classes because of no new functionality

---
 README  |  2 +-
 pike.c  | 40 +---------------------------------------
 test.sh | 15 +++------------
 3 files changed, 5 insertions(+), 52 deletions(-)

diff --git a/README b/README
index 24a9e53..fdafa67 100644
--- a/README
+++ b/README
@@ -41,7 +41,7 @@ TODO
 
 * Support for matching flags like case-insensitive, dot matches all,
 multiline, etc.
-* Support for more assertions like \A, \Z.
+* Support for wordend & wordbeg assertions
 
 Author and License
 ==================
diff --git a/pike.c b/pike.c
index d0c26d1..2b1ed9d 100644
--- a/pike.c
+++ b/pike.c
@@ -80,7 +80,6 @@ enum	/* rinst.opcode */
 	CHAR = 1,
 	ANY,
 	CLASS,
-	NAMEDCLASS,
 	// Assert position
 	BOL,
 	EOL,
@@ -150,24 +149,6 @@ int re_classmatch(const int *pc, const char *sp)
 	return !is_positive;
 }
 
-int re_namedclassmatch(const int *pc, const char *sp)
-{
-	// pc points to name of class
-	int off = (*pc >> 5) & 1;
-	if ((*pc | 0x20) == 'd') {
-		if (!(*sp >= '0' && *sp <= '9'))
-			off ^= 1;
-	} else if ((*pc | 0x20) == 's') {
-		if (!(*sp == ' ' || (*sp >= '\t' && *sp <= '\r')))
-			off ^= 1;
-	} else { // w
-		if (!((*sp >= 'A' && *sp <= 'Z') || (*sp >= 'a' && *sp <= 'z') ||
-			(*sp >= '0' && *sp <= '9') || *sp == '_'))
-			off ^= 1;
-	}
-	return off;
-}
-
 void re_dumpcode(rcode *prog)
 {
 	int pc = 0;
@@ -207,9 +188,6 @@ void re_dumpcode(rcode *prog)
 			}
 			printf("\n");
 			break;
-		case NAMEDCLASS:
-			printf("namedclass %c\n", code[pc++]);
-			break;
 		case MATCH:
 			printf("match\n");
 			break;
@@ -239,16 +217,6 @@ static int _compilecode(const char **re_loc, rcode *prog, int sizecode)
 		case '\\':
 			re++;
 			if (!*re) goto syntax_error; // Trailing backslash
-			c = *re | 0x20;
-			if (c == 'd' || c == 's' || c == 'w') {
-				term = PC;
-				EMIT(PC++, NAMEDCLASS);
-				EMIT(PC++, *re);
-				prog->len++;
-				break;
-			}
-			if ((c >= '0' && c <= '9') || (c >= 'a' && c <= 'z'))
-				goto unsupported_escape;
 		default:
 			term = PC;
 			EMIT(PC++, CHAR);
@@ -353,7 +321,6 @@ static int _compilecode(const char **re_loc, rcode *prog, int sizecode)
 				for (i = 0; i < size; i++)
 					switch (code[term]) {
 					case CLASS:
-					case NAMEDCLASS:
 					case JMP:
 					case SPLIT:
 					case RSPLIT:
@@ -591,7 +558,7 @@ int re_pikevm(rcode *prog, const char *s, const char **subp, int nsubp)
 	for(sp=s;; sp += l) {
 		if(clist->n == 0)
 			break;
-		gen++; uc_len(l, s)
+		gen++; uc_len(l, sp)
 		for(i=0; i<clist->n; i++) {
 			npc = clist->t[i].pc;
 			nsub = clist->t[i].sub;
@@ -614,11 +581,6 @@ int re_pikevm(rcode *prog, const char *s, const char **subp, int nsubp)
 					break;
 				npc += *(npc+1) * 2 + 2;
 				goto addthread;
-			case NAMEDCLASS:
-				if (!re_namedclassmatch(npc, sp))
-					break;
-				npc++;
-				goto addthread;
 			case MATCH:
 				matched = nsub;
 				for(i++; i < clist->n; i++)
diff --git a/test.sh b/test.sh
index 8ca4a2c..ee4b4eb 100755
--- a/test.sh
+++ b/test.sh
@@ -14,12 +14,9 @@ abc+
 ([0-9]*)(?:([a-z]*)(?:[0-9]*))
 (?:)
 1?:
-\\\d+
-\\\s+
-\\\w+
-(\\\w+)\\\s+(\\\w+)
-(\\\S+)\\\s+(\\\D+)
-(([0-9]*)([a-z]*)\\\d*)
+[0-9]+
+[a-zA-Z0-9_]+
+(([0-9]*)([a-z]*)[a-zA-Z0-9_]*)
 [a]*
 ([yab]*)(e*)([cd])
 ([yab]*)(e*)([^y]?)$
@@ -64,10 +61,7 @@ abcccdef
 1234hello568
 1:
 123abc456
- 	123abc456
 123abc_456 abc
-ABC 	123hello456 abc
-ABC 	helloabc456 abc
 123hello456
 a
 xyac
@@ -113,10 +107,7 @@ expect="\
 (0,0)
 (0,2)
 (0,3)
-(0,2)
 (0,10)
-(0,16)(0,3)(5,16)
-(0,13)(0,3)(5,13)
 (0,11)(0,11)(0,3)(3,8)
 (0,1)
 (1,4)(1,3)(3,3)(3,4)