extmod: Pull in upstream changes to re1.5; fixes bug, adds named class.

dpgeorge · dpgeorge · commit d09a5b51c2c9 · 2015-02-21T10:33:20.000Z
diff --git a/extmod/re1.5/charclass.c b/extmod/re1.5/charclass.c
@@ -11,3 +11,23 @@ int _re1_5_classmatch(const char *pc, const char *sp)
     }
     return !is_positive;
 }
+
+int _re1_5_namedclassmatch(const char *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;
+}
diff --git a/extmod/re1.5/compilecode.c b/extmod/re1.5/compilecode.c
@@ -78,6 +78,13 @@ const char *_compilecode(const char *re, ByteProg *prog)
         switch (*re) {
         case '\\':
             re++;
+            if ((*re | 0x20) == 'd' || (*re | 0x20) == 's' || (*re | 0x20) == 'w') {
+                term = pc;
+                EMIT(pc++, NamedClass);
+                EMIT(pc++, *re);
+                prog->len++;
+                break;
+            }
         default:
             term = pc;
             EMIT(pc++, Char);
@@ -112,22 +119,24 @@ const char *_compilecode(const char *re, ByteProg *prog)
             EMIT(term + 1, cnt);
             break;
         }
-        case '(':
+        case '(': {
             term = pc;
+            int sub = ++prog->sub;
 
             EMIT(pc++, Save);
-            EMIT(pc++, 2 * ++prog->sub);
+            EMIT(pc++, 2 * sub);
             prog->len++;
 
             prog->bytelen = pc;
             re = _compilecode(re + 1, prog);
             pc = prog->bytelen;
 
             EMIT(pc++, Save);
-            EMIT(pc++, 2 * prog->sub + 1);
+            EMIT(pc++, 2 * sub + 1);
             prog->len++;
 
             break;
+        }
         case '?':
             insert_code(code, term, 2, &pc);
             EMIT(term, Split);
diff --git a/extmod/re1.5/re1.5.h b/extmod/re1.5/re1.5.h
@@ -82,6 +82,7 @@ enum	/* Inst.opcode */
 	Any,
 	Class,
 	ClassNot,
+	NamedClass,
 
 	ASSERTS = 0x50,
 	Bol = ASSERTS,
@@ -145,5 +146,6 @@ int re1_5_compilecode(ByteProg *prog, const char *re);
 void re1_5_dumpcode(ByteProg *prog);
 void cleanmarks(ByteProg *prog);
 int _re1_5_classmatch(const char *pc, const char *sp);
+int _re1_5_namedclassmatch(const char *pc, const char *sp);
 
 #endif /*_RE1_5_REGEXP__H*/
diff --git a/extmod/re1.5/recursiveloop.c b/extmod/re1.5/recursiveloop.c
@@ -30,6 +30,12 @@ recursiveloop(char *pc, const char *sp, Subject *input, const char **subp, int n
 			pc += *(unsigned char*)pc * 2 + 1;
 			sp++;
 			continue;
+                case NamedClass:
+			if (!_re1_5_namedclassmatch(pc, sp))
+				return 0;
+			pc++;
+			sp++;
+			continue;
 		case Match:
 			return 1;
 		case Jmp:
