diff --git a/src/ucd-tools/data/espeak-ng/PropList.txt b/src/ucd-tools/data/espeak-ng/PropList.txt index 12387d87..a1aa9200 100644 --- a/src/ucd-tools/data/espeak-ng/PropList.txt +++ b/src/ucd-tools/data/espeak-ng/PropList.txt @@ -1,5 +1,5 @@ # espeak-ng/PropList-9.0.0.txt -# Date: 2017-04-17, 13:47:00 GMT +# Date: 2017-04-17, 16:18:00 GMT # Copyright (C) 2017 Reece H. Dunn # # This is an extension to the Unicode Character Database PropList.txt file, @@ -22,7 +22,7 @@ 0589 ; Full_Stop # Po ARMENIAN FULL STOP 06D4 ; Full_Stop # Po ARABIC FULL STOP 0701 ; Full_Stop # Po SYRIAC SUPRALINEAR FULL STOP -0702 ; Full_Stop # Po SYRIAC SUBLINEAR FULL STOP +0704 ; Full_Stop # Po SYRIAC SUBLINEAR COLON 0964 ; Full_Stop # Po DEVANAGARI DANDA 0DF4 ; Full_Stop # Po SINHALA PUNCTUATION KUNDDALIYA 0F0D ; Full_Stop # Po TIBETAN MARK SHAD @@ -58,6 +58,7 @@ E002E ; Full_Stop # Cf TAG FULL STOP 037E ; Question_Mark # Po GREEK QUESTION MARK 055F ; Question_Mark # Po ARMENIAN QUESTION MARK 061F ; Question_Mark # Po ARABIC QUESTION MARK +0709 ; Question_Mark # Po SYRIAC SUBLINEAR COLON SKEWED RIGHT 1367 ; Question_Mark # Po ETHIOPIC QUESTION MARK 1945 ; Question_Mark # Po LIMBU QUESTION MARK 2047..2049 ; Question_Mark # Po [3] DOUBLE QUESTION MARK..EXCLAMATION QUESTION MARK @@ -72,13 +73,14 @@ FF1F ; Question_Mark # Po FULLWIDTH QUESTION MARK 1E95F ; Question_Mark # Po ADLAM INITIAL QUESTION MARK E003F ; Question_Mark # Cf TAG QUESTION MARK -# Total code points: 22 +# Total code points: 23 # ================================================ 0021 ; Exclamation_Mark # Po EXCLAMATION MARK 00A1 ; Exclamation_Mark # Po INVERTED EXCLAMATION MARK 055B..055C ; Exclamation_Mark # Po [2] ARMENIAN EMPHASIS MARK..ARMENIAN EXCLAMATION MARK +0703 ; Exclamation_Mark # Po SYRIAC SUPRALINEAR COLON 07F9 ; Exclamation_Mark # Po NKO EXCLAMATION MARK 1944 ; Exclamation_Mark # Po LIMBU EXCLAMATION MARK 203C ; Exclamation_Mark # Po DOUBLE EXCLAMATION MARK @@ -92,13 +94,14 @@ FF01 ; Exclamation_Mark # Po FULLWIDTH EXCLAMATION MARK 1E95E ; Exclamation_Mark # Po ADLAM INITIAL EXCLAMATION MARK E0021 ; Exclamation_Mark # Cf TAG EXCLAMATION MARK -# Total code points: 16 +# Total code points: 17 # ================================================ 002C ; Comma # Po COMMA 055D ; Comma # Po ARMENIAN COMMA 060C ; Comma # Po ARABIC COMMA +0702 ; Comma # Po SYRIAC SUBLINEAR FULL STOP 07F8 ; Comma # Po NKO COMMA 0F14 ; Comma # Po TIBETAN MARK GTER TSHEG 1363 ; Comma # Po ETHIOPIC COMMA @@ -120,4 +123,21 @@ FF64 ; Comma # Po HALFWIDTH IDEOGRAPHIC COMMA 1F101..1F10A ; Comma # No [10] DIGIT ZERO COMMA..DIGIT NINE COMMA E002C ; Comma # Cf TAG COMMA -# Total code points: 34 +# Total code points: 35 + +# ================================================ + +003A ; Colon # Po COLON +0706..0707 ; Colon # Po [2] SYRIAC COLON SKEWED LEFT..SYRIAC COLON SKEWED RIGHT +1365..1366 ; Colon # Po [2] ETHIOPIC COLON..ETHIOPIC PREFACE COLON +1804 ; Colon # Po MONGOLIAN COLON +2982 ; Colon # Sm Z NOTATION TYPE COLON +A6F4 ; Colon # Po BAMUM COLON +FE13 ; Colon # Po PRESENTATION FORM FOR VERTICAL COLON +FE55 ; Colon # Po SMALL COLON +FF1A ; Colon # Po FULLWIDTH COLON +12471..12472 ; Colon # Po [2] CUNEIFORM PUNCTUATION SIGN VERTICAL COLON..CUNEIFORM PUNCTUATION SIGN DIAGONAL COLON +1DA8A ; Colon # Po SIGNWRITING COLON +E003A ; Colon # Cf TAG COLON + +# Total code points: 15 diff --git a/src/ucd-tools/src/include/ucd/ucd.h b/src/ucd-tools/src/include/ucd/ucd.h index 8115e740..2df96ad0 100644 --- a/src/ucd-tools/src/include/ucd/ucd.h +++ b/src/ucd-tools/src/include/ucd/ucd.h @@ -363,6 +363,7 @@ static const ucd_property UCD_PROPERTY_PATTERN_SYNTAX = 0x00 static const ucd_property UCD_PROPERTY_PREPENDED_CONCATENATION_MARK = 0x0000000100000000ull; /**< @brief Prepended_Concatenation_Mark */ // eSpeak NG extended properties: +static const ucd_property ESPEAKNG_PROPERTY_COLON = 0x0800000000000000ull; /**< @brief Colon */ static const ucd_property ESPEAKNG_PROPERTY_COMMA = 0x1000000000000000ull; /**< @brief Comma */ static const ucd_property ESPEAKNG_PROPERTY_EXCLAMATION_MARK = 0x2000000000000000ull; /**< @brief Exclamation_Mark */ static const ucd_property ESPEAKNG_PROPERTY_QUESTION_MARK = 0x4000000000000000ull; /**< @brief Question_Mark */ diff --git a/src/ucd-tools/src/proplist.c b/src/ucd-tools/src/proplist.c index 2efcb0dc..5a50e9b3 100644 --- a/src/ucd-tools/src/proplist.c +++ b/src/ucd-tools/src/proplist.c @@ -62,6 +62,7 @@ static ucd_property properties_Cf(codepoint_t c) if (c == 0x0E0021) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND | ESPEAKNG_PROPERTY_EXCLAMATION_MARK; if (c == 0x0E002C) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND | ESPEAKNG_PROPERTY_COMMA; if (c == 0x0E002E) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND | ESPEAKNG_PROPERTY_FULL_STOP; + if (c == 0x0E003A) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND | ESPEAKNG_PROPERTY_COLON; if (c == 0x0E003F) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND | ESPEAKNG_PROPERTY_QUESTION_MARK; if (c >= 0x0E0020 && c <= 0x0E007F) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND; break; @@ -1333,7 +1334,8 @@ static ucd_property properties_Po(codepoint_t c) if (c == 0x0027) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX; if (c == 0x002C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_COMMA; if (c == 0x002E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_FULL_STOP; - if (c >= 0x003A && c <= 0x003B) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX; + if (c == 0x003A) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_COLON; + if (c == 0x003B) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX; if (c == 0x003F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_QUESTION_MARK; if (c == 0x00A1) return UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_EXCLAMATION_MARK; if (c == 0x00B7) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER | UCD_PROPERTY_OTHER_ID_CONTINUE; @@ -1358,8 +1360,15 @@ static ucd_property properties_Po(codepoint_t c) break; case 0x0700: if (c == 0x0700) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL; - if (c >= 0x0701 && c <= 0x0702) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; - if (c >= 0x0703 && c <= 0x070A) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x0701) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; + if (c == 0x0702) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_COMMA; + if (c == 0x0703) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_EXCLAMATION_MARK; + if (c == 0x0704) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_FULL_STOP; + if (c == 0x0705) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c >= 0x0706 && c <= 0x0707) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; + if (c == 0x0708) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x0709) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_QUESTION_MARK; + if (c == 0x070A) return UCD_PROPERTY_TERMINAL_PUNCTUATION; if (c == 0x070C) return UCD_PROPERTY_TERMINAL_PUNCTUATION; if (c == 0x07F8) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; if (c == 0x07F9) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_EXCLAMATION_MARK; @@ -1391,7 +1400,8 @@ static ucd_property properties_Po(codepoint_t c) if (c == 0x1361) return UCD_PROPERTY_TERMINAL_PUNCTUATION; if (c == 0x1362) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; if (c == 0x1363) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; - if (c >= 0x1364 && c <= 0x1366) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x1364) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c >= 0x1365 && c <= 0x1366) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; if (c == 0x1367) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_QUESTION_MARK; if (c == 0x1368) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL; break; @@ -1408,7 +1418,9 @@ static ucd_property properties_Po(codepoint_t c) case 0x1800: if (c == 0x1803) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; if (c == 0x1802) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; - if (c >= 0x1803 && c <= 0x1805) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x1803) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x1804) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; + if (c == 0x1805) return UCD_PROPERTY_TERMINAL_PUNCTUATION; if (c == 0x1808) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; if (c == 0x1809) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; if (c == 0x180A) return UCD_PROPERTY_EXTENDER; @@ -1475,7 +1487,7 @@ static ucd_property properties_Po(codepoint_t c) if (c == 0xA60E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; if (c == 0xA60F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_QUESTION_MARK; if (c == 0xA6F3) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; - if (c == 0xA6F4) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0xA6F4) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; if (c == 0xA6F5) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; if (c == 0xA6F6) return UCD_PROPERTY_TERMINAL_PUNCTUATION; if (c == 0xA6F7) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_QUESTION_MARK; @@ -1501,12 +1513,14 @@ static ucd_property properties_Po(codepoint_t c) case 0xFE00: if (c >= 0xFE10 && c <= 0xFE11) return ESPEAKNG_PROPERTY_COMMA; if (c == 0xFE12) return ESPEAKNG_PROPERTY_FULL_STOP; + if (c == 0xFE13) return ESPEAKNG_PROPERTY_COLON; if (c == 0xFE15) return ESPEAKNG_PROPERTY_EXCLAMATION_MARK; if (c == 0xFE16) return ESPEAKNG_PROPERTY_QUESTION_MARK; if (c >= 0xFE45 && c <= 0xFE46) return UCD_PROPERTY_PATTERN_SYNTAX; if (c >= 0xFE50 && c <= 0xFE51) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; if (c == 0xFE52) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; - if (c >= 0xFE54 && c <= 0xFE55) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0xFE54) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0xFE55) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; if (c == 0xFE56) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_QUESTION_MARK; if (c == 0xFE57) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_EXCLAMATION_MARK; if (c == 0xFE61) return UCD_PROPERTY_OTHER_MATH; @@ -1520,7 +1534,8 @@ static ucd_property properties_Po(codepoint_t c) if (c == 0xFF0E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; if (c == 0xFF3C) return UCD_PROPERTY_OTHER_MATH; if (c == 0xFF65) return UCD_PROPERTY_HYPHEN; - if (c >= 0xFF1A && c <= 0xFF1B) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0xFF1A) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; + if (c == 0xFF1B) return UCD_PROPERTY_TERMINAL_PUNCTUATION; if (c == 0xFF1F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_QUESTION_MARK; if (c == 0xFF61) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; if (c == 0xFF64) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; @@ -1583,6 +1598,7 @@ static ucd_property properties_Po(codepoint_t c) if (c == 0x011C71) return UCD_PROPERTY_TERMINAL_PUNCTUATION; break; case 0x12400: + if (c >= 0x012471 && c <= 0x012472) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; if (c >= 0x012470 && c <= 0x012474) return UCD_PROPERTY_TERMINAL_PUNCTUATION; break; case 0x16A00: @@ -1600,7 +1616,8 @@ static ucd_property properties_Po(codepoint_t c) case 0x1DA00: if (c == 0x01DA87) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COMMA; if (c == 0x01DA88) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | ESPEAKNG_PROPERTY_FULL_STOP; - if (c >= 0x01DA89 && c <= 0x01DA8A) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x01DA89) return UCD_PROPERTY_TERMINAL_PUNCTUATION; + if (c == 0x01DA8A) return UCD_PROPERTY_TERMINAL_PUNCTUATION | ESPEAKNG_PROPERTY_COLON; break; case 0x1E900: if (c == 0x01E95E) return ESPEAKNG_PROPERTY_EXCLAMATION_MARK; @@ -1738,7 +1755,10 @@ static ucd_property properties_Sm(codepoint_t c) case 0x2500: case 0x2600: case 0x2700: + return UCD_PROPERTY_PATTERN_SYNTAX; case 0x2900: + if (c == 0x2982) return UCD_PROPERTY_PATTERN_SYNTAX | ESPEAKNG_PROPERTY_COLON; + return UCD_PROPERTY_PATTERN_SYNTAX; case 0x2A00: case 0x2B00: return UCD_PROPERTY_PATTERN_SYNTAX; diff --git a/src/ucd-tools/tools/printdata.py b/src/ucd-tools/tools/printdata.py index 95f470d8..e4c92c14 100755 --- a/src/ucd-tools/tools/printdata.py +++ b/src/ucd-tools/tools/printdata.py @@ -161,6 +161,7 @@ def properties(data): props += (2 ** 31) * data.get('Pattern_Syntax', 0) props += (2 ** 32) * data.get('Prepended_Concatenation_Mark', 0) # eSpeak NG extended properties: + props += (2 ** 59) * data.get('Colon', 0) props += (2 ** 60) * data.get('Comma', 0) props += (2 ** 61) * data.get('Exclamation_Mark', 0) props += (2 ** 62) * data.get('Question_Mark', 0)