mcp-scraper 0.40.1 → 0.40.2

package/dist/site-extract-repository-2SMMFKKL.js

@@ -0,0 +1,62 @@
+import {
+  abandonExtractSettlement,
+  claimFailedExtractJobForRefinalize,
+  completeExtractJob,
+  countSuccessfulPages,
+  createExtractJob,
+  createOrGetExtractJob,
+  extractJobLimitInfo,
+  failExtractJob,
+  failStaleRunningExtractJob,
+  failUnfundedExtractJob,
+  finishExtractJob,
+  getExtractJob,
+  getExtractJobByIdempotencyKey,
+  getExtractedImageLinks,
+  getExtractedPages,
+  getExtractedUrls,
+  listExtractJobs,
+  listFundedPendingExtractJobs,
+  listStaleRunningExtractJobs,
+  listUnsettledExtractJobs,
+  markExtractJobDispatchAttempt,
+  recordExtractJobDispatchFailure,
+  recordExtractSettlementFailure,
+  saveExtractPages,
+  setExtractJobTotal,
+  settleExtractJob,
+  terminalExtractJobStatus
+} from "./chunk-X2LKCX6H.js";
+import "./chunk-SBLGBZZB.js";
+import "./chunk-44HZLHDV.js";
+import "./chunk-GUVKHCKE.js";
+export {
+  abandonExtractSettlement,
+  claimFailedExtractJobForRefinalize,
+  completeExtractJob,
+  countSuccessfulPages,
+  createExtractJob,
+  createOrGetExtractJob,
+  extractJobLimitInfo,
+  failExtractJob,
+  failStaleRunningExtractJob,
+  failUnfundedExtractJob,
+  finishExtractJob,
+  getExtractJob,
+  getExtractJobByIdempotencyKey,
+  getExtractedImageLinks,
+  getExtractedPages,
+  getExtractedUrls,
+  listExtractJobs,
+  listFundedPendingExtractJobs,
+  listStaleRunningExtractJobs,
+  listUnsettledExtractJobs,
+  markExtractJobDispatchAttempt,
+  recordExtractJobDispatchFailure,
+  recordExtractSettlementFailure,
+  saveExtractPages,
+  setExtractJobTotal,
+  settleExtractJob,
+  terminalExtractJobStatus
+};
+//# sourceMappingURL=site-extract-repository-2SMMFKKL.js.map

package/dist/site-extract-repository-2SMMFKKL.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}

package/dist/worker-FXAGFYOE.js

@@ -0,0 +1,142 @@
+import {
+  classifyHarvestProblem,
+  createHarvestAttemptRecorder,
+  harvestProblemResponse,
+  serializeHarvestProblem
+} from "./chunk-NVXNEOUQ.js";
+import {
+  harvest
+} from "./chunk-7XBBFBYY.js";
+import {
+  browserServiceApiKey,
+  runWithCostContext
+} from "./chunk-3PIWJS6Y.js";
+import "./chunk-K443GQY5.js";
+import "./chunk-CB5C3BPB.js";
+import {
+  MC_COSTS,
+  serpActualCostMc
+} from "./chunk-SBLGBZZB.js";
+import "./chunk-44HZLHDV.js";
+import {
+  claimPendingJob,
+  completeJob,
+  creditMc,
+  debitMc,
+  failJob,
+  listHarvestAttempts
+} from "./chunk-GUVKHCKE.js";
+
+// src/api/webhook.ts
+async function deliverWebhook(url, payload, retries = 3) {
+  for (let attempt = 1; attempt <= retries; attempt++) {
+    try {
+      const res = await fetch(url, {
+        method: "POST",
+        headers: { "content-type": "application/json" },
+        body: JSON.stringify(payload),
+        signal: AbortSignal.timeout(1e4)
+      });
+      if (res.ok) return;
+      console.warn(`[webhook] attempt ${attempt} \u2192 ${res.status} from ${url}`);
+    } catch (err) {
+      console.warn(`[webhook] attempt ${attempt} failed:`, err instanceof Error ? err.message : err);
+    }
+    if (attempt < retries) await new Promise((r) => setTimeout(r, 1e3 * attempt * 2));
+  }
+  console.error(`[webhook] gave up after ${retries} attempts for ${url}`);
+}
+
+// src/api/worker.ts
+var MAX_CONCURRENT = 2;
+var running = 0;
+function countPaaQuestions(result) {
+  if (!result || typeof result !== "object") return 0;
+  const value = result;
+  if (typeof value.totalQuestions === "number") return value.totalQuestions;
+  return Array.isArray(value.flat) ? value.flat.length : 0;
+}
+function paaCostForQuestionCount(questionCount) {
+  return MC_COSTS.paa_base + Math.max(1, questionCount) * MC_COSTS.paa;
+}
+async function processJob(job) {
+  running++;
+  try {
+    const opts = typeof job.options === "string" ? JSON.parse(job.options) : job.options;
+    const headlessSentOut = { value: null };
+    const result = await runWithCostContext(
+      { op: opts.serpOnly ? "serp" : "paa", userId: Number(job.user_id), headlessSentOut },
+      () => harvest({
+        ...opts,
+        kernelApiKey: browserServiceApiKey(),
+        headless: true,
+        format: "json",
+        outputDir: "/tmp/paa-output-api",
+        onAttemptEvent: createHarvestAttemptRecorder(job.id, job.user_id)
+      })
+    );
+    await completeJob(job.id, result);
+    const attempts = await listHarvestAttempts(job.id, job.user_id);
+    if (!opts.serpOnly && typeof opts.billingHoldMc === "number") {
+      const actualCost = paaCostForQuestionCount(countPaaQuestions(result));
+      const diff = opts.billingHoldMc - actualCost;
+      if (diff > 0) await creditMc(job.user_id, diff, "paa_refund", "overestimate refund");
+      else if (diff < 0) await debitMc(job.user_id, -diff, "paa", opts.query ?? job.query);
+    } else if (opts.serpOnly && typeof opts.billingHoldMc === "number") {
+      const actualCost = serpActualCostMc(headlessSentOut.value);
+      const diff = opts.billingHoldMc - actualCost;
+      if (diff > 0) await creditMc(job.user_id, diff, "serp_refund", "headless-mode pricing settle");
+      else if (diff < 0) await debitMc(job.user_id, -diff, "serp", opts.query ?? job.query);
+    }
+    if (job.callback_url) {
+      await deliverWebhook(job.callback_url, { job_id: job.id, status: "done", result, attempts });
+    }
+  } catch (err) {
+    const problem = classifyHarvestProblem(err);
+    await failJob(job.id, serializeHarvestProblem(problem));
+    const attempts = await listHarvestAttempts(job.id, job.user_id);
+    try {
+      const opts = typeof job.options === "string" ? JSON.parse(job.options) : job.options;
+      if (typeof opts.billingHoldMc === "number" && opts.billingHoldMc > 0) {
+        await creditMc(job.user_id, opts.billingHoldMc, "refund", "failed call");
+      }
+    } catch {
+    }
+    if (job.callback_url) {
+      await deliverWebhook(job.callback_url, { job_id: job.id, status: "failed", ...harvestProblemResponse(problem), attempts });
+    }
+  } finally {
+    running--;
+  }
+}
+async function tickOnce() {
+  const job = await claimPendingJob();
+  if (!job) return { claimed: false };
+  const startedAt = Date.now();
+  await processJob(job);
+  return { claimed: true, jobId: job.id, completed: true, durationMs: Date.now() - startedAt };
+}
+async function drainQueue(budget) {
+  const results = [];
+  for (let i = 0; i < budget.maxJobs; i++) {
+    if (Date.now() >= budget.deadlineMs) break;
+    const r = await tickOnce();
+    results.push(r);
+    if (!r.claimed) break;
+  }
+  return results;
+}
+function startWorker() {
+  setInterval(async () => {
+    if (running >= MAX_CONCURRENT) return;
+    const job = await claimPendingJob();
+    if (job) void processJob(job);
+  }, 2e3);
+  console.log(`[worker] started \u2014 polling every 2s, max ${MAX_CONCURRENT} concurrent`);
+}
+export {
+  drainQueue,
+  startWorker,
+  tickOnce
+};
+//# sourceMappingURL=worker-FXAGFYOE.js.map

package/dist/worker-FXAGFYOE.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/api/webhook.ts","../src/api/worker.ts"],"sourcesContent":["export async function deliverWebhook(url: string, payload: object, retries = 3): Promise<void> {\n  for (let attempt = 1; attempt <= retries; attempt++) {\n    try {\n      const res = await fetch(url, {\n        method: 'POST',\n        headers: { 'content-type': 'application/json' },\n        body: JSON.stringify(payload),\n        signal: AbortSignal.timeout(10_000),\n      })\n      if (res.ok) return\n      console.warn(`[webhook] attempt ${attempt} → ${res.status} from ${url}`)\n    } catch (err) {\n      console.warn(`[webhook] attempt ${attempt} failed:`, err instanceof Error ? err.message : err)\n    }\n    if (attempt < retries) await new Promise((r) => setTimeout(r, 1000 * attempt * 2))\n  }\n  console.error(`[webhook] gave up after ${retries} attempts for ${url}`)\n}\n","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'\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, { job_id: job.id, status: 'done', result, attempts })\n    }\n  } catch (err) {\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, { job_id: job.id, status: 'failed', ...harvestProblemResponse(problem), attempts })\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":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,eAAsB,eAAe,KAAa,SAAiB,UAAU,GAAkB;AAC7F,WAAS,UAAU,GAAG,WAAW,SAAS,WAAW;AACnD,QAAI;AACF,YAAM,MAAM,MAAM,MAAM,KAAK;AAAA,QAC3B,QAAQ;AAAA,QACR,SAAS,EAAE,gBAAgB,mBAAmB;AAAA,QAC9C,MAAM,KAAK,UAAU,OAAO;AAAA,QAC5B,QAAQ,YAAY,QAAQ,GAAM;AAAA,MACpC,CAAC;AACD,UAAI,IAAI,GAAI;AACZ,cAAQ,KAAK,qBAAqB,OAAO,WAAM,IAAI,MAAM,SAAS,GAAG,EAAE;AAAA,IACzE,SAAS,KAAK;AACZ,cAAQ,KAAK,qBAAqB,OAAO,YAAY,eAAe,QAAQ,IAAI,UAAU,GAAG;AAAA,IAC/F;AACA,QAAI,UAAU,QAAS,OAAM,IAAI,QAAQ,CAAC,MAAM,WAAW,GAAG,MAAO,UAAU,CAAC,CAAC;AAAA,EACnF;AACA,UAAQ,MAAM,2BAA2B,OAAO,iBAAiB,GAAG,EAAE;AACxE;;;ACIA,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,EAAE,QAAQ,IAAI,IAAI,QAAQ,QAAQ,QAAQ,SAAS,CAAC;AAAA,IAC7F;AAAA,EACF,SAAS,KAAK;AACZ,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,EAAE,QAAQ,IAAI,IAAI,QAAQ,UAAU,GAAG,uBAAuB,OAAO,GAAG,SAAS,CAAC;AAAA,IAC3H;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

@@ -1,6 +1,6 @@
 {
   "name": "mcp-scraper",
-  "version": "0.40.1",
+  "version": "0.40.2",
   "description": "MCP server for MCP Scraper web intelligence tools",
   "type": "module",
   "main": "./dist/index.cjs",
@@ -62,7 +62,9 @@
     "email:workshop-reminders": "node --import tsx scripts/email/refresh-workshop-reminders.ts",
     "email:workshop-reminders:apply": "node --import tsx scripts/email/refresh-workshop-reminders.ts --apply",
     "billing:preview-credit-lot-repair": "node --import tsx scripts/billing/preview-credit-lot-repair.ts",
-    "billing:apply-credit-lot-repair": "node --import tsx scripts/billing/apply-credit-lot-repair.ts"
+    "billing:apply-credit-lot-repair": "node --import tsx scripts/billing/apply-credit-lot-repair.ts",
+    "prepack": "npm run build && npm run package:verify-entrypoints",
+    "package:verify-entrypoints": "node scripts/verify-package-entrypoints.mjs"
   },
   "dependencies": {
     "@anthropic-ai/sdk": "^0.96.0",