mcp-scraper 0.89.0 → 0.89.2

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 (38) hide show
  1. package/CHANGELOG.md +37 -0
  2. package/README.md +3 -1
  3. package/dist/{analytics-repository-2BMT5JNE.js → analytics-repository-X6J7TWTL.js} +1 -1
  4. package/dist/bin/api-server.js +1 -1
  5. package/dist/bin/mcp-scraper-cli.js +1 -1
  6. package/dist/bin/mcp-scraper-core.js +1 -1
  7. package/dist/bin/mcp-scraper-install.js +1 -1
  8. package/dist/bin/mcp-stdio-server.js +1 -1
  9. package/dist/bin/paa-harvest.js +1 -1
  10. package/dist/{chunk-4AI7DOS7.js → chunk-2O3UYISB.js} +1 -1
  11. package/dist/{chunk-PUHFVA7P.js → chunk-3MUEBDWY.js} +1 -1
  12. package/dist/{chunk-MASR22K4.js → chunk-D4NOJAUN.js} +1 -1
  13. package/dist/{chunk-WSCGYRWA.js → chunk-DEPJ4ZZ6.js} +1 -1
  14. package/dist/{chunk-HDPYG3XV.js → chunk-DMCYIWB3.js} +1 -1
  15. package/dist/{chunk-XPZVJIZ2.js → chunk-EN66HBAW.js} +1 -1
  16. package/dist/{chunk-E2WRWV3A.js → chunk-HYQDXTS7.js} +1 -1
  17. package/dist/{chunk-7YGVI5J4.js → chunk-J7EVLE6H.js} +1 -1
  18. package/dist/{chunk-TXENITMS.js → chunk-K7KMLT33.js} +1 -1
  19. package/dist/{chunk-L4CGLFPU.js → chunk-MKEMCMRO.js} +1 -1
  20. package/dist/{chunk-X54CQLK2.js → chunk-S7RA23GR.js} +1 -1
  21. package/dist/chunk-TW46NKVH.js +1 -0
  22. package/dist/{chunk-CFI6CXIV.js → chunk-VEJQ4XPI.js} +1 -1
  23. package/dist/{chunk-M22MM4N4.js → chunk-Z27SYWYD.js} +1 -1
  24. package/dist/{chunk-GMWKPIYX.js → chunk-Z66Z2HV4.js} +182 -79
  25. package/dist/{chunk-CCYWSJNG.js → chunk-ZWZ6OWY6.js} +1 -1
  26. package/dist/db-NLGNTQC3.js +1 -0
  27. package/dist/{extract-bundle-SNUIHM3J.js → extract-bundle-TXH3VNXH.js} +1 -1
  28. package/dist/{gmail-service-BZ3H75XC.js → gmail-service-J2DWSQJX.js} +1 -1
  29. package/dist/index.cjs +143 -32
  30. package/dist/index.js +1 -1
  31. package/dist/{lead-list-enrichment-repository-S2H3U7T7.js → lead-list-enrichment-repository-G7Z5DQZR.js} +1 -1
  32. package/dist/{location-data-repository-OTWHWMV6.js → location-data-repository-TEXZ3I2N.js} +1 -1
  33. package/dist/{server-RFR2A5UJ.js → server-LBOFC2EE.js} +849 -816
  34. package/dist/{site-extract-repository-SE776XDC.js → site-extract-repository-MBSCAP4D.js} +1 -1
  35. package/dist/{worker-XUDSM3AL.js → worker-SUCJKWU2.js} +1 -1
  36. package/package.json +1 -1
  37. package/dist/chunk-Z2QGQJS2.js +0 -1
  38. package/dist/db-F2MX63GI.js +0 -1
@@ -1,4 +1,4 @@
1
- var s={message:"Discover public Facebook Reel URLs, retain partial inventories, and transcribe selected videos. Facebook extraction now dismisses optional login dialogs and retries failed public acquisition."};var _={reset:"\x1B[0m",cyan:"\x1B[36m",lime:"\x1B[32m",amber:"\x1B[33m",red:"\x1B[31m",muted:"\x1B[90m",bold:"\x1B[1m"};function r(t,e,n){return n?`${_[e]}${t}${_.reset}`:t}function o(t,e,n){let a=t.padEnd(9," ");return` ${r(a,"muted",n)} ${e.join(r(" . ","muted",n))}`}function p(t){let e=t.color??!0,n=t.apiKeyConfigured?"$MCP_SCRAPER_API_KEY":"sk_live_your_key",a=String.raw`
1
+ var s={message:"MCP OAuth sign-in now fails closed on incomplete configuration, keeps registration limits with the active authorization store, and uses atomic token replay and rotation for more reliable connections."};var _={reset:"\x1B[0m",cyan:"\x1B[36m",lime:"\x1B[32m",amber:"\x1B[33m",red:"\x1B[31m",muted:"\x1B[90m",bold:"\x1B[1m"};function r(t,e,n){return n?`${_[e]}${t}${_.reset}`:t}function o(t,e,n){let a=t.padEnd(9," ");return` ${r(a,"muted",n)} ${e.join(r(" . ","muted",n))}`}function p(t){let e=t.color??!0,n=t.apiKeyConfigured?"$MCP_SCRAPER_API_KEY":"sk_live_your_key",a=String.raw`
2
2
  __ __ ____ ____
3
3
  | \/ |/ ___| _ \
4
4
  | |\/| | | | |_) |
@@ -0,0 +1 @@
1
+ import{$ as M,$a as Ma,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,Y as J,Ya as Ja,Z as K,Za as Ka,_ as L,_a as La,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-Z66Z2HV4.js";export{b as CORE_SCHEMA_VERSION,a as CREDIT_LOTS_EPOCH,mb as CREDIT_LOT_TTL,c as SITE_EXTRACT_COUNTER_REPAIR_VERSION,Ab as SiteAuditJobRowSchema,Bb as SiteAuditPhaseLogRowSchema,sa as advanceWorkflowRunStep,qa as appendWorkflowArtifacts,R as attachHarvestAttemptProviderSession,Qa as cancelJob,F as checkRateLimit,Ba as checkpointDurablePaaJob,Aa as checkpointJob,za as claimDurablePaaJob,sb as claimMonthlyFreeRefresh,ya as claimPendingJob,Sa as claimPendingKpoJob,q as claimStripeEvent,Da as completeDurablePaaJob,Ga as completeExpiredDurablePaaJob,Ma as completeJob,Va as completeKpoJob,Na as completeRunningJob,r as completeStripeEvent,na as completeWorkflowRunRecord,l as consumeCode,H as consumePasswordResetToken,xa as countActiveJobsForUser,wa as countActiveUsers,M as createJob,N as createJobWithId,Ra as createKpoJob,G as createPasswordResetToken,O as createRunningJob,E as createUser,ia as createWorkflowRun,ba as createWorkflowSchedule,pb as creditMc,qb as creditMcIdempotent,L as deactivateUser,tb as debitMc,ub as debitMcIdempotent,fa as deleteWorkflowSchedule,nb as expireOldLots,Ea as failDurablePaaJob,Ha as failExpiredDurablePaaJob,Oa as failJob,Wa as failKpoJob,Pa as failRunningJob,s as failStripeEvent,oa as failWorkflowRunRecord,S as finalizeExpiredHarvestAttempts,T as finishHarvestAttempt,u as generateApiKey,xb as getBillingLedgerSummary,j as getClient,eb as getConnectedAccountBillingState,d as getDb,Db as getInboxMessage,X as getJob,Ta as getKpoJob,Ya as getKpoPhaseLog,wb as getLedger,ib as getMemoryPlan,jb as getMemoryProvisioned,P as getOrCreateRunningJob,Z as getPageSnapshot,n as getRefresh,y as getUserByApiKey,x as getUserByEmail,z as getUserById,_a as getUserByStripeCustomerId,A as getUserMemoryCreds,D as getUserStats,ua as getWorkflowArtifact,ja as getWorkflowRun,ka as getWorkflowRunByIdempotencyKey,ra as getWorkflowRunStepState,ca as getWorkflowSchedule,v as hashPassword,Gb as insertInboxMessage,rb as ledgerExistsForOperation,yb as ledgerExistsForStripePI,Hb as listActiveUserIds,ga as listDueWorkflowSchedules,W as listHarvestAttempts,V as listHarvestAttemptsPendingProviderTelemetry,Cb as listInboxMessages,Y as listJobs,Za as listKpoJobs,Ia as listRecentTerminalDurablePaaJobs,Ka as listRecoverableDurablePaaJobs,aa as listRequestEvents,Ja as listStaleRunningDurablePaaJobs,K as listUsers,gb as listUsersDueForConnectedAccountBillingReconciliation,ta as listWorkflowArtifacts,la as listWorkflowRuns,da as listWorkflowSchedules,Xa as logKpoPhaseComplete,$ as logRequestEvent,Eb as markInboxMessageOpened,La as markJobBillingSettled,ma as markWorkflowRunRunning,ha as markWorkflowScheduleRan,h as migrate,ob as migrateExistingBalancesToLots,f as migrateSecuredMainOAuthSchema,ea as patchWorkflowSchedule,k as putCode,m as putRefresh,zb as reconcileBalanceMc,va as recordWorkflowWebhookDelivery,i as registerClient,Ca as renewDurablePaaLease,pa as replaceWorkflowArtifacts,Fa as requeueExpiredDurablePaaJob,lb as resolveUserByMemorySubscription,J as revokeApiKey,o as revokeRefresh,I as rotateApiKey,p as rotateRefresh,bb as setConcurrencyPackQuantity,cb as setConcurrencySubId,fb as setConnectedAccountBillingState,e as setDbForTesting,ab as setExtraConcurrencySlots,hb as setMemoryPlan,kb as setMemoryProvisioned,C as setPassword,$a as setStripeCustomerId,db as setSubscriptionTier,B as setUserMemoryCreds,vb as settleDebitMcIdempotent,Fb as softDeleteInboxMessage,Q as startHarvestAttempt,t as stripeEventAlreadyProcessed,U as updateHarvestAttemptProviderTelemetry,Ua as updateKpoJobState,_ as upsertPageSnapshot,w as verifyPassword,g as verifySecuredMainOAuthSchema};
@@ -1,4 +1,4 @@
1
- import{c as Je,h as ke,i as Ke,j as Qe,l as Ve}from"./chunk-TXENITMS.js";import{b as Ge,c as Ye,h as Xe,j as Ze}from"./chunk-CFI6CXIV.js";import"./chunk-YQZGZBB4.js";import"./chunk-QPWPR5XG.js";import{c as ze,e as Fe,f as He,g as We}from"./chunk-W2BVJ7S2.js";import"./chunk-TMB56NCA.js";import"./chunk-MZN4U5BL.js";import"./chunk-HE45FFBU.js";import{s as qe}from"./chunk-GMWKPIYX.js";import{createWriteStream as se,mkdirSync as ie,readFileSync as bt,readdirSync as xt,rmSync as kt,statSync as _t,writeFileSync as g}from"fs";import{createHash as G}from"crypto";import{basename as Pt,join as a}from"path";import{tmpdir as Mt}from"os";import{createGzip as It}from"zlib";import{once as Y}from"events";import{finished as ae}from"stream/promises";import{ZipFile as Ct}from"yazl";import At from"p-limit";import{createHash as _e}from"crypto";var pt=500,gt=.9,ht=1e4,ft=5e4;function ge(r,s){let n=10**s;return Math.round(r*n)/n}function yt(r,s){if(r.length===0||r.length!==s.length)throw new Error("Similarity vectors must have the same non-zero dimension.");let n=0,c=0,b=0;for(let w=0;w<r.length;w++)n+=r[w]*s[w],c+=r[w]*r[w],b+=s[w]*s[w];return c===0||b===0?0:Math.max(-1,Math.min(1,n/(Math.sqrt(c)*Math.sqrt(b))))}function wt(r){let s=_e("sha256");for(let n of[...r].sort((c,b)=>c.url.localeCompare(b.url)))s.update(n.url),s.update("\0"),s.update(n.contentHash),s.update("\0");return s.digest("hex")}function et(r){let s=r.replace(/!\[([^\]]*)\]\([^)]*\)/g,"$1").replace(/\[([^\]]+)\]\([^)]*\)/g,"$1").replace(/https?:\/\/\S+/gi," ").replace(/&(?:amp|nbsp|quot|#39);/gi," ").replace(/[#>*_`~|\\-]+/g," ").replace(/[^\p{L}\p{N}]+/gu," ").replace(/\s+/g," ").trim().toLowerCase();return s.length>=12?s:null}function St(r){let s=r.map(f=>f.bodyMarkdown.split(/\n{2,}/).map(M=>M.trim()).filter(Boolean)),n=r.length>=4?Math.max(3,Math.ceil(r.length*.5)):null,c=new Map;for(let f of s){let M=new Set(f.map(et).filter(I=>!!I));for(let I of M)c.set(I,(c.get(I)??0)+1)}let b=new Set;if(n!=null)for(let[f,M]of c)M>=n&&b.add(f);let w=0;return{texts:r.map((f,M)=>{let I=[];for(let _ of s[M]){let T=et(_);if(T&&b.has(T)){w+=_.length;continue}I.push(_)}let A=I.join(`
1
+ import{c as Je,h as ke,i as Ke,j as Qe,l as Ve}from"./chunk-K7KMLT33.js";import{b as Ge,c as Ye,h as Xe,j as Ze}from"./chunk-VEJQ4XPI.js";import"./chunk-YQZGZBB4.js";import"./chunk-QPWPR5XG.js";import{c as ze,e as Fe,f as He,g as We}from"./chunk-W2BVJ7S2.js";import"./chunk-TMB56NCA.js";import"./chunk-MZN4U5BL.js";import"./chunk-HE45FFBU.js";import{s as qe}from"./chunk-Z66Z2HV4.js";import{createWriteStream as se,mkdirSync as ie,readFileSync as bt,readdirSync as xt,rmSync as kt,statSync as _t,writeFileSync as g}from"fs";import{createHash as G}from"crypto";import{basename as Pt,join as a}from"path";import{tmpdir as Mt}from"os";import{createGzip as It}from"zlib";import{once as Y}from"events";import{finished as ae}from"stream/promises";import{ZipFile as Ct}from"yazl";import At from"p-limit";import{createHash as _e}from"crypto";var pt=500,gt=.9,ht=1e4,ft=5e4;function ge(r,s){let n=10**s;return Math.round(r*n)/n}function yt(r,s){if(r.length===0||r.length!==s.length)throw new Error("Similarity vectors must have the same non-zero dimension.");let n=0,c=0,b=0;for(let w=0;w<r.length;w++)n+=r[w]*s[w],c+=r[w]*r[w],b+=s[w]*s[w];return c===0||b===0?0:Math.max(-1,Math.min(1,n/(Math.sqrt(c)*Math.sqrt(b))))}function wt(r){let s=_e("sha256");for(let n of[...r].sort((c,b)=>c.url.localeCompare(b.url)))s.update(n.url),s.update("\0"),s.update(n.contentHash),s.update("\0");return s.digest("hex")}function et(r){let s=r.replace(/!\[([^\]]*)\]\([^)]*\)/g,"$1").replace(/\[([^\]]+)\]\([^)]*\)/g,"$1").replace(/https?:\/\/\S+/gi," ").replace(/&(?:amp|nbsp|quot|#39);/gi," ").replace(/[#>*_`~|\\-]+/g," ").replace(/[^\p{L}\p{N}]+/gu," ").replace(/\s+/g," ").trim().toLowerCase();return s.length>=12?s:null}function St(r){let s=r.map(f=>f.bodyMarkdown.split(/\n{2,}/).map(M=>M.trim()).filter(Boolean)),n=r.length>=4?Math.max(3,Math.ceil(r.length*.5)):null,c=new Map;for(let f of s){let M=new Set(f.map(et).filter(I=>!!I));for(let I of M)c.set(I,(c.get(I)??0)+1)}let b=new Set;if(n!=null)for(let[f,M]of c)M>=n&&b.add(f);let w=0;return{texts:r.map((f,M)=>{let I=[];for(let _ of s[M]){let T=et(_);if(T&&b.has(T)){w+=_.length;continue}I.push(_)}let A=I.join(`
2
2
 
3
3
  `).trim(),C=A.length>=60?A:f.bodyMarkdown;return`${f.title?.trim()||"Untitled page"}
4
4
 
@@ -1 +1 @@
1
- import{A as d,B as e,C as f,D as g,E as h,x as a,y as b,z as c}from"./chunk-4AI7DOS7.js";import"./chunk-M22MM4N4.js";import"./chunk-HUV2WTRW.js";import"./chunk-MZN4U5BL.js";import"./chunk-MASR22K4.js";import"./chunk-GMWKPIYX.js";export{d as GMAIL_ATTACHMENT_REF_TTL_MS,c as GMAIL_INLINE_BODY_BYTES,b as GMAIL_MAIN_CONTRACT_HASH,a as GMAIL_MAIN_CONTRACT_VERSION,h as GmailService,e as GmailServiceError,g as normalizeGmailMessage,f as parseGmailAddresses};
1
+ import{A as d,B as e,C as f,D as g,E as h,x as a,y as b,z as c}from"./chunk-2O3UYISB.js";import"./chunk-Z27SYWYD.js";import"./chunk-HUV2WTRW.js";import"./chunk-MZN4U5BL.js";import"./chunk-D4NOJAUN.js";import"./chunk-Z66Z2HV4.js";export{d as GMAIL_ATTACHMENT_REF_TTL_MS,c as GMAIL_INLINE_BODY_BYTES,b as GMAIL_MAIN_CONTRACT_HASH,a as GMAIL_MAIN_CONTRACT_VERSION,h as GmailService,e as GmailServiceError,g as normalizeGmailMessage,f as parseGmailAddresses};