修改蛋蛋赞电影抓取规则,如果当前数据已存在,则只更新播放链接
This commit is contained in:
parent
45d6c70775
commit
593bd3e758
@ -19,7 +19,8 @@ const parseOnePage = async (subLink, category) => {
|
|||||||
console.log(`>>>>>>>>>>>> ${url} not found`);
|
console.log(`>>>>>>>>>>>> ${url} not found`);
|
||||||
} else {
|
} else {
|
||||||
let resourceStr = stringUtil.getContentByReg(html,/links='(.+?)\|'/);
|
let resourceStr = stringUtil.getContentByReg(html,/links='(.+?)\|'/);
|
||||||
let idx = parseInt(subLink.replace('/dianying/', '').replace('.html', '')
|
let idx = parseInt(subLink.replace('/dianying/', '')
|
||||||
|
.replace('.html', '')
|
||||||
.replace('/dongman/', '')
|
.replace('/dongman/', '')
|
||||||
.replace('/dianshiju/', '')
|
.replace('/dianshiju/', '')
|
||||||
.replace('/zongyi/', ''));
|
.replace('/zongyi/', ''));
|
||||||
@ -36,22 +37,27 @@ const parseOnePage = async (subLink, category) => {
|
|||||||
let daoYan = $('meta[property="og:video:director"]').attr('content');
|
let daoYan = $('meta[property="og:video:director"]').attr('content');
|
||||||
let zhuYan = $('meta[property="og:video:actor"]').attr('content').replace(/ \/ /g, ',');
|
let zhuYan = $('meta[property="og:video:actor"]').attr('content').replace(/ \/ /g, ',');
|
||||||
let region = $('meta[property="og:video:area"]').attr('content');
|
let region = $('meta[property="og:video:area"]').attr('content');
|
||||||
let record = {
|
let record = await Movie.findByGid(idx);
|
||||||
gid: idx,
|
if (!record) {
|
||||||
name: $('meta[property="og:title"]').attr('content'),
|
record = {
|
||||||
type: typeArr,
|
gid: idx,
|
||||||
resources: resourceArr,
|
name: $('meta[property="og:title"]').attr('content'),
|
||||||
daoYan: daoYan ? daoYan.split(',') : [],
|
type: typeArr,
|
||||||
zhuYan: zhuYan ? zhuYan.split(',') : [],
|
resources: resourceArr,
|
||||||
score: Number($('meta[property="og:video:score"]').attr('content')),
|
daoYan: daoYan ? daoYan.split(',') : [],
|
||||||
img: $('meta[property="og:image"]').attr('content'),
|
zhuYan: zhuYan ? zhuYan.split(',') : [],
|
||||||
introduce: $('meta[property="og:description"]').attr('content'),
|
score: Number($('meta[property="og:video:score"]').attr('content')),
|
||||||
nameAlias: $('meta[property="og:video:alias"]').attr('content'),
|
img: $('meta[property="og:image"]').attr('content'),
|
||||||
region: region ? region.split(',') : [],
|
introduce: $('meta[property="og:description"]').attr('content'),
|
||||||
year: Number($('meta[property="og:video:release_date"]').attr('content')),
|
nameAlias: $('meta[property="og:video:alias"]').attr('content'),
|
||||||
category: category,
|
region: region ? region.split(',') : [],
|
||||||
|
year: Number($('meta[property="og:video:release_date"]').attr('content')),
|
||||||
|
category: category,
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
record.resources = resourceArr;
|
||||||
}
|
}
|
||||||
await Movie.updateOne(idx, record);
|
await record.save();
|
||||||
console.log(`@@@@@ ${subLink} @ ${record.name} saved`);
|
console.log(`@@@@@ ${subLink} @ ${record.name} saved`);
|
||||||
}
|
}
|
||||||
} catch (err) {
|
} catch (err) {
|
||||||
@ -124,34 +130,12 @@ const parseAllMovie = async (category) => {
|
|||||||
|
|
||||||
export default {
|
export default {
|
||||||
run: async () => {
|
run: async () => {
|
||||||
// await parseAllMovie('movie');
|
await parseAllMovie('movie');
|
||||||
// await parseAllMovie('tv');
|
// await parseAllMovie('tv');
|
||||||
// await parseAllMovie('show');
|
// await parseAllMovie('show');
|
||||||
// await parseAllMovie('cartoon');
|
// await parseAllMovie('cartoon');
|
||||||
// console.log('all done');
|
// console.log('all done');
|
||||||
let html = await netUtil.getData('https://wechat-test.kingsome.cn/', {})
|
// let html = await netUtil.getData('https://wechat-test.kingsome.cn/', {})
|
||||||
console.log(html);
|
// console.log(html);
|
||||||
// var ProxyLists = require('proxy-lists');
|
|
||||||
//
|
|
||||||
// var options = {
|
|
||||||
// countries: ['cn'],
|
|
||||||
// protocols: ['https'],
|
|
||||||
// };
|
|
||||||
//
|
|
||||||
// var gettingProxies = ProxyLists.getProxies(options);
|
|
||||||
// gettingProxies.on('data', function(proxies) {
|
|
||||||
// // Received some proxies.
|
|
||||||
// console.log(proxies);
|
|
||||||
// });
|
|
||||||
//
|
|
||||||
// gettingProxies.on('error', function(error) {
|
|
||||||
// // Some error has occurred.
|
|
||||||
// // console.error(error);
|
|
||||||
// });
|
|
||||||
//
|
|
||||||
// gettingProxies.once('end', function() {
|
|
||||||
// // Done getting proxies.
|
|
||||||
// console.log('finish get proxy');
|
|
||||||
// });
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
@ -11,8 +11,12 @@ export default {
|
|||||||
|
|
||||||
removeHtml(content, replceEnter) {
|
removeHtml(content, replceEnter) {
|
||||||
if (replceEnter) {
|
if (replceEnter) {
|
||||||
return content.replace(/<.+?>/g, '').replace(/\r\n/g, '<br/>').replace(/\s/g, '')
|
return content.replace(/<.+?>/g, '')
|
||||||
.replace(/(<br\/>)+/g, '<br/>').replace(/^<br\/>/, '').replace(/<br\/>$/, '');
|
.replace(/\r\n/g, '<br/>')
|
||||||
|
.replace(/\s/g, '')
|
||||||
|
.replace(/(<br\/>)+/g, '<br/>')
|
||||||
|
.replace(/^<br\/>/, '')
|
||||||
|
.replace(/<br\/>$/, '');
|
||||||
} else {
|
} else {
|
||||||
return content.replace(/<.+?>/g, '').replace(/\s/g, '');
|
return content.replace(/<.+?>/g, '').replace(/\s/g, '');
|
||||||
}
|
}
|
||||||
|
Loading…
x
Reference in New Issue
Block a user