mirror of
https://github.com/davidjohnbarton/crawler-google-places.git
synced 2025-12-12 16:38:45 +00:00
Fixed infinite scroll
This commit is contained in:
parent
84b4040a0b
commit
3b87790d07
|
|
@ -35,10 +35,10 @@ const getPageScrollInfo = (page, elementToScroll) => page.evaluate((elementToScr
|
||||||
module.exports = async (page, maxHeight, elementToScroll = 'body') => {
|
module.exports = async (page, maxHeight, elementToScroll = 'body') => {
|
||||||
const maybeResourceTypesInfiniteScroll = ['xhr', 'fetch', 'websocket', 'other'];
|
const maybeResourceTypesInfiniteScroll = ['xhr', 'fetch', 'websocket', 'other'];
|
||||||
const stringifyScrollInfo = (scrollInfo) => {
|
const stringifyScrollInfo = (scrollInfo) => {
|
||||||
return `scrollTop=${scrollInfo.scrollTop}, ` +
|
return `scrollTop=${scrollInfo.scrollTop}, `
|
||||||
`clientHeight=${scrollInfo.clientHeight}, ` +
|
+ `clientHeight=${scrollInfo.clientHeight}, `
|
||||||
`scrollHeight=${scrollInfo.scrollHeight}, ` +
|
+ `scrollHeight=${scrollInfo.scrollHeight}, `
|
||||||
`maxHeight=${maxHeight}`;
|
+ `maxHeight=${maxHeight}`;
|
||||||
};
|
};
|
||||||
const defaultScrollDelay = 3000;
|
const defaultScrollDelay = 3000;
|
||||||
const defaultElementTimeout = 60000;
|
const defaultElementTimeout = 60000;
|
||||||
|
|
@ -72,7 +72,6 @@ module.exports = async (page, maxHeight, elementToScroll = 'body') => {
|
||||||
}
|
}
|
||||||
});
|
});
|
||||||
|
|
||||||
try {
|
|
||||||
await page.waitForSelector(elementToScroll, { timeout: defaultElementTimeout });
|
await page.waitForSelector(elementToScroll, { timeout: defaultElementTimeout });
|
||||||
let scrollInfo = await getPageScrollInfo(page, elementToScroll);
|
let scrollInfo = await getPageScrollInfo(page, elementToScroll);
|
||||||
logInfo(`Infinite scroll started (${stringifyScrollInfo(scrollInfo)}).`);
|
logInfo(`Infinite scroll started (${stringifyScrollInfo(scrollInfo)}).`);
|
||||||
|
|
@ -121,8 +120,4 @@ module.exports = async (page, maxHeight, elementToScroll = 'body') => {
|
||||||
}
|
}
|
||||||
page.removeAllListeners('request');
|
page.removeAllListeners('request');
|
||||||
logInfo(`Infinite scroll finished (${stringifyScrollInfo(scrollInfo)} resourcesStats=${JSON.stringify(resourcesStats)})`);
|
logInfo(`Infinite scroll finished (${stringifyScrollInfo(scrollInfo)} resourcesStats=${JSON.stringify(resourcesStats)})`);
|
||||||
} catch (err) {
|
|
||||||
// Infinite scroll should not break whole crawler
|
|
||||||
logError('An exception thrown in infiniteScroll()', err);
|
|
||||||
}
|
|
||||||
};
|
};
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue
Block a user