Implement conditional casing for Greek letter sigma when title-casing text

Author: alexdowad

ext/mbstring/php_unicode.c

@@ -427,12 +427,34 @@ MBSTRING_API zend_string *php_unicode_convert_case(php_case_mode case_mode, cons
 					*p++ = w;
 					continue;
 				}
-				uint32_t w2 = title_mode ? php_unicode_tolower_raw(w, src_encoding) : php_unicode_totitle_raw(w, src_encoding);
+				uint32_t w2;
+				if (title_mode) {
+					if (w == 0x3A3) {
+						int j = i - 1;
+						while (j >= 0 && php_unicode_is_case_ignorable(wchar_buf[j])) {
+							j--;
+						}
+						if (j >= 0 ? php_unicode_is_cased(wchar_buf[j]) : scan_back_for_cased_letter(p, converted_end)) {
+							j = i + 1;
+							while (j < out_len && php_unicode_is_case_ignorable(wchar_buf[j])) {
+								j++;
+							}
+							if (j >= out_len ? !scan_ahead_for_cased_letter(in, in_len, state, src_encoding) : !php_unicode_is_cased(wchar_buf[j])) {
+								*p++ = 0x3C2;
+								goto set_title_mode;
+							}
+						}
+					}
+					w2 = php_unicode_tolower_raw(w, src_encoding);
+				} else {
+					w2 = php_unicode_totitle_raw(w, src_encoding);
+				}
 				if (UNEXPECTED(w2 > 0xFFFFFF)) {
 					p = emit_special_casing_sequence(w2, p);
 				} else {
 					*p++ = w2;
 				}
+set_title_mode:
 				if (!php_unicode_is_case_ignorable(w)) {
 					title_mode = php_unicode_is_cased(w);
 				}

ext/mbstring/tests/casemapping.phpt

@@ -115,7 +115,7 @@ dd
 dd
 69
 69
-Καλησπερα Σασ
+Καλησπερα Σας
 Καλησπερα Σασ
 καλησπερα σας
 καλησπερα σασ

ext/mbstring/tests/mb_convert_case_various_mode.phpt

@@ -21,6 +21,33 @@ try {
     echo $e->getMessage() . \PHP_EOL;
 }
 
+echo "\n-- Greek letter sigma --\n";
+var_dump(mb_convert_case("Σ", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("aΣ", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("aΣb", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("aΣ b", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case(" ΣΣΣΣ ", MB_CASE_TITLE, 'UTF-8'));
+
+// Apostrophe, full stop, colon, etc. are "case-ignorable"
+// When checking whether capital sigma is at the end of a word or not, we skip over
+// any number of case-ignorable characters, both when scanning back and when scanning forward
+var_dump(mb_convert_case("'Σ", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("ab'Σ", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("Σ'", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("Σ'a", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("a'Σ'a", MB_CASE_TITLE, 'UTF-8'));
+
+// We scan back by at least 63 characters when necessary,
+// but there is no guarantee that we will scan back further than that
+var_dump(mb_convert_case('a' . str_repeat('.', 63) . "Σ", MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case('a' . str_repeat('.', 64) . "Σ", MB_CASE_TITLE, 'UTF-8')); // Context-sensitive casing doesn't work here!
+
+// When scanning forward to confirm if capital sigma is at the end of a word or not,
+// there is no limit as to how far we will scan
+var_dump(mb_convert_case("abcΣ" . str_repeat('.', 64) . ' abc', MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("abcΣ" . str_repeat('.', 64) . 'a abc', MB_CASE_TITLE, 'UTF-8'));
+var_dump(mb_convert_case("abcΣ" . str_repeat('.', 256) . ' abc', MB_CASE_TITLE, 'UTF-8'));
+
 /* Regression test for new implementation;
  * When converting a codepoint, if we overwrite it with the converted version before
  * checking whether we should shift in/out of 'title mode', then the conversion will be incorrect */
@@ -38,5 +65,22 @@ string(13) "foo bar spaß"
 string(13) "Foo Bar Spaß"
 string(13) "foo bar spaß"
 mb_convert_case(): Argument #2 ($mode) must be one of the MB_CASE_* constants
+
+-- Greek letter sigma --
+string(2) "Σ"
+string(3) "Aς"
+string(4) "Aσb"
+string(5) "Aς B"
+string(10) " Σσσς "
+string(3) "'Σ"
+string(5) "Ab'ς"
+string(3) "Σ'"
+string(4) "Σ'a"
+string(6) "A'σ'a"
+string(66) "A...............................................................ς"
+string(67) "A................................................................σ"
+string(73) "Abcς................................................................ Abc"
+string(74) "Abcσ................................................................a Abc"
+string(265) "Abcς................................................................................................................................................................................................................................................................ Abc"
 string(12) "02bc004e012d"
 string(8) "0149012d"