mcp-scraper 0.42.3 → 0.43.2

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.
Files changed (64) hide show
  1. package/README.md +2 -2
  2. package/dist/bin/api-server.cjs +3284 -2130
  3. package/dist/bin/api-server.cjs.map +1 -1
  4. package/dist/bin/api-server.js +2 -2
  5. package/dist/bin/mcp-scraper-cli.cjs +1 -1
  6. package/dist/bin/mcp-scraper-cli.cjs.map +1 -1
  7. package/dist/bin/mcp-scraper-cli.js +3 -4
  8. package/dist/bin/mcp-scraper-cli.js.map +1 -1
  9. package/dist/bin/mcp-scraper-install.cjs +2 -2
  10. package/dist/bin/mcp-scraper-install.cjs.map +1 -1
  11. package/dist/bin/mcp-scraper-install.js +2 -2
  12. package/dist/bin/mcp-stdio-server.cjs +3361 -2817
  13. package/dist/bin/mcp-stdio-server.cjs.map +1 -1
  14. package/dist/bin/mcp-stdio-server.js +51 -24
  15. package/dist/bin/mcp-stdio-server.js.map +1 -1
  16. package/dist/bin/paa-harvest.cjs +21 -21
  17. package/dist/bin/paa-harvest.cjs.map +1 -1
  18. package/dist/bin/paa-harvest.js +2 -3
  19. package/dist/bin/paa-harvest.js.map +1 -1
  20. package/dist/{chunk-MJKVDLVR.js → chunk-2K74LVV7.js} +26 -2
  21. package/dist/{chunk-MJKVDLVR.js.map → chunk-2K74LVV7.js.map} +1 -1
  22. package/dist/{chunk-CB5C3BPB.js → chunk-3ZUBQQPQ.js} +23 -2
  23. package/dist/chunk-3ZUBQQPQ.js.map +1 -0
  24. package/dist/{chunk-6Q7V4IQV.js → chunk-6WLNXYRG.js} +19 -2
  25. package/dist/chunk-6WLNXYRG.js.map +1 -0
  26. package/dist/{chunk-G5MHJGKE.js → chunk-GUI33Q27.js} +2 -2
  27. package/dist/{chunk-DH42LIR3.js → chunk-H7LQ6PTI.js} +24 -16
  28. package/dist/chunk-H7LQ6PTI.js.map +1 -0
  29. package/dist/chunk-MD75M2L5.js +7 -0
  30. package/dist/chunk-MD75M2L5.js.map +1 -0
  31. package/dist/{chunk-ENJQ2NSX.js → chunk-TC4GDB2Q.js} +2 -2
  32. package/dist/{chunk-ENJQ2NSX.js.map → chunk-TC4GDB2Q.js.map} +1 -1
  33. package/dist/{chunk-IMUA3FJB.js → chunk-X5WDJ7D7.js} +6 -6
  34. package/dist/chunk-X5WDJ7D7.js.map +1 -0
  35. package/dist/{chunk-QKCDYMFT.js → chunk-YJNGNMDO.js} +2196 -1584
  36. package/dist/chunk-YJNGNMDO.js.map +1 -0
  37. package/dist/{chunk-FQI5PFE7.js → chunk-ZSJHRZY5.js} +3 -5
  38. package/dist/{chunk-FQI5PFE7.js.map → chunk-ZSJHRZY5.js.map} +1 -1
  39. package/dist/{extract-bundle-FH734F3R.js → extract-bundle-IOM45SFL.js} +5 -5
  40. package/dist/index.cjs +21 -21
  41. package/dist/index.cjs.map +1 -1
  42. package/dist/index.d.cts +21 -57
  43. package/dist/index.d.ts +21 -57
  44. package/dist/index.js +2 -3
  45. package/dist/index.js.map +1 -1
  46. package/dist/{server-B27DR4YB.js → server-BPEXEKA6.js} +800 -336
  47. package/dist/server-BPEXEKA6.js.map +1 -0
  48. package/dist/{site-extract-repository-5CVI3OWT.js → site-extract-repository-XNI4ZK2M.js} +3 -3
  49. package/dist/{worker-MQJOY4YP.js → worker-2LIOH744.js} +4 -5
  50. package/dist/{worker-MQJOY4YP.js.map → worker-2LIOH744.js.map} +1 -1
  51. package/package.json +10 -7
  52. package/dist/chunk-2YFTFT3V.js +0 -7
  53. package/dist/chunk-2YFTFT3V.js.map +0 -1
  54. package/dist/chunk-6Q7V4IQV.js.map +0 -1
  55. package/dist/chunk-CB5C3BPB.js.map +0 -1
  56. package/dist/chunk-DH42LIR3.js.map +0 -1
  57. package/dist/chunk-IMUA3FJB.js.map +0 -1
  58. package/dist/chunk-K443GQY5.js +0 -24
  59. package/dist/chunk-K443GQY5.js.map +0 -1
  60. package/dist/chunk-QKCDYMFT.js.map +0 -1
  61. package/dist/server-B27DR4YB.js.map +0 -1
  62. /package/dist/{chunk-G5MHJGKE.js.map → chunk-GUI33Q27.js.map} +0 -0
  63. /package/dist/{extract-bundle-FH734F3R.js.map → extract-bundle-IOM45SFL.js.map} +0 -0
  64. /package/dist/{site-extract-repository-5CVI3OWT.js.map → site-extract-repository-XNI4ZK2M.js.map} +0 -0
@@ -26,8 +26,8 @@ import {
26
26
  setExtractJobTotal,
27
27
  settleExtractJob,
28
28
  terminalExtractJobStatus
29
- } from "./chunk-G5MHJGKE.js";
30
- import "./chunk-IMUA3FJB.js";
29
+ } from "./chunk-GUI33Q27.js";
30
+ import "./chunk-X5WDJ7D7.js";
31
31
  import "./chunk-5UN33CGU.js";
32
32
  import "./chunk-FCFQ634B.js";
33
33
  export {
@@ -59,4 +59,4 @@ export {
59
59
  settleExtractJob,
60
60
  terminalExtractJobStatus
61
61
  };
62
- //# sourceMappingURL=site-extract-repository-5CVI3OWT.js.map
62
+ //# sourceMappingURL=site-extract-repository-XNI4ZK2M.js.map
@@ -7,7 +7,7 @@ import {
7
7
  } from "./chunk-ZBQ6EXZW.js";
8
8
  import {
9
9
  harvest
10
- } from "./chunk-DH42LIR3.js";
10
+ } from "./chunk-H7LQ6PTI.js";
11
11
  import {
12
12
  sanitizeAttempts,
13
13
  sanitizeHarvestResult
@@ -16,12 +16,11 @@ import {
16
16
  browserServiceApiKey,
17
17
  runWithCostContext
18
18
  } from "./chunk-DAS7L4SF.js";
19
- import "./chunk-K443GQY5.js";
20
- import "./chunk-CB5C3BPB.js";
19
+ import "./chunk-3ZUBQQPQ.js";
21
20
  import {
22
21
  MC_COSTS,
23
22
  serpActualCostMc
24
- } from "./chunk-IMUA3FJB.js";
23
+ } from "./chunk-X5WDJ7D7.js";
25
24
  import "./chunk-5UN33CGU.js";
26
25
  import {
27
26
  claimPendingJob,
@@ -138,4 +137,4 @@ export {
138
137
  startWorker,
139
138
  tickOnce
140
139
  };
141
- //# sourceMappingURL=worker-MQJOY4YP.js.map
140
+ //# sourceMappingURL=worker-2LIOH744.js.map
@@ -1 +1 @@
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, 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 await failJob(job.id, serializeHarvestProblem(problem))\n const attempts = await listHarvestAttempts(job.id, job.user_id)\n try {\n const opts = typeof job.options === 'string' ? JSON.parse(job.options) as { billingHoldMc?: number } : job.options as { billingHoldMc?: number }\n if (typeof opts.billingHoldMc === 'number' && opts.billingHoldMc > 0) {\n await creditMc(job.user_id, opts.billingHoldMc, 'refund', 'failed call')\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,QAAQ,IAAI,IAAI,wBAAwB,OAAO,CAAC;AACtD,UAAM,WAAW,MAAM,oBAAoB,IAAI,IAAI,IAAI,OAAO;AAC9D,QAAI;AACF,YAAM,OAAO,OAAO,IAAI,YAAY,WAAW,KAAK,MAAM,IAAI,OAAO,IAAkC,IAAI;AAC3G,UAAI,OAAO,KAAK,kBAAkB,YAAY,KAAK,gBAAgB,GAAG;AACpE,cAAM,SAAS,IAAI,SAAS,KAAK,eAAe,UAAU,aAAa;AAAA,MACzE;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":[]}
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, 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 await failJob(job.id, serializeHarvestProblem(problem))\n const attempts = await listHarvestAttempts(job.id, job.user_id)\n try {\n const opts = typeof job.options === 'string' ? JSON.parse(job.options) as { billingHoldMc?: number } : job.options as { billingHoldMc?: number }\n if (typeof opts.billingHoldMc === 'number' && opts.billingHoldMc > 0) {\n await creditMc(job.user_id, opts.billingHoldMc, 'refund', 'failed call')\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,QAAQ,IAAI,IAAI,wBAAwB,OAAO,CAAC;AACtD,UAAM,WAAW,MAAM,oBAAoB,IAAI,IAAI,IAAI,OAAO;AAC9D,QAAI;AACF,YAAM,OAAO,OAAO,IAAI,YAAY,WAAW,KAAK,MAAM,IAAI,OAAO,IAAkC,IAAI;AAC3G,UAAI,OAAO,KAAK,kBAAkB,YAAY,KAAK,gBAAgB,GAAG;AACpE,cAAM,SAAS,IAAI,SAAS,KAAK,eAAe,UAAU,aAAa;AAAA,MACzE;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":[]}
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "mcp-scraper",
3
- "version": "0.42.3",
3
+ "version": "0.43.2",
4
4
  "description": "MCP server for MCP Scraper web intelligence tools",
5
5
  "type": "module",
6
6
  "main": "./dist/index.cjs",
@@ -28,12 +28,12 @@
28
28
  ],
29
29
  "scripts": {
30
30
  "build": "tsup",
31
- "build:og-pages": "node scripts/build-og-pages.mjs",
32
31
  "build:md": "node scripts/build-md-mirror.mjs",
33
32
  "build:mcpb": "npm run build && node scripts/build-mcpb.mjs",
34
33
  "build:rates": "node --import tsx scripts/gen-rates.ts",
35
34
  "build:blog": "node scripts/run-ts-sandbox-safe.mjs scripts/build-blog.ts",
36
35
  "mcp:manifest": "node scripts/run-ts-sandbox-safe.mjs scripts/generate-mcp-tool-manifest.ts",
36
+ "mcp:measure-catalog": "node scripts/run-ts-sandbox-safe.mjs scripts/measure-mcp-catalog.ts",
37
37
  "mcp:lint-tools": "node scripts/run-ts-sandbox-safe.mjs scripts/lint-mcp-tool-craft.ts",
38
38
  "mcp:lint-tools:write": "node scripts/run-ts-sandbox-safe.mjs scripts/lint-mcp-tool-craft.ts --write",
39
39
  "api": "node scripts/run-ts-sandbox-safe.mjs bin/api-server.ts",
@@ -62,7 +62,8 @@
62
62
  "email:workshop-reminders:apply": "node --import tsx scripts/email/refresh-workshop-reminders.ts --apply",
63
63
  "billing:preview-credit-lot-repair": "node --import tsx scripts/billing/preview-credit-lot-repair.ts",
64
64
  "billing:apply-credit-lot-repair": "node --import tsx scripts/billing/apply-credit-lot-repair.ts",
65
- "prepack": "npm run build && npm run package:verify-entrypoints",
65
+ "mcp:verify": "npm run build && npm run mcp:manifest && npm run build:rates && npm run mcp:measure-catalog && npm run mcp:lint-tools && npm run typecheck && npm test && npm run package:verify-entrypoints",
66
+ "prepack": "npm run mcp:verify",
66
67
  "package:verify-entrypoints": "node scripts/verify-package-entrypoints.mjs"
67
68
  },
68
69
  "dependencies": {
@@ -70,19 +71,21 @@
70
71
  "@fal-ai/client": "^1.10.1",
71
72
  "@hono/node-server": "^2.0.2",
72
73
  "@libsql/client": "^0.17.3",
73
- "@modelcontextprotocol/sdk": "^1.29.0",
74
+ "@modelcontextprotocol/client": "^2.0.0",
75
+ "@modelcontextprotocol/node": "^2.0.0",
76
+ "@modelcontextprotocol/server": "^2.0.0",
74
77
  "@neondatabase/serverless": "^1.1.0",
75
78
  "@onkernel/sdk": "^0.52.0",
76
79
  "@resvg/resvg-js": "^2.6.2",
77
80
  "@vercel/blob": "^2.4.1",
78
81
  "commander": "^12.0.0",
79
82
  "diff": "^9.0.0",
80
- "hono": "^4.12.18",
83
+ "hono": "4.13.0",
81
84
  "inngest": "^4.4.0",
82
85
  "ipaddr.js": "^1.9.1",
83
86
  "jose": "^6.2.3",
84
87
  "marked": "^18.0.7",
85
- "mcpscraper-memory-tools": "^1.12.1",
88
+ "mcpscraper-memory-tools": "1.13.3",
86
89
  "p-limit": "^7.3.0",
87
90
  "papaparse": "^5.4.0",
88
91
  "playwright": "^1.44.0",
@@ -95,7 +98,7 @@
95
98
  "yauzl": "^3.4.0",
96
99
  "yazl": "^3.3.1",
97
100
  "youtube-transcript": "^1.3.1",
98
- "zod": "^3.23.0"
101
+ "zod": "^4.4.3"
99
102
  },
100
103
  "devDependencies": {
101
104
  "@onkernel/cli": "^0.19.2",
@@ -1,7 +0,0 @@
1
- // src/version.ts
2
- var PACKAGE_VERSION = "0.42.3";
3
-
4
- export {
5
- PACKAGE_VERSION
6
- };
7
- //# sourceMappingURL=chunk-2YFTFT3V.js.map
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/version.ts"],"sourcesContent":["export const PACKAGE_VERSION = '0.42.3'\n"],"mappings":";AAAO,IAAM,kBAAkB;","names":[]}
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/lib/media-extractor.ts","../src/api/site-extract-artifacts.ts"],"sourcesContent":["import { createWriteStream, mkdirSync, rmSync } from 'node:fs'\nimport { homedir } from 'node:os'\nimport { join, extname, basename } from 'node:path'\nimport { pipeline } from 'node:stream/promises'\nimport { Readable, Transform } from 'node:stream'\nimport { validatePublicHttpUrl } from '../api/url-utils.js'\n\nconst AD_PATTERNS: string[] = [\n 'doubleclick.net', 'googlesyndication.com', 'googletagmanager.com',\n 'google-analytics.com', 'googletagservices.com', 'adservice.google',\n 'googletag.', 'pagead2.googlesyndication',\n 'facebook.net/tr', 'connect.facebook.net', 'fbcdn.net/rsrc',\n 'analytics.twitter.com', 'static.ads-twitter.com', 'ads.twitter.com',\n 't.co/i/adsct', 'pixel.advertising.com',\n 'hotjar.com', 'clarity.ms', 'quantserve.com', 'scorecardresearch.com',\n 'newrelic.com', 'nr-data.net', 'segment.io', 'segment.com',\n 'amplitude.com', 'mixpanel.com', 'heap.io', 'fullstory.com',\n 'moatads.com', 'criteo.com', 'adsrvr.org', 'rubiconproject.com',\n 'pubmatic.com', 'openx.net', 'appnexus.com', 'amazon-adsystem.com',\n 'media.net', 'yieldmo.com', 'triplelift.com', 'sharethrough.com',\n 'prebid.', 'smaato.net', 'indexworm.com', 'casalemedia.com',\n 'outbrain.com', 'taboola.com', 'revcontent.com', 'mgid.com',\n 'tawk.to', 'intercom.io', 'drift.com', 'hs-scripts.com',\n 'zopim.com', 'livechatinc.com', 'userlike.com',\n 'onetrust.com', 'cookielaw.org', 'cookieinformation.com', 'trustarc.com',\n '/ads/', '/ad/', '/banner/', '/banners/', '/pixel/', '/beacon/',\n '/tracking/', '/tracker/', '/remarketing/', '/conversion/',\n '1x1.gif', 'spacer.gif', 'blank.gif', 'transparent.gif',\n]\n\nconst IMAGE_EXTS = new Set(['.jpg', '.jpeg', '.png', '.webp', '.gif', '.avif', '.svg', '.tiff'])\nconst VIDEO_EXTS = new Set(['.mp4', '.webm', '.mov', '.avi', '.m4v', '.ogv', '.mkv'])\nconst AUDIO_EXTS = new Set(['.mp3', '.wav', '.ogg', '.aac', '.m4a', '.flac', '.opus'])\n\nexport type MediaType = 'image' | 'video' | 'audio'\n\nexport interface MediaAsset {\n url: string\n type: MediaType\n mimeType: string | null\n filename: string\n savedPath: string | null\n sizeBytes: number | null\n}\n\nexport interface MediaManifest {\n pageUrl: string\n outputDir: string | null\n assets: MediaAsset[]\n filteredCount: number\n totalFound: number\n}\n\nexport interface MediaExtractOptions {\n types?: MediaType[]\n outputDir?: string | null\n}\n\nfunction isAdUrl(url: string): boolean {\n const lower = url.toLowerCase()\n return AD_PATTERNS.some(p => lower.includes(p))\n}\n\nfunction isDataUri(url: string): boolean {\n return url.startsWith('data:')\n}\n\nfunction typeFromUrl(url: string): MediaType | null {\n try {\n const ext = extname(new URL(url).pathname).toLowerCase()\n if (IMAGE_EXTS.has(ext)) return 'image'\n if (VIDEO_EXTS.has(ext)) return 'video'\n if (AUDIO_EXTS.has(ext)) return 'audio'\n } catch { /* non-parseable URL */ }\n return null\n}\n\nfunction typeFromMime(mime: string): MediaType | null {\n const lower = mime.toLowerCase()\n if (lower.startsWith('image/')) return 'image'\n if (lower.startsWith('video/')) return 'video'\n if (lower.startsWith('audio/')) return 'audio'\n return null\n}\n\nfunction resolveUrl(raw: string, base: string): string | null {\n if (!raw || isDataUri(raw)) return null\n try { return new URL(raw, base).href } catch { return null }\n}\n\nfunction safeFilename(url: string, index: number): string {\n try {\n const u = new URL(url)\n const base = basename(u.pathname).replace(/[^a-zA-Z0-9._-]/g, '_').slice(0, 80)\n return base || `asset-${index}`\n } catch {\n return `asset-${index}`\n }\n}\n\nexport function extractMediaUrls(html: string, baseUrl: string): string[] {\n const seen = new Set<string>()\n const urls: string[] = []\n\n const add = (raw: string) => {\n const resolved = resolveUrl(raw.trim(), baseUrl)\n if (!resolved || seen.has(resolved) || isAdUrl(resolved)) return\n seen.add(resolved)\n urls.push(resolved)\n }\n\n for (const m of html.matchAll(/<img\\s[^>]*>/gi)) {\n const tag = m[0]\n for (const attr of ['src', 'data-src', 'data-lazy-src', 'data-original']) {\n const v = (tag.match(new RegExp(`\\\\b${attr}\\\\s*=\\\\s*[\"']([^\"']+)[\"']`, 'i')) ?? [])[1]\n if (v) add(v)\n }\n const srcset = (tag.match(/\\bsrcset\\s*=\\s*[\"']([^\"']+)[\"']/i) ?? [])[1]\n if (srcset) {\n for (const part of srcset.split(',')) {\n const u = part.trim().split(/\\s+/)[0]\n if (u) add(u)\n }\n }\n }\n\n for (const m of html.matchAll(/<source\\s[^>]*>/gi)) {\n const tag = m[0]\n const src = (tag.match(/\\bsrc\\s*=\\s*[\"']([^\"']+)[\"']/i) ?? [])[1]\n if (src) add(src)\n const srcset = (tag.match(/\\bsrcset\\s*=\\s*[\"']([^\"']+)[\"']/i) ?? [])[1]\n if (srcset) {\n for (const part of srcset.split(',')) {\n const u = part.trim().split(/\\s+/)[0]\n if (u) add(u)\n }\n }\n }\n\n for (const m of html.matchAll(/<video\\s[^>]*>/gi)) {\n const tag = m[0]\n for (const attr of ['src', 'poster']) {\n const v = (tag.match(new RegExp(`\\\\b${attr}\\\\s*=\\\\s*[\"']([^\"']+)[\"']`, 'i')) ?? [])[1]\n if (v) add(v)\n }\n }\n\n for (const m of html.matchAll(/<audio\\s[^>]*>/gi)) {\n const tag = m[0]\n const v = (tag.match(/\\bsrc\\s*=\\s*[\"']([^\"']+)[\"']/i) ?? [])[1]\n if (v) add(v)\n }\n\n for (const m of html.matchAll(/<meta\\s[^>]*>/gi)) {\n const tag = m[0]\n const prop = (tag.match(/\\b(?:property|name)\\s*=\\s*[\"']([^\"']+)[\"']/i) ?? [])[1]?.toLowerCase()\n if (prop === 'og:image' || prop === 'twitter:image') {\n const content = (tag.match(/\\bcontent\\s*=\\s*[\"']([^\"']+)[\"']/i) ?? [])[1]\n if (content) add(content)\n }\n }\n\n for (const m of html.matchAll(/background(?:-image)?\\s*:\\s*url\\(\\s*[\"']?([^\"')]+)[\"']?\\s*\\)/gi)) {\n add(m[1])\n }\n\n return urls\n}\n\nexport async function downloadAsset(\n url: string,\n destDir: string,\n filename: string,\n options: {\n expectedType?: MediaType\n maxBytes?: number\n consumeBytes?: (bytes: number) => boolean\n } = {},\n): Promise<{ savedPath: string; sizeBytes: number; mimeType: string | null }> {\n const maxBytes = Math.max(1, Math.min(options.maxBytes ?? 50 * 1024 * 1024, 100 * 1024 * 1024))\n let target = url\n let res: Response | null = null\n for (let redirects = 0; redirects <= 5; redirects++) {\n const checked = await validatePublicHttpUrl(target, { field: 'media URL' })\n if (checked.error || !checked.parsed) throw new Error(checked.error ?? 'Media URL was rejected')\n res = await fetch(checked.parsed.href, {\n headers: { 'User-Agent': 'Mozilla/5.0 (compatible; ThorbitBot/1.0)' },\n signal: AbortSignal.timeout(15_000),\n redirect: 'manual',\n })\n if (res.status >= 300 && res.status < 400) {\n const location = res.headers.get('location')\n if (!location) throw new Error(`HTTP ${res.status} redirect did not include Location`)\n target = new URL(location, checked.parsed.href).href\n res = null\n continue\n }\n break\n }\n if (!res) throw new Error('Media download exceeded five redirects')\n if (!res.ok) throw new Error(`HTTP ${res.status}`)\n if (!res.body) throw new Error('Empty response body')\n\n const mimeType = res.headers.get('content-type')?.split(';')[0].trim() ?? null\n if (options.expectedType && (!mimeType || typeFromMime(mimeType) !== options.expectedType)) {\n throw new Error(`Expected ${options.expectedType} content but received ${mimeType ?? 'no content-type'}`)\n }\n const declaredBytes = Number(res.headers.get('content-length') ?? 0)\n if (Number.isFinite(declaredBytes) && declaredBytes > maxBytes) {\n throw new Error(`Media asset exceeds ${maxBytes} byte limit`)\n }\n\n let dest = join(destDir, filename)\n if (mimeType && !extname(filename)) {\n const mimeExt: Record<string, string> = {\n 'image/jpeg': '.jpg', 'image/png': '.png', 'image/webp': '.webp',\n 'image/gif': '.gif', 'image/svg+xml': '.svg', 'image/avif': '.avif',\n 'video/mp4': '.mp4', 'video/webm': '.webm',\n 'audio/mpeg': '.mp3', 'audio/ogg': '.ogg', 'audio/wav': '.wav',\n }\n const ext = mimeExt[mimeType]\n if (ext) dest = dest + ext\n }\n\n const writer = createWriteStream(dest)\n await new Promise<void>((resolve, reject) => {\n writer.once('open', () => resolve())\n writer.once('error', reject)\n })\n let streamedBytes = 0\n const limiter = new Transform({\n transform(chunk: Buffer, _encoding, callback) {\n const bytes = chunk.length\n if (streamedBytes + bytes > maxBytes) {\n callback(new Error(`Media asset exceeds ${maxBytes} byte limit`))\n return\n }\n if (options.consumeBytes && !options.consumeBytes(bytes)) {\n callback(new Error('Media export total byte limit exceeded'))\n return\n }\n streamedBytes += bytes\n callback(null, chunk)\n },\n })\n try {\n await pipeline(Readable.fromWeb(res.body as import('stream/web').ReadableStream), limiter, writer)\n } catch (error) {\n writer.destroy()\n rmSync(dest, { force: true })\n throw error\n }\n\n const { statSync } = await import('node:fs')\n const sizeBytes = statSync(dest).size\n\n return { savedPath: dest, sizeBytes, mimeType }\n}\n\nexport async function harvestPageMedia(\n html: string,\n pageUrl: string,\n options: MediaExtractOptions = {},\n): Promise<MediaManifest> {\n const types = options.types ?? ['image', 'video', 'audio']\n const typesSet = new Set(types)\n const rawUrls = extractMediaUrls(html, pageUrl)\n const totalFound = rawUrls.length\n\n const filteredUrls: string[] = []\n const kept: Array<{ url: string; type: MediaType }> = []\n\n for (const url of rawUrls) {\n const type = typeFromUrl(url)\n if (!type || !typesSet.has(type)) { filteredUrls.push(url); continue }\n kept.push({ url, type })\n }\n\n if (options.outputDir === null) {\n return {\n pageUrl,\n outputDir: null,\n assets: kept.map(({ url, type }, i) => ({\n url, type, mimeType: null, filename: safeFilename(url, i), savedPath: null, sizeBytes: null,\n })),\n filteredCount: filteredUrls.length,\n totalFound,\n }\n }\n\n const domain = (() => { try { return new URL(pageUrl).hostname.replace(/^www\\./, '') } catch { return 'unknown' } })()\n const stamp = new Date().toISOString().replace(/[:.]/g, '-').slice(0, 19)\n const outDir = options.outputDir ?? join(homedir(), 'Downloads', 'mcp-scraper', 'media', `${stamp}-${domain}`)\n mkdirSync(outDir, { recursive: true })\n\n const assets: MediaAsset[] = []\n\n await Promise.allSettled(\n kept.map(async ({ url, type }, i) => {\n const filename = safeFilename(url, i)\n try {\n const { savedPath, sizeBytes, mimeType } = await downloadAsset(url, outDir, filename)\n const resolvedType = mimeType ? (typeFromMime(mimeType) ?? type) : type\n assets.push({ url, type: resolvedType, mimeType, filename: basename(savedPath), savedPath, sizeBytes })\n } catch {\n assets.push({ url, type, mimeType: null, filename, savedPath: null, sizeBytes: null })\n }\n })\n )\n\n assets.sort((a, b) => {\n if (a.savedPath && !b.savedPath) return -1\n if (!a.savedPath && b.savedPath) return 1\n return a.url.localeCompare(b.url)\n })\n\n return { pageUrl, outputDir: outDir, assets, filteredCount: filteredUrls.length, totalFound }\n}\n","import {\n createPrivateArtifact,\n createPrivateArtifactFromStream,\n privateArtifactOwnerId,\n renewPrivateArtifactDownload,\n type PrivateArtifactPolicy,\n} from './private-artifacts.js'\nimport type { ExtractJobArtifact } from './site-extract-repository.js'\nimport type { Readable } from 'node:stream'\n\nexport const SITE_EXTRACT_ARTIFACT_PREFIX = 'site-extracts/'\nexport const SITE_EXTRACT_ARTIFACT_TTL_MS = 7 * 24 * 60 * 60 * 1_000\nexport const SITE_EXTRACT_DOWNLOAD_TTL_MS = 15 * 60 * 1_000\n\nfunction siteExtractArtifactToken(): string | null {\n return process.env.SITE_EXTRACT_ARTIFACT_READ_WRITE_TOKEN?.trim()\n || process.env.PRIVATE_ARTIFACT_READ_WRITE_TOKEN?.trim()\n || process.env.CONNECTED_DATA_READ_WRITE_TOKEN?.trim()\n || process.env.CONNECTED_DATA_BLOB_READ_WRITE_TOKEN?.trim()\n || null\n}\n\nfunction hostedByEnvironment(): boolean {\n return process.env.VERCEL === '1' || process.env.NODE_ENV === 'production'\n}\n\nfunction policy(): PrivateArtifactPolicy {\n return {\n prefix: SITE_EXTRACT_ARTIFACT_PREFIX,\n artifactTtlMs: SITE_EXTRACT_ARTIFACT_TTL_MS,\n downloadTtlMs: SITE_EXTRACT_DOWNLOAD_TTL_MS,\n token: siteExtractArtifactToken(),\n }\n}\n\nexport function siteExtractArtifactOwnerId(artifactId: string): string | null {\n return privateArtifactOwnerId(artifactId, SITE_EXTRACT_ARTIFACT_PREFIX)\n}\n\nexport async function createSiteExtractBundleArtifact(args: {\n ownerId: string\n jobId: string\n createdAt: Date | string | number\n content: Buffer\n}): Promise<ExtractJobArtifact> {\n const pointer = await createPrivateArtifact({\n policy: policy(),\n ownerId: args.ownerId,\n artifactKey: `${args.jobId}.zip`,\n createdAt: args.createdAt,\n filename: `${args.jobId}-site-export.zip`,\n contentType: 'application/zip',\n content: args.content,\n })\n return {\n key: pointer.artifactId,\n url: pointer.downloadUrl ?? '',\n bytes: pointer.bytes,\n contentType: pointer.contentType,\n filename: pointer.filename,\n sha256: pointer.sha256,\n expiresAt: pointer.expiresAt,\n downloadUrlExpiresAt: pointer.downloadUrlExpiresAt,\n }\n}\n\nexport async function createSiteExtractBundleArtifactStream(args: {\n ownerId: string\n jobId: string\n createdAt: Date | string | number\n content: Readable\n}): Promise<ExtractJobArtifact> {\n const pointer = await createPrivateArtifactFromStream({\n policy: policy(),\n ownerId: args.ownerId,\n artifactKey: `${args.jobId}.zip`,\n createdAt: args.createdAt,\n filename: `${args.jobId}-site-export.zip`,\n contentType: 'application/zip',\n content: args.content,\n })\n return {\n key: pointer.artifactId,\n url: pointer.downloadUrl ?? '',\n bytes: pointer.bytes,\n contentType: pointer.contentType,\n filename: pointer.filename,\n sha256: pointer.sha256,\n expiresAt: pointer.expiresAt,\n downloadUrlExpiresAt: pointer.downloadUrlExpiresAt,\n }\n}\n\nexport async function renewSiteExtractArtifactDownload(args: {\n artifactId: string\n ownerId: string\n}): Promise<{ downloadUrl: string; downloadUrlExpiresAt: string; expiresAt: string } | null> {\n return renewPrivateArtifactDownload({\n policy: policy(),\n artifactId: args.artifactId,\n ownerId: args.ownerId,\n })\n}\n\nexport async function cleanupExpiredSiteExtractArtifacts(args: {\n now?: Date\n force?: boolean\n} = {}): Promise<{ deleted: number; store: 'private-vercel-blob' | 'local' | 'none'; skipped?: boolean }> {\n const now = args.now ?? new Date()\n const token = siteExtractArtifactToken()\n if (!token) return { deleted: 0, store: hostedByEnvironment() ? 'none' : 'local' }\n if (!args.force && !(now.getUTCHours() === 3 && now.getUTCMinutes() === 21)) {\n return { deleted: 0, store: 'private-vercel-blob', skipped: true }\n }\n const cutoff = now.getTime() - SITE_EXTRACT_ARTIFACT_TTL_MS\n const { list, del } = await import('@vercel/blob')\n let cursor: string | undefined\n let deleted = 0\n for (let page = 0; page < 20; page += 1) {\n const result = await list({ prefix: SITE_EXTRACT_ARTIFACT_PREFIX, token, limit: 1_000, cursor })\n const expired = result.blobs.filter(blob => new Date(blob.uploadedAt).getTime() <= cutoff)\n if (expired.length > 0) {\n await del(expired.map(blob => blob.pathname), { token })\n deleted += expired.length\n }\n if (!result.hasMore || !result.cursor) break\n cursor = result.cursor\n }\n return { deleted, store: 'private-vercel-blob' }\n}\n"],"mappings":";;;;;;;AAAA,SAAS,mBAAmB,WAAW,cAAc;AACrD,SAAS,eAAe;AACxB,SAAS,MAAM,SAAS,gBAAgB;AACxC,SAAS,gBAAgB;AACzB,SAAS,UAAU,iBAAiB;AAGpC,IAAM,cAAwB;AAAA,EAC5B;AAAA,EAAmB;AAAA,EAAyB;AAAA,EAC5C;AAAA,EAAwB;AAAA,EAAyB;AAAA,EACjD;AAAA,EAAc;AAAA,EACd;AAAA,EAAmB;AAAA,EAAwB;AAAA,EAC3C;AAAA,EAAyB;AAAA,EAA0B;AAAA,EACnD;AAAA,EAAgB;AAAA,EAChB;AAAA,EAAc;AAAA,EAAc;AAAA,EAAkB;AAAA,EAC9C;AAAA,EAAgB;AAAA,EAAe;AAAA,EAAc;AAAA,EAC7C;AAAA,EAAiB;AAAA,EAAgB;AAAA,EAAW;AAAA,EAC5C;AAAA,EAAe;AAAA,EAAc;AAAA,EAAc;AAAA,EAC3C;AAAA,EAAgB;AAAA,EAAa;AAAA,EAAgB;AAAA,EAC7C;AAAA,EAAa;AAAA,EAAe;AAAA,EAAkB;AAAA,EAC9C;AAAA,EAAW;AAAA,EAAc;AAAA,EAAiB;AAAA,EAC1C;AAAA,EAAgB;AAAA,EAAe;AAAA,EAAkB;AAAA,EACjD;AAAA,EAAW;AAAA,EAAe;AAAA,EAAa;AAAA,EACvC;AAAA,EAAa;AAAA,EAAmB;AAAA,EAChC;AAAA,EAAgB;AAAA,EAAiB;AAAA,EAAyB;AAAA,EAC1D;AAAA,EAAS;AAAA,EAAQ;AAAA,EAAY;AAAA,EAAa;AAAA,EAAW;AAAA,EACrD;AAAA,EAAc;AAAA,EAAa;AAAA,EAAiB;AAAA,EAC5C;AAAA,EAAW;AAAA,EAAc;AAAA,EAAa;AACxC;AAEA,IAAM,aAAc,oBAAI,IAAI,CAAC,QAAQ,SAAS,QAAQ,SAAS,QAAQ,SAAS,QAAQ,OAAO,CAAC;AAChG,IAAM,aAAc,oBAAI,IAAI,CAAC,QAAQ,SAAS,QAAQ,QAAQ,QAAQ,QAAQ,MAAM,CAAC;AACrF,IAAM,aAAc,oBAAI,IAAI,CAAC,QAAQ,QAAQ,QAAQ,QAAQ,QAAQ,SAAS,OAAO,CAAC;AA0BtF,SAAS,QAAQ,KAAsB;AACrC,QAAM,QAAQ,IAAI,YAAY;AAC9B,SAAO,YAAY,KAAK,OAAK,MAAM,SAAS,CAAC,CAAC;AAChD;AAEA,SAAS,UAAU,KAAsB;AACvC,SAAO,IAAI,WAAW,OAAO;AAC/B;AAEA,SAAS,YAAY,KAA+B;AAClD,MAAI;AACF,UAAM,MAAM,QAAQ,IAAI,IAAI,GAAG,EAAE,QAAQ,EAAE,YAAY;AACvD,QAAI,WAAW,IAAI,GAAG,EAAG,QAAO;AAChC,QAAI,WAAW,IAAI,GAAG,EAAG,QAAO;AAChC,QAAI,WAAW,IAAI,GAAG,EAAG,QAAO;AAAA,EAClC,QAAQ;AAAA,EAA0B;AAClC,SAAO;AACT;AAEA,SAAS,aAAa,MAAgC;AACpD,QAAM,QAAQ,KAAK,YAAY;AAC/B,MAAI,MAAM,WAAW,QAAQ,EAAG,QAAO;AACvC,MAAI,MAAM,WAAW,QAAQ,EAAG,QAAO;AACvC,MAAI,MAAM,WAAW,QAAQ,EAAG,QAAO;AACvC,SAAO;AACT;AAEA,SAAS,WAAW,KAAa,MAA6B;AAC5D,MAAI,CAAC,OAAO,UAAU,GAAG,EAAG,QAAO;AACnC,MAAI;AAAE,WAAO,IAAI,IAAI,KAAK,IAAI,EAAE;AAAA,EAAK,QAAQ;AAAE,WAAO;AAAA,EAAK;AAC7D;AAEA,SAAS,aAAa,KAAa,OAAuB;AACxD,MAAI;AACF,UAAM,IAAI,IAAI,IAAI,GAAG;AACrB,UAAM,OAAO,SAAS,EAAE,QAAQ,EAAE,QAAQ,oBAAoB,GAAG,EAAE,MAAM,GAAG,EAAE;AAC9E,WAAO,QAAQ,SAAS,KAAK;AAAA,EAC/B,QAAQ;AACN,WAAO,SAAS,KAAK;AAAA,EACvB;AACF;AAEO,SAAS,iBAAiB,MAAc,SAA2B;AACxE,QAAM,OAAO,oBAAI,IAAY;AAC7B,QAAM,OAAiB,CAAC;AAExB,QAAM,MAAM,CAAC,QAAgB;AAC3B,UAAM,WAAW,WAAW,IAAI,KAAK,GAAG,OAAO;AAC/C,QAAI,CAAC,YAAY,KAAK,IAAI,QAAQ,KAAK,QAAQ,QAAQ,EAAG;AAC1D,SAAK,IAAI,QAAQ;AACjB,SAAK,KAAK,QAAQ;AAAA,EACpB;AAEA,aAAW,KAAK,KAAK,SAAS,gBAAgB,GAAG;AAC/C,UAAM,MAAM,EAAE,CAAC;AACf,eAAW,QAAQ,CAAC,OAAO,YAAY,iBAAiB,eAAe,GAAG;AACxE,YAAM,KAAK,IAAI,MAAM,IAAI,OAAO,MAAM,IAAI,6BAA6B,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC;AACrF,UAAI,EAAG,KAAI,CAAC;AAAA,IACd;AACA,UAAM,UAAU,IAAI,MAAM,kCAAkC,KAAK,CAAC,GAAG,CAAC;AACtE,QAAI,QAAQ;AACV,iBAAW,QAAQ,OAAO,MAAM,GAAG,GAAG;AACpC,cAAM,IAAI,KAAK,KAAK,EAAE,MAAM,KAAK,EAAE,CAAC;AACpC,YAAI,EAAG,KAAI,CAAC;AAAA,MACd;AAAA,IACF;AAAA,EACF;AAEA,aAAW,KAAK,KAAK,SAAS,mBAAmB,GAAG;AAClD,UAAM,MAAM,EAAE,CAAC;AACf,UAAM,OAAO,IAAI,MAAM,+BAA+B,KAAK,CAAC,GAAG,CAAC;AAChE,QAAI,IAAK,KAAI,GAAG;AAChB,UAAM,UAAU,IAAI,MAAM,kCAAkC,KAAK,CAAC,GAAG,CAAC;AACtE,QAAI,QAAQ;AACV,iBAAW,QAAQ,OAAO,MAAM,GAAG,GAAG;AACpC,cAAM,IAAI,KAAK,KAAK,EAAE,MAAM,KAAK,EAAE,CAAC;AACpC,YAAI,EAAG,KAAI,CAAC;AAAA,MACd;AAAA,IACF;AAAA,EACF;AAEA,aAAW,KAAK,KAAK,SAAS,kBAAkB,GAAG;AACjD,UAAM,MAAM,EAAE,CAAC;AACf,eAAW,QAAQ,CAAC,OAAO,QAAQ,GAAG;AACpC,YAAM,KAAK,IAAI,MAAM,IAAI,OAAO,MAAM,IAAI,6BAA6B,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC;AACrF,UAAI,EAAG,KAAI,CAAC;AAAA,IACd;AAAA,EACF;AAEA,aAAW,KAAK,KAAK,SAAS,kBAAkB,GAAG;AACjD,UAAM,MAAM,EAAE,CAAC;AACf,UAAM,KAAK,IAAI,MAAM,+BAA+B,KAAK,CAAC,GAAG,CAAC;AAC9D,QAAI,EAAG,KAAI,CAAC;AAAA,EACd;AAEA,aAAW,KAAK,KAAK,SAAS,iBAAiB,GAAG;AAChD,UAAM,MAAM,EAAE,CAAC;AACf,UAAM,QAAQ,IAAI,MAAM,6CAA6C,KAAK,CAAC,GAAG,CAAC,GAAG,YAAY;AAC9F,QAAI,SAAS,cAAc,SAAS,iBAAiB;AACnD,YAAM,WAAW,IAAI,MAAM,mCAAmC,KAAK,CAAC,GAAG,CAAC;AACxE,UAAI,QAAS,KAAI,OAAO;AAAA,IAC1B;AAAA,EACF;AAEA,aAAW,KAAK,KAAK,SAAS,gEAAgE,GAAG;AAC/F,QAAI,EAAE,CAAC,CAAC;AAAA,EACV;AAEA,SAAO;AACT;AAEA,eAAsB,cACpB,KACA,SACA,UACA,UAII,CAAC,GACuE;AAC5E,QAAM,WAAW,KAAK,IAAI,GAAG,KAAK,IAAI,QAAQ,YAAY,KAAK,OAAO,MAAM,MAAM,OAAO,IAAI,CAAC;AAC9F,MAAI,SAAS;AACb,MAAI,MAAuB;AAC3B,WAAS,YAAY,GAAG,aAAa,GAAG,aAAa;AACnD,UAAM,UAAU,MAAM,sBAAsB,QAAQ,EAAE,OAAO,YAAY,CAAC;AAC1E,QAAI,QAAQ,SAAS,CAAC,QAAQ,OAAQ,OAAM,IAAI,MAAM,QAAQ,SAAS,wBAAwB;AAC/F,UAAM,MAAM,MAAM,QAAQ,OAAO,MAAM;AAAA,MACrC,SAAS,EAAE,cAAc,2CAA2C;AAAA,MACpE,QAAQ,YAAY,QAAQ,IAAM;AAAA,MAClC,UAAU;AAAA,IACZ,CAAC;AACD,QAAI,IAAI,UAAU,OAAO,IAAI,SAAS,KAAK;AACzC,YAAM,WAAW,IAAI,QAAQ,IAAI,UAAU;AAC3C,UAAI,CAAC,SAAU,OAAM,IAAI,MAAM,QAAQ,IAAI,MAAM,oCAAoC;AACrF,eAAS,IAAI,IAAI,UAAU,QAAQ,OAAO,IAAI,EAAE;AAChD,YAAM;AACN;AAAA,IACF;AACA;AAAA,EACF;AACA,MAAI,CAAC,IAAK,OAAM,IAAI,MAAM,wCAAwC;AAClE,MAAI,CAAC,IAAI,GAAI,OAAM,IAAI,MAAM,QAAQ,IAAI,MAAM,EAAE;AACjD,MAAI,CAAC,IAAI,KAAM,OAAM,IAAI,MAAM,qBAAqB;AAEpD,QAAM,WAAW,IAAI,QAAQ,IAAI,cAAc,GAAG,MAAM,GAAG,EAAE,CAAC,EAAE,KAAK,KAAK;AAC1E,MAAI,QAAQ,iBAAiB,CAAC,YAAY,aAAa,QAAQ,MAAM,QAAQ,eAAe;AAC1F,UAAM,IAAI,MAAM,YAAY,QAAQ,YAAY,yBAAyB,YAAY,iBAAiB,EAAE;AAAA,EAC1G;AACA,QAAM,gBAAgB,OAAO,IAAI,QAAQ,IAAI,gBAAgB,KAAK,CAAC;AACnE,MAAI,OAAO,SAAS,aAAa,KAAK,gBAAgB,UAAU;AAC9D,UAAM,IAAI,MAAM,uBAAuB,QAAQ,aAAa;AAAA,EAC9D;AAEA,MAAI,OAAO,KAAK,SAAS,QAAQ;AACjC,MAAI,YAAY,CAAC,QAAQ,QAAQ,GAAG;AAClC,UAAM,UAAkC;AAAA,MACtC,cAAc;AAAA,MAAQ,aAAa;AAAA,MAAQ,cAAc;AAAA,MACzD,aAAa;AAAA,MAAQ,iBAAiB;AAAA,MAAQ,cAAc;AAAA,MAC5D,aAAa;AAAA,MAAQ,cAAc;AAAA,MACnC,cAAc;AAAA,MAAQ,aAAa;AAAA,MAAQ,aAAa;AAAA,IAC1D;AACA,UAAM,MAAM,QAAQ,QAAQ;AAC5B,QAAI,IAAK,QAAO,OAAO;AAAA,EACzB;AAEA,QAAM,SAAS,kBAAkB,IAAI;AACrC,QAAM,IAAI,QAAc,CAAC,SAAS,WAAW;AAC3C,WAAO,KAAK,QAAQ,MAAM,QAAQ,CAAC;AACnC,WAAO,KAAK,SAAS,MAAM;AAAA,EAC7B,CAAC;AACD,MAAI,gBAAgB;AACpB,QAAM,UAAU,IAAI,UAAU;AAAA,IAC5B,UAAU,OAAe,WAAW,UAAU;AAC5C,YAAM,QAAQ,MAAM;AACpB,UAAI,gBAAgB,QAAQ,UAAU;AACpC,iBAAS,IAAI,MAAM,uBAAuB,QAAQ,aAAa,CAAC;AAChE;AAAA,MACF;AACA,UAAI,QAAQ,gBAAgB,CAAC,QAAQ,aAAa,KAAK,GAAG;AACxD,iBAAS,IAAI,MAAM,wCAAwC,CAAC;AAC5D;AAAA,MACF;AACA,uBAAiB;AACjB,eAAS,MAAM,KAAK;AAAA,IACtB;AAAA,EACF,CAAC;AACD,MAAI;AACF,UAAM,SAAS,SAAS,QAAQ,IAAI,IAA2C,GAAG,SAAS,MAAM;AAAA,EACnG,SAAS,OAAO;AACd,WAAO,QAAQ;AACf,WAAO,MAAM,EAAE,OAAO,KAAK,CAAC;AAC5B,UAAM;AAAA,EACR;AAEA,QAAM,EAAE,SAAS,IAAI,MAAM,OAAO,IAAS;AAC3C,QAAM,YAAY,SAAS,IAAI,EAAE;AAEjC,SAAO,EAAE,WAAW,MAAM,WAAW,SAAS;AAChD;AAEA,eAAsB,iBACpB,MACA,SACA,UAA+B,CAAC,GACR;AACxB,QAAM,QAAW,QAAQ,SAAS,CAAC,SAAS,SAAS,OAAO;AAC5D,QAAM,WAAW,IAAI,IAAI,KAAK;AAC9B,QAAM,UAAW,iBAAiB,MAAM,OAAO;AAC/C,QAAM,aAAa,QAAQ;AAE3B,QAAM,eAAyB,CAAC;AAChC,QAAM,OAAgD,CAAC;AAEvD,aAAW,OAAO,SAAS;AACzB,UAAM,OAAO,YAAY,GAAG;AAC5B,QAAI,CAAC,QAAQ,CAAC,SAAS,IAAI,IAAI,GAAG;AAAE,mBAAa,KAAK,GAAG;AAAG;AAAA,IAAS;AACrE,SAAK,KAAK,EAAE,KAAK,KAAK,CAAC;AAAA,EACzB;AAEA,MAAI,QAAQ,cAAc,MAAM;AAC9B,WAAO;AAAA,MACL;AAAA,MACA,WAAW;AAAA,MACX,QAAQ,KAAK,IAAI,CAAC,EAAE,KAAK,KAAK,GAAG,OAAO;AAAA,QACtC;AAAA,QAAK;AAAA,QAAM,UAAU;AAAA,QAAM,UAAU,aAAa,KAAK,CAAC;AAAA,QAAG,WAAW;AAAA,QAAM,WAAW;AAAA,MACzF,EAAE;AAAA,MACF,eAAe,aAAa;AAAA,MAC5B;AAAA,IACF;AAAA,EACF;AAEA,QAAM,UAAW,MAAM;AAAE,QAAI;AAAE,aAAO,IAAI,IAAI,OAAO,EAAE,SAAS,QAAQ,UAAU,EAAE;AAAA,IAAE,QAAQ;AAAE,aAAO;AAAA,IAAU;AAAA,EAAE,GAAG;AACtH,QAAM,SAAU,oBAAI,KAAK,GAAE,YAAY,EAAE,QAAQ,SAAS,GAAG,EAAE,MAAM,GAAG,EAAE;AAC1E,QAAM,SAAU,QAAQ,aAAa,KAAK,QAAQ,GAAG,aAAa,eAAe,SAAS,GAAG,KAAK,IAAI,MAAM,EAAE;AAC9G,YAAU,QAAQ,EAAE,WAAW,KAAK,CAAC;AAErC,QAAM,SAAuB,CAAC;AAE9B,QAAM,QAAQ;AAAA,IACZ,KAAK,IAAI,OAAO,EAAE,KAAK,KAAK,GAAG,MAAM;AACnC,YAAM,WAAW,aAAa,KAAK,CAAC;AACpC,UAAI;AACF,cAAM,EAAE,WAAW,WAAW,SAAS,IAAI,MAAM,cAAc,KAAK,QAAQ,QAAQ;AACpF,cAAM,eAAe,WAAY,aAAa,QAAQ,KAAK,OAAQ;AACnE,eAAO,KAAK,EAAE,KAAK,MAAM,cAAc,UAAU,UAAU,SAAS,SAAS,GAAG,WAAW,UAAU,CAAC;AAAA,MACxG,QAAQ;AACN,eAAO,KAAK,EAAE,KAAK,MAAM,UAAU,MAAM,UAAU,WAAW,MAAM,WAAW,KAAK,CAAC;AAAA,MACvF;AAAA,IACF,CAAC;AAAA,EACH;AAEA,SAAO,KAAK,CAAC,GAAG,MAAM;AACpB,QAAI,EAAE,aAAa,CAAC,EAAE,UAAW,QAAO;AACxC,QAAI,CAAC,EAAE,aAAa,EAAE,UAAW,QAAO;AACxC,WAAO,EAAE,IAAI,cAAc,EAAE,GAAG;AAAA,EAClC,CAAC;AAED,SAAO,EAAE,SAAS,WAAW,QAAQ,QAAQ,eAAe,aAAa,QAAQ,WAAW;AAC9F;;;ACnTO,IAAM,+BAA+B;AACrC,IAAM,+BAA+B,IAAI,KAAK,KAAK,KAAK;AACxD,IAAM,+BAA+B,KAAK,KAAK;AAEtD,SAAS,2BAA0C;AACjD,SAAO,QAAQ,IAAI,wCAAwC,KAAK,KAC3D,QAAQ,IAAI,mCAAmC,KAAK,KACpD,QAAQ,IAAI,iCAAiC,KAAK,KAClD,QAAQ,IAAI,sCAAsC,KAAK,KACvD;AACP;AAEA,SAAS,sBAA+B;AACtC,SAAO,QAAQ,IAAI,WAAW,OAAO,QAAQ,IAAI,aAAa;AAChE;AAEA,SAAS,SAAgC;AACvC,SAAO;AAAA,IACL,QAAQ;AAAA,IACR,eAAe;AAAA,IACf,eAAe;AAAA,IACf,OAAO,yBAAyB;AAAA,EAClC;AACF;AAiCA,eAAsB,sCAAsC,MAK5B;AAC9B,QAAM,UAAU,MAAM,gCAAgC;AAAA,IACpD,QAAQ,OAAO;AAAA,IACf,SAAS,KAAK;AAAA,IACd,aAAa,GAAG,KAAK,KAAK;AAAA,IAC1B,WAAW,KAAK;AAAA,IAChB,UAAU,GAAG,KAAK,KAAK;AAAA,IACvB,aAAa;AAAA,IACb,SAAS,KAAK;AAAA,EAChB,CAAC;AACD,SAAO;AAAA,IACL,KAAK,QAAQ;AAAA,IACb,KAAK,QAAQ,eAAe;AAAA,IAC5B,OAAO,QAAQ;AAAA,IACf,aAAa,QAAQ;AAAA,IACrB,UAAU,QAAQ;AAAA,IAClB,QAAQ,QAAQ;AAAA,IAChB,WAAW,QAAQ;AAAA,IACnB,sBAAsB,QAAQ;AAAA,EAChC;AACF;AAEA,eAAsB,iCAAiC,MAGsC;AAC3F,SAAO,6BAA6B;AAAA,IAClC,QAAQ,OAAO;AAAA,IACf,YAAY,KAAK;AAAA,IACjB,SAAS,KAAK;AAAA,EAChB,CAAC;AACH;AAEA,eAAsB,mCAAmC,OAGrD,CAAC,GAAqG;AACxG,QAAM,MAAM,KAAK,OAAO,oBAAI,KAAK;AACjC,QAAM,QAAQ,yBAAyB;AACvC,MAAI,CAAC,MAAO,QAAO,EAAE,SAAS,GAAG,OAAO,oBAAoB,IAAI,SAAS,QAAQ;AACjF,MAAI,CAAC,KAAK,SAAS,EAAE,IAAI,YAAY,MAAM,KAAK,IAAI,cAAc,MAAM,KAAK;AAC3E,WAAO,EAAE,SAAS,GAAG,OAAO,uBAAuB,SAAS,KAAK;AAAA,EACnE;AACA,QAAM,SAAS,IAAI,QAAQ,IAAI;AAC/B,QAAM,EAAE,MAAM,IAAI,IAAI,MAAM,OAAO,cAAc;AACjD,MAAI;AACJ,MAAI,UAAU;AACd,WAAS,OAAO,GAAG,OAAO,IAAI,QAAQ,GAAG;AACvC,UAAM,SAAS,MAAM,KAAK,EAAE,QAAQ,8BAA8B,OAAO,OAAO,KAAO,OAAO,CAAC;AAC/F,UAAM,UAAU,OAAO,MAAM,OAAO,UAAQ,IAAI,KAAK,KAAK,UAAU,EAAE,QAAQ,KAAK,MAAM;AACzF,QAAI,QAAQ,SAAS,GAAG;AACtB,YAAM,IAAI,QAAQ,IAAI,UAAQ,KAAK,QAAQ,GAAG,EAAE,MAAM,CAAC;AACvD,iBAAW,QAAQ;AAAA,IACrB;AACA,QAAI,CAAC,OAAO,WAAW,CAAC,OAAO,OAAQ;AACvC,aAAS,OAAO;AAAA,EAClB;AACA,SAAO,EAAE,SAAS,OAAO,sBAAsB;AACjD;","names":[]}
@@ -1 +0,0 @@
1
- {"version":3,"sources":["../src/schemas.ts"],"sourcesContent":["import { z } from 'zod'\n\nexport const DEFAULT_PROXY_MODE = 'none' as const\nexport const DEFAULT_MAPS_PROXY_MODE = 'none' as const\n\nexport const HarvestOptionsSchema = z.object({\n query: z.string().min(1),\n location: z.string().optional(),\n gl: z.string().length(2).default('us'),\n hl: z.string().length(2).default('en'),\n device: z.enum(['desktop', 'mobile']).default('desktop'),\n proxyMode: z.enum(['location', 'configured', 'none']).default(DEFAULT_PROXY_MODE),\n proxyZip: z.string().regex(/^\\d{5}$/).optional(),\n keepDefaultProxy: z.boolean().optional(),\n requireUsEgress: z.boolean().optional(),\n maxAttempts: z.number().int().min(1).max(12).optional(),\n debug: z.boolean().default(false),\n depth: z.number().int().min(1).max(30).default(3),\n maxQuestions: z.number().int().min(1).max(1000).default(100),\n headless: z.boolean().default(false),\n profileDir: z.string().optional(),\n proxy: z.string().url().optional(),\n kernelApiKey: z.string().optional(),\n kernelProxyId: z.string().optional(),\n kernelProxyResolution: z.unknown().optional(),\n outputDir: z.string().default('./paa-output'),\n format: z.enum(['json', 'csv', 'both']).default('both'),\n serpOnly: z.boolean().default(false),\n pages: z.number().int().min(1).max(2).default(1),\n recency: z.enum(['day', 'week', 'month', 'year']).optional(),\n softDeadlineMs: z.number().optional(),\n})\n\nexport function recencyToTbs(recency: 'day' | 'week' | 'month' | 'year' | undefined): string | undefined {\n switch (recency) {\n case 'day': return 'qdr:d'\n case 'week': return 'qdr:w'\n case 'month': return 'qdr:m'\n case 'year': return 'qdr:y'\n default: return undefined\n }\n}\n\nexport const MapsPlaceOptionsSchema = z.object({\n businessName: z.string().min(1),\n location: z.string().min(1),\n gl: z.string().length(2).default('us'),\n hl: z.string().length(2).default('en'),\n includeReviews: z.boolean().default(false),\n maxReviews: z.number().int().min(1).max(500).default(50),\n includeServices: z.boolean().default(false),\n kernelApiKey: z.string().optional(),\n kernelProxyId: z.string().optional(),\n headless: z.boolean().default(true),\n})\n\nexport const MapsSearchOptionsSchema = z.object({\n query: z.string().min(1),\n location: z.string().optional(),\n gl: z.string().length(2).default('us'),\n hl: z.string().length(2).default('en'),\n maxResults: z.number().int().min(1).max(50).default(10),\n includeServices: z.boolean().default(false),\n proxyMode: z.enum(['location', 'configured', 'none']).default(DEFAULT_MAPS_PROXY_MODE),\n proxyZip: z.string().regex(/^\\d{5}$/).optional(),\n serpRedirect: z.boolean().optional(),\n forceDirectEgress: z.boolean().optional(),\n debug: z.boolean().default(false),\n kernelApiKey: z.string().optional(),\n kernelProxyId: z.string().optional(),\n kernelProxyResolution: z.unknown().optional(),\n headless: z.boolean().default(true),\n})\n\nexport const RawPAAItemSchema = z.object({\n question: z.string().min(1),\n answer: z.string().optional(),\n sourceTitle: z.string().optional(),\n sourceSite: z.string().optional(),\n sourceCite: z.string().optional(),\n})\n\nexport const RawMapsOverviewSchema = z.object({\n name: z.string().nullable(),\n rating: z.string().nullable(),\n reviewCount: z.string().nullable(),\n category: z.string().nullable(),\n address: z.string().nullable(),\n hoursSummary: z.string().nullable(),\n phone: z.string().nullable(),\n phoneDisplay: z.string().nullable(),\n website: z.string().nullable(),\n plusCode: z.string().nullable(),\n bookingUrl: z.string().nullable(),\n})\n\nexport const RawMapsHoursRowSchema = z.object({\n day: z.string(),\n hours: z.string(),\n})\n\nexport const RawMapsReviewStatsSchema = z.object({\n reviewHistogram: z.array(z.object({\n stars: z.number(),\n count: z.string(),\n })),\n reviewTopics: z.array(z.object({\n label: z.string(),\n count: z.string(),\n })),\n})\n\nexport const RawMapsReviewCardSchema = z.object({\n reviewId: z.string(),\n author: z.string().nullable(),\n stars: z.string().nullable(),\n date: z.string().nullable(),\n text: z.string().nullable(),\n ownerResponse: z.string().nullable(),\n})\n\nexport const RawMapsAboutAttributeSchema = z.object({\n section: z.string(),\n attribute: z.string(),\n})\n"],"mappings":";AAAA,SAAS,SAAS;AAEX,IAAM,qBAAqB;AAC3B,IAAM,0BAA0B;AAEhC,IAAM,uBAAuB,EAAE,OAAO;AAAA,EAC3C,OAAc,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC9B,UAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,IAAc,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAC/C,IAAc,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAC/C,QAAc,EAAE,KAAK,CAAC,WAAW,QAAQ,CAAC,EAAE,QAAQ,SAAS;AAAA,EAC7D,WAAc,EAAE,KAAK,CAAC,YAAY,cAAc,MAAM,CAAC,EAAE,QAAQ,kBAAkB;AAAA,EACnF,UAAc,EAAE,OAAO,EAAE,MAAM,SAAS,EAAE,SAAS;AAAA,EACnD,kBAAkB,EAAE,QAAQ,EAAE,SAAS;AAAA,EACvC,iBAAiB,EAAE,QAAQ,EAAE,SAAS;AAAA,EACtC,aAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,SAAS;AAAA,EACvD,OAAc,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACvC,OAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,QAAQ,CAAC;AAAA,EACvD,cAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,GAAI,EAAE,QAAQ,GAAG;AAAA,EAC3D,UAAc,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACvC,YAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,OAAc,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS;AAAA,EACxC,cAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,eAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,uBAAuB,EAAE,QAAQ,EAAE,SAAS;AAAA,EAC5C,WAAc,EAAE,OAAO,EAAE,QAAQ,cAAc;AAAA,EAC/C,QAAc,EAAE,KAAK,CAAC,QAAQ,OAAO,MAAM,CAAC,EAAE,QAAQ,MAAM;AAAA,EAC5D,UAAc,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACvC,OAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,CAAC,EAAE,QAAQ,CAAC;AAAA,EACtD,SAAc,EAAE,KAAK,CAAC,OAAO,QAAQ,SAAS,MAAM,CAAC,EAAE,SAAS;AAAA,EAChE,gBAAgB,EAAE,OAAO,EAAE,SAAS;AACtC,CAAC;AAEM,SAAS,aAAa,SAA4E;AACvG,UAAQ,SAAS;AAAA,IACf,KAAK;AAAO,aAAO;AAAA,IACnB,KAAK;AAAQ,aAAO;AAAA,IACpB,KAAK;AAAS,aAAO;AAAA,IACrB,KAAK;AAAQ,aAAO;AAAA,IACpB;AAAS,aAAO;AAAA,EAClB;AACF;AAEO,IAAM,yBAAyB,EAAE,OAAO;AAAA,EAC7C,cAAe,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/B,UAAe,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/B,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,gBAAgB,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACzC,YAAe,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,GAAG,EAAE,QAAQ,EAAE;AAAA,EAC1D,iBAAiB,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EAC1C,cAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,eAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,UAAe,EAAE,QAAQ,EAAE,QAAQ,IAAI;AACzC,CAAC;AAEM,IAAM,0BAA0B,EAAE,OAAO;AAAA,EAC9C,OAAe,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/B,UAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,YAAe,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE;AAAA,EACzD,iBAAiB,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EAC1C,WAAe,EAAE,KAAK,CAAC,YAAY,cAAc,MAAM,CAAC,EAAE,QAAQ,uBAAuB;AAAA,EACzF,UAAe,EAAE,OAAO,EAAE,MAAM,SAAS,EAAE,SAAS;AAAA,EACpD,cAAe,EAAE,QAAQ,EAAE,SAAS;AAAA,EACpC,mBAAmB,EAAE,QAAQ,EAAE,SAAS;AAAA,EACxC,OAAe,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACxC,cAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,eAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,uBAAuB,EAAE,QAAQ,EAAE,SAAS;AAAA,EAC5C,UAAe,EAAE,QAAQ,EAAE,QAAQ,IAAI;AACzC,CAAC;AAEM,IAAM,mBAAmB,EAAE,OAAO;AAAA,EACvC,UAAa,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC7B,QAAa,EAAE,OAAO,EAAE,SAAS;AAAA,EACjC,aAAa,EAAE,OAAO,EAAE,SAAS;AAAA,EACjC,YAAa,EAAE,OAAO,EAAE,SAAS;AAAA,EACjC,YAAa,EAAE,OAAO,EAAE,SAAS;AACnC,CAAC;AAEM,IAAM,wBAAwB,EAAE,OAAO;AAAA,EAC5C,MAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,QAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,aAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,UAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,SAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,cAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,OAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,cAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,SAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,UAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,YAAc,EAAE,OAAO,EAAE,SAAS;AACpC,CAAC;AAEM,IAAM,wBAAwB,EAAE,OAAO;AAAA,EAC5C,KAAO,EAAE,OAAO;AAAA,EAChB,OAAO,EAAE,OAAO;AAClB,CAAC;AAEM,IAAM,2BAA2B,EAAE,OAAO;AAAA,EAC/C,iBAAiB,EAAE,MAAM,EAAE,OAAO;AAAA,IAChC,OAAO,EAAE,OAAO;AAAA,IAChB,OAAO,EAAE,OAAO;AAAA,EAClB,CAAC,CAAC;AAAA,EACF,cAAc,EAAE,MAAM,EAAE,OAAO;AAAA,IAC7B,OAAO,EAAE,OAAO;AAAA,IAChB,OAAO,EAAE,OAAO;AAAA,EAClB,CAAC,CAAC;AACJ,CAAC;AAEM,IAAM,0BAA0B,EAAE,OAAO;AAAA,EAC9C,UAAe,EAAE,OAAO;AAAA,EACxB,QAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,OAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,MAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,MAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,eAAe,EAAE,OAAO,EAAE,SAAS;AACrC,CAAC;AAEM,IAAM,8BAA8B,EAAE,OAAO;AAAA,EAClD,SAAW,EAAE,OAAO;AAAA,EACpB,WAAW,EAAE,OAAO;AACtB,CAAC;","names":[]}