mcp-scraper 0.96.0 → 0.96.1

package/dist/{chunk-LFRAA3MY.js → chunk-RN4VG3VH.js}

@@ -1,4 +1,4 @@
-import{$ as E,B as R,D as S,La as C,Ma as P,Y as v,_ as h,ha as T,ja as w,ka as _,la as y,x as u,y as f,z as g}from"./chunk-ARSQZCTJ.js";import{a as x}from"./chunk-VAC3SQ5T.js";import{a as c}from"./chunk-7UHDOIUE.js";import{readFileSync as L}from"fs";import{homedir as B}from"os";import{join as U}from"path";import{serveStdio as Y}from"@modelcontextprotocol/server/stdio";import{McpServer as J}from"@modelcontextprotocol/server";var I=new Map(_.map(r=>[r.upstreamName,r.id]));function N(r){let s=r.split(`
+import{$ as E,B as R,D as S,La as C,Ma as P,Y as v,_ as h,ha as T,ja as w,ka as _,la as y,x as u,y as f,z as g}from"./chunk-RIHI3KH7.js";import{a as x}from"./chunk-FB4KBB5W.js";import{a as c}from"./chunk-CVQUHVHX.js";import{readFileSync as L}from"fs";import{homedir as B}from"os";import{join as U}from"path";import{serveStdio as Y}from"@modelcontextprotocol/server/stdio";import{McpServer as J}from"@modelcontextprotocol/server";var I=new Map(_.map(r=>[r.upstreamName,r.id]));function N(r){let s=r.split(`
 `).filter(t=>t.startsWith("data:")).map(t=>t.slice(5).trim()).filter(Boolean);if(!s.length)return JSON.parse(r);for(let t of s){let e=JSON.parse(t);if(e.result||e.error)return e}throw new Error("hosted MCP returned no JSON-RPC result")}function l(r){return{content:[{type:"text",text:JSON.stringify({ok:!1,error:r})}],isError:!0}}var p=class{baseUrl;apiKey;constructor(s,t){this.baseUrl=s.replace(/\/$/,""),this.apiKey=t}async callMemoryTool(s,t){let e=I.get(s);if(!e)return l(`unknown memory tool: ${s}`);try{let n=await fetch(`${this.baseUrl}/mcp`,{method:"POST",headers:{"content-type":"application/json",accept:"application/json, text/event-stream","x-api-key":this.apiKey},body:JSON.stringify({jsonrpc:"2.0",id:`memory:${e}`,method:"tools/call",params:{name:e,arguments:t}})}),i=await n.text();if(!n.ok)return l(`hosted memory ${e} failed (HTTP ${n.status})`);let o=N(i);return o.error?l(o.error.message??`hosted memory ${e} failed`):o.result??l(`hosted memory ${e} returned no result`)}catch(n){return l(n instanceof Error?n.message:`hosted memory ${e} call failed`)}}};function M(r,s){if(s.length===0)throw new Error("Restricted MCP tool allowlist must contain at least one tool name");let t=new Set;for(let o of s){if(typeof o!="string"||o.trim()!==o||o.length===0)throw new Error("Restricted MCP tool allowlist contains a malformed tool name");if(t.has(o))throw new Error(`Restricted MCP tool allowlist contains duplicate tool name: ${o}`);t.add(o)}let e=new Set,n=r,i=n.registerTool.bind(r);return n.registerTool=(o,a,m)=>{if(t.has(o))return e.add(o),i(o,a,m)},{assertComplete(){let o=[...t].filter(a=>!e.has(a));if(o.length>0)throw new Error(`Restricted MCP tool allowlist contains unknown or unavailable tool names: ${o.join(", ")}`)},registeredToolNames(){return[...e]}}}import{Client as b,StreamableHTTPClientTransport as k}from"@modelcontextprotocol/client";var $=new Set(["artifact_read","artifact_describe","artifact_download","report_artifact_read"]),A=()=>({content:[{type:"text",text:"Retained file storage is unavailable. Preserve the artifactId and retry retrieval; do not repeat acquisition."}],isError:!0});function O(r,s,t={}){return async(e,n)=>{if(!$.has(e))return A();let i=new b({name:"mcp-scraper-artifact-client",version:c},{versionNegotiation:{mode:"auto"}}),o=AbortSignal.timeout(3e4);try{let a=new k(new URL(`${r.replace(/\/$/,"")}/mcp`),{requestInit:{signal:o,headers:{Authorization:`Bearer ${s}`,"x-api-key":s,...t.localNetworkAccess?{"x-mcp-scraper-local-network":"1"}:{}}}});return await i.connect(a,{signal:o,timeout:3e4}),await i.callTool({name:e,arguments:n},{signal:o,timeout:3e4})}catch{return A()}finally{await i.close().catch(()=>{})}}}var G=new Set(["paa","serp","browser-agent","scheduled-results","memory"]);function K(){let s=[process.env.MCP_SCRAPER_KEY_PATH?.trim(),U(B(),".mcp-scraper-key")].filter(Boolean);for(let t of s)try{let e=L(t,"utf8").trim();if(e)return e}catch{}}function j(r,s={}){let t=s.toolsets??G,e=process.env.MCP_SCRAPER_BASE_URL?.trim()||process.env.MCP_BASE_URL?.trim()||"https://mcpscraper.dev",n=R(),i=S({deploymentProfile:n,transportProfile:"stdio",baseUrl:e,explicitlyEnabled:process.env.MCP_SCRAPER_ALLOW_PRIVATE_NETWORK==="1"}),o=process.env.BROWSER_AGENT_CONSOLE_URL?.trim()||e,a=new J({name:"mcp-scraper",version:c},{instructions:f,cacheHints:{"server/discover":{ttlMs:3e5,cacheScope:"private"},"tools/list":{ttlMs:3e5,cacheScope:"private"},"prompts/list":{ttlMs:3e5,cacheScope:"private"},"resources/list":{ttlMs:3e5,cacheScope:"private"},"resources/templates/list":{ttlMs:3e5,cacheScope:"private"},"resources/read":{ttlMs:6e4,cacheScope:"private"}}});g(a);let m=s.allowedToolNames?M(a,s.allowedToolNames):void 0,d=s.httpExecutor??new T(e,r,{localNetworkAccess:i,analyticsIdentity:()=>E(a.server.getClientVersion()?.name)});return t.has("paa")&&h(a,d,{ownerId:v(r),deploymentProfile:n,transportProfile:"stdio",baseUrl:e,localNetworkAccess:i,taskHandleSecret:r,remoteArtifactAccess:O(e,r,{localNetworkAccess:i})}),t.has("serp")&&u(a,d,{exposeDevelopmentDiagnostics:n==="development"||n==="test"}),t.has("browser-agent")&&w(a,{baseUrl:e,apiKey:r,consoleBaseUrl:o}),t.has("scheduled-results")&&P(a,new C(e,r)),t.has("memory")&&y(a,new p(e,r)),m?.assertComplete(),a}function Re(r={}){let s=process.argv.includes("--stdio")||process.env.MCP_SCRAPER_FORCE_STDIO==="1",t=!!(process.stdin.isTTY&&process.stdout.isTTY),e=process.argv.includes("--help")||process.argv.includes("-h");if(!s&&(t||e)){let i=process.argv.includes("--no-color")||process.env.NO_COLOR!==void 0||process.env.FORCE_COLOR==="0"||!process.stdout.isTTY;process.stdout.write(x({version:c,color:!i,apiKeyConfigured:!!process.env.MCP_SCRAPER_API_KEY?.trim()})),process.exit(0)}let n=(process.env.MCP_SCRAPER_API_KEY??process.env.MCP_SCRAPER_KEY??process.env.MCP_API_KEY??K())?.trim();n||(process.stderr.write(`MCP_SCRAPER_API_KEY env var or ~/.mcp-scraper-key is required
 `),process.exit(1)),Y(()=>j(n,r),{legacy:"serve",onerror(i){process.stderr.write(`${i.message}
 `)}})}export{Re as a};

package/dist/{chunk-SNJR5N36.js → chunk-RYX4FJVH.js}

@@ -1 +1 @@
-import{c as j}from"./chunk-DGAW7FD7.js";import{k as X}from"./chunk-FRXKMOXI.js";import{c as A,d as $,g as b,h as C,n as K,o as U,p as G,q as M,r as D,s as W,y as q,z as F}from"./chunk-6SNOUSSZ.js";import{a as B,b as L,c as H}from"./chunk-QER2ELZP.js";import{b as k,c as S,d as y,e as x,f as O,g as I,h,i as m,k as E,o as N}from"./chunk-HUNSEGUN.js";import{h as v,i as z}from"./chunk-C7MGJUTU.js";import{a as d,b as J,c as T,d as R}from"./chunk-HNNYXMSI.js";import{randomUUID as oe}from"crypto";import{createHash as Q,randomUUID as V}from"crypto";function _(e,i=new Set){if(e===null||typeof e=="string"||typeof e=="boolean"||typeof e=="number"&&Number.isFinite(e))return e;if(typeof e!="object"||!e)throw new d("Operation inputs must be finite JSON data");if(i.has(e))throw new d("Operation inputs cannot be cyclic");if(!Array.isArray(e)&&![Object.prototype,null].includes(Object.getPrototypeOf(e)))throw new d("Operation inputs must be plain JSON data");i.add(e);try{return Array.isArray(e)?e.map(t=>_(t,i)):Object.fromEntries(Object.keys(e).sort().map(t=>[t,_(e[t],i)]))}finally{i.delete(e)}}async function P(e){if(!Number.isSafeInteger(e.userId)||e.userId<=0)throw new d("Authenticated operation principal is required");if(e.jobId!=null&&(!e.jobId.trim()||e.jobId.length>512))throw new d("Operation job ID is invalid");let i=e.tool.trim();if(!i||i.length>100)throw new d("Operation tool is required and bounded");if(e.idempotencyKey!=null&&(!e.idempotencyKey.trim()||e.idempotencyKey.length>512))throw new d("Operation idempotency key is invalid");let t=e.leaseMs??3e5;if(!Number.isSafeInteger(t)||t<1e3||t>36e5)throw new d("Execution lease must be between one second and one hour");let n=JSON.stringify(_(e.normalizedInput));if(Buffer.byteLength(n)>64*1024)throw new d("Operation inputs exceed 64 KiB");let r=e.idempotencyKey?.trim()??V();return E({userId:e.userId,ownerScope:`user:${e.userId}`,tool:i,jobId:e.jobId??null,normalizedFlags:e.normalizedInput,idempotencyKey:r,requestFingerprint:Q("sha256").update(n).digest("hex")},V(),t,e.storeGate)}async function Y(e){let i=JSON.stringify(_(e.normalizedInput));if(!Number.isSafeInteger(e.userId)||e.userId<=0||Buffer.byteLength(i)>64*1024)throw new d("Invalid operation receipt principal/input");let{readOperationIntentResult:t}=await import("./repository-4EC5EXVN.js"),{operationIdentityHash:n}=await import("./contracts-GUXGFCVV.js");return t(e.userId,n(`user:${e.userId}\0${e.tool}\0${e.idempotencyKey}`),Q("sha256").update(i).digest("hex"))}async function we(e){let i=Date.now()+27e4,t=e.idempotencyKey?.trim()??oe();if(!t||t.length>512)return{status:400,body:{error_code:"invalid_request",message:"Idempotency-Key must contain 1 to 512 characters"}};let n=(a,c,p)=>({...a,body:{...a.body,operation:{runId:c,idempotencyKey:t,state:p}}}),r=(a,c,p)=>n({status:c==="pending"?202:503,headers:{"Retry-After":"2"},body:{...v({errorCode:"response_lost",retryable:!0,chargeStatus:p,details:{retry_guidance:"Retry the identical request with the returned idempotencyKey. Pending work is not started again; a saved response replays. Unknown work permits compensation only."}}),message:c==="pending"?"This operation is still running. Retry with the same idempotencyKey.":"The execution outcome is unknown. Retry with the same idempotencyKey for result or compensation recovery."}},a,c),s;try{s=await P({...e,idempotencyKey:t,leaseMs:3e5})}catch(a){return a instanceof d?n({status:400,body:{error_code:"invalid_request",message:a.message,retryable:!1}},null,"not_started"):a instanceof J?n({status:409,body:{error_code:"idempotency_conflict",message:a.message,retryable:!1}},null,"conflict"):r(null,"unknown")}let o=s.kind==="claimed"?s.claim.runId:s.runId;if(s.kind==="replay"){let a=JSON.parse(s.result.json);return!a||typeof a.status!="number"||!a.body||typeof a.body!="object"?r(o,"unknown"):n(a,o,"replayed")}if(s.kind==="pending")return r(o,"pending");if(s.kind==="unknown")try{let a=await y(o,e.userId);if(!a)return r(o,"unknown");let c=await e.recover(a);if(c.recover)return await h(a),n(c.reply,o,"unknown");let p=n(c.reply,o,"recovered_unknown"),l={json:JSON.stringify(p),status:"unknown"};return await O(a,l)?p:r(o,"unknown")}catch{return r(o,"unknown","refund_pending")}if(s.kind!=="claimed")return r(o,"unknown");let u=s.claim,f;try{if(f=await e.prepare(),f.rejection)return await I(u),n(f.rejection,o,"not_started");if(!await k(u))return r(o,"unknown");let a={claim:u,deadlineAt:i,assertActive:async()=>{if(B(i),!await m(u))throw Error("Execution ownership is no longer current")}},c;try{c=await e.execute(a,f.value)}catch{return await h(u),r(o,"unknown","refund_pending")}if(c.recover)return await h(u),n(c.reply,o,"unknown");let p=n(c.reply,o,c.reply.status<400?"succeeded":"failed");try{return await x(u,{json:JSON.stringify(p),status:p.status<400?"succeeded":"failed"})?p:r(o,"unknown")}catch{return await h(u).catch(()=>{}),r(o,"unknown")}}catch{return await I(u).catch(()=>{}),r(o,"unknown")}finally{await f?.release().catch(()=>{})}}var g=3e5;function ee(e){if(!e?.trim()||e.length>512)throw new d("A bounded framework workflow ID is required");return T(`workflow:${e}`)}async function ye(e){if(!Number.isSafeInteger(e.userId)||e.userId<=0)throw new d("Workflow principal is required");let i=ee(e.workflowId),t=await E(e,i,g);if(t.kind==="claimed"){let n;try{n=await k(t.claim)}catch(r){if(!await m(t.claim))throw r;n=!0}if(!n)throw Error("workflow_execution_start_unknown");return t}if(t.kind==="pending"){let n={runId:t.runId,userId:e.userId,token:i};if(await m(n)&&await S(n,g))return{kind:"claimed",claim:n}}return t}var te=e=>`${e.workflow.runId}:${e.phase}`;async function he(e){if(!await S(e.workflow,g))throw Error("workflow_execution_not_current");let i=te(e),t=await P({userId:e.workflow.userId,tool:e.tool,jobId:e.jobId,normalizedInput:e.normalizedInput,idempotencyKey:i,leaseMs:g});if(t.kind==="replay"){if(t.result.status!=="succeeded")throw Error(`operation_phase_unknown:${t.runId}`);return JSON.parse(t.result.json)}if(t.kind!=="claimed")throw Error(`operation_phase_${t.kind}:${t.runId}`);let n=t.claim;try{await e.prepare()}catch(r){throw await I(n).catch(()=>{}),r}try{let r;try{r=await k(n)}catch(a){if(!await m(n))throw a;r=!0}if(!r)throw Error("operation_phase_start_unknown");let s=Date.now()+g,o=async()=>{if(Date.now()>=s)throw new L(e.phase);if(!await m(e.workflow)||!await m(n))throw Error("operation_phase_claim_not_current")};await o();let u=await H(s,()=>e.execute({claim:n,deadlineAt:s,assertActive:o}),e.phase);await o();let f=JSON.stringify(u);if(f==null||!await x(n,{status:"succeeded",json:f}))throw Error("operation_phase_publication_unknown");return u}catch(r){await h(n).catch(()=>{});let s=await y(n.runId,n.userId).catch(()=>null);throw s&&await O(s,{status:"unknown",json:JSON.stringify({phase:e.phase,outcome:"unknown"})}).catch(()=>{}),r}}async function be(e,i){let t=await Y({userId:e.workflow.userId,tool:e.tool,idempotencyKey:te(e),normalizedInput:e.normalizedInput});if(!t||t.result.status!=="succeeded"||R({value:JSON.parse(t.result.json)})!==R({value:i}))throw Error("operation_phase_cache_receipt_unknown")}async function ge(e,i,t){let n={runId:e,userId:i,token:ee(t)};if(await m(n))return{claim:n,recovery:!1};let r=await y(e,i);return r?{claim:r,recovery:!0}:null}async function re(e,i,t){let n={json:JSON.stringify(i),status:e.recovery?"unknown":t};if(!(e.recovery?await O(e.claim,n,e.settlementGate):await x(e.claim,n)))throw Error("workflow_result_publication_unknown")}async function je(e){if(!await b(e))return{kind:"missing",body:{error:"job not found"}};let t=await M(e);if(!t)return{kind:"conflict",body:{error:"Only failed or expired recovery jobs can be re-finalized.",status:(await b(e))?.status??"missing"}};let n=t.executionGeneration;if(!await D(e,n))return{kind:"conflict",body:{error:"job state changed during re-finalization claim"}};let r;try{r=await X(t,{})}catch(l){throw await F(e,l instanceof Error?l.message:String(l),void 0,n).catch(()=>!1),l}let s=A(t),o=$(t,s.creditTruncated),u=o==="failed"?`No pages were extracted successfully (${t.failedUrls} failed, ${t.remainingUrls} remaining).`:o==="partial"?`${t.successfulUrls} pages succeeded; ${t.failedUrls} failed and ${t.remainingUrls} remain.${s.creditTruncated?` Credit availability limited this request from ${s.requestedMaxPages} to ${s.effectiveMaxPages} pages.`:""}`:null,f;try{f=await q(e,r,o,u,!1,null,n)}catch(l){let w=await b(e).catch(()=>null);if(!w||w.id!==e||w.executionGeneration!==n||w.status!==o||JSON.stringify(w.artifacts)!==JSON.stringify(r)||w.error!==(u?z(u).slice(0,2e3):null)||w.publicError!==null)throw l;f=!0}if(!f)return{kind:"conflict",body:{error:"job state changed during re-finalization"}};let a=await b(e);if(!a||a.executionGeneration!==n||a.status!==o)return{kind:"conflict",body:{error:"job state changed before re-finalization settlement"}};let c="already_settled_or_refunded";if(a.billedMc==null&&a.userId!=null){let l=await j(a);if(!l)throw Error("extract_settlement_acknowledgment_unknown");c=`settled: billed ${l.netChargeMc} mc, refunded ${l.refundMc} mc${l.expiredRefundMc?`, expired refund ${l.expiredRefundMc} mc`:""}`}let p=r.find(l=>l.contentType==="application/zip"||l.filename?.endsWith(".zip"));return{kind:"complete",body:{ok:!0,jobId:e,artifacts:r.length,settlement:c,bundleUrl:p?.url??null,bundleBytes:p?.bytes??null}}}async function Pe(e=25){let i=await U(e);for(let r of i)await G(r.id).catch(()=>!1);let t=await C(e),n={checked:new Set([...i,...t].map(r=>r.id)).size,settled:0,failed:0};for(let r of t){if(r.userId==null){await W(r.id,"Settlement abandoned because the legacy job has no user owner.",r.executionGeneration??0),n.failed+=1;continue}try{if(!await j(r))throw Error("extract_settlement_acknowledgment_unknown");if(r.executionGeneration===0&&await N(`extract-site:${r.id}`,r.userId)){let o=await y(`extract-site:${r.id}`,r.userId);o&&await re({claim:o,recovery:!0},{jobId:r.id,status:r.status,artifacts:r.artifacts??[]},"unknown")}n.settled+=1}catch(s){n.failed+=1;let o=s instanceof Error?s.message:String(s);await K(r.id,o,r.executionGeneration??0).catch(()=>{}),console.warn("[site-extract/reconcile]",r.id,o)}}return n}export{we as a,g as b,ee as c,ye as d,he as e,be as f,ge as g,re as h,je as i,Pe as j};
+import{c as j}from"./chunk-BPKTUCV7.js";import{k as X}from"./chunk-ZTMQEDIB.js";import{c as A,d as $,g as b,h as C,n as K,o as U,p as G,q as M,r as D,s as W,y as q,z as F}from"./chunk-7WPCGRD7.js";import{a as B,b as L,c as H}from"./chunk-QER2ELZP.js";import{b as k,c as S,d as y,e as x,f as O,g as I,h,i as m,k as E,o as N}from"./chunk-HUNSEGUN.js";import{h as v,i as z}from"./chunk-C7MGJUTU.js";import{a as d,b as J,c as T,d as R}from"./chunk-HNNYXMSI.js";import{randomUUID as oe}from"crypto";import{createHash as Q,randomUUID as V}from"crypto";function _(e,i=new Set){if(e===null||typeof e=="string"||typeof e=="boolean"||typeof e=="number"&&Number.isFinite(e))return e;if(typeof e!="object"||!e)throw new d("Operation inputs must be finite JSON data");if(i.has(e))throw new d("Operation inputs cannot be cyclic");if(!Array.isArray(e)&&![Object.prototype,null].includes(Object.getPrototypeOf(e)))throw new d("Operation inputs must be plain JSON data");i.add(e);try{return Array.isArray(e)?e.map(t=>_(t,i)):Object.fromEntries(Object.keys(e).sort().map(t=>[t,_(e[t],i)]))}finally{i.delete(e)}}async function P(e){if(!Number.isSafeInteger(e.userId)||e.userId<=0)throw new d("Authenticated operation principal is required");if(e.jobId!=null&&(!e.jobId.trim()||e.jobId.length>512))throw new d("Operation job ID is invalid");let i=e.tool.trim();if(!i||i.length>100)throw new d("Operation tool is required and bounded");if(e.idempotencyKey!=null&&(!e.idempotencyKey.trim()||e.idempotencyKey.length>512))throw new d("Operation idempotency key is invalid");let t=e.leaseMs??3e5;if(!Number.isSafeInteger(t)||t<1e3||t>36e5)throw new d("Execution lease must be between one second and one hour");let n=JSON.stringify(_(e.normalizedInput));if(Buffer.byteLength(n)>64*1024)throw new d("Operation inputs exceed 64 KiB");let r=e.idempotencyKey?.trim()??V();return E({userId:e.userId,ownerScope:`user:${e.userId}`,tool:i,jobId:e.jobId??null,normalizedFlags:e.normalizedInput,idempotencyKey:r,requestFingerprint:Q("sha256").update(n).digest("hex")},V(),t,e.storeGate)}async function Y(e){let i=JSON.stringify(_(e.normalizedInput));if(!Number.isSafeInteger(e.userId)||e.userId<=0||Buffer.byteLength(i)>64*1024)throw new d("Invalid operation receipt principal/input");let{readOperationIntentResult:t}=await import("./repository-4EC5EXVN.js"),{operationIdentityHash:n}=await import("./contracts-GUXGFCVV.js");return t(e.userId,n(`user:${e.userId}\0${e.tool}\0${e.idempotencyKey}`),Q("sha256").update(i).digest("hex"))}async function we(e){let i=Date.now()+27e4,t=e.idempotencyKey?.trim()??oe();if(!t||t.length>512)return{status:400,body:{error_code:"invalid_request",message:"Idempotency-Key must contain 1 to 512 characters"}};let n=(a,c,p)=>({...a,body:{...a.body,operation:{runId:c,idempotencyKey:t,state:p}}}),r=(a,c,p)=>n({status:c==="pending"?202:503,headers:{"Retry-After":"2"},body:{...v({errorCode:"response_lost",retryable:!0,chargeStatus:p,details:{retry_guidance:"Retry the identical request with the returned idempotencyKey. Pending work is not started again; a saved response replays. Unknown work permits compensation only."}}),message:c==="pending"?"This operation is still running. Retry with the same idempotencyKey.":"The execution outcome is unknown. Retry with the same idempotencyKey for result or compensation recovery."}},a,c),s;try{s=await P({...e,idempotencyKey:t,leaseMs:3e5})}catch(a){return a instanceof d?n({status:400,body:{error_code:"invalid_request",message:a.message,retryable:!1}},null,"not_started"):a instanceof J?n({status:409,body:{error_code:"idempotency_conflict",message:a.message,retryable:!1}},null,"conflict"):r(null,"unknown")}let o=s.kind==="claimed"?s.claim.runId:s.runId;if(s.kind==="replay"){let a=JSON.parse(s.result.json);return!a||typeof a.status!="number"||!a.body||typeof a.body!="object"?r(o,"unknown"):n(a,o,"replayed")}if(s.kind==="pending")return r(o,"pending");if(s.kind==="unknown")try{let a=await y(o,e.userId);if(!a)return r(o,"unknown");let c=await e.recover(a);if(c.recover)return await h(a),n(c.reply,o,"unknown");let p=n(c.reply,o,"recovered_unknown"),l={json:JSON.stringify(p),status:"unknown"};return await O(a,l)?p:r(o,"unknown")}catch{return r(o,"unknown","refund_pending")}if(s.kind!=="claimed")return r(o,"unknown");let u=s.claim,f;try{if(f=await e.prepare(),f.rejection)return await I(u),n(f.rejection,o,"not_started");if(!await k(u))return r(o,"unknown");let a={claim:u,deadlineAt:i,assertActive:async()=>{if(B(i),!await m(u))throw Error("Execution ownership is no longer current")}},c;try{c=await e.execute(a,f.value)}catch{return await h(u),r(o,"unknown","refund_pending")}if(c.recover)return await h(u),n(c.reply,o,"unknown");let p=n(c.reply,o,c.reply.status<400?"succeeded":"failed");try{return await x(u,{json:JSON.stringify(p),status:p.status<400?"succeeded":"failed"})?p:r(o,"unknown")}catch{return await h(u).catch(()=>{}),r(o,"unknown")}}catch{return await I(u).catch(()=>{}),r(o,"unknown")}finally{await f?.release().catch(()=>{})}}var g=3e5;function ee(e){if(!e?.trim()||e.length>512)throw new d("A bounded framework workflow ID is required");return T(`workflow:${e}`)}async function ye(e){if(!Number.isSafeInteger(e.userId)||e.userId<=0)throw new d("Workflow principal is required");let i=ee(e.workflowId),t=await E(e,i,g);if(t.kind==="claimed"){let n;try{n=await k(t.claim)}catch(r){if(!await m(t.claim))throw r;n=!0}if(!n)throw Error("workflow_execution_start_unknown");return t}if(t.kind==="pending"){let n={runId:t.runId,userId:e.userId,token:i};if(await m(n)&&await S(n,g))return{kind:"claimed",claim:n}}return t}var te=e=>`${e.workflow.runId}:${e.phase}`;async function he(e){if(!await S(e.workflow,g))throw Error("workflow_execution_not_current");let i=te(e),t=await P({userId:e.workflow.userId,tool:e.tool,jobId:e.jobId,normalizedInput:e.normalizedInput,idempotencyKey:i,leaseMs:g});if(t.kind==="replay"){if(t.result.status!=="succeeded")throw Error(`operation_phase_unknown:${t.runId}`);return JSON.parse(t.result.json)}if(t.kind!=="claimed")throw Error(`operation_phase_${t.kind}:${t.runId}`);let n=t.claim;try{await e.prepare()}catch(r){throw await I(n).catch(()=>{}),r}try{let r;try{r=await k(n)}catch(a){if(!await m(n))throw a;r=!0}if(!r)throw Error("operation_phase_start_unknown");let s=Date.now()+g,o=async()=>{if(Date.now()>=s)throw new L(e.phase);if(!await m(e.workflow)||!await m(n))throw Error("operation_phase_claim_not_current")};await o();let u=await H(s,()=>e.execute({claim:n,deadlineAt:s,assertActive:o}),e.phase);await o();let f=JSON.stringify(u);if(f==null||!await x(n,{status:"succeeded",json:f}))throw Error("operation_phase_publication_unknown");return u}catch(r){await h(n).catch(()=>{});let s=await y(n.runId,n.userId).catch(()=>null);throw s&&await O(s,{status:"unknown",json:JSON.stringify({phase:e.phase,outcome:"unknown"})}).catch(()=>{}),r}}async function be(e,i){let t=await Y({userId:e.workflow.userId,tool:e.tool,idempotencyKey:te(e),normalizedInput:e.normalizedInput});if(!t||t.result.status!=="succeeded"||R({value:JSON.parse(t.result.json)})!==R({value:i}))throw Error("operation_phase_cache_receipt_unknown")}async function ge(e,i,t){let n={runId:e,userId:i,token:ee(t)};if(await m(n))return{claim:n,recovery:!1};let r=await y(e,i);return r?{claim:r,recovery:!0}:null}async function re(e,i,t){let n={json:JSON.stringify(i),status:e.recovery?"unknown":t};if(!(e.recovery?await O(e.claim,n,e.settlementGate):await x(e.claim,n)))throw Error("workflow_result_publication_unknown")}async function je(e){if(!await b(e))return{kind:"missing",body:{error:"job not found"}};let t=await M(e);if(!t)return{kind:"conflict",body:{error:"Only failed or expired recovery jobs can be re-finalized.",status:(await b(e))?.status??"missing"}};let n=t.executionGeneration;if(!await D(e,n))return{kind:"conflict",body:{error:"job state changed during re-finalization claim"}};let r;try{r=await X(t,{})}catch(l){throw await F(e,l instanceof Error?l.message:String(l),void 0,n).catch(()=>!1),l}let s=A(t),o=$(t,s.creditTruncated),u=o==="failed"?`No pages were extracted successfully (${t.failedUrls} failed, ${t.remainingUrls} remaining).`:o==="partial"?`${t.successfulUrls} pages succeeded; ${t.failedUrls} failed and ${t.remainingUrls} remain.${s.creditTruncated?` Credit availability limited this request from ${s.requestedMaxPages} to ${s.effectiveMaxPages} pages.`:""}`:null,f;try{f=await q(e,r,o,u,!1,null,n)}catch(l){let w=await b(e).catch(()=>null);if(!w||w.id!==e||w.executionGeneration!==n||w.status!==o||JSON.stringify(w.artifacts)!==JSON.stringify(r)||w.error!==(u?z(u).slice(0,2e3):null)||w.publicError!==null)throw l;f=!0}if(!f)return{kind:"conflict",body:{error:"job state changed during re-finalization"}};let a=await b(e);if(!a||a.executionGeneration!==n||a.status!==o)return{kind:"conflict",body:{error:"job state changed before re-finalization settlement"}};let c="already_settled_or_refunded";if(a.billedMc==null&&a.userId!=null){let l=await j(a);if(!l)throw Error("extract_settlement_acknowledgment_unknown");c=`settled: billed ${l.netChargeMc} mc, refunded ${l.refundMc} mc${l.expiredRefundMc?`, expired refund ${l.expiredRefundMc} mc`:""}`}let p=r.find(l=>l.contentType==="application/zip"||l.filename?.endsWith(".zip"));return{kind:"complete",body:{ok:!0,jobId:e,artifacts:r.length,settlement:c,bundleUrl:p?.url??null,bundleBytes:p?.bytes??null}}}async function Pe(e=25){let i=await U(e);for(let r of i)await G(r.id).catch(()=>!1);let t=await C(e),n={checked:new Set([...i,...t].map(r=>r.id)).size,settled:0,failed:0};for(let r of t){if(r.userId==null){await W(r.id,"Settlement abandoned because the legacy job has no user owner.",r.executionGeneration??0),n.failed+=1;continue}try{if(!await j(r))throw Error("extract_settlement_acknowledgment_unknown");if(r.executionGeneration===0&&await N(`extract-site:${r.id}`,r.userId)){let o=await y(`extract-site:${r.id}`,r.userId);o&&await re({claim:o,recovery:!0},{jobId:r.id,status:r.status,artifacts:r.artifacts??[]},"unknown")}n.settled+=1}catch(s){n.failed+=1;let o=s instanceof Error?s.message:String(s);await K(r.id,o,r.executionGeneration??0).catch(()=>{}),console.warn("[site-extract/reconcile]",r.id,o)}}return n}export{we as a,g as b,ee as c,ye as d,he as e,be as f,ge as g,re as h,je as i,Pe as j};

package/dist/{chunk-JW4GDY7S.js → chunk-SE4VKLEC.js}

@@ -1,4 +1,4 @@
-import{S as K}from"./chunk-TPN25AFZ.js";import{a as X}from"./chunk-NALXKTUU.js";import{h as J,i as H}from"./chunk-RDHGQ2WZ.js";import{a as f}from"./chunk-WNUXMPVX.js";import{createHash as U,createHmac as re,timingSafeEqual as he}from"crypto";import{createHash as fe,randomUUID as Y}from"crypto";var ge=1440*60*1e3,F=5e3,h=class extends Error{constructor(n,i,o=400){super(i);this.code=n;this.status=o;this.name="GmailSelectionError"}code;status},k=null,P=null;function $(e){return e===null||typeof e!="object"?JSON.stringify(e):Array.isArray(e)?`[${e.map($).join(",")}]`:`{${Object.entries(e).sort(([t],[n])=>t.localeCompare(n)).map(([t,n])=>`${JSON.stringify(t)}:${$(n)}`).join(",")}}`}function b(e){return fe("sha256").update($(e)).digest("hex")}async function x(){let e=f();return k&&P===e||(P=e,k=(async()=>{await e.execute(`
+import{S as K}from"./chunk-SFUOSJ3A.js";import{a as X}from"./chunk-NALXKTUU.js";import{h as J,i as H}from"./chunk-RDHGQ2WZ.js";import{a as f}from"./chunk-WNUXMPVX.js";import{createHash as U,createHmac as re,timingSafeEqual as he}from"crypto";import{createHash as fe,randomUUID as Y}from"crypto";var ge=1440*60*1e3,F=5e3,h=class extends Error{constructor(n,i,o=400){super(i);this.code=n;this.status=o;this.name="GmailSelectionError"}code;status},k=null,P=null;function $(e){return e===null||typeof e!="object"?JSON.stringify(e):Array.isArray(e)?`[${e.map($).join(",")}]`:`{${Object.entries(e).sort(([t],[n])=>t.localeCompare(n)).map(([t,n])=>`${JSON.stringify(t)}:${$(n)}`).join(",")}}`}function b(e){return fe("sha256").update($(e)).digest("hex")}async function x(){let e=f();return k&&P===e||(P=e,k=(async()=>{await e.execute(`
       CREATE TABLE IF NOT EXISTS gmail_selections (
         selection_id TEXT PRIMARY KEY,
         owner_identity TEXT NOT NULL,

package/dist/{chunk-TPN25AFZ.js → chunk-SFUOSJ3A.js}

@@ -1,4 +1,4 @@
-import{a as M,b as q,h as L}from"./chunk-R5LJL6IR.js";import{o as b,r as J,s as pe,t as _e,u as H,w as ge}from"./chunk-QWE7VKIX.js";import{b as I,d as F}from"./chunk-ZPQNY5TI.js";import{g as j}from"./chunk-VVSHZV5S.js";import{b as $}from"./chunk-4ZVO6GX4.js";import{a as l}from"./chunk-WNUXMPVX.js";import{createHash as O,randomUUID as Oe}from"crypto";import{Client as tt,StreamableHTTPClientTransport as nt}from"@modelcontextprotocol/client";import{randomUUID as Q}from"crypto";import{createHash as Y,randomBytes as W,scryptSync as fe,timingSafeEqual as We}from"crypto";function Ee(){return"sk_"+W(24).toString("hex")}function ye(e){let t=W(16).toString("hex"),n=fe(e,t,64);return`${t}:${n.toString("hex")}`}function Ct(e,t){let[n,i]=t.split(":"),r=Buffer.from(i,"hex"),o=fe(e,n,64);return We(r,o)}function K(e){return{id:Number(e.id),email:String(e.email),name:e.name!=null?String(e.name):null,api_key:String(e.api_key),key_active:Number(e.key_active??1),password_hash:e.password_hash!=null?String(e.password_hash):null,created_at:String(e.created_at),active:Number(e.active),stripe_customer_id:e.stripe_customer_id!=null?String(e.stripe_customer_id):null,balance_mc:Number(e.balance_mc??0),extra_concurrency_slots:Number(e.extra_concurrency_slots??0),concurrency_stripe_sub_id:e.concurrency_stripe_sub_id!=null?String(e.concurrency_stripe_sub_id):null,concurrency_pack_quantity:Number(e.concurrency_pack_quantity??0),subscription_concurrency:Number(e.subscription_concurrency??0),subscription_tier:e.subscription_tier!=null?String(e.subscription_tier):null,subscription_id:e.subscription_id!=null?String(e.subscription_id):null,memory_plan:e.memory_plan!=null?String(e.memory_plan):"free",memory_subscription_id:e.memory_subscription_id!=null?String(e.memory_subscription_id):null,memory_provisioned:Number(e.memory_provisioned??0)}}async function X(e){let t=await l().execute({sql:"SELECT * FROM users WHERE email = ? AND active = 1",args:[e]});return t.rows[0]?K(t.rows[0]):void 0}async function St(e){let t=await l().execute({sql:"SELECT * FROM users WHERE api_key = ? AND active = 1 AND key_active = 1",args:[e]});return t.rows[0]?K(t.rows[0]):void 0}async function Nt(e){let t=await l().execute({sql:"SELECT * FROM users WHERE id = ? AND active = 1",args:[e]});return t.rows[0]?K(t.rows[0]):void 0}async function ht(e,t){await l().execute({sql:"UPDATE users SET password_hash = ? WHERE id = ?",args:[ye(t),e]})}async function Rt(e){let t=l(),[n,i,r,o,s]=await Promise.all([t.execute({sql:"SELECT COUNT(*) as n FROM jobs WHERE user_id = ?",args:[e]}),t.execute({sql:"SELECT COUNT(*) as n FROM jobs WHERE user_id = ? AND status = 'done'",args:[e]}),t.execute({sql:"SELECT COUNT(*) as n FROM jobs WHERE user_id = ? AND status = 'failed'",args:[e]}),t.execute({sql:"SELECT COALESCE(SUM(json_extract(result, '$.totalQuestions')), 0) as n FROM jobs WHERE user_id = ? AND status = 'done'",args:[e]}),t.execute({sql:"SELECT MAX(created_at) as t FROM jobs WHERE user_id = ?",args:[e]})]);return{total:Number(n.rows[0]?.n??0),done:Number(i.rows[0]?.n??0),failed:Number(r.rows[0]?.n??0),totalQuestions:Number(o.rows[0]?.n??0),lastUsed:s.rows[0]?.t!=null?String(s.rows[0].t):null}}async function At(e,t,n,i){let r=l(),o=Ee(),s=n??W(6).toString("hex"),a=ye(s);return{id:(await r.execute({sql:"INSERT INTO users (email, name, api_key, password_hash, stripe_customer_id) VALUES (?, ?, ?, ?, ?)",args:[e,t??null,o,a,i??null]})).lastInsertRowid,email:e,name:t,api_key:o,password:s}}async function xt(e,t,n,i){let r=Math.floor(Date.now()/1e3),o=Math.floor(r/i)*i,s=Y("sha256").update(t).digest("hex"),d=await l().execute({sql:`
+import{a as M,b as q,f as L}from"./chunk-6BD75VKS.js";import{o as b,r as J,s as pe,t as _e,u as H,w as ge}from"./chunk-CORWWDHL.js";import{f as I,h as F}from"./chunk-5LYINW7Z.js";import{g as j}from"./chunk-VVSHZV5S.js";import{b as $}from"./chunk-4ZVO6GX4.js";import{a as l}from"./chunk-WNUXMPVX.js";import{createHash as O,randomUUID as Oe}from"crypto";import{Client as tt,StreamableHTTPClientTransport as nt}from"@modelcontextprotocol/client";import{randomUUID as Q}from"crypto";import{createHash as Y,randomBytes as W,scryptSync as fe,timingSafeEqual as We}from"crypto";function Ee(){return"sk_"+W(24).toString("hex")}function ye(e){let t=W(16).toString("hex"),n=fe(e,t,64);return`${t}:${n.toString("hex")}`}function Ct(e,t){let[n,i]=t.split(":"),r=Buffer.from(i,"hex"),o=fe(e,n,64);return We(r,o)}function K(e){return{id:Number(e.id),email:String(e.email),name:e.name!=null?String(e.name):null,api_key:String(e.api_key),key_active:Number(e.key_active??1),password_hash:e.password_hash!=null?String(e.password_hash):null,created_at:String(e.created_at),active:Number(e.active),stripe_customer_id:e.stripe_customer_id!=null?String(e.stripe_customer_id):null,balance_mc:Number(e.balance_mc??0),extra_concurrency_slots:Number(e.extra_concurrency_slots??0),concurrency_stripe_sub_id:e.concurrency_stripe_sub_id!=null?String(e.concurrency_stripe_sub_id):null,concurrency_pack_quantity:Number(e.concurrency_pack_quantity??0),subscription_concurrency:Number(e.subscription_concurrency??0),subscription_tier:e.subscription_tier!=null?String(e.subscription_tier):null,subscription_id:e.subscription_id!=null?String(e.subscription_id):null,memory_plan:e.memory_plan!=null?String(e.memory_plan):"free",memory_subscription_id:e.memory_subscription_id!=null?String(e.memory_subscription_id):null,memory_provisioned:Number(e.memory_provisioned??0)}}async function X(e){let t=await l().execute({sql:"SELECT * FROM users WHERE email = ? AND active = 1",args:[e]});return t.rows[0]?K(t.rows[0]):void 0}async function St(e){let t=await l().execute({sql:"SELECT * FROM users WHERE api_key = ? AND active = 1 AND key_active = 1",args:[e]});return t.rows[0]?K(t.rows[0]):void 0}async function Nt(e){let t=await l().execute({sql:"SELECT * FROM users WHERE id = ? AND active = 1",args:[e]});return t.rows[0]?K(t.rows[0]):void 0}async function ht(e,t){await l().execute({sql:"UPDATE users SET password_hash = ? WHERE id = ?",args:[ye(t),e]})}async function Rt(e){let t=l(),[n,i,r,o,s]=await Promise.all([t.execute({sql:"SELECT COUNT(*) as n FROM jobs WHERE user_id = ?",args:[e]}),t.execute({sql:"SELECT COUNT(*) as n FROM jobs WHERE user_id = ? AND status = 'done'",args:[e]}),t.execute({sql:"SELECT COUNT(*) as n FROM jobs WHERE user_id = ? AND status = 'failed'",args:[e]}),t.execute({sql:"SELECT COALESCE(SUM(json_extract(result, '$.totalQuestions')), 0) as n FROM jobs WHERE user_id = ? AND status = 'done'",args:[e]}),t.execute({sql:"SELECT MAX(created_at) as t FROM jobs WHERE user_id = ?",args:[e]})]);return{total:Number(n.rows[0]?.n??0),done:Number(i.rows[0]?.n??0),failed:Number(r.rows[0]?.n??0),totalQuestions:Number(o.rows[0]?.n??0),lastUsed:s.rows[0]?.t!=null?String(s.rows[0].t):null}}async function At(e,t,n,i){let r=l(),o=Ee(),s=n??W(6).toString("hex"),a=ye(s);return{id:(await r.execute({sql:"INSERT INTO users (email, name, api_key, password_hash, stripe_customer_id) VALUES (?, ?, ?, ?, ?)",args:[e,t??null,o,a,i??null]})).lastInsertRowid,email:e,name:t,api_key:o,password:s}}async function xt(e,t,n,i){let r=Math.floor(Date.now()/1e3),o=Math.floor(r/i)*i,s=Y("sha256").update(t).digest("hex"),d=await l().execute({sql:`
       INSERT INTO rate_limits (scope, key_hash, window_start, count)
       VALUES (?, ?, ?, 1)
       ON CONFLICT(scope, key_hash, window_start)

package/dist/{chunk-GVYGAZV7.js → chunk-UXNR3K6I.js}

@@ -1 +1 @@
-import{g as c,j as d,k as p,l as w,m}from"./chunk-6SNOUSSZ.js";import{i as u}from"./chunk-C7MGJUTU.js";import{Inngest as x}from"inngest";var h=new x({id:"mcp-scraper"});function b(r){return Math.min(3600,Math.max(15,15*2**Math.min(8,Math.max(0,r-1))))}async function f(r){let t={claimed:r.rows.length,dispatched:0,failed:0},e=()=>{t.unknown=(t.unknown??0)+1},o=async a=>{let i=await r.read(a).catch(()=>null);return i?.id===a.id?i:null};for(let a of r.rows){let i=a.claimToken;if(!i||!await r.isCurrent(a,i).catch(()=>!1)){e();continue}try{await r.send(a)}catch(n){let l=(n instanceof Error?n.message:String(n)).slice(0,2e3);if(await r.retainRetry(a,i,l,b(a.attempts)).catch(async()=>{let s=await o(a);return s?.status==="failed"&&s.lastError===l}))t.failed+=1;else{let s=await o(a);s?.status==="dispatched"&&s.dispatchedAt?t.dispatched+=1:e()}continue}if(await r.acknowledge(a,i).catch(()=>!1))t.dispatched+=1;else{let n=await o(a);n?.status==="dispatched"&&n.dispatchedAt?t.dispatched+=1:e()}}return t}async function k(r){return f({rows:r,isCurrent:(t,e)=>p(t.id,e),send:async t=>{try{return await h.send({id:t.id,name:"mcp-scraper/extract.requested",data:{jobId:t.id}})}catch(e){throw Error(u(e instanceof Error?e.message:String(e)).slice(0,2e3))}},acknowledge:(t,e)=>w(t.id,e),retainRetry:(t,e,o,a)=>m(t.id,e,o,a),read:async t=>{let e=await c(t.id);return e?{id:e.id,status:e.dispatchedAt?"dispatched":e.dispatchClaimToken?"dispatching":e.dispatchError?"failed":"pending",dispatchedAt:e.dispatchedAt??null,lastError:e.dispatchError??null}:null}})}async function A(r){if((await c(r))?.dispatchedAt)return{claimed:0,dispatched:1,failed:0};let e=await d({jobId:r});return e.length?k(e):{claimed:0,dispatched:0,failed:0,unknown:1}}async function S(r=25){let t=await k(await d({limit:r}));return{checked:t.claimed,dispatched:t.dispatched,failed:t.failed,...t.unknown?{unknown:t.unknown}:{}}}export{h as a,f as b,A as c,S as d};
+import{g as c,j as d,k as p,l as w,m}from"./chunk-7WPCGRD7.js";import{i as u}from"./chunk-C7MGJUTU.js";import{Inngest as x}from"inngest";var h=new x({id:"mcp-scraper"});function b(r){return Math.min(3600,Math.max(15,15*2**Math.min(8,Math.max(0,r-1))))}async function f(r){let t={claimed:r.rows.length,dispatched:0,failed:0},e=()=>{t.unknown=(t.unknown??0)+1},o=async a=>{let i=await r.read(a).catch(()=>null);return i?.id===a.id?i:null};for(let a of r.rows){let i=a.claimToken;if(!i||!await r.isCurrent(a,i).catch(()=>!1)){e();continue}try{await r.send(a)}catch(n){let l=(n instanceof Error?n.message:String(n)).slice(0,2e3);if(await r.retainRetry(a,i,l,b(a.attempts)).catch(async()=>{let s=await o(a);return s?.status==="failed"&&s.lastError===l}))t.failed+=1;else{let s=await o(a);s?.status==="dispatched"&&s.dispatchedAt?t.dispatched+=1:e()}continue}if(await r.acknowledge(a,i).catch(()=>!1))t.dispatched+=1;else{let n=await o(a);n?.status==="dispatched"&&n.dispatchedAt?t.dispatched+=1:e()}}return t}async function k(r){return f({rows:r,isCurrent:(t,e)=>p(t.id,e),send:async t=>{try{return await h.send({id:t.id,name:"mcp-scraper/extract.requested",data:{jobId:t.id}})}catch(e){throw Error(u(e instanceof Error?e.message:String(e)).slice(0,2e3))}},acknowledge:(t,e)=>w(t.id,e),retainRetry:(t,e,o,a)=>m(t.id,e,o,a),read:async t=>{let e=await c(t.id);return e?{id:e.id,status:e.dispatchedAt?"dispatched":e.dispatchClaimToken?"dispatching":e.dispatchError?"failed":"pending",dispatchedAt:e.dispatchedAt??null,lastError:e.dispatchError??null}:null}})}async function A(r){if((await c(r))?.dispatchedAt)return{claimed:0,dispatched:1,failed:0};let e=await d({jobId:r});return e.length?k(e):{claimed:0,dispatched:0,failed:0,unknown:1}}async function S(r=25){let t=await k(await d({limit:r}));return{checked:t.claimed,dispatched:t.dispatched,failed:t.failed,...t.unknown?{unknown:t.unknown}:{}}}export{h as a,f as b,A as c,S as d};

package/dist/{chunk-FRXKMOXI.js → chunk-ZTMQEDIB.js}

@@ -1,4 +1,4 @@
-import{a as ot}from"./chunk-U4WIJZCX.js";import{F as fe,a as Qe,c as Ve}from"./chunk-6SNOUSSZ.js";import{g as Ke,h as Ze}from"./chunk-KXJCGV3H.js";import{c as st}from"./chunk-JMZIDHYQ.js";import{c as tt,e as nt,f as rt,g as it}from"./chunk-CNSVZTTT.js";import{c as et}from"./chunk-5BKSKLHX.js";import{h as Ge}from"./chunk-C7MGJUTU.js";import{a as se}from"./chunk-WNUXMPVX.js";import{createWriteStream as ae,mkdtempSync as Yt,mkdirSync as le,readFileSync as Wt,readdirSync as Gt,rmSync as Kt,statSync as Zt,writeFileSync as y}from"fs";import{createHash as W}from"crypto";import{join as l}from"path";import{tmpdir as Qt}from"os";import{createGzip as Vt}from"zlib";import{once as G}from"events";import{finished as ue}from"stream/promises";import{ZipFile as en}from"yazl";import tn from"p-limit";var vt=new Set(["browser_navigation_blocked","browser_navigation_failed","browser_timeout","browser_session_failed","browser_result_missing"]);function Mt(e){return e instanceof Error?e.code:void 0}function At(e){return e instanceof Error?e.message:String(e)}function at(e){let r=e?.match(/^browser_http_(\d+)$/);return r?Number(r[1]):null}function It(e){return e===404?"page_not_found":e===403?"page_forbidden":e===429?"page_rate_limited":e>=500&&e<=599?"page_server_error":"page_http_error"}function Rt(e){let r=At(e),i=Mt(e);if(ot(e,r))return{errorCode:"vendor_unavailable",httpStatus:null,retryable:!0};let a=at(i)??at(r.match(/^(browser_http_\d+)/)?.[1]);return a!=null?{errorCode:It(a),httpStatus:a,retryable:a===429||a>=500}:i==="browser_challenge_unresolved"||/^browser_challenge_unresolved/.test(r)?{errorCode:"bot_check_unresolved",httpStatus:null,retryable:!0}:i==="browser_response_too_large"||/exceeds \d+ byte limit/i.test(r)?{errorCode:"page_too_large",httpStatus:null,retryable:!1}:/browser (?:has been )?closed|context closed|target page, context or browser has been closed|session closed/i.test(r)?{errorCode:"browser_session_interrupted",httpStatus:null,retryable:!0}:vt.has(i??"")||/timeout|navigation_failed|ERR_/i.test(r)?{errorCode:"page_unreachable",httpStatus:null,retryable:!0}:{errorCode:"extraction_failed",httpStatus:null,retryable:!1}}function lt(e,r){switch(e){case"page_not_found":return`The page could not be found (HTTP ${r??404}). It may have been moved or deleted.`;case"page_forbidden":return`The site refused the request (HTTP ${r??403}).`;case"page_rate_limited":return"The site is rate-limiting requests right now (HTTP 429). Retrying after a short wait usually works.";case"page_server_error":return`The page's own server returned an error (HTTP ${r}). This is on the target site's side, not something a retry here can fix.`;case"page_http_error":return`The page returned an unexpected HTTP ${r} response.`;case"bot_check_unresolved":return"This site has automated-traffic protection (a bot/CAPTCHA check) that could not be resolved in time. Some sites are simply not extractable this way.";case"page_too_large":return"The page is larger than we can safely process.";case"page_unreachable":return"The page did not respond in time, or the connection could not be completed.";case"browser_session_interrupted":return"The browser session closed before extraction completed. A retry uses a fresh browser session.";case"vendor_unavailable":return"MCP Scraper's internal browser services require MCP Scraper team attention. Servers or IPs are down until this is fixed \u2014 this is not caused by your request. Please retry in a few minutes.";case"extraction_failed":return"The page could not be extracted. Check retryable before attempting it again; contact support if this persists for the same URL."}}function on(e,r={}){let i=lt(e.errorCode,e.httpStatus);return{...Ge({errorCode:e.errorCode,retryable:e.retryable,chargeStatus:r.chargeStatus,details:{...r.details??{},...e.httpStatus!=null?{http_status:e.httpStatus}:{}}}),message:i,error:i}}function an(e){return e.httpStatus===404?404:e.httpStatus===403?403:e.httpStatus===429?429:e.httpStatus!=null&&e.httpStatus>=500&&e.httpStatus<=599?500:e.httpStatus!=null?502:e.errorCode==="vendor_unavailable"?503:502}var Dt="wayback_capture_missing";function Me(e,r){if(e==null)return{failureCode:null,failureReason:null};if(e===Dt)return{failureCode:e,failureReason:r??null};let i=new Error(r??e);i.code=e;let a=Rt(i);return{failureCode:a.errorCode,failureReason:lt(a.errorCode,a.httpStatus)}}import{createHash as Ie}from"crypto";import{neon as $t}from"@neondatabase/serverless";import{waitUntil as Ut}from"@vercel/functions";var jt="jina",ye=null,ut=!1;function be(){return(process.env.JINA_EMBED_MODEL??"jina-embeddings-v5-omni-small").trim()}function _e(){return Number((process.env.JINA_EMBED_DIM??"1024").trim())}function Ae(){return!!(process.env.JINA_API_KEY?.trim()&&process.env.MEMORY_DATABASE_URL?.trim())}function oe(){if(ye)return ye;let e=process.env.MEMORY_DATABASE_URL?.trim();if(!e)throw new Error("MEMORY_DATABASE_URL is not set; Commons semantic search needs the shared Postgres.");return ye=$t(e),ye}async function mt(){if(ut)return;let e=_e();await oe().query("CREATE EXTENSION IF NOT EXISTS vector"),await oe().query(`
+import{a as ot}from"./chunk-U4WIJZCX.js";import{F as fe,a as Qe,c as Ve}from"./chunk-7WPCGRD7.js";import{g as Ke,h as Ze}from"./chunk-KXJCGV3H.js";import{c as st}from"./chunk-JMZIDHYQ.js";import{c as tt,e as nt,f as rt,g as it}from"./chunk-CNSVZTTT.js";import{c as et}from"./chunk-5BKSKLHX.js";import{h as Ge}from"./chunk-C7MGJUTU.js";import{a as se}from"./chunk-WNUXMPVX.js";import{createWriteStream as ae,mkdtempSync as Yt,mkdirSync as le,readFileSync as Wt,readdirSync as Gt,rmSync as Kt,statSync as Zt,writeFileSync as y}from"fs";import{createHash as W}from"crypto";import{join as l}from"path";import{tmpdir as Qt}from"os";import{createGzip as Vt}from"zlib";import{once as G}from"events";import{finished as ue}from"stream/promises";import{ZipFile as en}from"yazl";import tn from"p-limit";var vt=new Set(["browser_navigation_blocked","browser_navigation_failed","browser_timeout","browser_session_failed","browser_result_missing"]);function Mt(e){return e instanceof Error?e.code:void 0}function At(e){return e instanceof Error?e.message:String(e)}function at(e){let r=e?.match(/^browser_http_(\d+)$/);return r?Number(r[1]):null}function It(e){return e===404?"page_not_found":e===403?"page_forbidden":e===429?"page_rate_limited":e>=500&&e<=599?"page_server_error":"page_http_error"}function Rt(e){let r=At(e),i=Mt(e);if(ot(e,r))return{errorCode:"vendor_unavailable",httpStatus:null,retryable:!0};let a=at(i)??at(r.match(/^(browser_http_\d+)/)?.[1]);return a!=null?{errorCode:It(a),httpStatus:a,retryable:a===429||a>=500}:i==="browser_challenge_unresolved"||/^browser_challenge_unresolved/.test(r)?{errorCode:"bot_check_unresolved",httpStatus:null,retryable:!0}:i==="browser_response_too_large"||/exceeds \d+ byte limit/i.test(r)?{errorCode:"page_too_large",httpStatus:null,retryable:!1}:/browser (?:has been )?closed|context closed|target page, context or browser has been closed|session closed/i.test(r)?{errorCode:"browser_session_interrupted",httpStatus:null,retryable:!0}:vt.has(i??"")||/timeout|navigation_failed|ERR_/i.test(r)?{errorCode:"page_unreachable",httpStatus:null,retryable:!0}:{errorCode:"extraction_failed",httpStatus:null,retryable:!1}}function lt(e,r){switch(e){case"page_not_found":return`The page could not be found (HTTP ${r??404}). It may have been moved or deleted.`;case"page_forbidden":return`The site refused the request (HTTP ${r??403}).`;case"page_rate_limited":return"The site is rate-limiting requests right now (HTTP 429). Retrying after a short wait usually works.";case"page_server_error":return`The page's own server returned an error (HTTP ${r}). This is on the target site's side, not something a retry here can fix.`;case"page_http_error":return`The page returned an unexpected HTTP ${r} response.`;case"bot_check_unresolved":return"This site has automated-traffic protection (a bot/CAPTCHA check) that could not be resolved in time. Some sites are simply not extractable this way.";case"page_too_large":return"The page is larger than we can safely process.";case"page_unreachable":return"The page did not respond in time, or the connection could not be completed.";case"browser_session_interrupted":return"The browser session closed before extraction completed. A retry uses a fresh browser session.";case"vendor_unavailable":return"MCP Scraper's internal browser services require MCP Scraper team attention. Servers or IPs are down until this is fixed \u2014 this is not caused by your request. Please retry in a few minutes.";case"extraction_failed":return"The page could not be extracted. Check retryable before attempting it again; contact support if this persists for the same URL."}}function on(e,r={}){let i=lt(e.errorCode,e.httpStatus);return{...Ge({errorCode:e.errorCode,retryable:e.retryable,chargeStatus:r.chargeStatus,details:{...r.details??{},...e.httpStatus!=null?{http_status:e.httpStatus}:{}}}),message:i,error:i}}function an(e){return e.httpStatus===404?404:e.httpStatus===403?403:e.httpStatus===429?429:e.httpStatus!=null&&e.httpStatus>=500&&e.httpStatus<=599?500:e.httpStatus!=null?502:e.errorCode==="vendor_unavailable"?503:502}var Dt="wayback_capture_missing";function Me(e,r){if(e==null)return{failureCode:null,failureReason:null};if(e===Dt)return{failureCode:e,failureReason:r??null};let i=new Error(r??e);i.code=e;let a=Rt(i);return{failureCode:a.errorCode,failureReason:lt(a.errorCode,a.httpStatus)}}import{createHash as Ie}from"crypto";import{neon as $t}from"@neondatabase/serverless";import{waitUntil as Ut}from"@vercel/functions";var jt="jina",ye=null,ut=!1;function be(){return(process.env.JINA_EMBED_MODEL??"jina-embeddings-v5-omni-small").trim()}function _e(){return Number((process.env.JINA_EMBED_DIM??"1024").trim())}function Ae(){return!!(process.env.JINA_API_KEY?.trim()&&process.env.MEMORY_DATABASE_URL?.trim())}function oe(){if(ye)return ye;let e=process.env.MEMORY_DATABASE_URL?.trim();if(!e)throw new Error("MEMORY_DATABASE_URL is not set; Commons semantic search needs the shared Postgres.");return ye=$t(e),ye}async function mt(){if(ut)return;let e=_e();await oe().query("CREATE EXTENSION IF NOT EXISTS vector"),await oe().query(`
     CREATE TABLE IF NOT EXISTS commons_index_vectors (
       document_id   TEXT PRIMARY KEY,
       entity_id     TEXT NOT NULL,

package/dist/{dispatch-3HY4735M.js → dispatch-P7IY75LH.js}

@@ -1 +1 @@
-import{c as a,d as b}from"./chunk-GVYGAZV7.js";import"./chunk-6SNOUSSZ.js";import"./chunk-KXJCGV3H.js";import"./chunk-HUNSEGUN.js";import"./chunk-SEZETSB3.js";import"./chunk-SDQW7IJN.js";import"./chunk-C7MGJUTU.js";import"./chunk-4ZVO6GX4.js";import"./chunk-WNUXMPVX.js";import"./chunk-OILOHQNL.js";import"./chunk-HNNYXMSI.js";export{a as dispatchSiteExtractJob,b as redispatchFundedSiteExtractJobs};
+import{c as a,d as b}from"./chunk-UXNR3K6I.js";import"./chunk-7WPCGRD7.js";import"./chunk-KXJCGV3H.js";import"./chunk-HUNSEGUN.js";import"./chunk-SEZETSB3.js";import"./chunk-SDQW7IJN.js";import"./chunk-C7MGJUTU.js";import"./chunk-4ZVO6GX4.js";import"./chunk-WNUXMPVX.js";import"./chunk-OILOHQNL.js";import"./chunk-HNNYXMSI.js";export{a as dispatchSiteExtractJob,b as redispatchFundedSiteExtractJobs};