Merge pull request #1412 from RaiKoHoff/Dev_NewFeatures

Oniguruma: Update and CRLF fix
2026-06-14 21:09:05 +08:00 · 2019-07-09 11:54:47 +02:00 · 2019-07-09 11:54:47 +02:00 · 4632f50b58
commit 4632f50b58
parent 0781844c77 09b3318b56
10 changed files with 58 additions and 17 deletions
--- a/oniguruma/doc/API
+++ b/oniguruma/doc/API
@ -1,4 +1,4 @@
-Oniguruma API  Version 6.9.2  2019/03/25
+Oniguruma API  Version 6.9.3  2019/07/06

 #include <oniguruma.h>

@ -168,6 +168,9 @@ Oniguruma API  Version 6.9.2  2019/03/25
 # int onig_new_deluxe(regex_t** reg, const UChar* pattern, const UChar* pattern_end,
                      OnigCompileInfo* ci, OnigErrorInfo* einfo)

+  This function is deprecate, and it does not allow the case where
+  the encoding of pattern and target is different.
+
  Create a regex object.
  This function is deluxe version of onig_new().

@ -299,6 +302,7 @@ Oniguruma API  Version 6.9.2  2019/03/25
                   const UChar* range, OnigRegion* region, OnigOptionType option)

  Search string and return search result and matching region.
+  Do not pass invalid byte string in the regex character encoding.

  normal return: match position offset (i.e.  p - str >= 0)
  not found:     ONIG_MISMATCH (< 0)
@ -323,15 +327,19 @@ Oniguruma API  Version 6.9.2  2019/03/25
                   const UChar* start, const UChar* range, OnigRegion* region,
                   OnigOptionType option, OnigMatchParam* mp)

-   arguments
-   1-7:  same as onig_search()
-   8 mp: match parameter values (match_stack_limit, retry_limit_in_match)
+  Search string and return search result and matching region.
+  Do not pass invalid byte string in the regex character encoding.
+
+  arguments
+  1-7:  same as onig_search()
+  8 mp: match parameter values (match_stack_limit, retry_limit_in_match)


 # int onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at,
                 OnigRegion* region, OnigOptionType option)

  Match string and return result and matching region.
+  Do not pass invalid byte string in the regex character encoding.

  normal return: match length  (>= 0)
  not match:     ONIG_MISMATCH ( < 0)
@ -353,6 +361,9 @@ Oniguruma API  Version 6.9.2  2019/03/25
                            const UChar* at, OnigRegion* region,
                            OnigOptionType option, OnigMatchParam* mp)

+  Match string and return result and matching region.
+  Do not pass invalid byte string in the regex character encoding.
+
   arguments
   1-6:  same as onig_match()
   7 mp: match parameter values (match_stack_limit, retry_limit_in_match)
@ -364,6 +375,7 @@ Oniguruma API  Version 6.9.2  2019/03/25
                void* callback_arg)

  Scan string and callback with matching region.
+  Do not pass invalid byte string in the regex character encoding.

  normal return: number of matching times
  error:         error code
@ -611,14 +623,20 @@ Oniguruma API  Version 6.9.2  2019/03/25


 # int onigenc_strlen(OnigEncoding enc, const UChar* s, const UChar* end)
+
+  Return number of characters in the string.
+
+
 # int onigenc_strlen_null(OnigEncoding enc, const UChar* s)

  Return number of characters in the string.
+  Do not pass invalid byte string in the character encoding.


 # int onigenc_str_bytelen_null(OnigEncoding enc, const UChar* s)

  Return number of bytes in the string.
+  Do not pass invalid byte string in the character encoding.


 # int onig_set_default_syntax(OnigSyntaxType* syntax)
--- a/oniguruma/src/ascii.c
+++ b/oniguruma/src/ascii.c
@ -1,4 +1,4 @@
-/**********************************************************************
+/**********************************************************************
  ascii.c -  Oniguruma (regular expression library)
 **********************************************************************/
 /*-
@ -98,7 +98,11 @@ OnigEncodingType OnigEncodingASCII = {
  "US-ASCII",  /* name */
  1,           /* max enc length */
  1,           /* min enc length */
+#ifdef USE_CRNL_AS_LINE_TERMINATOR
+  onigenc_is_mbc_newline_0x0a_or_0x0d,
+#else
  onigenc_is_mbc_newline_0x0a,
+#endif
  onigenc_single_byte_mbc_to_code,
  onigenc_single_byte_code_to_mbclen,
  onigenc_single_byte_code_to_mbc,
--- a/oniguruma/src/regenc.c
+++ b/oniguruma/src/regenc.c
@ -1,4 +1,4 @@
-/**********************************************************************
+/**********************************************************************
  regenc.c -  Oniguruma (regular expression library)
 **********************************************************************/
 /*-
@ -694,6 +694,17 @@ onigenc_is_mbc_newline_0x0a(const UChar* p, const UChar* end)
  return 0;
 }

+#ifdef USE_CRNL_AS_LINE_TERMINATOR
+extern int
+onigenc_is_mbc_newline_0x0a_or_0x0d(const UChar* p, const UChar* end)
+{
+  if (p < end) {
+    if ((*p == 0x0a) || (*p == 0x0d)) return 1;
+  }
+  return 0;
+}
+#endif
+
 /* for single byte encodings */
 extern int
 onigenc_ascii_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED, const UChar** p,
--- a/oniguruma/src/regenc.h
+++ b/oniguruma/src/regenc.h
@ -1,4 +1,4 @@
-#ifndef REGENC_H
+#ifndef REGENC_H
 #define REGENC_H
 /**********************************************************************
  regenc.h -  Oniguruma (regular expression library)
@ -111,15 +111,16 @@ struct PropertyNameCtype {
  int ctype;
 };

-/* #define USE_CRNL_AS_LINE_TERMINATOR */
+#define USE_CRNL_AS_LINE_TERMINATOR
 #define USE_UNICODE_PROPERTIES
 #define USE_UNICODE_EXTENDED_GRAPHEME_CLUSTER
 #define USE_UNICODE_WORD_BREAK
 /* #define USE_UNICODE_CASE_FOLD_TURKISH_AZERI */
-/* #define USE_UNICODE_ALL_LINE_TERMINATORS */  /* see Unicode.org UTS #18 */
+/* #define USE_UNICODE_ALL_LINE_TERMINATORS */ /* see Unicode.org UTS #18 */


-#define ONIG_ENCODING_INIT_DEFAULT           ONIG_ENCODING_ASCII
+//#define ONIG_ENCODING_INIT_DEFAULT           ONIG_ENCODING_ASCII
+#define ONIG_ENCODING_INIT_DEFAULT           ONIG_ENCODING_UTF8


 #define ENC_SKIP_OFFSET_1_OR_0             7
@ -146,7 +147,9 @@ extern int onigenc_apply_all_case_fold_with_map P_((int map_size, const OnigPair
 extern int onigenc_get_case_fold_codes_by_str_with_map P_((int map_size, const OnigPairCaseFoldCodes map[], int ess_tsett_flag, OnigCaseFoldType flag, const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem items[]));
 extern int onigenc_not_support_get_ctype_code_range P_((OnigCtype ctype, OnigCodePoint* sb_out, const OnigCodePoint* ranges[]));
 extern int onigenc_is_mbc_newline_0x0a P_((const UChar* p, const UChar* end));
-
+#ifdef USE_CRNL_AS_LINE_TERMINATOR
+extern int onigenc_is_mbc_newline_0x0a_or_0x0d P_((const UChar* p, const UChar* end));
+#endif

 /* methods for single byte encoding */
 extern int onigenc_ascii_mbc_case_fold P_((OnigCaseFoldType flag, const UChar** p, const UChar* end, UChar* lower));
--- a/oniguruma/src/regexec.c
+++ b/oniguruma/src/regexec.c
@ -1,4 +1,4 @@
-/**********************************************************************
+/**********************************************************************
  regexec.c -  Oniguruma (regular expression library)
 **********************************************************************/
 /*-
--- a/oniguruma/src/regext.c
+++ b/oniguruma/src/regext.c
@ -29,6 +29,7 @@

 #include "regint.h"

+#if 0
 static void
 conv_ext0be32(const UChar* s, const UChar* end, UChar* conv)
 {
@ -95,7 +96,6 @@ static int
 conv_encoding(OnigEncoding from, OnigEncoding to, const UChar* s, const UChar* end,
              UChar** conv, UChar** conv_end)
 {
-#if 0
  int len = (int )(end - s);

  if (to == ONIG_ENCODING_UTF16_BE) {
@ -156,9 +156,9 @@ conv_encoding(OnigEncoding from, OnigEncoding to, const UChar* s, const UChar* e
      goto swap32;
    }
  }
-#endif
  return ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION;
 }
+#endif

 extern int
 onig_new_deluxe(regex_t** reg, const UChar* pattern, const UChar* pattern_end,
--- a/oniguruma/src/regint.h
+++ b/oniguruma/src/regint.h
@ -1,4 +1,4 @@
-#ifndef REGINT_H
+#ifndef REGINT_H
 #define REGINT_H
 /**********************************************************************
  regint.h -  Oniguruma (regular expression library)
--- a/oniguruma/src/utf8.c
+++ b/oniguruma/src/utf8.c
@ -1,4 +1,4 @@
-/**********************************************************************
+/**********************************************************************
  utf8.c -  Oniguruma (regular expression library)
 **********************************************************************/
 /*-
@ -297,7 +297,11 @@ OnigEncodingType OnigEncodingUTF8 = {
  6,
 #endif
  1,           /* min enc length */
+#ifdef USE_CRNL_AS_LINE_TERMINATOR
+  onigenc_is_mbc_newline_0x0a_or_0x0d,
+#else
  onigenc_is_mbc_newline_0x0a,
+#endif
  mbc_to_code,
  code_to_mbclen,
  code_to_mbc,
--- a/oniguruma/version.txt
+++ b/oniguruma/version.txt
@ -1 +1 @@
-6.9.2
+6.9.3
--- a/scioniguruma/OnigurumaRegExEngine.cxx
+++ b/scioniguruma/OnigurumaRegExEngine.cxx
@ -75,6 +75,7 @@ static void SetSimpleOptions(OnigOptionType& onigOptions,
  ONIG_OPTION_OFF(onigOptions, ONIG_OPTION_EXTEND);
  ONIG_OPTION_OFF(onigOptions, ONIG_OPTION_SINGLELINE);
  ONIG_OPTION_ON(onigOptions, ONIG_OPTION_NEGATE_SINGLELINE);
+
  //ONIG_OPTION_OFF(onigOptions, ONIG_OPTION_ASCII_RANGE);
  //ONIG_OPTION_OFF(onigOptions, ONIG_OPTION_CAPTURE_GROUP);
 @ -1 +1 @@
 .9.2
 .9.3