修改抓取重试逻辑
This commit is contained in:
parent
d9f106d9ef
commit
6e4ef6793a
@ -25,23 +25,13 @@ const parseOnePage = async ({subLink, category, sortIdx}) => {
|
||||
lastStatus: true,
|
||||
})
|
||||
} catch(err) {
|
||||
console.log('parse page with network error, try again :', url);
|
||||
try {
|
||||
html = await proxyUtil.getDataProxy(url)
|
||||
await CrawlRecord.updateRecord({url: url,
|
||||
className: 'dandanzan',
|
||||
methodName: 'parseOnePage',
|
||||
params: {subLink, category, sortIdx},
|
||||
lastStatus: true,
|
||||
})
|
||||
} catch (err2) {
|
||||
await CrawlRecord.updateRecord({url: url,
|
||||
className: 'dandanzan',
|
||||
methodName: 'parseOnePage',
|
||||
params: {subLink, category, sortIdx},
|
||||
lastStatus: false,
|
||||
})
|
||||
}
|
||||
console.log('parse page with network error:', url);
|
||||
await CrawlRecord.updateRecord({url: url,
|
||||
className: 'dandanzan',
|
||||
methodName: 'parseOnePage',
|
||||
params: {subLink, category, sortIdx},
|
||||
lastStatus: false,
|
||||
})
|
||||
}
|
||||
if (!html) {
|
||||
return false;
|
||||
@ -127,23 +117,13 @@ const parseListPage = async ({idx, category}) => {
|
||||
lastStatus: true,
|
||||
})
|
||||
} catch (err) {
|
||||
console.log('parse page with network error, try again :', url);
|
||||
try {
|
||||
html = await proxyUtil.getDataProxy(url)
|
||||
await CrawlRecord.updateRecord({url: url,
|
||||
className: 'dandanzan',
|
||||
methodName: 'parseListPage',
|
||||
params: {idx, category},
|
||||
lastStatus: true,
|
||||
})
|
||||
} catch (err2) {
|
||||
await CrawlRecord.updateRecord({url: url,
|
||||
className: 'dandanzan',
|
||||
methodName: 'parseListPage',
|
||||
params: {idx, category},
|
||||
lastStatus: false,
|
||||
})
|
||||
}
|
||||
console.log('parse page with network error:', url);
|
||||
await CrawlRecord.updateRecord({url: url,
|
||||
className: 'dandanzan',
|
||||
methodName: 'parseListPage',
|
||||
params: {idx, category},
|
||||
lastStatus: false,
|
||||
})
|
||||
}
|
||||
if (html) {
|
||||
const $ = cheerio.load(html);
|
||||
|
@ -36,19 +36,43 @@ export default {
|
||||
}
|
||||
const proxy = 'http://' + proxys[stringUtil.randomNum(0, proxys.length - 1)].link;
|
||||
return new Promise(async (resolve, reject) => {
|
||||
let response
|
||||
try {
|
||||
let response = await request.get(url)
|
||||
response = await request.get(url)
|
||||
.set('User-Agent', random_useragent.getRandom())
|
||||
.proxy(proxy)
|
||||
.retry(2)
|
||||
.timeout(15000);
|
||||
if(response.statusCode === 200 ){
|
||||
resolve(response.text);
|
||||
} else {
|
||||
reject(new Error(' server response code: ' + response.statusCode));
|
||||
}
|
||||
} catch (err) {
|
||||
reject(err);
|
||||
console.log('parse page with network error, try again :', url);
|
||||
try {
|
||||
response = await request.get(url)
|
||||
.set('User-Agent', random_useragent.getRandom())
|
||||
.proxy(proxy)
|
||||
.retry(2)
|
||||
.timeout(15000);
|
||||
} catch (err2) {
|
||||
reject(err2)
|
||||
}
|
||||
}
|
||||
if(response.statusCode === 200 ){
|
||||
resolve(response.text);
|
||||
} else {
|
||||
console.log('parse page with statusCode: ', statusCode, url);
|
||||
try {
|
||||
response = await request.get(url)
|
||||
.set('User-Agent', random_useragent.getRandom())
|
||||
.proxy(proxy)
|
||||
.retry(2)
|
||||
.timeout(15000);
|
||||
if(response.statusCode === 200 ){
|
||||
resolve(response.text);
|
||||
} else {
|
||||
reject(new Error('parse page with error statusCode: ' + response.statusCode))
|
||||
}
|
||||
} catch (err2) {
|
||||
reject(err2)
|
||||
}
|
||||
}
|
||||
})
|
||||
},
|
||||
|
Loading…
x
Reference in New Issue
Block a user