Androz2091/deepl-scraper
1
0
Fork 0
forked from KaKi87/deepl-scraper
Code Issues Pull requests Projects Releases Wiki Activity

Update glossary support #1

Merged
Androz2091 merged 9 commits from KaKi87/deepl-scraper:glossary-v2 into glossary-v2 2021-03-18 17:13:53 +01:00
Conversation 1 Commits 9 Files changed 2 +99 -41
2 changed files with 99 additions and 41 deletions
Download patch file Download diff file Expand all files Collapse all files

60
README.md
View file

@ -30,7 +30,14 @@ sudo sysctl -w kernel.unprivileged_userns_clone=1
### Usage
```js
const { translate, getSupportedLanguages, setGlossary, quit } = require('deepl-scraper');
const {
translate,
getSupportedLanguages,
setGlossary,
clearGlossary,
getSupportedGlossaryLanguages,
quit
} = require('deepl-scraper');
translate(sentence, source, target).then(console.log).catch(console.error);
```
@ -93,19 +100,56 @@ DeepL allows you to define a glossary, so you can add your own words in the tran
You can add new words in the glossary using:
```js
setGlossary({
channel: 'salon'
}, 'EN', 'FR');
await setGlossary({
'channel': 'salon'
}, 'en', 'fr');
```
Once a word is added, you can update it by calling the function again:
```js
setGlossary({
channel: 'test'
}, 'EN', 'FR');
await setGlossary({
'channel': 'test'
}, 'en', 'fr');
```
Note: at this time, the only way to reset the glossary is to quit the browser.
Glossary is cleared when you quit the browser, although it can be cleared earlier using the following method :
```js
await clearGlossary();
```
As glossary doesn't support all translation-supported languages, you can get the list using the following method :
```js
await getSupportedGlossaryLanguages();
```
```json
[
{
"sourceLanguage": "en",
"targetLanguage": "de"
},
{
"sourceLanguage": "de",
"targetLanguage": "en"
},
{
"sourceLanguage": "en",
"targetLanguage": "fr"
},
{
"sourceLanguage": "fr",
"targetLanguage": "en"
},
{
"sourceLanguage": "en",
"targetLanguage": "es"
},
{
"sourceLanguage": "es",
"targetLanguage": "en"
}
]
```
#### Quitting

80
index.js
View file

@ -11,7 +11,9 @@ const getBrowser = async () => {
const getNewPage = async () => await (await getBrowser()).newPage();
const glossaryButton = `.lmt__glossary_onOffSwitch_label`;
const
glossaryButton = `.lmt__glossary_onOffSwitch_label`,
glossaryLangSelect = `.lmt__glossary_newEntry_langButton`;
module.exports = {
translate: async (sentence, sourceLanguage = 'auto', targetLanguage) => {
@ -82,7 +84,7 @@ module.exports = {
await page.waitForSelector('.lmt--active_translation_request');
await page.waitForSelector('.lmt--active_translation_request', { hidden: true });
_res.target.translation = await page.$eval(targetSentenceField, el => el.value);
page.close().catch(() => {});
await page.close();
return _res;
},
clearGlossary: async () => {
@ -90,18 +92,14 @@ module.exports = {
await page.goto(homepage);
await page.waitForSelector(glossaryButton, { visible: true });
await page.click(glossaryButton);
await page.evaluate(() => {
Array.from(document.querySelectorAll('button[dl-test=glossary-entry-delete-button]')).forEach((element) => {
element.click();
});
});
for(const button of await page.$$('button[dl-test=glossary-entry-delete-button]'))
await (await page.$('button[dl-test=glossary-entry-delete-button]')).click();
},
setGlossary: async (glossary, sourceLanguage, targetLanguage) => {
sourceLanguage = sourceLanguage.toUpperCase();
targetLanguage = targetLanguage.toUpperCase();
const
langSelect = `.lmt__glossary_newEntry_langButton`,
sourceGlossaryInput = `.lmt__glossary_newEntry_inputSource`,
targetGlossaryInput = `.lmt__glossary_newEntry_inputTarget`,
glossaryAddButton = `.lmt__glossary_acceptButton`;
@ -111,38 +109,32 @@ module.exports = {
await page.goto(homepage);
await page.waitForSelector(glossaryButton, { visible: true });
await page.click(glossaryButton);
await page.waitForSelector(langSelect, { visible: true });
await page.click(langSelect);
try {
await page.evaluate(({
sourceLanguage,
targetLanguage
}) => {
const langButtons = Array.from(document.querySelectorAll('button[dl-test=glossary-newentry-lang-dropdown-option]'));
const button = langButtons.find((btn) => {
const childrens = Array.from(Array.from(btn.children)[1].children);
const buttonSourceLanguage = childrens[0].textContent;
const buttonTargetLanguage = childrens[2].textContent;
return buttonSourceLanguage === sourceLanguage && buttonTargetLanguage === targetLanguage;
});
button.click();
}, {
sourceLanguage,
targetLanguage
});
}
catch(_){
throw new Error('UNSUPPORTED_GLOSSARY_LANGUAGE');
await page.waitForSelector(glossaryLangSelect, { visible: true });
await page.click(glossaryLangSelect);
{
let isSet = false;
for(const button of await page.$$('button[dl-test=glossary-newentry-lang-dropdown-option]')){
const
_sourceLanguage = await (await (await button.$('.langCode')).getProperty('textContent')).jsonValue(),
_targetLanguage = await (await (await button.$('.langCode:nth-of-type(2)')).getProperty('textContent')).jsonValue();
if(sourceLanguage === _sourceLanguage && targetLanguage === _targetLanguage){
await button.click();
isSet = true;
break;
}
}
if(!isSet)
throw new Error('UNSUPPORTED_GLOSSARY_LANGUAGE');
}
await page.waitForSelector(sourceGlossaryInput, { visible: true });
const words = Object.keys(glossary);
for (let word of words) {
await page.type(sourceGlossaryInput, word);
await page.type(targetGlossaryInput, glossary[word]);
await page.click(glossaryAddButton);
}
page.close().catch(() => {});
await page.close();
},
quit: async () => {
if(browser)
@ -179,6 +171,28 @@ module.exports = {
}
}
}
return res;
},
getSupportedGlossaryLanguages: async () => {
const page = await getNewPage();
await page.goto(homepage);
await page.waitForSelector(glossaryButton, { visible: true });
await page.click(glossaryButton);
await page.waitForSelector(glossaryLangSelect, { visible: true });
await page.click(glossaryLangSelect);
const res = [];
for(const button of await page.$$('button[dl-test=glossary-newentry-lang-dropdown-option]')){
res.push({
sourceLanguage: (await (await (await button.$('.langCode')).getProperty('textContent')).jsonValue()).toLowerCase(),
targetLanguage: (await (await (await button.$('.langCode:nth-of-type(2)')).getProperty('textContent')).jsonValue()).toLowerCase()
});
}
await page.close();
return res;
}
};