AMAZEHOME/TypeWords
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: update dict

Browse Source
This commit is contained in:
王念超
2024-06-12 18:35:34 +08:00
parent 0dd415c2f3
commit 0a6ccda581
72 changed files with 109 additions and 486 deletions
Download Patch File Download Diff File Expand all files Collapse all files

0
js_node/f.cjs → js_node/axiosGetYoudao.cjs
View File

65
js_node/checkRepeat.cjs
View File

@@ -1,65 +0,0 @@
let path = require("path");
let fs = require("fs");
const str = fs.readFileSync("../public/dicts/en/common/translate/en2zh_CN.json", "utf8");
let translateDict = JSON.parse(str);
let ts = []
console.log(translateDict.length)
let pathName = "../public/dicts/en/common";
// let pathName = "./d";
//判断是不是目录
const dirs = fs.readdirSync(pathName)
dirs.forEach(dictName => {
if (!dictName.includes('en2zh_CN.json') || !dictName.includes('ts.json')) {
let dictPath = path.join(pathName, dictName)
// console.log('d', dictPath)
formatDict(dictPath)
}
})
fs.writeFileSync(
"../public/dicts/en/common/translate/ts.json",
JSON.stringify(ts, null, 2)
);
console.log(ts.length)
function formatDict(path) {
try {
const str = fs.readFileSync(path, "utf8");
let dicts = JSON.parse(str);
dicts = dicts.map(v => {
if (!translateDict.find(w => w.name === v.name)) {
if (!v.trans) {
console.log(path, v)
}
let data = {
"name": "",
"trans": [],
usphone: '',
ukphone: '',
...v
}
translateDict.push(data)
ts.push(data)
}
// delete v.trans
// delete v.usphone
// delete v.ukphone
return v.name
})
// let newDicts = Array.from(new Set(dicts))
// console.log(dicts.length);
// console.log(newDicts.length);
// fs.writeFileSync(
// path,
// JSON.stringify(newDicts, null, 2)
// );
} catch (e) {
// console.log('err', e)
}
}

103
js_node/format.cjs
View File

@@ -1,103 +0,0 @@
let path = require("path");
let fs = require("fs");
let fileName = 'CET4_1.json'
let read = './dict/'
let save = "./format/";
//判断是不是目录
const dirs = fs.readdirSync(read)
dirs.forEach(dictName => {
formatDict(read, dictName)
})
// formatDict(read, fileName)
function formatDict(path, name) {
try {
let newObj = []
let str = fs.readFileSync(path + name, "utf8");
let replace = [
['é', 'e'], ['ê', 'e'], ['è', 'e'], ['ë', 'e'], ['à', 'a'], ['â', 'a'], ['ç', 'c'], ['î', 'i'], ['ï', 'i'], ['ô', 'o'], ['ù', 'u'], ['û', 'u'], ['ü', 'u'], ['ÿ', 'y']
]
replace.map(v => {
str = str.replaceAll(v[0], v[1])
})
let list = str.split('\n')
list.map(v => {
if (!v) return
let item = JSON.parse(v)
// console.log('v', item.headWord)
let data = {
id: item.wordRank,
word: item.headWord,
trans: [],
phonetic0: item.content.word.content.usphone,
phonetic1: item.content.word.content.ukphone,
sentences: [],
relWords: [],
phrases: [],
synos: [],
memory: ''
}
if (item.content.word.content.trans) {
data.trans = item.content.word.content.trans.map(s => {
return {
pos: s.pos,
cn: s.tranCn,
en: s.tranOther
}
})
}
if (item.content.word.content.sentence) {
data.sentences = item.content.word.content.sentence.sentences.map(s => {
return {
v: s.sCn, tran: s.sContent,
}
})
}
if (item.content.word.content.remMethod) {
data.memory = item.content.word.content.remMethod.val
}
if (item.content.word.content.relWord) {
data.relWords = item.content.word.content.relWord.rels.map(s => {
return {
"pos": s.pos,
"ws": s.words.map(a => {
return {
w: a.hwd,
tran: a.tran
}
})
}
})
}
if (item.content.word.content.phrase) {
data.phrases = item.content.word.content.phrase.phrases.map(s => {
return {
v: s.pCn,
tran: s.pContent,
}
})
}
if (item.content.word.content.syno) {
data.synos = item.content.word.content.syno.synos.map(s => {
return {
pos: s.pos,
tran: s.tran,
ws: s.hwds
}
})
}
newObj.push(data)
})
fs.writeFileSync(save + name, JSON.stringify(newObj, null, 2));
// fs.writeFileSync(save + name.replace('.json', '.min.json'), JSON.stringify(newObj));
console.log(name, newObj.length)
} catch (e) {
console.log('err', name, e)
}
}

58
js_node/format5.cjs
View File

@@ -1,58 +0,0 @@
let path = require("path");
let fs = require("fs");
let fileName = '../public/translate/en2zh_CN.json'
let fileNameWords = '../public/translate/en2zh_CN.words.json'
let read = '../public/dicts/en/word/common/'
let save = "./format2/";
let not = []
let str = fs.readFileSync(fileNameWords, "utf8");
let words = JSON.parse(str)
str = fs.readFileSync(fileName, "utf8");
let allWords = JSON.parse(str)
//判断是不是目录
// const dirs = fs.readdirSync(read)
// dirs.forEach(dictName => {
// formatDict(read, dictName)
// })
// formatDict(read, '2024HongBao_T1.json')
fs.writeFileSync(save + 'not.json', JSON.stringify(not.filter(v => v.list.length), null, 2));
function formatDict(path, name) {
try {
let newObj = []
let str = fs.readFileSync(path + name, "utf8");
let list = JSON.parse(str)
let notListCurrent = {
name,
list: []
}
list.map(a => {
let rIndex = words.findIndex(v => v === a.word)
if (rIndex === -1) {
notListCurrent.list.push(a.word)
newObj.push(a)
} else {
let data = {
...a,
...allWords[rIndex]
}
newObj.push(data)
}
})
not.push(notListCurrent)
fs.writeFileSync(save + name, JSON.stringify(newObj, null, 2));
// fs.writeFileSync(save + name.replace('.json', '.min.json'), JSON.stringify(newObj));
console.log(`当前字典${name},长度：${list.length},没有翻译的：${notListCurrent.list.length}`)
} catch (e) {
console.log('err', name, e)
}
}

122
js_node/formatBigDict.cjs
View File

@@ -1,122 +0,0 @@
const { log } = require("console");
const fs = require("fs");
try {
const str = fs.readFileSync("./public/dicts/coca20000.json", "utf8");
let dicts = JSON.parse(str);
console.log(dicts.length);
// let a = [
// "vt.",
// "vi.",
// "pron.",
// "adj.",
// "adv.",
// "num.",
// "interj.",
// "art.",
// "aux.",
// "conj.",
// "prep.",
// // "n.",
// // "v.",
// ];
// dicts.map((v) => {
// v.trans = v.trans.map((s) => {
// s = s.replaceAll(",", "，");
// a.map((b) => {
// s = s.replaceAll(b, `^${b}`);
// });
// if (s[0] === "^") {
// s = s.substr(1);
// }
// s = s.split("^").filter((v) => v);
// let last = null;
// s = s.reduce((p, c, i, a) => {
// if (last) {
// c = last + c;
// p.push(c);
// last = null;
// } else {
// if (c.includes("& ")) {
// last = c;
// } else {
// p.push(c);
// last = null;
// }
// }
// return p;
// }, []);
// s = s.map((d) => {
// let r1 = d.indexOf("adv.");
// if (r1 > -1) {
// let t2 = d.substr(r1 + 4).replaceAll("v.", `^v.`);
// t2 = t2.split("^").filter((v) => v);
// t2[0] = d.substr(0, r1 + 4) + t2[0];
// d = t2;
// } else {
// let t2 = d.replaceAll("v.", `^v.`);
// t2 = t2.split("^").filter((v) => v);
// d = t2;
// }
// return d;
// });
// s = s.flat();
// s = s.map((d) => {
// let r1 = d.indexOf("pron.");
// if (r1 > -1) {
// let t2 = d.substr(r1 + 5).replaceAll("n.", `^n.`);
// t2 = t2.split("^").filter((v) => v);
// t2[0] = d.substr(0, r1 + 5) + t2[0];
// d = t2;
// } else {
// let t2 = d.replaceAll("n.", `^n.`);
// t2 = t2.split("^").filter((v) => v);
// d = t2;
// }
// return d;
// });
// s = s.flat();
// return s;
// });
// v.trans = v.trans.flat();
// return v;
// });
// console.log(dicts);
// let newDict = [];
// dicts.map((v) => {
// if (!newDict.find((s) => s.name === v.name)) {
// newDict.push(v);
// }
// });
// console.log(newDict.length);
dicts.map((v) => {
v.trans = v.trans.map((d) => {
let t2 = d.replaceAll("int.", `^int.`);
t2 = t2.split("^").filter((v) => v);
d = t2;
return d;
});
v.trans = v.trans.flat();
return v;
});
fs.writeFileSync(
"./public/dicts/coca20000.json",
JSON.stringify(dicts, null, 2)
);
} catch (err) {
console.error(err);
}

42
js_node/formatDictSimple.cjs
View File

@@ -1,42 +0,0 @@
let path = require("path");
let fs = require("fs");
const str = fs.readFileSync("../public/dicts/en/common/translate/en2zh_CN.json", "utf8");
let translateDict = JSON.parse(str);
let pathName = "../public/dicts/en/common";
// let pathName = "./d";
const dirs = fs.readdirSync(pathName)
dirs.forEach(dictName => {
if (!dictName.includes('en2zh_CN.json') || !dictName.includes('ts.json')) {
let dictPath = path.join(pathName, dictName)
// console.log('d', dictPath)
formatDict(dictPath)
}
})
function formatDict(path) {
try {
const str = fs.readFileSync(path, "utf8");
let dicts = JSON.parse(str);
dicts = dicts.map(v => {
let data = {
"name": v.name,
"trans": [],
usphone: '',
ukphone: '',
}
return data
})
let newDicts = Array.from(new Set(dicts))
fs.writeFileSync(
path,
JSON.stringify(newDicts, null, 2)
);
} catch (e) {
// console.log('err', e)
}
}

1
js_node/res/allNew2.min.json
View File

File diff suppressed because one or more lines are too long

1
js_node/res/minNew2.min.json
View File

File diff suppressed because one or more lines are too long

1
js_node/res/notContain.min.json
View File

File diff suppressed because one or more lines are too long

1
js_node/save/minNew.min.json
View File

File diff suppressed because one or more lines are too long

58
js_node/template.js
View File

@@ -1,58 +0,0 @@
let path = require("path");
let fs = require("fs");
let fileName = '../public/translate/en2zh_CN.json'
let fileNameWords = '../public/translate/en2zh_CN.words.json'
let read = '../format/'
let save = "./format/";
let not = []
let str = fs.readFileSync(fileNameWords, "utf8");
let words = JSON.parse(str)
str = fs.readFileSync(fileName, "utf8");
let allWords = JSON.parse(str)
//判断是不是目录
// const dirs = fs.readdirSync(read)
// dirs.forEach(dictName => {
// formatDict(read, dictName)
// })
// formatDict(read, '2024HongBao_T1.json')
fs.writeFileSync(save + 'not.json', JSON.stringify(not.filter(v => v.list.length), null, 2));
function formatDict(path, name) {
try {
let newObj = []
let str = fs.readFileSync(path + name, "utf8");
let list = JSON.parse(str)
let notListCurrent = {
name,
list: []
}
list.map(a => {
let rIndex = words.findIndex(v => v === a.word)
if (rIndex === -1) {
notListCurrent.list.push(a.word)
newObj.push(a)
} else {
let data = {
...a,
...allWords[rIndex]
}
newObj.push(data)
}
})
not.push(notListCurrent)
fs.writeFileSync(save + name, JSON.stringify(newObj, null, 2));
// fs.writeFileSync(save + name.replace('.json', '.min.json'), JSON.stringify(newObj));
console.log(`当前字典${name},长度：${list.length},没有翻译的：${notListCurrent.list.length}`)
} catch (e) {
console.log('err', name, e)
}
}

32
js_node/template2.js
View File

@@ -1,32 +0,0 @@
let path = require("path");
let fs = require("fs");
let fileName = '../public/translate/en2zh_CN.json'
let fileNameWords = '../public/translate/en2zh_CN.words.json'
let read = './format/'
let save = "./format2/";
let standardDictNames = []
//判断是不是目录
const dirs = fs.readdirSync(read)
dirs.forEach(dictName => {
formatDict(read, dictName)
})
// formatDict(read, 'BEC_2.min.json')
fs.writeFileSync(save + 'standardDictNames.json', JSON.stringify(standardDictNames, null, 2));
function formatDict(path, name) {
try {
let str = fs.readFileSync(path + name, "utf8");
let list = JSON.parse(str)
let standardName = name.replace('.min', '')
standardDictNames.push(standardName)
fs.writeFileSync(save + standardName, JSON.stringify(list, null, 2));
fs.writeFileSync(save + standardName.replace('.json', '.min.json'), JSON.stringify(list));
} catch (e) {
console.log('err', name, e)
}
}

38
js_node/test.cjs
View File

@@ -2,6 +2,40 @@ let path = require("path");
let fs = require("fs");
const axios = require('axios')
let str = fs.readFileSync('./save/allNew.min.json', "utf8");
let str = fs.readFileSync('./failWord.txt', "utf8");
let failList = []
let list = JSON.parse(str)
console.log('str',JSON.parse(str)[0])
async function sleep(val) {
return new Promise(resolve => {
setTimeout(resolve, val)
})
}
function fail(word) {
failList.push(word)
console.log('失败：', failList.length)
fs.writeFileSync('./fail.txt', JSON.stringify(failList, null, 2));
}
async function test() {
for (let i = 0; i < list.length; i++) {
let v = list[i]
console.log('进度', i)
await sleep(100)
axios({
url: 'http://localhost/index.php/v1/support/test',
method: 'post',
data: v
}).then(r => {
if (!r.data.success) {
fail(v.word)
}
}).catch(r => {
fail(v.word)
})
}
}
test()

38
js_node/test2.cjs Normal file
View File

@@ -0,0 +1,38 @@
let path = require("path");
let fs = require("fs");
const axios = require('axios')
let str = fs.readFileSync('./save/allNew.min.json', "utf8");
let failStr = fs.readFileSync('./fail.txt', "utf8");
let failList = JSON.parse(failStr)
let list = JSON.parse(str)
async function sleep(val) {
return new Promise(resolve => {
setTimeout(resolve, val)
})
}
let s = []
function fail(word) {
s.push(word)
console.log('失败：', s.length)
fs.writeFileSync('./failWord.txt', JSON.stringify(s, null, 2));
}
// console.log('failList',failList)
async function test() {
for (let i = 0; i < list.length; i++) {
// for (let i = 0; i < 3000; i++) {
let v = list[i]
if (failList.includes(v.word)) {
console.log('进度', v.word)
fail(v)
}
}
}
test()

35
js_node/从原始词典提取单词.cjs Normal file
View File

@@ -0,0 +1,35 @@
let path = require("path");
let fs = require("fs");
let read = './dict/'
let save = "./res/";
//判断是不是目录
const dirs = fs.readdirSync(read)
dirs.forEach(dictName => {
formatDict(read, dictName)
})
// formatDict(read, 'BEC_2.json')
function formatDict(path, name) {
try {
let newObj = []
let str = fs.readFileSync(path + name, "utf8");
let list = str.split('\n')
list.map(v => {
if (!v) return
let item = JSON.parse(v)
// console.log('v', item.headWord)
newObj.push(item.headWord)
})
// fs.writeFileSync(save + name, JSON.stringify(newObj, null, 2));
fs.writeFileSync(save + name.replace('.json', '_word.json'), JSON.stringify(newObj));
console.log(name, newObj.length)
} catch (e) {
console.log('err', name, e)
}
}