modure: Update to re1.5 v0.6.1, fixed and extended character class support.

pfalcon · pfalcon · commit 297d8469b8fc · 2014-10-17T22:25:18.000+03:00
diff --git a/extmod/re1.5/charclass.c b/extmod/re1.5/charclass.c
@@ -3,9 +3,11 @@
 int _re1_5_classmatch(const char *pc, const char *sp)
 {
     // pc points to "cnt" byte after opcode
+    int is_positive = (pc[-1] == Class);
     int cnt = *pc++;
     while (cnt--) {
-        if (!(*sp >= *pc && *sp <= pc[1])) return 0;
+        if (*sp >= *pc && *sp <= pc[1]) return is_positive;
+        pc += 2;
     }
-    return 1;
-}
+    return !is_positive;
+}
diff --git a/extmod/re1.5/compilecode.c b/extmod/re1.5/compilecode.c
@@ -48,6 +48,7 @@ int re1_5_sizecode(const char *re)
         case '[': {
             pc += 2;
             re++;
+            if (*re == '^') re++;
             while (*re != ']') {
                 if (!*re) return -1;
                 if (re[1] == '-') {
@@ -91,10 +92,15 @@ const char *_compilecode(const char *re, ByteProg *prog)
         case '[': {
             int cnt;
             term = pc;
-            EMIT(pc++, Class);
+            re++;
+            if (*re == '^') {
+                EMIT(pc++, ClassNot);
+                re++;
+            } else {
+                EMIT(pc++, Class);
+            }
             pc++; // Skip # of pair byte
             prog->len++;
-            re++;
             for (cnt = 0; *re != ']'; re++, cnt++) {
                 if (!*re) return NULL;
                 EMIT(pc++, *re);
diff --git a/extmod/re1.5/dumpcode.c b/extmod/re1.5/dumpcode.c
@@ -32,9 +32,11 @@ void re1_5_dumpcode(ByteProg *prog)
                 case Any:
                         printf("any\n");
                         break;
-                case Class: {
-                        int num = code[pc++];
-                        printf("class %d", num);
+                case Class:
+                case ClassNot: {
+                        int num = code[pc];
+                        printf("class%s %d", (code[pc - 1] == ClassNot ? "not" : ""), num);
+                        pc++;
                         while (num--) {
                             printf(" 0x%02x-0x%02x", code[pc], code[pc + 1]);
                             pc += 2;
diff --git a/extmod/re1.5/re1.5.h b/extmod/re1.5/re1.5.h
@@ -81,6 +81,7 @@ enum	/* Inst.opcode */
 	Char = CONSUMERS,
 	Any,
 	Class,
+	ClassNot,
 
 	ASSERTS = 0x50,
 	Bol = ASSERTS,
diff --git a/extmod/re1.5/recursiveloop.c b/extmod/re1.5/recursiveloop.c
@@ -24,6 +24,7 @@ recursiveloop(char *pc, const char *sp, Subject *input, const char **subp, int n
 			sp++;
 			continue;
 		case Class:
+		case ClassNot:
 			if (!_re1_5_classmatch(pc, sp))
 				return 0;
 			pc += *(unsigned char*)pc * 2 + 1;
diff --git a/tests/extmod/ure1.py b/tests/extmod/ure1.py
@@ -20,13 +20,27 @@
 except IndexError:
     print("IndexError")
 
-r = re.compile("[a-c]")
+r = re.compile("[a-cu-z]")
 m = r.match("a")
 print(m.group(0))
+m = r.match("z")
+print(m.group(0))
 m = r.match("d")
 print(m)
 m = r.match("A")
 print(m)
+print("===")
+
+r = re.compile("[^a-cu-z]")
+m = r.match("a")
+print(m)
+m = r.match("z")
+print(m)
+m = r.match("d")
+print(m.group(0))
+m = r.match("A")
+print(m.group(0))
+
 
 r = re.compile("o+")
 m = r.search("foobar")
