mcp-scraper 0.89.5 → 0.89.6
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 +14 -0
- package/README.md +1 -1
- package/dist/{analytics-repository-RG7CFK6W.js → analytics-repository-4UA3WI3U.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-WYCXAFID.js → chunk-3H7HU5JI.js} +1 -1
- package/dist/{chunk-SEHBXTMX.js → chunk-C6JXNOSF.js} +1 -1
- package/dist/chunk-CULFQBFR.js +1 -0
- package/dist/chunk-D6BTAT5Y.js +1 -0
- package/dist/{chunk-3UPA3XHF.js → chunk-FKSTNWP6.js} +1 -1
- package/dist/{chunk-UB5N2ICD.js → chunk-FVSXX43Z.js} +1 -1
- package/dist/{chunk-DBDMD7V7.js → chunk-G6GAIXXA.js} +2 -2
- package/dist/{chunk-4SQ4RKDX.js → chunk-H2I63VAD.js} +1 -1
- package/dist/chunk-H3QW5U67.js +41 -0
- package/dist/chunk-ISZIA6NE.js +820 -0
- package/dist/{chunk-O2QKRYLS.js → chunk-KKI2SNB2.js} +1 -1
- package/dist/{chunk-A2CZ7WRV.js → chunk-PPOI4C3Q.js} +1 -1
- package/dist/{chunk-RZSJC7FX.js → chunk-VZTFJLOK.js} +1 -1
- package/dist/{chunk-53HIX3X7.js → chunk-WUVMD6ZX.js} +1 -1
- package/dist/{chunk-JJDYKM5I.js → chunk-WXOLIRFJ.js} +1 -1
- package/dist/chunk-Y3ITV2OK.js +19 -0
- package/dist/{chunk-NGQYSQQA.js → chunk-YCZKKSTT.js} +1 -1
- package/dist/{chunk-7JAMFXNL.js → chunk-YQ5U5C2X.js} +121 -73
- package/dist/{chunk-4A5F7IUA.js → chunk-Z45GM22P.js} +190 -190
- package/dist/db-IZ5UF33U.js +1 -0
- package/dist/{extract-bundle-RVOBMGU2.js → extract-bundle-BTZQJTYV.js} +1 -1
- package/dist/gmail-service-ZZDF52KK.js +1 -0
- package/dist/index.cjs +19 -19
- package/dist/index.js +1 -1
- package/dist/{lead-list-enrichment-repository-GKVCX7BM.js → lead-list-enrichment-repository-K4H5DMSF.js} +1 -1
- package/dist/{location-data-repository-WK27TSHZ.js → location-data-repository-KALKE3FG.js} +1 -1
- package/dist/{server-7TSXB24M.js → server-YVNOU3RK.js} +727 -1546
- package/dist/{site-extract-repository-FAHOTQGX.js → site-extract-repository-FTTIOWAL.js} +1 -1
- package/dist/stripe-event-worker-CURL3RYL.js +1 -0
- package/dist/{worker-ROG7SO3R.js → worker-MUHILD2W.js} +1 -1
- package/package.json +1 -1
- package/dist/chunk-2ICI4TIX.js +0 -59
- package/dist/chunk-SCHTLEMZ.js +0 -1
- package/dist/db-DV4HHCYP.js +0 -1
- package/dist/gmail-service-FJTBCZPH.js +0 -1
|
@@ -0,0 +1 @@
|
|
|
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};
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import{c as Je,h as ke,i as Ke,j as Qe,l as Ve}from"./chunk-
|
|
1
|
+
import{c as Je,h as ke,i as Ke,j as Qe,l as Ve}from"./chunk-3H7HU5JI.js";import{b as Ge,c as Ye,h as Xe,j as Ze}from"./chunk-WXOLIRFJ.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{t as qe}from"./chunk-YQ5U5C2X.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
|
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
import{e as a,f as b,g as c,h as d,i as e,j as f,k as g,l as h}from"./chunk-H3QW5U67.js";import"./chunk-Y3ITV2OK.js";import"./chunk-YCZKKSTT.js";import"./chunk-HUV2WTRW.js";import"./chunk-MZN4U5BL.js";import"./chunk-WUVMD6ZX.js";import"./chunk-YQ5U5C2X.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};
|