mcp-scraper 0.89.6 → 0.89.9
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +51 -1
- package/README.md +3 -3
- package/dist/{analytics-repository-4UA3WI3U.js → analytics-repository-UX6VU22H.js} +1 -1
- package/dist/bin/api-server.js +1 -1
- package/dist/bin/mcp-scraper-cli.js +1 -1
- package/dist/bin/mcp-scraper-core.js +1 -1
- package/dist/bin/mcp-scraper-install.js +1 -1
- package/dist/bin/mcp-stdio-server.js +1 -1
- package/dist/bin/paa-harvest.js +1 -1
- package/dist/{chunk-YCZKKSTT.js → chunk-2GRE4BAO.js} +1 -1
- package/dist/chunk-2PXP7TMY.js +1 -0
- package/dist/{chunk-3GP5CYZX.js → chunk-2TZBO52D.js} +1 -1
- package/dist/{chunk-H2I63VAD.js → chunk-2WWKTVCO.js} +1 -1
- package/dist/{chunk-ISZIA6NE.js → chunk-52XUWR3X.js} +1 -1
- package/dist/{chunk-Y3ITV2OK.js → chunk-5B42QVDC.js} +1 -1
- package/dist/{chunk-KKI2SNB2.js → chunk-7AC66V3T.js} +1 -1
- package/dist/{chunk-YQ5U5C2X.js → chunk-7BEWBRMT.js} +74 -55
- package/dist/{chunk-WUVMD6ZX.js → chunk-DKQZNSYM.js} +1 -1
- package/dist/{chunk-MZN4U5BL.js → chunk-E4ME7J6G.js} +1 -1
- package/dist/{chunk-C6JXNOSF.js → chunk-F4LIB772.js} +8 -8
- package/dist/{chunk-PPOI4C3Q.js → chunk-HR2A2WTL.js} +1 -1
- package/dist/{chunk-XLWNEVUZ.js → chunk-JHMI6HEO.js} +3 -3
- package/dist/{chunk-FVSXX43Z.js → chunk-JSNFZPPB.js} +1 -1
- package/dist/{chunk-3H7HU5JI.js → chunk-JVU2THSY.js} +1 -1
- package/dist/{chunk-VZTFJLOK.js → chunk-LZTDANNZ.js} +1 -1
- package/dist/{chunk-QPWPR5XG.js → chunk-M5QHXNFZ.js} +1 -1
- package/dist/{chunk-FKSTNWP6.js → chunk-PHBU6X5B.js} +1 -1
- package/dist/chunk-PVTKGAZL.js +1 -0
- package/dist/{chunk-H3QW5U67.js → chunk-Q4A2KIPZ.js} +3 -3
- package/dist/{chunk-G6GAIXXA.js → chunk-S3N55WHK.js} +4 -4
- package/dist/{chunk-WXOLIRFJ.js → chunk-VT6SLBOU.js} +1 -1
- package/dist/{chunk-Z45GM22P.js → chunk-XATD7HDY.js} +227 -227
- package/dist/{chunk-YQZGZBB4.js → chunk-XG6GCEUE.js} +1 -1
- package/dist/db-UKQ74OVT.js +1 -0
- package/dist/{extract-bundle-BTZQJTYV.js → extract-bundle-DI7OVLT5.js} +5 -5
- package/dist/{gmail-service-ZZDF52KK.js → gmail-service-ASX7YBC3.js} +1 -1
- package/dist/index.cjs +45 -45
- package/dist/index.d.cts +16 -0
- package/dist/index.d.ts +16 -0
- package/dist/index.js +1 -1
- package/dist/{lead-list-enrichment-repository-K4H5DMSF.js → lead-list-enrichment-repository-GPDHJVM3.js} +1 -1
- package/dist/{location-data-repository-KALKE3FG.js → location-data-repository-MXUNVDHW.js} +1 -1
- package/dist/{server-YVNOU3RK.js → server-EYSV2RRD.js} +702 -702
- package/dist/{site-extract-repository-FTTIOWAL.js → site-extract-repository-WADCAIYB.js} +1 -1
- package/dist/{stripe-event-worker-CURL3RYL.js → stripe-event-worker-ZPNWXX2T.js} +1 -1
- package/dist/worker-577GJULQ.js +1 -0
- package/package.json +1 -1
- package/dist/chunk-CULFQBFR.js +0 -1
- package/dist/db-IZ5UF33U.js +0 -1
- package/dist/worker-MUHILD2W.js +0 -1
|
@@ -1 +1 @@
|
|
|
1
|
-
import{A as s,B as t,C as u,D as v,E as w,F as x,G as y,H as z,I as A,J as B,K as C,L as D,M as E,i as a,j as b,k as c,l as d,m as e,n as f,o as g,p as h,q as i,r as j,s as k,t as l,u as m,v as n,w as o,x as p,y as q,z as r}from"./chunk-
|
|
1
|
+
import{A as s,B as t,C as u,D as v,E as w,F as x,G as y,H as z,I as A,J as B,K as C,L as D,M as E,i as a,j as b,k as c,l as d,m as e,n as f,o as g,p as h,q as i,r as j,s as k,t as l,u as m,v as n,w as o,x as p,y as q,z as r}from"./chunk-VT6SLBOU.js";import"./chunk-TMB56NCA.js";import"./chunk-E4ME7J6G.js";import"./chunk-7BEWBRMT.js";export{a as XRAY_ENTITLED_EXTRACT_BILLING_CLASS,q as abandonExtractSettlement,p as claimFailedExtractJobForRefinalize,z as completeExtractJob,w as countSuccessfulPages,d as createExtractJob,f as createOrGetExtractJob,b as extractJobLimitInfo,A as failExtractJob,o as failStaleRunningExtractJob,B as failUnfundedExtractJob,E as finalizeXRayEntitledExtractJobCharge,y as finishExtractJob,g as getExtractJob,e as getExtractJobByIdempotencyKey,v as getExtractedImageLinks,u as getExtractedPages,x as getExtractedUrls,h as listExtractJobs,t as listExtractPages,j as listFundedPendingExtractJobs,n as listStaleRunningExtractJobs,i as listUnsettledExtractJobs,k as markExtractJobDispatchAttempt,l as recordExtractJobDispatchFailure,m as recordExtractSettlementFailure,s as saveExtractPages,C as setExtractJobPublicError,r as setExtractJobTotal,D as settleExtractJob,c as terminalExtractJobStatus};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
import{ab as S,cb as h}from"./chunk-
|
|
1
|
+
import{ab as S,cb as h}from"./chunk-52XUWR3X.js";import"./chunk-5B42QVDC.js";import"./chunk-HR2A2WTL.js";import"./chunk-2GRE4BAO.js";import"./chunk-D6BTAT5Y.js";import"./chunk-E4ME7J6G.js";import"./chunk-DKQZNSYM.js";import{H as d,I as u,J as l,K as f,L as v}from"./chunk-7BEWBRMT.js";var g=12;function b(e){return e&&typeof e=="object"&&"code"in e&&typeof e.code=="string"?e.code:e instanceof Error?e.name:"unknown_error"}function k(e){return Math.min(21600,30*2**Math.max(0,e-1))}async function x(e={}){let w=e.stripeClient??S,E=Math.max(1,Math.min(100,e.limit??100)),r=500,i=0,s=0,c=0,m=0;for(;i<r;){let a=await d(Math.min(E,r-i),e.leaseSeconds??300);if(a.length===0)break;i+=a.length;for(let t of a)try{let n=await w.events.retrieve(t.eventId);await h(n),await u(t.eventId,t.token)&&(s+=1)}catch(n){let p=n instanceof Error?n.message:String(n),o=b(n);t.attemptCount>=g||o==="resource_missing"?await f(t.eventId,t.token,p,o)&&(m+=1):await l(t.eventId,t.token,p,o,k(t.attemptCount))&&(c+=1)}}return{claimed:i,processed:s,retried:c,deadLettered:m,capped:i>=r,health:await v()}}export{x as drainStripeEventQueue};
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
import{a as O,b as y,c as D,d as g,f as A,g as _}from"./chunk-PHBU6X5B.js";import"./chunk-XG6GCEUE.js";import{u as k}from"./chunk-F4LIB772.js";import"./chunk-M5QHXNFZ.js";import{b as f,c as H}from"./chunk-7AC66V3T.js";import{a as m,b}from"./chunk-E4ME7J6G.js";import{a as v}from"./chunk-4FROKQJN.js";import{b as M}from"./chunk-DKQZNSYM.js";import"./chunk-2TZBO52D.js";import"./chunk-2PXP7TMY.js";import{Mb as l,Qb as p,Sb as d,Ua as u,gb as h,ib as c,na as o}from"./chunk-7BEWBRMT.js";var P=2,w=0;function C(e){if(!e||typeof e!="object")return 0;let t=e;return typeof t.totalQuestions=="number"?t.totalQuestions:Array.isArray(t.flat)?t.flat.length:0}function R(e){return m.paa_base+Math.max(1,e)*m.paa}async function N(e){w++;try{let t=typeof e.options=="string"?JSON.parse(e.options):e.options,r={value:null},i=await M({op:t.serpOnly?"serp":"paa",userId:Number(e.user_id),headlessSentOut:r},()=>k({...t,\u006b\u0065\u0072\u006e\u0065\u006cApiKey:v(),headless:!0,format:"json",outputDir:"/tmp/paa-output-api",onAttemptEvent:A(e.id,e.user_id)}));await h(e.id,i);let s=await o(e.id,e.user_id);if(typeof t.billingHoldMc=="number"&&t.billingDebitKey){let n=t.serpOnly?b(r.value):R(C(i));await d(Number(e.user_id),t.billingDebitKey,n,t.serpOnly?"serp_refund":"paa_refund",t.serpOnly?"SERP search settlement":"PAA harvest settlement",t.serpOnly?"serp_search":"paa_harvest")}else if(!t.serpOnly&&typeof t.billingHoldMc=="number"){let n=R(C(i)),a=t.billingHoldMc-n;a>0?await l(e.user_id,a,"paa_refund","overestimate refund"):a<0&&await p(e.user_id,-a,"paa",t.query??e.query)}else if(t.serpOnly&&typeof t.billingHoldMc=="number"){let n=b(r.value),a=t.billingHoldMc-n;a>0?await l(e.user_id,a,"serp_refund","headless-mode pricing settle"):a<0&&await p(e.user_id,-a,"serp",t.query??e.query)}e.callback_url&&await _(e.callback_url,{job_id:e.id,status:"done",result:H(i),attempts:f(s)})}catch(t){console.error("[harvest/worker] failed",{jobId:e.id,error:t instanceof Error?t.message:String(t)});let r=O(t),i=typeof e.options=="string"?JSON.parse(e.options):e.options,s=typeof i.billingHoldMc=="number"&&i.billingHoldMc>0;await c(e.id,y(r),g(r,{chargeStatus:s?"refund_pending":"not_charged"}));let n=await o(e.id,e.user_id);try{s&&(i.billingDebitKey?await d(Number(e.user_id),i.billingDebitKey,0,i.serpOnly?"serp_refund":"paa_refund","failed call",i.serpOnly?"serp_search":"paa_harvest"):await l(e.user_id,i.billingHoldMc,"refund","failed call"),await c(e.id,y(r),g(r,{chargeStatus:"refunded"})))}catch{}e.callback_url&&await _(e.callback_url,{job_id:e.id,status:"failed",...D(r),attempts:f(n)})}finally{w--}}async function x(){let e=await u();if(!e)return{claimed:!1};let t=Date.now();return await N(e),{claimed:!0,jobId:e.id,completed:!0,durationMs:Date.now()-t}}async function W(e){let t=[];for(let r=0;r<e.maxJobs&&!(Date.now()>=e.deadlineMs);r++){let i=await x();if(t.push(i),!i.claimed)break}return t}function B(){setInterval(async()=>{if(w>=P)return;let e=await u();e&&N(e)},2e3),console.log(`[worker] started \u2014 polling every 2s, max ${P} concurrent`)}export{W as drainQueue,B as startWorker,x as tickOnce};
|
package/package.json
CHANGED
package/dist/chunk-CULFQBFR.js
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
var o="0.89.6";export{o as a};
|
package/dist/db-IZ5UF33U.js
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
import{$ as M,$a as Ma,$b as Mb,A as l,Aa as la,Ab as lb,B as m,Ba as ma,Bb as mb,C as n,Ca as na,Cb as nb,D as o,Da as oa,Db as ob,E as p,Ea as pa,Eb as pb,F as q,Fa as qa,Fb as qb,G as r,Ga as ra,Gb as rb,H as s,Ha as sa,Hb as sb,I as t,Ia as ta,Ib as tb,J as u,Ja as ua,Jb as ub,K as v,Ka as va,Kb as vb,L as w,La as wa,Lb as wb,M as x,Ma as xa,Mb as xb,N as y,Na as ya,Nb as yb,O as z,Oa as za,Ob as zb,P as A,Pa as Aa,Pb as Ab,Q as B,Qa as Ba,Qb as Bb,R as C,Ra as Ca,Rb as Cb,S as D,Sa as Da,Sb as Db,T as E,Ta as Ea,Tb as Eb,U as F,Ua as Fa,Ub as Fb,V as G,Va as Ga,Vb as Gb,W as H,Wa as Ha,Wb as Hb,X as I,Xa as Ia,Xb as Ib,Y as J,Ya as Ja,Yb as Jb,Z as K,Za as Ka,Zb as Kb,_ as L,_a as La,_b as Lb,aa as N,ab as Na,ba as O,bb as Oa,ca as P,cb as Pa,da as Q,db as Qa,ea as R,eb as Ra,fa as S,fb as Sa,ga as T,gb as Ta,ha as U,hb as Ua,ia as V,ib as Va,ja as W,jb as Wa,ka as X,kb as Xa,la as Y,lb as Ya,ma as Z,mb as Za,na as _,nb as _a,oa as $,ob as $a,p as a,pa as aa,pb as ab,q as b,qa as ba,qb as bb,r as c,ra as ca,rb as cb,s as d,sa as da,sb as db,t as e,ta as ea,tb as eb,u as f,ua as fa,ub as fb,v as g,va as ga,vb as gb,w as h,wa as ha,wb as hb,x as i,xa as ia,xb as ib,y as j,ya as ja,yb as jb,z as k,za as ka,zb as kb}from"./chunk-YQ5U5C2X.js";export{b as CORE_SCHEMA_VERSION,a as CREDIT_LOTS_EPOCH,rb as CREDIT_LOT_TTL,c as SITE_EXTRACT_COUNTER_REPAIR_VERSION,Fb as SiteAuditJobRowSchema,Gb as SiteAuditPhaseLogRowSchema,wa as advanceWorkflowRunStep,ua as appendWorkflowArtifacts,ib as applySubscriptionInvoiceGrant,V as attachHarvestAttemptProviderSession,Ua as cancelJob,J as checkRateLimit,Fa as checkpointDurablePaaJob,Ea as checkpointJob,s as claimDueStripeEvents,Da as claimDurablePaaJob,xb as claimMonthlyFreeRefresh,Ca as claimPendingJob,Wa as claimPendingKpoJob,Ha as completeDurablePaaJob,Ka as completeExpiredDurablePaaJob,Qa as completeJob,Za as completeKpoJob,Ra as completeRunningJob,t as completeStripeEvent,ra as completeWorkflowRunRecord,m as consumeCode,L as consumePasswordResetToken,Ba as countActiveJobsForUser,Aa as countActiveUsers,Q as createJob,R as createJobWithId,Va as createKpoJob,K as createPasswordResetToken,S as createRunningJob,I as createUser,ma as createWorkflowRun,fa as createWorkflowSchedule,ub as creditMc,vb as creditMcIdempotent,P as deactivateUser,v as deadLetterStripeEvent,yb as debitMc,zb as debitMcIdempotent,ja as deleteWorkflowSchedule,r as enqueueStripeEvent,sb as expireOldLots,Ia as failDurablePaaJob,La as failExpiredDurablePaaJob,Sa as failJob,_a as failKpoJob,Ta as failRunningJob,sa as failWorkflowRunRecord,W as finalizeExpiredHarvestAttempts,X as finishHarvestAttempt,y as generateApiKey,Cb as getBillingLedgerSummary,k as getClient,jb as getConnectedAccountBillingState,e as getDb,Ib as getInboxMessage,$ as getJob,Xa as getKpoJob,ab as getKpoPhaseLog,Bb as getLedger,nb as getMemoryPlan,ob as getMemoryProvisioned,T as getOrCreateRunningJob,ba as getPageSnapshot,o as getRefresh,w as getStripeEventQueueHealth,C as getUserByApiKey,B as getUserByEmail,D as getUserById,cb as getUserByStripeCustomerId,E as getUserMemoryCreds,H as getUserStats,ya as getWorkflowArtifact,na as getWorkflowRun,oa as getWorkflowRunByIdempotencyKey,va as getWorkflowRunStepState,ga as getWorkflowSchedule,z as hashPassword,Lb as insertInboxMessage,d as inspectCurrentCoreSchema,wb as ledgerExistsForOperation,Db as ledgerExistsForStripePI,Mb as listActiveUserIds,ka as listDueWorkflowSchedules,_ as listHarvestAttempts,Z as listHarvestAttemptsPendingProviderTelemetry,Hb as listInboxMessages,aa as listJobs,bb as listKpoJobs,Ma as listRecentTerminalDurablePaaJobs,Oa as listRecoverableDurablePaaJobs,ea as listRequestEvents,Na as listStaleRunningDurablePaaJobs,O as listUsers,lb as listUsersDueForConnectedAccountBillingReconciliation,xa as listWorkflowArtifacts,pa as listWorkflowRuns,ha as listWorkflowSchedules,$a as logKpoPhaseComplete,da as logRequestEvent,Jb as markInboxMessageOpened,Pa as markJobBillingSettled,qa as markWorkflowRunRunning,la as markWorkflowScheduleRan,i as migrate,tb as migrateExistingBalancesToLots,g as migrateSecuredMainOAuthSchema,ia as patchWorkflowSchedule,l as putCode,n as putRefresh,Eb as reconcileBalanceMc,za as recordWorkflowWebhookDelivery,j as registerClient,Ga as renewDurablePaaLease,ta as replaceWorkflowArtifacts,Ja as requeueExpiredDurablePaaJob,qb as resolveUserByMemorySubscription,u as retryStripeEvent,N as revokeApiKey,p as revokeRefresh,M as rotateApiKey,q as rotateRefresh,fb as setConcurrencyPackQuantity,gb as setConcurrencySubId,kb as setConnectedAccountBillingState,f as setDbForTesting,eb as setExtraConcurrencySlots,mb as setMemoryPlan,pb as setMemoryProvisioned,G as setPassword,db as setStripeCustomerId,hb as setSubscriptionTier,F as setUserMemoryCreds,Ab as settleDebitMcIdempotent,Kb as softDeleteInboxMessage,U as startHarvestAttempt,x as stripeEventAlreadyProcessed,Y as updateHarvestAttemptProviderTelemetry,Ya as updateKpoJobState,ca as upsertPageSnapshot,A as verifyPassword,h as verifySecuredMainOAuthSchema};
|
package/dist/worker-MUHILD2W.js
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
import{a as h,b as f,c as O,d as y,f as A,g as b}from"./chunk-FKSTNWP6.js";import"./chunk-YQZGZBB4.js";import{u as k}from"./chunk-C6JXNOSF.js";import"./chunk-QPWPR5XG.js";import{b as p,c as M}from"./chunk-KKI2SNB2.js";import{a as m,b as H}from"./chunk-MZN4U5BL.js";import{a as w}from"./chunk-4FROKQJN.js";import{b as _}from"./chunk-WUVMD6ZX.js";import"./chunk-3GP5CYZX.js";import{Jb as o,Nb as d,Ra as u,db as v,fb as c,na as l}from"./chunk-YQ5U5C2X.js";var C=2,g=0;function R(e){if(!e||typeof e!="object")return 0;let t=e;return typeof t.totalQuestions=="number"?t.totalQuestions:Array.isArray(t.flat)?t.flat.length:0}function x(e){return m.paa_base+Math.max(1,e)*m.paa}async function P(e){g++;try{let t=typeof e.options=="string"?JSON.parse(e.options):e.options,r={value:null},a=await _({op:t.serpOnly?"serp":"paa",userId:Number(e.user_id),headlessSentOut:r},()=>k({...t,\u006b\u0065\u0072\u006e\u0065\u006cApiKey:w(),headless:!0,format:"json",outputDir:"/tmp/paa-output-api",onAttemptEvent:A(e.id,e.user_id)}));await v(e.id,a);let s=await l(e.id,e.user_id);if(!t.serpOnly&&typeof t.billingHoldMc=="number"){let n=x(R(a)),i=t.billingHoldMc-n;i>0?await o(e.user_id,i,"paa_refund","overestimate refund"):i<0&&await d(e.user_id,-i,"paa",t.query??e.query)}else if(t.serpOnly&&typeof t.billingHoldMc=="number"){let n=H(r.value),i=t.billingHoldMc-n;i>0?await o(e.user_id,i,"serp_refund","headless-mode pricing settle"):i<0&&await d(e.user_id,-i,"serp",t.query??e.query)}e.callback_url&&await b(e.callback_url,{job_id:e.id,status:"done",result:M(a),attempts:p(s)})}catch(t){console.error("[harvest/worker] failed",{jobId:e.id,error:t instanceof Error?t.message:String(t)});let r=h(t),a=typeof e.options=="string"?JSON.parse(e.options):e.options,s=typeof a.billingHoldMc=="number"&&a.billingHoldMc>0;await c(e.id,f(r),y(r,{chargeStatus:s?"refund_pending":"not_charged"}));let n=await l(e.id,e.user_id);try{s&&(await o(e.user_id,a.billingHoldMc,"refund","failed call"),await c(e.id,f(r),y(r,{chargeStatus:"refunded"})))}catch{}e.callback_url&&await b(e.callback_url,{job_id:e.id,status:"failed",...O(r),attempts:p(n)})}finally{g--}}async function J(){let e=await u();if(!e)return{claimed:!1};let t=Date.now();return await P(e),{claimed:!0,jobId:e.id,completed:!0,durationMs:Date.now()-t}}async function W(e){let t=[];for(let r=0;r<e.maxJobs&&!(Date.now()>=e.deadlineMs);r++){let a=await J();if(t.push(a),!a.claimed)break}return t}function B(){setInterval(async()=>{if(g>=C)return;let e=await u();e&&P(e)},2e3),console.log(`[worker] started \u2014 polling every 2s, max ${C} concurrent`)}export{W as drainQueue,B as startWorker,J as tickOnce};
|