diff options
author | Peng Huang <shawn.p.huang@gmail.com> | 2010-05-12 15:35:56 +0800 |
---|---|---|
committer | Peng Huang <shawn.p.huang@gmail.com> | 2010-05-12 15:35:56 +0800 |
commit | 0fe0c8fdb5c2ce9090394034a505427b1aeab176 (patch) | |
tree | b5ddc1452839468c7db17ce8b0e6037d198361e1 /scripts | |
parent | dbd5b6db458766db85d25e26303c2ad8490f0d53 (diff) | |
download | ibus-libpinyin-0fe0c8fdb5c2ce9090394034a505427b1aeab176.tar.gz |
Generate PinyinParserTable with bopomofo
Diffstat (limited to 'scripts')
-rw-r--r-- | scripts/bopomofo.py | 449 | ||||
-rw-r--r-- | scripts/genpytable.py | 24 |
2 files changed, 468 insertions, 5 deletions
diff --git a/scripts/bopomofo.py b/scripts/bopomofo.py new file mode 100644 index 0000000..c4c0a65 --- /dev/null +++ b/scripts/bopomofo.py @@ -0,0 +1,449 @@ +# vim:set et sts=4: +# -*- coding: utf-8 -*- + +''' +const static gunichar bopomofo_char[] = { + L'\0',L'ㄅ',L'ㄆ',L'ㄇ',L'ㄈ',L'ㄉ',L'ㄊ',L'ㄋ',L'ㄌ',L'ㄍ',L'ㄎ', + L'ㄏ',L'ㄐ',L'ㄑ',L'ㄒ',L'ㄓ',L'ㄔ',L'ㄕ',L'ㄖ',L'ㄗ',L'ㄘ',L'ㄙ', + + L'ㄧ',L'ㄨ',L'ㄩ',L'ㄚ',L'ㄛ',L'ㄜ',L'ㄝ',L'ㄞ',L'ㄟ',L'ㄠ',L'ㄡ', + L'ㄢ',L'ㄣ',L'ㄤ',L'ㄥ',L'ㄦ', + + L'ˊ',L'ˇ',L'ˋ',L'˙', +}; +''' + +bopomofo_pinyin_map = { + "ㄅ" : "b", + "ㄅㄚ" : "ba", + "ㄅㄛ" : "bo", + "ㄅㄞ" : "bai", + "ㄅㄟ" : "bei", + "ㄅㄠ" : "bao", + "ㄅㄢ" : "ban", + "ㄅㄣ" : "ben", + "ㄅㄤ" : "bang", + "ㄅㄥ" : "beng", + "ㄅㄧ" : "bi", + "ㄅㄧㄝ" : "bie", + "ㄅㄧㄠ" : "biao", + "ㄅㄧㄢ" : "bian", + "ㄅㄧㄣ" : "bin", + "ㄅㄧㄥ" : "bing", + "ㄅㄨ" : "bu", + "ㄆ" : "p", + "ㄆㄚ" : "pa", + "ㄆㄛ" : "po", + "ㄆㄞ" : "pai", + "ㄆㄟ" : "pei", + "ㄆㄠ" : "pao", + "ㄆㄡ" : "pou", + "ㄆㄢ" : "pan", + "ㄆㄣ" : "pen", + "ㄆㄤ" : "pang", + "ㄆㄥ" : "peng", + "ㄆㄧ" : "pi", + "ㄆㄧㄝ" : "pie", + "ㄆㄧㄠ" : "piao", + "ㄆㄧㄢ" : "pian", + "ㄆㄧㄣ" : "pin", + "ㄆㄧㄥ" : "ping", + "ㄆㄨ" : "pu", + "ㄇ" : "m", + "ㄇㄚ" : "ma", + "ㄇㄛ" : "mo", + "ㄇㄜ" : "me", + "ㄇㄞ" : "mai", + "ㄇㄟ" : "mei", + "ㄇㄠ" : "mao", + "ㄇㄡ" : "mou", + "ㄇㄢ" : "man", + "ㄇㄣ" : "men", + "ㄇㄤ" : "mang", + "ㄇㄥ" : "meng", + "ㄇㄧ" : "mi", + "ㄇㄧㄝ" : "mie", + "ㄇㄧㄠ" : "miao", + "ㄇㄧㄡ" : "miu", + "ㄇㄧㄢ" : "mian", + "ㄇㄧㄣ" : "min", + "ㄇㄧㄥ" : "ming", + "ㄇㄨ" : "mu", + "ㄈ" : "f", + "ㄈㄚ" : "fa", + "ㄈㄛ" : "fo", + "ㄈㄜ" : "fe", + "ㄈㄟ" : "fei", + "ㄈㄡ" : "fou", + "ㄈㄢ" : "fan", + "ㄈㄣ" : "fen", + "ㄈㄤ" : "fang", + "ㄈㄥ" : "feng", + "ㄈㄨ" : "fu", + "ㄉ" : "d", + "ㄉㄚ" : "da", + "ㄉㄜ" : "de", + "ㄉㄞ" : "dai", + "ㄉㄟ" : "dei", + "ㄉㄠ" : "dao", + "ㄉㄡ" : "dou", + "ㄉㄢ" : "dan", + "ㄉㄣ" : "den", + "ㄉㄤ" : "dang", + "ㄉㄥ" : "deng", + "ㄉㄧ" : "di", + "ㄉㄧㄚ" : "dia", + "ㄉㄧㄝ" : "die", + "ㄉㄧㄠ" : "diao", + "ㄉㄧㄡ" : "diu", + "ㄉㄧㄢ" : "dian", + "ㄉㄧㄣ" : "din", + "ㄉㄧㄥ" : "ding", + "ㄉㄨ" : "du", + "ㄉㄨㄛ" : "duo", + "ㄉㄨㄟ" : "dui", + "ㄉㄨㄢ" : "duan", + "ㄉㄨㄣ" : "dun", + "ㄉㄨㄥ" : "dong", + "ㄊ" : "t", + "ㄊㄚ" : "ta", + "ㄊㄜ" : "te", + "ㄊㄞ" : "tai", + "ㄊㄠ" : "tao", + "ㄊㄡ" : "tou", + "ㄊㄢ" : "tan", + "ㄊㄤ" : "tang", + "ㄊㄥ" : "teng", + "ㄊㄧ" : "ti", + "ㄊㄧㄝ" : "tie", + "ㄊㄧㄠ" : "tiao", + "ㄊㄧㄢ" : "tian", + "ㄊㄧㄥ" : "ting", + "ㄊㄨ" : "tu", + "ㄊㄨㄛ" : "tuo", + "ㄊㄨㄟ" : "tui", + "ㄊㄨㄢ" : "tuan", + "ㄊㄨㄣ" : "tun", + "ㄊㄨㄥ" : "tong", + "ㄋ" : "n", + "ㄋㄚ" : "na", + "ㄋㄜ" : "ne", + "ㄋㄞ" : "nai", + "ㄋㄟ" : "nei", + "ㄋㄠ" : "nao", + "ㄋㄡ" : "nou", + "ㄋㄢ" : "nan", + "ㄋㄣ" : "nen", + "ㄋㄤ" : "nang", + "ㄋㄥ" : "neng", + "ㄋㄧ" : "ni", + "ㄋㄧㄚ" : "nia", + "ㄋㄧㄝ" : "nie", + "ㄋㄧㄠ" : "niao", + "ㄋㄧㄡ" : "niu", + "ㄋㄧㄢ" : "nian", + "ㄋㄧㄣ" : "nin", + "ㄋㄧㄤ" : "niang", + "ㄋㄧㄥ" : "ning", + "ㄋㄨ" : "nu", + "ㄋㄨㄛ" : "nuo", + "ㄋㄨㄢ" : "nuan", + "ㄋㄨㄣ" : "nun", + "ㄋㄨㄥ" : "nong", + "ㄋㄩ" : "nv", + "ㄋㄩㄝ" : "nve", + "ㄌ" : "l", + "ㄌㄚ" : "la", + "ㄌㄛ" : "lo", + "ㄌㄜ" : "le", + "ㄌㄞ" : "lai", + "ㄌㄟ" : "lei", + "ㄌㄠ" : "lao", + "ㄌㄡ" : "lou", + "ㄌㄢ" : "lan", + "ㄌㄣ" : "len", + "ㄌㄤ" : "lang", + "ㄌㄥ" : "leng", + "ㄌㄧ" : "li", + "ㄌㄧㄚ" : "lia", + "ㄌㄧㄝ" : "lie", + "ㄌㄧㄠ" : "liao", + "ㄌㄧㄡ" : "liu", + "ㄌㄧㄢ" : "lian", + "ㄌㄧㄣ" : "lin", + "ㄌㄧㄤ" : "liang", + "ㄌㄧㄥ" : "ling", + "ㄌㄨ" : "lu", + "ㄌㄨㄛ" : "luo", + "ㄌㄨㄢ" : "luan", + "ㄌㄨㄣ" : "lun", + "ㄌㄨㄥ" : "long", + "ㄌㄩ" : "lv", + "ㄌㄩㄝ" : "lve", + "ㄍ" : "g", + "ㄍㄚ" : "ga", + "ㄍㄜ" : "ge", + "ㄍㄞ" : "gai", + "ㄍㄟ" : "gei", + "ㄍㄠ" : "gao", + "ㄍㄡ" : "gou", + "ㄍㄢ" : "gan", + "ㄍㄣ" : "gen", + "ㄍㄤ" : "gang", + "ㄍㄥ" : "geng", + "ㄍㄨ" : "gu", + "ㄍㄨㄚ" : "gua", + "ㄍㄨㄛ" : "guo", + "ㄍㄨㄞ" : "guai", + "ㄍㄨㄟ" : "gui", + "ㄍㄨㄢ" : "guan", + "ㄍㄨㄣ" : "gun", + "ㄍㄨㄤ" : "guang", + "ㄍㄨㄥ" : "gong", + "ㄎ" : "k", + "ㄎㄚ" : "ka", + "ㄎㄜ" : "ke", + "ㄎㄞ" : "kai", + "ㄎㄟ" : "kei", + "ㄎㄠ" : "kao", + "ㄎㄡ" : "kou", + "ㄎㄢ" : "kan", + "ㄎㄣ" : "ken", + "ㄎㄤ" : "kang", + "ㄎㄥ" : "keng", + "ㄎㄨ" : "ku", + "ㄎㄨㄚ" : "kua", + "ㄎㄨㄛ" : "kuo", + "ㄎㄨㄞ" : "kuai", + "ㄎㄨㄟ" : "kui", + "ㄎㄨㄢ" : "kuan", + "ㄎㄨㄣ" : "kun", + "ㄎㄨㄤ" : "kuang", + "ㄎㄨㄥ" : "kong", + "ㄏ" : "h", + "ㄏㄚ" : "ha", + "ㄏㄜ" : "he", + "ㄏㄞ" : "hai", + "ㄏㄟ" : "hei", + "ㄏㄠ" : "hao", + "ㄏㄡ" : "hou", + "ㄏㄢ" : "han", + "ㄏㄣ" : "hen", + "ㄏㄤ" : "hang", + "ㄏㄥ" : "heng", + "ㄏㄨ" : "hu", + "ㄏㄨㄚ" : "hua", + "ㄏㄨㄛ" : "huo", + "ㄏㄨㄞ" : "huai", + "ㄏㄨㄟ" : "hui", + "ㄏㄨㄢ" : "huan", + "ㄏㄨㄣ" : "hun", + "ㄏㄨㄤ" : "huang", + "ㄏㄨㄥ" : "hong", + "ㄐ" : "j", + "ㄐㄧ" : "ji", + "ㄐㄧㄚ" : "jia", + "ㄐㄧㄝ" : "jie", + "ㄐㄧㄠ" : "jiao", + "ㄐㄧㄡ" : "jiu", + "ㄐㄧㄢ" : "jian", + "ㄐㄧㄣ" : "jin", + "ㄐㄧㄤ" : "jiang", + "ㄐㄧㄥ" : "jing", + "ㄐㄩ" : "ju", + "ㄐㄩㄝ" : "jue", + "ㄐㄩㄢ" : "juan", + "ㄐㄩㄣ" : "jun", + "ㄐㄩㄥ" : "jiong", + "ㄑ" : "q", + "ㄑㄧ" : "qi", + "ㄑㄧㄚ" : "qia", + "ㄑㄧㄝ" : "qie", + "ㄑㄧㄠ" : "qiao", + "ㄑㄧㄡ" : "qiu", + "ㄑㄧㄢ" : "qian", + "ㄑㄧㄣ" : "qin", + "ㄑㄧㄤ" : "qiang", + "ㄑㄧㄥ" : "qing", + "ㄑㄩ" : "qu", + "ㄑㄩㄝ" : "que", + "ㄑㄩㄢ" : "quan", + "ㄑㄩㄣ" : "qun", + "ㄑㄩㄥ" : "qiong", + "ㄒ" : "x", + "ㄒㄧ" : "xi", + "ㄒㄧㄚ" : "xia", + "ㄒㄧㄝ" : "xie", + "ㄒㄧㄠ" : "xiao", + "ㄒㄧㄡ" : "xiu", + "ㄒㄧㄢ" : "xian", + "ㄒㄧㄣ" : "xin", + "ㄒㄧㄤ" : "xiang", + "ㄒㄧㄥ" : "xing", + "ㄒㄩ" : "xu", + "ㄒㄩㄝ" : "xue", + "ㄒㄩㄢ" : "xuan", + "ㄒㄩㄣ" : "xun", + "ㄒㄩㄥ" : "xiong", + "ㄓ" : "zhi", + "ㄓㄚ" : "zha", + "ㄓㄜ" : "zhe", + "ㄓㄞ" : "zhai", + "ㄓㄟ" : "zhei", + "ㄓㄠ" : "zhao", + "ㄓㄡ" : "zhou", + "ㄓㄢ" : "zhan", + "ㄓㄣ" : "zhen", + "ㄓㄤ" : "zhang", + "ㄓㄥ" : "zheng", + "ㄓㄨ" : "zhu", + "ㄓㄨㄚ" : "zhua", + "ㄓㄨㄛ" : "zhuo", + "ㄓㄨㄞ" : "zhuai", + "ㄓㄨㄟ" : "zhui", + "ㄓㄨㄢ" : "zhuan", + "ㄓㄨㄣ" : "zhun", + "ㄓㄨㄤ" : "zhuang", + "ㄓㄨㄥ" : "zhong", + "ㄔ" : "chi", + "ㄔㄚ" : "cha", + "ㄔㄜ" : "che", + "ㄔㄞ" : "chai", + "ㄔㄠ" : "chao", + "ㄔㄡ" : "chou", + "ㄔㄢ" : "chan", + "ㄔㄣ" : "chen", + "ㄔㄤ" : "chang", + "ㄔㄥ" : "cheng", + "ㄔㄨ" : "chu", + "ㄔㄨㄚ" : "chua", + "ㄔㄨㄛ" : "chuo", + "ㄔㄨㄞ" : "chuai", + "ㄔㄨㄟ" : "chui", + "ㄔㄨㄢ" : "chuan", + "ㄔㄨㄣ" : "chun", + "ㄔㄨㄤ" : "chuang", + "ㄔㄨㄥ" : "chong", + "ㄕ" : "shi", + "ㄕㄚ" : "sha", + "ㄕㄜ" : "she", + "ㄕㄞ" : "shai", + "ㄕㄟ" : "shei", + "ㄕㄠ" : "shao", + "ㄕㄡ" : "shou", + "ㄕㄢ" : "shan", + "ㄕㄣ" : "shen", + "ㄕㄤ" : "shang", + "ㄕㄥ" : "sheng", + "ㄕㄨ" : "shu", + "ㄕㄨㄚ" : "shua", + "ㄕㄨㄛ" : "shuo", + "ㄕㄨㄞ" : "shuai", + "ㄕㄨㄟ" : "shui", + "ㄕㄨㄢ" : "shuan", + "ㄕㄨㄣ" : "shun", + "ㄕㄨㄤ" : "shuang", + "ㄖ" : "ri", + "ㄖㄜ" : "re", + "ㄖㄠ" : "rao", + "ㄖㄡ" : "rou", + "ㄖㄢ" : "ran", + "ㄖㄣ" : "ren", + "ㄖㄤ" : "rang", + "ㄖㄥ" : "reng", + "ㄖㄨ" : "ru", + "ㄖㄨㄚ" : "rua", + "ㄖㄨㄛ" : "ruo", + "ㄖㄨㄟ" : "rui", + "ㄖㄨㄢ" : "ruan", + "ㄖㄨㄣ" : "run", + "ㄖㄨㄥ" : "rong", + "ㄗ" : "zi", + "ㄗㄚ" : "za", + "ㄗㄜ" : "ze", + "ㄗㄞ" : "zai", + "ㄗㄟ" : "zei", + "ㄗㄠ" : "zao", + "ㄗㄡ" : "zou", + "ㄗㄢ" : "zan", + "ㄗㄣ" : "zen", + "ㄗㄤ" : "zang", + "ㄗㄥ" : "zeng", + "ㄗㄨ" : "zu", + "ㄗㄨㄛ" : "zuo", + "ㄗㄨㄟ" : "zui", + "ㄗㄨㄢ" : "zuan", + "ㄗㄨㄣ" : "zun", + "ㄗㄨㄥ" : "zong", + "ㄘ" : "ci", + "ㄘㄚ" : "ca", + "ㄘㄜ" : "ce", + "ㄘㄞ" : "cai", + "ㄘㄠ" : "cao", + "ㄘㄡ" : "cou", + "ㄘㄢ" : "can", + "ㄘㄣ" : "cen", + "ㄘㄤ" : "cang", + "ㄘㄥ" : "ceng", + "ㄘㄨ" : "cu", + "ㄘㄨㄛ" : "cuo", + "ㄘㄨㄟ" : "cui", + "ㄘㄨㄢ" : "cuan", + "ㄘㄨㄣ" : "cun", + "ㄘㄨㄥ" : "cong", + "ㄙ" : "si", + "ㄙㄚ" : "sa", + "ㄙㄜ" : "se", + "ㄙㄞ" : "sai", + "ㄙㄠ" : "sao", + "ㄙㄡ" : "sou", + "ㄙㄢ" : "san", + "ㄙㄣ" : "sen", + "ㄙㄤ" : "sang", + "ㄙㄥ" : "seng", + "ㄙㄨ" : "su", + "ㄙㄨㄛ" : "suo", + "ㄙㄨㄟ" : "sui", + "ㄙㄨㄢ" : "suan", + "ㄙㄨㄣ" : "sun", + "ㄙㄨㄥ" : "song", + "ㄚ" : "a", + "ㄛ" : "o", + "ㄜ" : "e", + "ㄞ" : "ai", + "ㄟ" : "ei", + "ㄠ" : "ao", + "ㄡ" : "ou", + "ㄢ" : "an", + "ㄣ" : "en", + "ㄤ" : "ang", + "ㄥ" : "eng", + "ㄦ" : "er", + "ㄧ" : "yi", + "ㄧㄚ" : "ya", + "ㄧㄛ" : "yo", + "ㄧㄝ" : "ye", + "ㄧㄞ" : "yai", + "ㄧㄠ" : "yao", + "ㄧㄡ" : "you", + "ㄧㄢ" : "yan", + "ㄧㄣ" : "yin", + "ㄧㄤ" : "yang", + "ㄧㄥ" : "ying", + "ㄨ" : "wu", + "ㄨㄚ" : "wa", + "ㄨㄛ" : "wo", + "ㄨㄞ" : "wai", + "ㄨㄟ" : "wei", + "ㄨㄢ" : "wan", + "ㄨㄣ" : "wen", + "ㄨㄤ" : "wang", + "ㄨㄥ" : "weng", + "ㄩ" : "yu", + "ㄩㄝ" : "yue", + "ㄩㄢ" : "yuan", + "ㄩㄣ" : "yun", + "ㄩㄥ" : "yong", +} + +pinyin_bopomofo_map = dict([(v, k) for k, v in bopomofo_pinyin_map.items()]) diff --git a/scripts/genpytable.py b/scripts/genpytable.py index 4e031ba..5021718 100644 --- a/scripts/genpytable.py +++ b/scripts/genpytable.py @@ -2,6 +2,7 @@ # -*- coding: utf-8 -*- from pydict import * +from bopomofo import * def str_cmp(a, b): if len(a) == len(b): @@ -219,7 +220,9 @@ def get_pinyin_with_fuzzy(): (fs2 and fs2 + fy2 not in pinyin_list): fy2 = "" - yield text, s, y, fs1, fy1, fs2, fy2, l, flags + bopomofo = pinyin_bopomofo_map.get(text, "") + + yield text, bopomofo, s, y, fs1, fy1, fs2, fy2, l, flags def gen_header(): @@ -269,13 +272,13 @@ def union_dups(a): na.sort() return na -def gen_tables(): +def gen_table(): pinyins = list(get_pinyin_with_fuzzy()) pinyins = union_dups(pinyins) print 'static const Pinyin pinyin_table[] = {' - for i, (text, s, y, fs1, fy1, fs2, fy2, l, flags) in enumerate(pinyins): + for i, (text, bopomofo, s, y, fs1, fy1, fs2, fy2, l, flags) in enumerate(pinyins): s_id = "PINYIN_ID_%s" % s.upper() if s else "PINYIN_ID_ZERO" y_id = "PINYIN_ID_%s" % y.upper() if y else "PINYIN_ID_ZERO" fs1_id = "PINYIN_ID_%s" % fs1.upper() if fs1 else "PINYIN_ID_ZERO" @@ -286,6 +289,7 @@ def gen_tables(): # args = (i, ) + tuple(['"%s"' % s for s in p[:3]]) + tuple(["PINYIN_ID_%s" % s.upper() if s else "PINYIN_ID_ZERO" for s in p[3:9]]) + p[9:-1] + (str(p[-1]), ) print ''' { /* %d */ text : "%s", + bopomofo : "%s", sheng : "%s", yun : "%s", sheng_id : %s, @@ -296,13 +300,22 @@ def gen_tables(): fyun_id_2 : %s, len : %d, flags : %s - },''' % (i, text, s, y.replace("v", "ü"), s_id, y_id, fs1_id, fy1_id, fs2_id, fy2_id, l, flags) + },''' % (i, text, bopomofo, s, y.replace("v", "ü"), s_id, y_id, fs1_id, fy1_id, fs2_id, fy2_id, l, flags) print '};' print return pinyins +def gen_bopomofo_table(pinyins): + bopomofo_table = [ (i, p) for i, p in enumerate(pinyins)] + bopomofo_table.sort(lambda a, b: cmp(a[1][1], b[1][1])) + print 'static const Pinyin *bopomofo_table[] = {' + for i, p in bopomofo_table: + if p[1]: + print ' %-20s %s' % ('&pinyin_table[%d],' % i, '// "%s" => "%s"' % (p[1], p[0])) + print '};' + def get_all_special(pinyins): for p in pinyins: if p[-1] in ["n", "g", "r"]: @@ -394,8 +407,9 @@ def gen_special_table(pinyins): def main(): # gen_header() # gen_macros() - pinyins = gen_tables() + pinyins = gen_table() # gen_full_pinyin_table (pinyins) + gen_bopomofo_table(pinyins) gen_special_table(pinyins) # gen_option_check("pinyin_option_check_sheng", fuzzy_shengmu) # gen_option_check("pinyin_option_check_yun", fuzzy_yunmu) |