mirror of
https://git.mirrors.martin98.com/https://github.com/mendableai/firecrawl
synced 2025-08-14 17:36:00 +08:00
feat(crawl-status): same for v0
This commit is contained in:
parent
95ce3c3b71
commit
a7b56ab87c
@ -9,31 +9,53 @@ import * as Sentry from "@sentry/node";
|
|||||||
import { configDotenv } from "dotenv";
|
import { configDotenv } from "dotenv";
|
||||||
import { Job } from "bullmq";
|
import { Job } from "bullmq";
|
||||||
import { toLegacyDocument } from "../v1/types";
|
import { toLegacyDocument } from "../v1/types";
|
||||||
|
import type { DBJob, PseudoJob } from "../v1/crawl-status";
|
||||||
configDotenv();
|
configDotenv();
|
||||||
|
|
||||||
export async function getJobs(crawlId: string, ids: string[]) {
|
export async function getJobs(crawlId: string, ids: string[]): Promise<PseudoJob<any>[]> {
|
||||||
const jobs = (
|
const [bullJobs, dbJobs] = await Promise.all([
|
||||||
await Promise.all(ids.map((x) => getScrapeQueue().getJob(x)))
|
Promise.all(ids.map((x) => getScrapeQueue().getJob(x))).then(x => x.filter(x => x)) as Promise<(Job<any, any, string> & { id: string })[]>,
|
||||||
).filter((x) => x) as Job[];
|
process.env.USE_DB_AUTHENTICATION === "true" ? await supabaseGetJobsByCrawlId(crawlId) : [],
|
||||||
|
]);
|
||||||
|
|
||||||
if (process.env.USE_DB_AUTHENTICATION === "true") {
|
const bullJobMap = new Map<string, PseudoJob<any>>();
|
||||||
const supabaseData = await supabaseGetJobsByCrawlId(crawlId);
|
const dbJobMap = new Map<string, DBJob>();
|
||||||
|
|
||||||
supabaseData.forEach((x) => {
|
for (const job of bullJobs) {
|
||||||
const job = jobs.find((y) => y.id === x.job_id);
|
bullJobMap.set(job.id, job);
|
||||||
if (job) {
|
}
|
||||||
job.returnvalue = x.docs;
|
|
||||||
|
for (const job of dbJobs) {
|
||||||
|
dbJobMap.set(job.job_id, job);
|
||||||
|
}
|
||||||
|
|
||||||
|
const jobs: PseudoJob<any>[] = [];
|
||||||
|
|
||||||
|
for (const id of ids) {
|
||||||
|
const bullJob = bullJobMap.get(id);
|
||||||
|
const dbJob = dbJobMap.get(id);
|
||||||
|
|
||||||
|
if (!bullJob && !dbJob) continue;
|
||||||
|
|
||||||
|
const data = dbJob?.docs ?? bullJob?.returnvalue;
|
||||||
|
|
||||||
|
const job: PseudoJob<any> = {
|
||||||
|
id,
|
||||||
|
getState: bullJob ? (() => bullJob.getState()) : (() => dbJob!.success ? "completed" : "failed"),
|
||||||
|
returnvalue: Array.isArray(data)
|
||||||
|
? data[0]
|
||||||
|
: data,
|
||||||
|
data: {
|
||||||
|
scrapeOptions: bullJob ? bullJob.data.scrapeOptions : dbJob!.page_options,
|
||||||
|
},
|
||||||
|
timestamp: bullJob ? bullJob.timestamp : new Date(dbJob!.date_added).valueOf(),
|
||||||
|
failedReason: (bullJob ? bullJob.failedReason : dbJob!.message) || undefined,
|
||||||
}
|
}
|
||||||
});
|
|
||||||
}
|
|
||||||
|
|
||||||
jobs.forEach((job) => {
|
jobs.push(job);
|
||||||
job.returnvalue = Array.isArray(job.returnvalue)
|
}
|
||||||
? job.returnvalue[0]
|
|
||||||
: job.returnvalue;
|
|
||||||
});
|
|
||||||
|
|
||||||
return jobs;
|
return jobs;
|
||||||
}
|
}
|
||||||
|
|
||||||
export async function crawlStatusController(req: Request, res: Response) {
|
export async function crawlStatusController(req: Request, res: Response) {
|
||||||
@ -93,8 +115,9 @@ export async function crawlStatusController(req: Request, res: Response) {
|
|||||||
if (
|
if (
|
||||||
jobs.length > 0 &&
|
jobs.length > 0 &&
|
||||||
jobs[0].data &&
|
jobs[0].data &&
|
||||||
jobs[0].data.pageOptions &&
|
jobs[0].data.scrapeOptions &&
|
||||||
!jobs[0].data.pageOptions.includeRawHtml
|
jobs[0].data.scrapeOptions.formats &&
|
||||||
|
!jobs[0].data.scrapeOptions.formats.includes("rawHtml")
|
||||||
) {
|
) {
|
||||||
data.forEach((item) => {
|
data.forEach((item) => {
|
||||||
if (item) {
|
if (item) {
|
||||||
|
@ -25,7 +25,7 @@ import { logger } from "../../lib/logger";
|
|||||||
import { supabase_service } from "../../services/supabase";
|
import { supabase_service } from "../../services/supabase";
|
||||||
configDotenv();
|
configDotenv();
|
||||||
|
|
||||||
type PseudoJob<T> = {
|
export type PseudoJob<T> = {
|
||||||
id: string,
|
id: string,
|
||||||
getState(): Promise<JobState | "unknown"> | JobState | "unknown",
|
getState(): Promise<JobState | "unknown"> | JobState | "unknown",
|
||||||
returnvalue: T | null,
|
returnvalue: T | null,
|
||||||
@ -33,9 +33,10 @@ type PseudoJob<T> = {
|
|||||||
data: {
|
data: {
|
||||||
scrapeOptions: any,
|
scrapeOptions: any,
|
||||||
},
|
},
|
||||||
|
failedReason?: string,
|
||||||
}
|
}
|
||||||
|
|
||||||
type DBJob = { docs: any, success: boolean, page_options: any, date_added: any }
|
export type DBJob = { docs: any, success: boolean, page_options: any, date_added: any, message: string | null }
|
||||||
|
|
||||||
export async function getJob(id: string): Promise<PseudoJob<any> | null> {
|
export async function getJob(id: string): Promise<PseudoJob<any> | null> {
|
||||||
const [bullJob, dbJob] = await Promise.all([
|
const [bullJob, dbJob] = await Promise.all([
|
||||||
@ -57,6 +58,7 @@ export async function getJob(id: string): Promise<PseudoJob<any> | null> {
|
|||||||
scrapeOptions: bullJob ? bullJob.data.scrapeOptions : dbJob!.page_options,
|
scrapeOptions: bullJob ? bullJob.data.scrapeOptions : dbJob!.page_options,
|
||||||
},
|
},
|
||||||
timestamp: bullJob ? bullJob.timestamp : new Date(dbJob!.date_added).valueOf(),
|
timestamp: bullJob ? bullJob.timestamp : new Date(dbJob!.date_added).valueOf(),
|
||||||
|
failedReason: (bullJob ? bullJob.failedReason : dbJob!.message) || undefined,
|
||||||
}
|
}
|
||||||
|
|
||||||
return job;
|
return job;
|
||||||
@ -99,6 +101,7 @@ export async function getJobs(ids: string[]): Promise<PseudoJob<any>[]> {
|
|||||||
scrapeOptions: bullJob ? bullJob.data.scrapeOptions : dbJob!.page_options,
|
scrapeOptions: bullJob ? bullJob.data.scrapeOptions : dbJob!.page_options,
|
||||||
},
|
},
|
||||||
timestamp: bullJob ? bullJob.timestamp : new Date(dbJob!.date_added).valueOf(),
|
timestamp: bullJob ? bullJob.timestamp : new Date(dbJob!.date_added).valueOf(),
|
||||||
|
failedReason: (bullJob ? bullJob.failedReason : dbJob!.message) || undefined,
|
||||||
}
|
}
|
||||||
|
|
||||||
jobs.push(job);
|
jobs.push(job);
|
||||||
|
Loading…
x
Reference in New Issue
Block a user