Revert r41786 "broken utf-8"

revert unexpected commit

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@41855 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

enc/utf_8.c

@@ -233,12 +233,12 @@ mbc_enc_len(const UChar* p, const UChar* e, OnigEncoding enc ARG_UNUSED)
   if (p == e) return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(EncLen_UTF8[firstbyte]-2);
   s = trans[s][*p++];
   if (s < 0) return s == ACCEPT ? ONIGENC_CONSTRUCT_MBCLEN_CHARFOUND(3) :
-                                  ONIGENC_CONSTRUCT_MBCLEN_INVALID2(2);
+                                  ONIGENC_CONSTRUCT_MBCLEN_INVALID();
 
   if (p == e) return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(EncLen_UTF8[firstbyte]-3);
   s = trans[s][*p++];
   return s == ACCEPT ? ONIGENC_CONSTRUCT_MBCLEN_CHARFOUND(4) :
-                       ONIGENC_CONSTRUCT_MBCLEN_INVALID2(3);
+                       ONIGENC_CONSTRUCT_MBCLEN_INVALID();
 }
 
 static int

include/ruby/oniguruma.h

@@ -246,14 +246,11 @@ ONIG_EXTERN OnigEncodingType OnigEncodingASCII;
 #define ONIGENC_MBCLEN_CHARFOUND_P(r)           (0 < (r))
 #define ONIGENC_MBCLEN_CHARFOUND_LEN(r)         (r)
 
-#define ONIGENC_INVALID_NUM                     0x10000
-#define ONIGENC_CONSTRUCT_MBCLEN_INVALID()      (-1-ONIGENC_INVALID_NUM)
-#define ONIGENC_CONSTRUCT_MBCLEN_INVALID2(n)    (-(n)-ONIGENC_INVALID_NUM)
-#define ONIGENC_MBCLEN_INVALID_P(r)             ((r) < -ONIGENC_INVALID_NUM)
-#define ONIGENC_MBCLEN_INVALID_LEN(r)           (-(r)-ONIGENC_INVALID_NUM)
+#define ONIGENC_CONSTRUCT_MBCLEN_INVALID()      (-1)
+#define ONIGENC_MBCLEN_INVALID_P(r)             ((r) == -1)
 
 #define ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(n)    (-1-(n))
-#define ONIGENC_MBCLEN_NEEDMORE_P(r)            (-ONIGENC_INVALID_NUM <= (r) && (r) < -1)
+#define ONIGENC_MBCLEN_NEEDMORE_P(r)            ((r) < -1)
 #define ONIGENC_MBCLEN_NEEDMORE_LEN(r)          (-1-(r))
 
 #define ONIGENC_PRECISE_MBC_ENC_LEN(enc,p,e)   (enc)->precise_mbc_enc_len(p,e,enc)

regenc.c

@@ -55,12 +55,10 @@ extern int
 onigenc_mbclen_approximate(const OnigUChar* p,const OnigUChar* e, struct OnigEncodingTypeST* enc)
 {
   int ret = ONIGENC_PRECISE_MBC_ENC_LEN(enc,p,e);
-  if (ONIGENC_MBCLEN_NEEDMORE_P(ret))
-    return (int)(e-p)+ONIGENC_MBCLEN_NEEDMORE_LEN(ret);
-  else if (ONIGENC_MBCLEN_INVALID_P(ret))
-    return (int)(e-p)+ONIGENC_MBCLEN_INVALID_LEN(ret);
-  else if (ONIGENC_MBCLEN_CHARFOUND_P(ret))
+  if (ONIGENC_MBCLEN_CHARFOUND_P(ret))
     return ONIGENC_MBCLEN_CHARFOUND_LEN(ret);
+  else if (ONIGENC_MBCLEN_NEEDMORE_P(ret))
+    return (int)(e-p)+ONIGENC_MBCLEN_NEEDMORE_LEN(ret);
   return 1;
 }