mirror of
https://github.com/webrecorder/browsertrix-crawler.git
synced 2025-10-19 06:23:16 +00:00
instead of puppeteer.exposeFunction(), use cdp function bindings directly to avoid issues custom toJSON overrides:
- add Runtime.addBinding for each exposed function, handle in one place with Runtime.bindingCalled - convert binding names to BxFunctionBindings enum - update to browsertrix-behaviors 0.7.1 to avoid waiting for return value - fixes #770
This commit is contained in:
parent
c25c6771a8
commit
cf3a4dcb44
5 changed files with 58 additions and 31 deletions
|
@ -18,7 +18,7 @@
|
|||
"dependencies": {
|
||||
"@novnc/novnc": "1.4.0",
|
||||
"@webrecorder/wabac": "^2.20.8",
|
||||
"browsertrix-behaviors": "^0.7.0",
|
||||
"browsertrix-behaviors": "^0.7.1",
|
||||
"client-zip": "^2.4.5",
|
||||
"css-selector-parser": "^3.0.5",
|
||||
"fetch-socks": "^1.3.0",
|
||||
|
|
|
@ -40,9 +40,7 @@ import { collectCustomBehaviors, getInfoString } from "./util/file_reader.js";
|
|||
import { Browser } from "./util/browser.js";
|
||||
|
||||
import {
|
||||
ADD_LINK_FUNC,
|
||||
BEHAVIOR_LOG_FUNC,
|
||||
FETCH_FUNC,
|
||||
BxFunctionBindings,
|
||||
DISPLAY,
|
||||
ExtractSelector,
|
||||
PAGE_OP_TIMEOUT_SECS,
|
||||
|
@ -768,17 +766,43 @@ export class Crawler {
|
|||
await this.screencaster.screencastPage(page, cdp, workerid);
|
||||
}
|
||||
|
||||
await page.exposeFunction(
|
||||
ADD_LINK_FUNC,
|
||||
(url: string) => callbacks.addLink && callbacks.addLink(url),
|
||||
);
|
||||
cdp.on("Runtime.bindingCalled", (params) => {
|
||||
const { name, payload } = params;
|
||||
|
||||
switch (name as BxFunctionBindings) {
|
||||
case BxFunctionBindings.AddLinkFunc:
|
||||
callbacks.addLink && callbacks.addLink(payload);
|
||||
break;
|
||||
|
||||
case BxFunctionBindings.BehaviorLogFunc:
|
||||
{
|
||||
const logdata: { data: string; type: string } = JSON.parse(payload);
|
||||
this._behaviorLog(logdata, page.url(), workerid);
|
||||
}
|
||||
break;
|
||||
|
||||
case BxFunctionBindings.FetchFunc:
|
||||
if (recorder) {
|
||||
recorder.addExternalFetch(payload, cdp);
|
||||
}
|
||||
break;
|
||||
|
||||
case BxFunctionBindings.AddToSeenSet:
|
||||
this.crawlState
|
||||
.addToUserSet(payload)
|
||||
.catch((e) => logger.warn("Adding to URL set error", e));
|
||||
break;
|
||||
}
|
||||
});
|
||||
|
||||
await cdp.send("Runtime.addBinding", {
|
||||
name: BxFunctionBindings.AddLinkFunc,
|
||||
});
|
||||
|
||||
if (this.params.behaviorOpts) {
|
||||
await page.exposeFunction(
|
||||
BEHAVIOR_LOG_FUNC,
|
||||
(logdata: { data: string; type: string }) =>
|
||||
this._behaviorLog(logdata, page.url(), workerid),
|
||||
);
|
||||
await cdp.send("Runtime.addBinding", {
|
||||
name: BxFunctionBindings.BehaviorLogFunc,
|
||||
});
|
||||
await this.browser.addInitScript(page, behaviors);
|
||||
|
||||
const initScript = `
|
||||
|
@ -791,9 +815,11 @@ self.__bx_behaviors.selectMainBehavior();
|
|||
this.behaviorsChecked = true;
|
||||
}
|
||||
|
||||
await page.exposeFunction(FETCH_FUNC, (url: string) => {
|
||||
return recorder ? recorder.addExternalFetch(url, cdp) : true;
|
||||
});
|
||||
if (recorder) {
|
||||
await cdp.send("Runtime.addBinding", {
|
||||
name: BxFunctionBindings.FetchFunc,
|
||||
});
|
||||
}
|
||||
|
||||
await this.browser.addInitScript(page, initScript);
|
||||
}
|
||||
|
@ -873,11 +899,9 @@ self.__bx_behaviors.selectMainBehavior();
|
|||
}
|
||||
}
|
||||
|
||||
await page.exposeFunction("__bx_addSet", (data: string) =>
|
||||
this.crawlState.addToUserSet(data),
|
||||
);
|
||||
|
||||
// await page.exposeFunction("__bx_hasSet", (data: string) => this.crawlState.hasUserSet(data));
|
||||
await cdp.send("Runtime.addBinding", {
|
||||
name: BxFunctionBindings.AddToSeenSet,
|
||||
});
|
||||
}
|
||||
|
||||
async setupExecContextEvents(
|
||||
|
@ -2295,7 +2319,7 @@ self.__bx_behaviors.selectMainBehavior();
|
|||
selector,
|
||||
extract,
|
||||
isAttribute,
|
||||
addLinkFunc: ADD_LINK_FUNC,
|
||||
addLinkFunc: BxFunctionBindings.AddLinkFunc,
|
||||
})
|
||||
.catch((e) =>
|
||||
logger.warn("Link Extraction failed in frame", {
|
||||
|
|
|
@ -10,7 +10,6 @@ import { hideBin } from "yargs/helpers";
|
|||
import { createParser } from "css-selector-parser";
|
||||
|
||||
import {
|
||||
BEHAVIOR_LOG_FUNC,
|
||||
WAIT_UNTIL_OPTS,
|
||||
EXTRACT_TEXT_TYPES,
|
||||
SERVICE_WORKER_OPTS,
|
||||
|
@ -18,6 +17,7 @@ import {
|
|||
BEHAVIOR_TYPES,
|
||||
ExtractSelector,
|
||||
DEFAULT_MAX_RETRIES,
|
||||
BxFunctionBindings,
|
||||
} from "./constants.js";
|
||||
import { ScopedSeed } from "./seeds.js";
|
||||
import { interpolateFilename } from "./storage.js";
|
||||
|
@ -721,7 +721,7 @@ class ArgParser {
|
|||
);
|
||||
}
|
||||
});
|
||||
behaviorOpts.log = BEHAVIOR_LOG_FUNC;
|
||||
behaviorOpts.log = BxFunctionBindings.BehaviorLogFunc;
|
||||
behaviorOpts.startEarly = true;
|
||||
behaviorOpts.clickSelector = argv.clickSelector;
|
||||
argv.behaviorOpts = JSON.stringify(behaviorOpts);
|
||||
|
|
|
@ -22,9 +22,12 @@ export const DETECT_SITEMAP = "<detect>";
|
|||
|
||||
export const EXTRACT_TEXT_TYPES = ["to-pages", "to-warc", "final-to-warc"];
|
||||
|
||||
export const BEHAVIOR_LOG_FUNC = "__bx_log";
|
||||
export const ADD_LINK_FUNC = "__bx_addLink";
|
||||
export const FETCH_FUNC = "__bx_fetch";
|
||||
export enum BxFunctionBindings {
|
||||
BehaviorLogFunc = "__bx_log",
|
||||
AddLinkFunc = "__bx_addLink",
|
||||
FetchFunc = "__bx_fetch",
|
||||
AddToSeenSet = "__bx_addSet",
|
||||
}
|
||||
|
||||
export const MAX_DEPTH = 1000000;
|
||||
export const DEFAULT_MAX_RETRIES = 2;
|
||||
|
|
|
@ -1460,10 +1460,10 @@ browserslist@^4.24.0:
|
|||
node-releases "^2.0.18"
|
||||
update-browserslist-db "^1.1.1"
|
||||
|
||||
browsertrix-behaviors@^0.7.0:
|
||||
version "0.7.0"
|
||||
resolved "https://registry.yarnpkg.com/browsertrix-behaviors/-/browsertrix-behaviors-0.7.0.tgz#a08b7d3e9cd449d0d76b14a438e28472124fd1a4"
|
||||
integrity sha512-t0X74puXJsH8sVkkVZwEdo8L5E1PYtzX/RkVXM4fwwBIL804bOB8WIV+5Dfwov/odaukhB67KZhM00hN60SiBA==
|
||||
browsertrix-behaviors@^0.7.1:
|
||||
version "0.7.1"
|
||||
resolved "https://registry.yarnpkg.com/browsertrix-behaviors/-/browsertrix-behaviors-0.7.1.tgz#dcb30c038e4060ef2393eb001ce9f10e3ce71c39"
|
||||
integrity sha512-tZ7Bv/IAWzLTNORf/yQqGHpPAQ4tP8sxql8YT491VHlCk939F1YIUrQ36XJOaSyfjmmm2WV9nCMXkDpCsw6zQg==
|
||||
dependencies:
|
||||
query-selector-shadow-dom "^1.0.1"
|
||||
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue