New upstream version 6.9.5upstream/6.9.5

1 files changed, 41 insertions, 25 deletions

diff --git a/src/regenc.c b/src/regenc.c
index 16ac313..dbfbc89 100644
--- a/src/regenc.c
+++ b/src/regenc.c
@@ -29,6 +29,9 @@
 
 #include "regint.h"
 
+#define LARGE_S   0x53
+#define SMALL_S   0x73
+
 OnigEncoding OnigEncDefaultCharEncoding = ONIG_ENCODING_INIT_DEFAULT;
 
 #define INITED_LIST_SIZE  20
@@ -549,7 +552,7 @@ static int
 ss_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED,
                        OnigApplyAllCaseFoldFunc f, void* arg)
 {
-  static OnigCodePoint ss[] = { 0x73, 0x73 };
+  static OnigCodePoint ss[] = { SMALL_S, SMALL_S };
 
   return (*f)((OnigCodePoint )0xdf, ss, 2, arg);
 }
@@ -588,35 +591,48 @@ onigenc_get_case_fold_codes_by_str_with_map(int map_size,
     int ess_tsett_flag, OnigCaseFoldType flag ARG_UNUSED,
     const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem items[])
 {
-  if (0x41 <= *p && *p <= 0x5a) {
+  int i, j, n;
+  static OnigUChar sa[] = { LARGE_S, SMALL_S };
+
+  if (0x41 <= *p && *p <= 0x5a) { /* A - Z */
+    if (*p == LARGE_S && ess_tsett_flag != 0 && end > p + 1
+        && (*(p+1) == LARGE_S || *(p+1) == SMALL_S)) { /* SS */
+    ss_combination:
+      items[0].byte_len = 2;
+      items[0].code_len = 1;
+      items[0].code[0] = (OnigCodePoint )0xdf;
+
+      n = 1;
+      for (i = 0; i < 2; i++) {
+        for (j = 0; j < 2; j++) {
+          if (sa[i] == *p && sa[j] == *(p+1))
+            continue;
+
+          items[n].byte_len = 2;
+          items[n].code_len = 2;
+          items[n].code[0] = (OnigCodePoint )sa[i];
+          items[n].code[1] = (OnigCodePoint )sa[j];
+          n++;
+        }
+      }
+      return 4;
+    }
+
     items[0].byte_len = 1;
     items[0].code_len = 1;
     items[0].code[0] = (OnigCodePoint )(*p + 0x20);
-    if (*p == 0x53 && ess_tsett_flag != 0 && end > p + 1
-        && (*(p+1) == 0x53 || *(p+1) == 0x73)) {
-      /* SS */
-      items[1].byte_len = 2;
-      items[1].code_len = 1;
-      items[1].code[0] = (OnigCodePoint )0xdf;
-      return 2;
-    }
-    else
-      return 1;
+    return 1;
   }
-  else if (0x61 <= *p && *p <= 0x7a) {
+  else if (0x61 <= *p && *p <= 0x7a) { /* a - z */
+    if (*p == SMALL_S && ess_tsett_flag != 0 && end > p + 1
+        && (*(p+1) == SMALL_S || *(p+1) == LARGE_S)) {
+      goto ss_combination;
+    }
+
     items[0].byte_len = 1;
     items[0].code_len = 1;
     items[0].code[0] = (OnigCodePoint )(*p - 0x20);
-    if (*p == 0x73 && ess_tsett_flag != 0 && end > p + 1
-        && (*(p+1) == 0x73 || *(p+1) == 0x53)) {
-      /* ss */
-      items[1].byte_len = 2;
-      items[1].code_len = 1;
-      items[1].code[0] = (OnigCodePoint )0xdf;
-      return 2;
-    }
-    else
-      return 1;
+    return 1;
   }
   else if (*p == 0xdf && ess_tsett_flag != 0) {
     items[0].byte_len = 1;
@@ -676,7 +692,7 @@ extern int
 onigenc_is_mbc_newline_0x0a(const UChar* p, const UChar* end)
 {
   if (p < end) {
-    if (*p == 0x0a) return 1;
+    if (*p == NEWLINE_CODE) return 1;
   }
   return 0;
 }
@@ -887,7 +903,7 @@ onigenc_is_mbc_word_ascii(OnigEncoding enc, UChar* s, const UChar* end)
 {
   OnigCodePoint code = ONIGENC_MBC_TO_CODE(enc, s, end);
 
-  if (code > 127) return 0;
+  if (code > ASCII_LIMIT) return 0;
 
   return ONIGENC_IS_ASCII_CODE_WORD(code);
 }