summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authornaruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2008-01-13 17:16:09 +0000
committernaruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2008-01-13 17:16:09 +0000
commit8f15b8128cf47ff679d4f0d3767605d0d0afcc3c ()
tree643bd4938ff2bccdecdb8e63702ce8bd4e55a10a
parent21671b558cfd8d90647f40a8594fd9e6db038768 (diff)
* include/ruby/oniguruma.h: remove ONIG_ENCODING_* and OnigEncoding*
which are not builtin. * regenc.{c,h} (onigenc_mb2_code_to_mbclen, onigenc_mb4_code_to_mbclen): fix . * enc/big5.c, enc/euc_kr.c, enc/euc_tw.c, enc/gb18030.c, enc/koi8_r.c, enc/windows_1251.c: imported from Oniguruma. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@15026 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
-rw-r--r--ChangeLog11
-rw-r--r--enc/big5.c164
-rw-r--r--enc/euc_jp.c8
-rw-r--r--enc/euc_kr.c141
-rw-r--r--enc/euc_tw.c140
-rw-r--r--enc/gb18030.c500
-rw-r--r--enc/iso_8859_2.c2
-rw-r--r--enc/iso_8859_5.c2
-rw-r--r--enc/koi8_r.c212
-rw-r--r--enc/shift_jis.c10
-rw-r--r--enc/windows_1251.c201
-rw-r--r--include/ruby/oniguruma.h58
-rw-r--r--regenc.c4
-rw-r--r--regenc.h4
14 files changed, 1383 insertions, 74 deletions
@@ -1,3 +1,14 @@
Sun Jan 13 22:47:28 2008 NARUSE, Yui <[email protected]>
* enc/make_encdb.h: sort encoding names by original name.
@@ -0,0 +1,164 @@
@@ -139,7 +139,7 @@ mbc_to_code(const UChar* p, const UChar* end, OnigEncoding enc)
int c, i, len;
OnigCodePoint n;
- len = enclen(&encoding_EUC_JP, p, end);
n = (OnigCodePoint )*p++;
if (len == 1) return n;
@@ -191,7 +191,7 @@ code_to_mbc(OnigCodePoint code, UChar *buf, OnigEncoding enc)
*p++ = (UChar )(code & 0xff);
#if 1
- if (enclen(&encoding_EUC_JP, buf, p) != (p - buf))
return ONIGERR_INVALID_CODE_POINT_VALUE;
#endif
return p - buf;
@@ -213,7 +213,7 @@ mbc_case_fold(OnigCaseFoldType flag,
else {
int i;
- len = enclen(&encoding_EUC_JP, p, end);
for (i = 0; i < len; i++) {
*lower++ = *p++;
}
@@ -235,7 +235,7 @@ left_adjust_char_head(const UChar* start, const UChar* s, OnigEncoding enc)
p = s;
while (!eucjp_islead(*p) && p > start) p--;
- len = enclen(&encoding_EUC_JP, p, s);
if (p + len > s) return (UChar* )p;
p += len;
return (UChar* )(p + ((s - p) & ~1));
@@ -0,0 +1,141 @@
@@ -0,0 +1,140 @@
@@ -0,0 +1,500 @@
@@ -238,5 +238,5 @@ OnigEncodingDefine(iso_8859_2, ISO_8859_2) = {
onigenc_always_true_is_allowed_reverse_match
};
ENC_ALIAS("ISO8859-2", "ISO-8859-2");
-ENC_REPLICATE("Windowws-1250", "ISO-8859-2");
ENC_ALIAS("CP1250", "Windows-1250");
@@ -228,5 +228,3 @@ OnigEncodingDefine(iso_8859_5, ISO_8859_5) = {
onigenc_always_true_is_allowed_reverse_match
};
ENC_ALIAS("ISO8859-5", "ISO-8859-5");
-ENC_REPLICATE("Windows-1251", "ISO-8859-5");
-ENC_ALIAS("CP1251", "Windows-1251");
@@ -0,0 +1,212 @@
@@ -152,7 +152,7 @@ mbc_to_code(const UChar* p, const UChar* end, OnigEncoding enc)
int c, i, len;
OnigCodePoint n;
- len = enclen(&encoding_Shift_JIS, p, end);
c = *p++;
n = c;
if (len == 1) return n;
@@ -174,7 +174,7 @@ code_to_mbc(OnigCodePoint code, UChar *buf, OnigEncoding enc)
*p++ = (UChar )(code & 0xff);
#if 0
- if (enclen(&encoding_Shift_JIS, buf) != (p - buf))
return REGERR_INVALID_CODE_POINT_VALUE;
#endif
return p - buf;
@@ -194,7 +194,7 @@ mbc_case_fold(OnigCaseFoldType flag,
}
else {
int i;
- int len = enclen(&encoding_Shift_JIS, p, end);
for (i = 0; i < len; i++) {
*lower++ = *p++;
@@ -209,7 +209,7 @@ static int
is_mbc_ambiguous(OnigCaseFoldType flag,
const UChar** pp, const UChar* end)
{
- return onigenc_mbn_is_mbc_ambiguous(&encoding_Shift_JIS, flag, pp, end);
}
#endif
@@ -247,7 +247,7 @@ left_adjust_char_head(const UChar* start, const UChar* s, OnigEncoding enc)
}
}
}
- len = enclen(&encoding_Shift_JIS, p, s);
if (p + len > s) return (UChar* )p;
p += len;
return (UChar* )(p + ((s - p) & ~1));
@@ -0,0 +1,201 @@
@@ -173,68 +173,10 @@ typedef struct OnigEncodingTypeST {
typedef OnigEncodingType* OnigEncoding;
ONIG_EXTERN OnigEncodingType OnigEncodingASCII;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_1;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_2;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_3;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_4;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_5;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_6;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_7;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_8;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_9;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_10;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_11;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_13;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_14;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_15;
-ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_16;
ONIG_EXTERN OnigEncodingType OnigEncodingUTF8;
-ONIG_EXTERN OnigEncodingType OnigEncodingUTF_16BE;
-ONIG_EXTERN OnigEncodingType OnigEncodingUTF_16LE;
-ONIG_EXTERN OnigEncodingType OnigEncodingUTF_32BE;
-ONIG_EXTERN OnigEncodingType OnigEncodingUTF_32LE;
-ONIG_EXTERN OnigEncodingType OnigEncodingEUC_JP;
-ONIG_EXTERN OnigEncodingType OnigEncodingEUC_TW;
-ONIG_EXTERN OnigEncodingType OnigEncodingEUC_KR;
-ONIG_EXTERN OnigEncodingType OnigEncodingEUC_CN;
-ONIG_EXTERN OnigEncodingType OnigEncodingSJIS;
-ONIG_EXTERN OnigEncodingType OnigEncodingKOI8;
-ONIG_EXTERN OnigEncodingType OnigEncodingKOI8_R;
-ONIG_EXTERN OnigEncodingType OnigEncodingCP1251;
-ONIG_EXTERN OnigEncodingType OnigEncodingBIG5;
-ONIG_EXTERN OnigEncodingType OnigEncodingGB18030;
#define ONIG_ENCODING_ASCII (&OnigEncodingASCII)
-#define ONIG_ENCODING_ISO_8859_1 (&OnigEncodingISO_8859_1)
-#define ONIG_ENCODING_ISO_8859_2 (&OnigEncodingISO_8859_2)
-#define ONIG_ENCODING_ISO_8859_3 (&OnigEncodingISO_8859_3)
-#define ONIG_ENCODING_ISO_8859_4 (&OnigEncodingISO_8859_4)
-#define ONIG_ENCODING_ISO_8859_5 (&OnigEncodingISO_8859_5)
-#define ONIG_ENCODING_ISO_8859_6 (&OnigEncodingISO_8859_6)
-#define ONIG_ENCODING_ISO_8859_7 (&OnigEncodingISO_8859_7)
-#define ONIG_ENCODING_ISO_8859_8 (&OnigEncodingISO_8859_8)
-#define ONIG_ENCODING_ISO_8859_9 (&OnigEncodingISO_8859_9)
-#define ONIG_ENCODING_ISO_8859_10 (&OnigEncodingISO_8859_10)
-#define ONIG_ENCODING_ISO_8859_11 (&OnigEncodingISO_8859_11)
-#define ONIG_ENCODING_ISO_8859_13 (&OnigEncodingISO_8859_13)
-#define ONIG_ENCODING_ISO_8859_14 (&OnigEncodingISO_8859_14)
-#define ONIG_ENCODING_ISO_8859_15 (&OnigEncodingISO_8859_15)
-#define ONIG_ENCODING_ISO_8859_16 (&OnigEncodingISO_8859_16)
#define ONIG_ENCODING_UTF8 (&OnigEncodingUTF8)
-#define ONIG_ENCODING_UTF_16BE (&OnigEncodingUTF_16BE)
-#define ONIG_ENCODING_UTF_16LE (&OnigEncodingUTF_16LE)
-#define ONIG_ENCODING_UTF_32BE (&OnigEncodingUTF_32BE)
-#define ONIG_ENCODING_UTF_32LE (&OnigEncodingUTF_32LE)
-#define ONIG_ENCODING_EUC_JP (&OnigEncodingEUC_JP)
-#define ONIG_ENCODING_EUC_TW (&OnigEncodingEUC_TW)
-#define ONIG_ENCODING_EUC_KR (&OnigEncodingEUC_KR)
-#define ONIG_ENCODING_EUC_CN (&OnigEncodingEUC_CN)
-#define ONIG_ENCODING_SJIS (&OnigEncodingSJIS)
-#define ONIG_ENCODING_KOI8 (&OnigEncodingKOI8)
-#define ONIG_ENCODING_KOI8_R (&OnigEncodingKOI8_R)
-#define ONIG_ENCODING_CP1251 (&OnigEncodingCP1251)
-#define ONIG_ENCODING_BIG5 (&OnigEncodingBIG5)
-#define ONIG_ENCODING_GB18030 (&OnigEncodingGB18030)
#define ONIG_ENCODING_UNDEF ((OnigEncoding )0)
@@ -712,14 +712,14 @@ onigenc_mbn_is_mbc_ambiguous(OnigEncoding enc, OnigCaseFoldType flag,
#endif
extern int
-onigenc_mb2_code_to_mbclen(OnigCodePoint code)
{
if ((code & 0xff00) != 0) return 2;
else return 1;
}
extern int
-onigenc_mb4_code_to_mbclen(OnigCodePoint code)
{
if ((code & 0xff000000) != 0) return 4;
else if ((code & 0xff0000) != 0) return 3;
@@ -133,12 +133,12 @@ ONIG_EXTERN int onigenc_ascii_is_code_ctype P_((OnigCodePoint code, unsigned int
/* methods for multi byte encoding */
ONIG_EXTERN OnigCodePoint onigenc_mbn_mbc_to_code P_((OnigEncoding enc, const UChar* p, const UChar* end));
ONIG_EXTERN int onigenc_mbn_mbc_case_fold P_((OnigEncoding enc, OnigCaseFoldType flag, const UChar** p, const UChar* end, UChar* lower));
-ONIG_EXTERN int onigenc_mb2_code_to_mbclen P_((OnigCodePoint code));
ONIG_EXTERN int onigenc_mb2_code_to_mbc P_((OnigEncoding enc, OnigCodePoint code, UChar *buf));
ONIG_EXTERN int onigenc_minimum_property_name_to_ctype P_((OnigEncoding enc, UChar* p, UChar* end));
ONIG_EXTERN int onigenc_unicode_property_name_to_ctype P_((OnigEncoding enc, UChar* p, UChar* end));
ONIG_EXTERN int onigenc_mb2_is_code_ctype P_((OnigEncoding enc, OnigCodePoint code, unsigned int ctype));
-ONIG_EXTERN int onigenc_mb4_code_to_mbclen P_((OnigCodePoint code));
ONIG_EXTERN int onigenc_mb4_code_to_mbc P_((OnigEncoding enc, OnigCodePoint code, UChar *buf));
ONIG_EXTERN int onigenc_mb4_is_code_ctype P_((OnigEncoding enc, OnigCodePoint code, unsigned int ctype));