From 6470cbe4679efe89d68b11c2eb660f1bfe42b1d7 Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 4 Jun 2015 11:53:36 -0400 Subject: [PATCH] [numex] Catalan and Chinese numex rules converted from RBNF, now covering top 15 languages in OSM addresses --- data/numex/rules/ca.json | 829 +++++++++++++++++++++++++++++++++++++++ data/numex/rules/zh.json | 210 ++++++++++ 2 files changed, 1039 insertions(+) create mode 100644 data/numex/rules/ca.json create mode 100644 data/numex/rules/zh.json diff --git a/data/numex/rules/ca.json b/data/numex/rules/ca.json new file mode 100644 index 00000000..16f68398 --- /dev/null +++ b/data/numex/rules/ca.json @@ -0,0 +1,829 @@ +{ + "rules": [ + { + "name": "zero", + "value": 0, + "type": "cardinal" + }, + { + "name": "un", + "value": 1, + "type": "cardinal" + }, + { + "name": "una", + "value": 1, + "type": "cardinal", + "gender": "f" + }, + { + "name": "dos", + "value": 2, + "type": "cardinal" + }, + { + "name": "dues", + "value": 2, + "type": "cardinal", + "gender": "f" + }, + { + "name": "tres", + "value": 3, + "type": "cardinal" + }, + { + "name": "quatre", + "value": 4, + "type": "cardinal" + }, + { + "name": "cinc", + "value": 5, + "type": "cardinal" + }, + { + "name": "sis", + "value": 6, + "type": "cardinal" + }, + { + "name": "set", + "value": 7, + "type": "cardinal" + }, + { + "name": "vuit", + "value": 8, + "type": "cardinal" + }, + { + "name": "nou", + "value": 9, + "type": "cardinal" + }, + { + "name": "deu", + "value": 10, + "type": "cardinal" + }, + { + "name": "onze", + "value": 11, + "type": "cardinal" + }, + { + "name": "dotze", + "value": 12, + "type": "cardinal" + }, + { + "name": "tretze", + "value": 13, + "type": "cardinal" + }, + { + "name": "catorze", + "value": 14, + "type": "cardinal" + }, + { + "name": "quinze", + "value": 15, + "type": "cardinal" + }, + { + "name": "setze", + "value": 16, + "type": "cardinal" + }, + { + "name": "disset", + "value": 17, + "type": "cardinal" + }, + { + "name": "divuit", + "value": 18, + "type": "cardinal" + }, + { + "name": "dinou", + "value": 19, + "type": "cardinal" + }, + { + "name": "vint", + "value": 20, + "type": "cardinal", + "right": "add" + }, + { + "name": "trenta", + "value": 30, + "type": "cardinal", + "right": "add" + }, + { + "name": "quaranta", + "value": 40, + "type": "cardinal", + "right": "add" + }, + { + "name": "cinquanta", + "value": 50, + "type": "cardinal", + "right": "add" + }, + { + "name": "seixanta", + "value": 60, + "type": "cardinal", + "right": "add" + }, + { + "name": "setanta", + "value": 70, + "type": "cardinal", + "right": "add" + }, + { + "name": "vuitanta", + "value": 80, + "type": "cardinal", + "right": "add" + }, + { + "name": "noranta", + "value": 90, + "type": "cardinal", + "right": "add" + }, + { + "name": "cent", + "value": 100, + "type": "cardinal", + "right": "add" + }, + { + "name": "cents", + "value": 100, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "mil", + "value": 1000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "milió", + "value": 1000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "milio", + "value": 1000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "milions", + "value": 1000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "miliard", + "value": 1000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "miliards", + "value": 1000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "bilió", + "value": 1000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "bilions", + "value": 1000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "biliard", + "value": 1000000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "biliards", + "value": 1000000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "zeroè", + "value": 0, + "type": "ordinal" + }, + { + "name": "zeroe", + "value": 0, + "type": "ordinal" + }, + { + "name": "primer", + "value": 1, + "type": "ordinal" + }, + { + "name": "segon", + "value": 2, + "type": "ordinal" + }, + { + "name": "tercer", + "value": 3, + "type": "ordinal" + }, + { + "name": "quart", + "value": 4, + "type": "ordinal" + }, + { + "name": "cinquè", + "value": 5, + "type": "ordinal" + }, + { + "name": "cinque", + "value": 5, + "type": "ordinal" + }, + { + "name": "sisè", + "value": 6, + "type": "ordinal" + }, + { + "name": "sise", + "value": 6, + "type": "ordinal" + }, + { + "name": "setè", + "value": 7, + "type": "ordinal" + }, + { + "name": "sete", + "value": 7, + "type": "ordinal" + }, + { + "name": "vuitè", + "value": 8, + "type": "ordinal" + }, + { + "name": "vuite", + "value": 8, + "type": "ordinal" + }, + { + "name": "novè", + "value": 9, + "type": "ordinal" + }, + { + "name": "nove", + "value": 9, + "type": "ordinal" + }, + { + "name": "desè", + "value": 10, + "type": "ordinal" + }, + { + "name": "dese", + "value": 10, + "type": "ordinal" + }, + { + "name": "onzè", + "value": 11, + "type": "ordinal" + }, + { + "name": "onze", + "value": 11, + "type": "ordinal" + }, + { + "name": "dotzè", + "value": 12, + "type": "ordinal" + }, + { + "name": "dotze", + "value": 12, + "type": "ordinal" + }, + { + "name": "tretzè", + "value": 13, + "type": "ordinal" + }, + { + "name": "tretze", + "value": 13, + "type": "ordinal" + }, + { + "name": "catorzè", + "value": 14, + "type": "ordinal" + }, + { + "name": "catorze", + "value": 14, + "type": "ordinal" + }, + { + "name": "quinzè", + "value": 15, + "type": "ordinal" + }, + { + "name": "quinze", + "value": 15, + "type": "ordinal" + }, + { + "name": "setzè", + "value": 16, + "type": "ordinal" + }, + { + "name": "setze", + "value": 16, + "type": "ordinal" + }, + { + "name": "dissetè", + "value": 17, + "type": "ordinal" + }, + { + "name": "dissete", + "value": 17, + "type": "ordinal" + }, + { + "name": "divuitè", + "value": 18, + "type": "ordinal" + }, + { + "name": "divuite", + "value": 18, + "type": "ordinal" + }, + { + "name": "dinovè", + "value": 19, + "type": "ordinal" + }, + { + "name": "dinove", + "value": 19, + "type": "ordinal" + }, + { + "name": "vintè", + "value": 20, + "type": "ordinal" + }, + { + "name": "vinte", + "value": 20, + "type": "ordinal" + }, + { + "name": "trentè", + "value": 30, + "type": "ordinal" + }, + { + "name": "trente", + "value": 30, + "type": "ordinal" + }, + { + "name": "quarantè", + "value": 40, + "type": "ordinal" + }, + { + "name": "quarante", + "value": 40, + "type": "ordinal" + }, + { + "name": "cinquantè", + "value": 50, + "type": "ordinal" + }, + { + "name": "cinquante", + "value": 50, + "type": "ordinal" + }, + { + "name": "seixantè", + "value": 60, + "type": "ordinal" + }, + { + "name": "seixante", + "value": 60, + "type": "ordinal" + }, + { + "name": "setantè", + "value": 70, + "type": "ordinal" + }, + { + "name": "setante", + "value": 70, + "type": "ordinal" + }, + { + "name": "vuitantè", + "value": 80, + "type": "ordinal" + }, + { + "name": "vuitante", + "value": 80, + "type": "ordinal" + }, + { + "name": "norantè", + "value": 90, + "type": "ordinal" + }, + { + "name": "norante", + "value": 90, + "type": "ordinal" + }, + { + "name": "centè", + "value": 100, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "cente", + "value": 100, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "milè", + "value": 1000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "mile", + "value": 1000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "milionè", + "value": 1000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "milione", + "value": 1000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "miliardè", + "value": 1000000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "miliarde", + "value": 1000000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "bilionè", + "value": 1000000000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "bilione", + "value": 1000000000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "biliardè", + "value": 1000000000000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "biliarde", + "value": 1000000000000000, + "type": "ordinal", + "left": "multiply" + }, + { + "name": "zerona", + "value": 0, + "type": "ordinal", + "gender": "f" + }, + { + "name": "primera", + "value": 1, + "type": "ordinal", + "gender": "f" + }, + { + "name": "segona", + "value": 2, + "type": "ordinal", + "gender": "f" + }, + { + "name": "tercera", + "value": 3, + "type": "ordinal", + "gender": "f" + }, + { + "name": "quarta", + "value": 4, + "type": "ordinal", + "gender": "f" + }, + { + "name": "cinquena", + "value": 5, + "type": "ordinal", + "gender": "f" + }, + { + "name": "sisena", + "value": 6, + "type": "ordinal", + "gender": "f" + }, + { + "name": "setena", + "value": 7, + "type": "ordinal", + "gender": "f" + }, + { + "name": "vuitena", + "value": 8, + "type": "ordinal", + "gender": "f" + }, + { + "name": "novena", + "value": 9, + "type": "ordinal", + "gender": "f" + }, + { + "name": "desena", + "value": 10, + "type": "ordinal", + "gender": "f" + }, + { + "name": "onzena", + "value": 11, + "type": "ordinal", + "gender": "f" + }, + { + "name": "dotzena", + "value": 12, + "type": "ordinal", + "gender": "f" + }, + { + "name": "tretzena", + "value": 13, + "type": "ordinal", + "gender": "f" + }, + { + "name": "catorzena", + "value": 14, + "type": "ordinal", + "gender": "f" + }, + { + "name": "quinzena", + "value": 15, + "type": "ordinal", + "gender": "f" + }, + { + "name": "setzena", + "value": 16, + "type": "ordinal", + "gender": "f" + }, + { + "name": "dissetena", + "value": 17, + "type": "ordinal", + "gender": "f" + }, + { + "name": "divuitena", + "value": 18, + "type": "ordinal", + "gender": "f" + }, + { + "name": "dinovena", + "value": 19, + "type": "ordinal", + "gender": "f" + }, + { + "name": "vintena", + "value": 20, + "type": "ordinal", + "gender": "f" + }, + { + "name": "trentena", + "value": 30, + "type": "ordinal", + "gender": "f" + }, + { + "name": "quarantena", + "value": 40, + "type": "ordinal", + "gender": "f" + }, + { + "name": "cinquantena", + "value": 50, + "type": "ordinal", + "gender": "f" + }, + { + "name": "seixantena", + "value": 60, + "type": "ordinal", + "gender": "f" + }, + { + "name": "setantena", + "value": 70, + "type": "ordinal", + "gender": "f" + }, + { + "name": "vuitantena", + "value": 80, + "type": "ordinal", + "gender": "f" + }, + { + "name": "norantena", + "value": 90, + "type": "ordinal", + "gender": "f" + }, + { + "name": "centena", + "value": 100, + "type": "ordinal", + "gender": "f", + "left": "multiply" + }, + { + "name": "milena", + "value": 1000, + "type": "ordinal", + "gender": "f", + "left": "multiply" + }, + { + "name": "milionena", + "value": 1000000, + "type": "ordinal", + "gender": "f", + "left": "multiply" + }, + { + "name": "miliardena", + "value": 1000000000, + "type": "ordinal", + "gender": "f", + "left": "multiply" + }, + { + "name": "bilionena", + "value": 1000000000000, + "type": "ordinal", + "gender": "f", + "left": "multiply" + }, + { + "name": "biliardena", + "value": 1000000000000000, + "type": "ordinal", + "gender": "f", + "left": "multiply" + } + ], + "ordinal_indicators": [ + { + "suffixes": { + "0": ["è", "e"], + "1": ["r", "er"], + "11": ["è", "e"], + "2": ["n"], + "12": ["è", "e"], + "3": ["r", "er"], + "13": ["è", "e"], + "4": ["t"], + "14": ["è", "e"], + "5": ["è", "e"], + "6": ["è", "e"], + "7": ["è", "e"], + "8": ["è", "e"], + "9": ["è", "e"] + } + }, + { + "gender": "f", + "suffixes": { + "0": ["a"], + "1": ["a", "ra", "era"], + "2": ["a"], + "3": ["a"], + "4": ["a"], + "5": ["a"], + "6": ["a"], + "7": ["a"], + "8": ["a"], + "9": ["a"] + } + + } + + ], + "stopwords": [ + "i" + ] +} \ No newline at end of file diff --git a/data/numex/rules/zh.json b/data/numex/rules/zh.json new file mode 100644 index 00000000..21006369 --- /dev/null +++ b/data/numex/rules/zh.json @@ -0,0 +1,210 @@ +{ + "rules": [ + { + "name": "〇", + "value": 0, + "type": "cardinal" + }, + { + "name": "零", + "value": 0, + "type": "cardinal" + }, + { + "name": "一", + "value": 1, + "type": "cardinal" + }, + { + "name": "二", + "value": 2, + "type": "cardinal" + }, + { + "name": "两", + "value": 2, + "type": "cardinal" + }, + { + "name": "三", + "value": 3, + "type": "cardinal" + }, + { + "name": "四", + "value": 4, + "type": "cardinal" + }, + { + "name": "五", + "value": 5, + "type": "cardinal" + }, + { + "name": "六", + "value": 6, + "type": "cardinal" + }, + { + "name": "七", + "value": 7, + "type": "cardinal" + }, + { + "name": "八", + "value": 8, + "type": "cardinal" + }, + { + "name": "九", + "value": 9, + "type": "cardinal" + }, + { + "name": "十", + "value": 10, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "百", + "value": 100, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "百〇", + "value": 100, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "百零", + "value": 100, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "千", + "value": 1000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "千〇", + "value": 1000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "千零", + "value": 1000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "万", + "value": 10000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "万〇", + "value": 10000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "万零", + "value": 10000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "亿", + "value": 100000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "亿〇", + "value": 100000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "亿零", + "value": 100000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "兆", + "value": 1000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "兆〇", + "value": 1000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "兆零", + "value": 1000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "亿零", + "value": 100000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "京", + "value": 10000000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "京〇", + "value": 10000000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "京零", + "value": 10000000000000000, + "type": "cardinal", + "left": "multiply", + "right": "add" + }, + { + "name": "第", + "value": 0, + "type": "ordinal_indicator", + "right": "add" + } + ] +} \ No newline at end of file