mcp-scraper 0.66.0 → 0.66.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +32 -1
- package/README.md +2 -2
- package/dist/{analytics-repository-IKE53S3E.js → analytics-repository-RUXBFSFX.js} +1 -2
- package/dist/bin/api-server.cjs +2898 -2075
- package/dist/bin/api-server.js +4 -5
- package/dist/bin/mcp-scraper-cli.cjs +1 -2
- package/dist/bin/mcp-scraper-cli.js +6 -7
- package/dist/bin/mcp-scraper-install.cjs +7 -3
- package/dist/bin/mcp-scraper-install.js +2 -3
- package/dist/bin/mcp-stdio-server.cjs +1669 -97
- package/dist/bin/mcp-stdio-server.js +11 -12
- package/dist/bin/paa-harvest.cjs +0 -1
- package/dist/bin/paa-harvest.js +4 -5
- package/dist/{chunk-VIWT35S6.js → chunk-3EYALCYK.js} +1 -2
- package/dist/{chunk-4LINSJNI.js → chunk-3TKKJQB4.js} +1 -2
- package/dist/{chunk-4VG3NUUY.js → chunk-55B3T5L2.js} +23 -28
- package/dist/{chunk-FG6LJGJ7.js → chunk-77LQV73B.js} +0 -1
- package/dist/{chunk-N54JS5WP.js → chunk-ABYBPODM.js} +1 -2
- package/dist/chunk-ALEBMPOW.js +6 -0
- package/dist/{chunk-2XUXVKT4.js → chunk-AYJHI2FK.js} +0 -1
- package/dist/{chunk-AAGVOPP5.js → chunk-CETDPNMZ.js} +24 -25
- package/dist/{chunk-5BEPOFKG.js → chunk-CNLLTD5B.js} +0 -1
- package/dist/{chunk-P7YFUNKR.js → chunk-DBWWIK7O.js} +3 -4
- package/dist/{chunk-EKJ65E3Z.js → chunk-DNM65UCK.js} +0 -1
- package/dist/{chunk-3YSEJTMC.js → chunk-GGZEC22A.js} +0 -1
- package/dist/{chunk-AJ4YGHG2.js → chunk-K7Y27ND3.js} +3 -14
- package/dist/{chunk-XPFRP5QT.js → chunk-KBTUPRTF.js} +3 -4
- package/dist/{chunk-5PIO7QBG.js → chunk-KXRWY7DZ.js} +3 -4
- package/dist/{chunk-DCWXVAQT.js → chunk-OZDCTNUV.js} +6 -2
- package/dist/{chunk-BIUQ6O2Y.js → chunk-OZJMVCDK.js} +0 -1
- package/dist/{chunk-X7UG2JCF.js → chunk-PGJQDMC2.js} +1 -2
- package/dist/{chunk-DKIG2TEE.js → chunk-PMCT2YQM.js} +2 -3
- package/dist/{chunk-G3P3ZDB4.js → chunk-RJ7JVYKU.js} +0 -1
- package/dist/{chunk-WD73OFL3.js → chunk-VXLU74YZ.js} +1 -2
- package/dist/{chunk-SRRI2HJL.js → chunk-YKL3O7IU.js} +1742 -104
- package/dist/{db-BRI465XQ.js → db-4RVZ3NYZ.js} +9 -10
- package/dist/{extract-bundle-QAI7JYV2.js → extract-bundle-EA5URLXN.js} +8 -9
- package/dist/index.cjs +0 -1
- package/dist/index.js +4 -5
- package/dist/{lead-list-enrichment-repository-ODWGRVCQ.js → lead-list-enrichment-repository-IBXNWM6H.js} +2 -3
- package/dist/{location-data-repository-67MMEVZQ.js → location-data-repository-WHHQUCUI.js} +3 -4
- package/dist/{server-MTKWBSAR.js → server-Q7VYGSGQ.js} +450 -1264
- package/dist/{site-extract-repository-J3RH3MOS.js → site-extract-repository-NTM4MDQN.js} +4 -5
- package/dist/{worker-K7MPPQ7I.js → worker-UTBUWEXQ.js} +8 -9
- package/package.json +4 -1
- package/dist/analytics-repository-IKE53S3E.js.map +0 -1
- package/dist/bin/api-server.cjs.map +0 -1
- package/dist/bin/api-server.js.map +0 -1
- package/dist/bin/mcp-scraper-cli.cjs.map +0 -1
- package/dist/bin/mcp-scraper-cli.js.map +0 -1
- package/dist/bin/mcp-scraper-install.cjs.map +0 -1
- package/dist/bin/mcp-scraper-install.js.map +0 -1
- package/dist/bin/mcp-stdio-server.cjs.map +0 -1
- package/dist/bin/mcp-stdio-server.js.map +0 -1
- package/dist/bin/paa-harvest.cjs.map +0 -1
- package/dist/bin/paa-harvest.js.map +0 -1
- package/dist/chunk-2XUXVKT4.js.map +0 -1
- package/dist/chunk-3YSEJTMC.js.map +0 -1
- package/dist/chunk-4LINSJNI.js.map +0 -1
- package/dist/chunk-4VG3NUUY.js.map +0 -1
- package/dist/chunk-5BEPOFKG.js.map +0 -1
- package/dist/chunk-5PIO7QBG.js.map +0 -1
- package/dist/chunk-AAGVOPP5.js.map +0 -1
- package/dist/chunk-AJ4YGHG2.js.map +0 -1
- package/dist/chunk-BIUQ6O2Y.js.map +0 -1
- package/dist/chunk-DCWXVAQT.js.map +0 -1
- package/dist/chunk-DKIG2TEE.js.map +0 -1
- package/dist/chunk-EKJ65E3Z.js.map +0 -1
- package/dist/chunk-FG6LJGJ7.js.map +0 -1
- package/dist/chunk-G3P3ZDB4.js.map +0 -1
- package/dist/chunk-HDYOVGC6.js +0 -7
- package/dist/chunk-HDYOVGC6.js.map +0 -1
- package/dist/chunk-N54JS5WP.js.map +0 -1
- package/dist/chunk-P7YFUNKR.js.map +0 -1
- package/dist/chunk-SRRI2HJL.js.map +0 -1
- package/dist/chunk-VIWT35S6.js.map +0 -1
- package/dist/chunk-WD73OFL3.js.map +0 -1
- package/dist/chunk-X7UG2JCF.js.map +0 -1
- package/dist/chunk-XPFRP5QT.js.map +0 -1
- package/dist/db-BRI465XQ.js.map +0 -1
- package/dist/extract-bundle-QAI7JYV2.js.map +0 -1
- package/dist/index.cjs.map +0 -1
- package/dist/index.js.map +0 -1
- package/dist/lead-list-enrichment-repository-ODWGRVCQ.js.map +0 -1
- package/dist/location-data-repository-67MMEVZQ.js.map +0 -1
- package/dist/server-MTKWBSAR.js.map +0 -1
- package/dist/site-extract-repository-J3RH3MOS.js.map +0 -1
- package/dist/worker-K7MPPQ7I.js.map +0 -1
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/api/worker.ts"],"sourcesContent":["import { claimPendingJob, completeJob, failJob, creditMc, debitMc, listHarvestAttempts } from './db.js'\nimport { browserServiceApiKey } from '../lib/browser-service-env.js'\nimport { harvest } from '../harvest.js'\nimport { deliverWebhook } from './webhook.js'\nimport type { HarvestOptions } from '../types.js'\nimport { MC_COSTS, serpActualCostMc } from './rates.js'\nimport { classifyHarvestProblem, harvestProblemEnvelope, harvestProblemResponse, serializeHarvestProblem } from './harvest-problems.js'\nimport { createHarvestAttemptRecorder } from './harvest-attempt-events.js'\nimport { runWithCostContext } from './cost-context.js'\nimport { sanitizeAttempts, sanitizeHarvestResult } from './outbound-sanitize.js'\n\nexport type TickResult = {\n claimed: boolean\n jobId?: string\n completed?: boolean\n durationMs?: number\n}\nexport type DrainBudget = {\n maxJobs: number\n deadlineMs: number\n}\n\nconst MAX_CONCURRENT = 2\nlet running = 0\n\nfunction countPaaQuestions(result: unknown): number {\n if (!result || typeof result !== 'object') return 0\n const value = result as { totalQuestions?: unknown; flat?: unknown }\n if (typeof value.totalQuestions === 'number') return value.totalQuestions\n return Array.isArray(value.flat) ? value.flat.length : 0\n}\n\nfunction paaCostForQuestionCount(questionCount: number): number {\n return MC_COSTS.paa_base + Math.max(1, questionCount) * MC_COSTS.paa\n}\n\nasync function processJob(job: Awaited<ReturnType<typeof claimPendingJob>> & object) {\n running++\n try {\n const opts = typeof job.options === 'string' ? JSON.parse(job.options) as Partial<HarvestOptions> & { billingHoldMc?: number } : job.options as Partial<HarvestOptions> & { billingHoldMc?: number }\n const headlessSentOut: { value: boolean | null } = { value: null }\n const result = await runWithCostContext(\n { op: opts.serpOnly ? 'serp' : 'paa', userId: Number(job.user_id), headlessSentOut },\n () => harvest({\n ...opts,\n kernelApiKey: browserServiceApiKey(),\n headless: true,\n format: 'json',\n outputDir: '/tmp/paa-output-api',\n onAttemptEvent: createHarvestAttemptRecorder(job.id, job.user_id),\n }),\n )\n await completeJob(job.id, result)\n const attempts = await listHarvestAttempts(job.id, job.user_id)\n if (!opts.serpOnly && typeof opts.billingHoldMc === 'number') {\n const actualCost = paaCostForQuestionCount(countPaaQuestions(result))\n const diff = opts.billingHoldMc - actualCost\n if (diff > 0) await creditMc(job.user_id, diff, 'paa_refund', 'overestimate refund')\n else if (diff < 0) await debitMc(job.user_id, -diff, 'paa', opts.query ?? job.query)\n } else if (opts.serpOnly && typeof opts.billingHoldMc === 'number') {\n const actualCost = serpActualCostMc(headlessSentOut.value)\n const diff = opts.billingHoldMc - actualCost\n if (diff > 0) await creditMc(job.user_id, diff, 'serp_refund', 'headless-mode pricing settle')\n else if (diff < 0) await debitMc(job.user_id, -diff, 'serp', opts.query ?? job.query)\n }\n if (job.callback_url) {\n await deliverWebhook(job.callback_url, {\n job_id: job.id,\n status: 'done',\n result: sanitizeHarvestResult(result as unknown as Record<string, unknown>),\n attempts: sanitizeAttempts(attempts),\n })\n }\n } catch (err) {\n console.error('[harvest/worker] failed', {\n jobId: job.id,\n error: err instanceof Error ? err.message : String(err),\n })\n const problem = classifyHarvestProblem(err)\n const failedOptions = typeof job.options === 'string'\n ? JSON.parse(job.options) as { billingHoldMc?: number }\n : job.options as { billingHoldMc?: number }\n const hasHold = typeof failedOptions.billingHoldMc === 'number' && failedOptions.billingHoldMc > 0\n await failJob(job.id, serializeHarvestProblem(problem), harvestProblemEnvelope(problem, {\n chargeStatus: hasHold ? 'refund_pending' : 'not_charged',\n }))\n const attempts = await listHarvestAttempts(job.id, job.user_id)\n try {\n if (hasHold) {\n await creditMc(job.user_id, failedOptions.billingHoldMc!, 'refund', 'failed call')\n await failJob(job.id, serializeHarvestProblem(problem), harvestProblemEnvelope(problem, { chargeStatus: 'refunded' }))\n }\n } catch {}\n if (job.callback_url) {\n await deliverWebhook(job.callback_url, {\n job_id: job.id,\n status: 'failed',\n ...harvestProblemResponse(problem),\n attempts: sanitizeAttempts(attempts),\n })\n }\n } finally {\n running--\n }\n}\n\nexport async function tickOnce(): Promise<TickResult> {\n const job = await claimPendingJob()\n if (!job) return { claimed: false }\n const startedAt = Date.now()\n await processJob(job as NonNullable<typeof job>)\n return { claimed: true, jobId: (job as { id: string }).id, completed: true, durationMs: Date.now() - startedAt }\n}\n\nexport async function drainQueue(budget: DrainBudget): Promise<TickResult[]> {\n const results: TickResult[] = []\n for (let i = 0; i < budget.maxJobs; i++) {\n if (Date.now() >= budget.deadlineMs) break\n const r = await tickOnce()\n results.push(r)\n if (!r.claimed) break\n }\n return results\n}\n\nexport function startWorker(): void {\n setInterval(async () => {\n if (running >= MAX_CONCURRENT) return\n const job = await claimPendingJob()\n if (job) void processJob(job as NonNullable<typeof job>)\n }, 2000)\n console.log(`[worker] started — polling every 2s, max ${MAX_CONCURRENT} concurrent`)\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAsBA,IAAM,iBAAiB;AACvB,IAAI,UAAU;AAEd,SAAS,kBAAkB,QAAyB;AAClD,MAAI,CAAC,UAAU,OAAO,WAAW,SAAU,QAAO;AAClD,QAAM,QAAQ;AACd,MAAI,OAAO,MAAM,mBAAmB,SAAU,QAAO,MAAM;AAC3D,SAAO,MAAM,QAAQ,MAAM,IAAI,IAAI,MAAM,KAAK,SAAS;AACzD;AAEA,SAAS,wBAAwB,eAA+B;AAC9D,SAAO,SAAS,WAAW,KAAK,IAAI,GAAG,aAAa,IAAI,SAAS;AACnE;AAEA,eAAe,WAAW,KAA2D;AACnF;AACA,MAAI;AACF,UAAM,OAAO,OAAO,IAAI,YAAY,WAAW,KAAK,MAAM,IAAI,OAAO,IAA4D,IAAI;AACrI,UAAM,kBAA6C,EAAE,OAAO,KAAK;AACjE,UAAM,SAAS,MAAM;AAAA,MACnB,EAAE,IAAI,KAAK,WAAW,SAAS,OAAO,QAAQ,OAAO,IAAI,OAAO,GAAG,gBAAgB;AAAA,MACnF,MAAM,QAAQ;AAAA,QACZ,GAAG;AAAA,QACH,cAAc,qBAAqB;AAAA,QACnC,UAAU;AAAA,QACV,QAAQ;AAAA,QACR,WAAW;AAAA,QACX,gBAAgB,6BAA6B,IAAI,IAAI,IAAI,OAAO;AAAA,MAClE,CAAC;AAAA,IACH;AACA,UAAM,YAAY,IAAI,IAAI,MAAM;AAChC,UAAM,WAAW,MAAM,oBAAoB,IAAI,IAAI,IAAI,OAAO;AAC9D,QAAI,CAAC,KAAK,YAAY,OAAO,KAAK,kBAAkB,UAAU;AAC5D,YAAM,aAAa,wBAAwB,kBAAkB,MAAM,CAAC;AACpE,YAAM,OAAO,KAAK,gBAAgB;AAClC,UAAI,OAAO,EAAG,OAAM,SAAS,IAAI,SAAS,MAAM,cAAc,qBAAqB;AAAA,eAC1E,OAAO,EAAG,OAAM,QAAQ,IAAI,SAAS,CAAC,MAAM,OAAO,KAAK,SAAS,IAAI,KAAK;AAAA,IACrF,WAAW,KAAK,YAAY,OAAO,KAAK,kBAAkB,UAAU;AAClE,YAAM,aAAa,iBAAiB,gBAAgB,KAAK;AACzD,YAAM,OAAO,KAAK,gBAAgB;AAClC,UAAI,OAAO,EAAG,OAAM,SAAS,IAAI,SAAS,MAAM,eAAe,8BAA8B;AAAA,eACpF,OAAO,EAAG,OAAM,QAAQ,IAAI,SAAS,CAAC,MAAM,QAAQ,KAAK,SAAS,IAAI,KAAK;AAAA,IACtF;AACA,QAAI,IAAI,cAAc;AACpB,YAAM,eAAe,IAAI,cAAc;AAAA,QACrC,QAAQ,IAAI;AAAA,QACZ,QAAQ;AAAA,QACR,QAAQ,sBAAsB,MAA4C;AAAA,QAC1E,UAAU,iBAAiB,QAAQ;AAAA,MACrC,CAAC;AAAA,IACH;AAAA,EACF,SAAS,KAAK;AACZ,YAAQ,MAAM,2BAA2B;AAAA,MACvC,OAAO,IAAI;AAAA,MACX,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAAA,IACxD,CAAC;AACD,UAAM,UAAU,uBAAuB,GAAG;AAC1C,UAAM,gBAAgB,OAAO,IAAI,YAAY,WACzC,KAAK,MAAM,IAAI,OAAO,IACtB,IAAI;AACR,UAAM,UAAU,OAAO,cAAc,kBAAkB,YAAY,cAAc,gBAAgB;AACjG,UAAM,QAAQ,IAAI,IAAI,wBAAwB,OAAO,GAAG,uBAAuB,SAAS;AAAA,MACtF,cAAc,UAAU,mBAAmB;AAAA,IAC7C,CAAC,CAAC;AACF,UAAM,WAAW,MAAM,oBAAoB,IAAI,IAAI,IAAI,OAAO;AAC9D,QAAI;AACF,UAAI,SAAS;AACX,cAAM,SAAS,IAAI,SAAS,cAAc,eAAgB,UAAU,aAAa;AACjF,cAAM,QAAQ,IAAI,IAAI,wBAAwB,OAAO,GAAG,uBAAuB,SAAS,EAAE,cAAc,WAAW,CAAC,CAAC;AAAA,MACvH;AAAA,IACF,QAAQ;AAAA,IAAC;AACT,QAAI,IAAI,cAAc;AACpB,YAAM,eAAe,IAAI,cAAc;AAAA,QACrC,QAAQ,IAAI;AAAA,QACZ,QAAQ;AAAA,QACR,GAAG,uBAAuB,OAAO;AAAA,QACjC,UAAU,iBAAiB,QAAQ;AAAA,MACrC,CAAC;AAAA,IACH;AAAA,EACF,UAAE;AACA;AAAA,EACF;AACF;AAEA,eAAsB,WAAgC;AACpD,QAAM,MAAM,MAAM,gBAAgB;AAClC,MAAI,CAAC,IAAK,QAAO,EAAE,SAAS,MAAM;AAClC,QAAM,YAAY,KAAK,IAAI;AAC3B,QAAM,WAAW,GAA8B;AAC/C,SAAO,EAAE,SAAS,MAAM,OAAQ,IAAuB,IAAI,WAAW,MAAM,YAAY,KAAK,IAAI,IAAI,UAAU;AACjH;AAEA,eAAsB,WAAW,QAA4C;AAC3E,QAAM,UAAwB,CAAC;AAC/B,WAAS,IAAI,GAAG,IAAI,OAAO,SAAS,KAAK;AACvC,QAAI,KAAK,IAAI,KAAK,OAAO,WAAY;AACrC,UAAM,IAAI,MAAM,SAAS;AACzB,YAAQ,KAAK,CAAC;AACd,QAAI,CAAC,EAAE,QAAS;AAAA,EAClB;AACA,SAAO;AACT;AAEO,SAAS,cAAoB;AAClC,cAAY,YAAY;AACtB,QAAI,WAAW,eAAgB;AAC/B,UAAM,MAAM,MAAM,gBAAgB;AAClC,QAAI,IAAK,MAAK,WAAW,GAA8B;AAAA,EACzD,GAAG,GAAI;AACP,UAAQ,IAAI,iDAA4C,cAAc,aAAa;AACrF;","names":[]}
|