@smart-cloud/publisher-exporter 1.1.1 → 1.1.2
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.
- package/dist/crawl.js +3 -3
- package/package.json +1 -1
package/dist/crawl.js
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import {chromium}from'playwright';import S from'fs/promises';import f from'path';import {gunzipSync}from'zlib';import bt from'crypto';import Ct from'fast-glob';import {availableParallelism}from'os';import {Worker}from'worker_threads';var gt=".deploy-plan.json",ht="deploy-plan.json";function ft(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.resolve(e):""}function Ae(e){let t=String(e||"").replace(/\\/g,"/").replace(/^\/+/,"").trim();return !t||t==="."||t.startsWith("../")||t===".."?null:t}function
|
|
1
|
+
import {chromium}from'playwright';import S from'fs/promises';import f from'path';import {gunzipSync}from'zlib';import bt from'crypto';import Ct from'fast-glob';import {availableParallelism}from'os';import {Worker}from'worker_threads';var gt=".deploy-plan.json",ht="deploy-plan.json";function ft(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.resolve(e):""}function Ae(e){let t=String(e||"").replace(/\\/g,"/").replace(/^\/+/,"").trim();return !t||t==="."||t.startsWith("../")||t===".."?null:t}function ie(e){return [...new Set(e.map(Ae).filter(t=>t!==null))]}function mt(e,t){return {schemaVersion:1,generatedAt:String(t.generatedAt||new Date().toISOString()),outputDir:f.resolve(e.outputDir),runMode:t.runMode,crawlMode:t.crawlMode,fullSyncRequired:!!t.fullSyncRequired,changedFiles:ie(t.changedFiles||[]),deletedFiles:ie(t.deletedFiles||[]),rewriteTargets:ie(t.rewriteTargets||[])}}function wt(e){let t=f.join(f.resolve(e.outputDir),gt),r=ft();return r?[f.join(r,ht),t]:[t]}function V(e,t){let r=f.resolve(e),s=f.resolve(t),n=f.relative(r,s).replace(/\\/g,"/");return Ae(n)}async function ke(e,t){let r=mt(e,t);for(let s of wt(e))await S.mkdir(f.dirname(s),{recursive:true}),await S.writeFile(s,JSON.stringify(r,null,2),"utf8");}function Te(e){let t=e.trim();return t==="."?".":t.replace(/\/$/,"")}function yt(e){let t=String(e??"").trim();if(!t)return "";let r=t;if(/^https?:\/\//i.test(r))try{r=new URL(r).pathname;}catch{return ""}if(r=r.split(/[?#]/,1)[0]?.replace(/\\/g,"/").trim()??"",!r)return "";let s=r.endsWith("/"),n=r.replace(/^\/+/,"").split("/").map(o=>o.replace(/[^A-Za-z0-9._-]/g,"")).filter(o=>o.length>0&&o!=="."&&o!=="..");if(n.length===0)return "";let i=`/${n.join("/")}`;return s&&f.extname(i)===""&&(i+="/"),i==="/"?"":i}function ae(e){return !e||typeof e!="object"?{}:Object.fromEntries(Object.entries(e).map(([t,r])=>[t.trim(),String(r??"")]).filter(([t])=>t.length>0))}function St(e){if(!e||typeof e!="object")return {};let t={};for(let[r,s]of Object.entries(e)){let n=r.trim();if(!n||!s||typeof s!="object")continue;let i=s,o={};if(typeof i.targetOrigin=="string"){let a=Te(i.targetOrigin);a&&(o.targetOrigin=a);}let u=ae(i.extraReplacements);if(Object.keys(u).length>0&&(o.extraReplacements=u),i.s3&&typeof i.s3=="object"){let a={},c=i.s3;typeof c.bucket=="string"&&(a.bucket=c.bucket.trim()),typeof c.prefix=="string"&&(a.prefix=c.prefix.trim()),typeof c.region=="string"&&(a.region=c.region.trim()),typeof c.htmlCacheControl=="string"&&(a.htmlCacheControl=c.htmlCacheControl.trim()),typeof c.assetCacheControl=="string"&&(a.assetCacheControl=c.assetCacheControl.trim()),Object.keys(a).length>0&&(o.s3=a);}if(i.cloudFront&&typeof i.cloudFront=="object"){let a={},c=i.cloudFront;typeof c.distributionId=="string"&&(a.distributionId=c.distributionId.trim()),Array.isArray(c.invalidationPaths)&&(a.invalidationPaths=c.invalidationPaths.map(l=>String(l??"").trim()).filter(l=>l.length>0)),Object.keys(a).length>0&&(o.cloudFront=a);}t[n]=o;}return t}function oe(e,t){let s=String(e||"").replace(/\\/g,"/").trim().replace(/^\/+|\/+$/g,"");if(!s)return t;let n=s.split("/").map(i=>i.trim()).filter(i=>i.length>0&&i!=="."&&i!=="..");return n.length>0?n.join("/"):t}function vt(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.resolve(e):""}function Me(e,t,r){let s=String(t||"").trim();return s&&f.isAbsolute(s)?f.resolve(s):f.resolve(e,oe(s,r))}async function ue(){let e=process.env.PUBLISHER_CONFIG||"publisher.config.json",t=await S.readFile(e,"utf8"),r=JSON.parse(t);r.sourceOrigin=r.sourceOrigin.replace(/\/$/,""),r.targetOrigin=Te(r.targetOrigin),r.ignoreHttpsErrors??=false,r.outputDir=String(r.outputDir||"export").trim()||"export",r.urlRewriteMode||=r.targetOrigin==="."?"relative":"absolute",r.noJavaScriptRenderPathPrefixes||=[],r.seedPaths||=[],r.generated404RequestPath=yt(r.generated404RequestPath),r.sitemapPaths||=["/sitemap_index.xml","/sitemap.xml"],r.allowedAssetHosts||=[],r.assetPathPrefixes||=["/wp-content/","/wp-includes/","/static/","/assets/","/build/","/_next/","/docs/","/sitemap","/robots.txt","/llms.txt"],r.blockedPathPrefixes||=[],r.blockedSearchFragments||=[],r.concurrency||=1,r.maxPages||=0,r.extraReplacements=ae(r.extraReplacements),r.postCrawlCopyMap=ae(r.postCrawlCopyMap),r.logDir=String(r.logDir||"logs").trim()||"logs",r.verbose??=false,r.logLevel||=r.verbose?"debug":"info",r.s3SyncMode||="sdk-upload-delete",r.readiness||={waitForSelector:null,waitForFunction:null,timeoutMs:1500,fallbackWaitMs:1500},r.readiness.timeoutMs??=1500,r.readiness.fallbackWaitMs??=1500,r.viewport||={width:1440,height:1200},r.navigationTimeoutMs||=3e4,r.scheduler||={enabled:false,timezone:"UTC",rules:[]},r.scheduler.enabled??=false,r.scheduler.timezone||="UTC",r.scheduler.rules||=[],r.deploymentProfiles=St(r.deploymentProfiles),r.assetDownloadConcurrency=Number(r.assetDownloadConcurrency)>0?Number(r.assetDownloadConcurrency):r.concurrency,r.rewriteConcurrency=Number(r.rewriteConcurrency)>0?Number(r.rewriteConcurrency):r.assetDownloadConcurrency,r.defaultDeploymentProfile=String(r.defaultDeploymentProfile??"").trim(),r.defaultDeploymentProfile&&!r.deploymentProfiles[r.defaultDeploymentProfile]&&(r.defaultDeploymentProfile="");let s=vt(),n=s?f.resolve(s,".."):"";return n?(r.outputDir=Me(n,r.outputDir,"export"),r.logDir=Me(n,r.logDir,"logs")):(f.isAbsolute(r.outputDir)||(r.outputDir=oe(r.outputDir,"export")),f.isAbsolute(r.logDir)||(r.logDir=oe(r.logDir,"logs"))),r}function le(e,t=10){return bt.createHash("sha1").update(e).digest("hex").slice(0,t)}async function ce(e){await S.mkdir(f.dirname(e),{recursive:true});}function $(e){return [".html",".htm",".css",".js",".mjs",".json",".xml",".xsl",".txt",".svg",".map",".enc",".jws"].includes(f.extname(e).toLowerCase())}function F(e){return e.replace(/"/g,'"').replace(/"/g,'"').replace(/'/g,"'").replace(/'/g,"'").replace(/&/g,"&").replace(/</g,"<").replace(/>/g,">")}function Q(e){return e.replace(/&/g,"&").replace(/"/g,""").replace(/'/g,"'").replace(/</g,"<").replace(/>/g,">")}function G(e){return e.replace(/\//g,"\\/")}function Le(e){return e.replace(/\//g,"\\\\/")}function K(e,t,r){if(!t)return;e[t]=r;let s=G(t),n=G(r);e[s]=n;let i=Le(t),o=Le(r);e[i]=o;let u=Q(t),a=Q(r);e[u]=a;let c=Q(s),l=Q(n);e[c]=l;let p=Q(i),d=Q(o);e[p]=d;}function de(e,t){let r=t.startsWith("/")?t:`/${t}`,s=(e||"").replace(/\/$/,"");return !s||s==="."||s==="/"?r:`${s}${r}`}function De(e,t){try{return new URL(e,t==="."?"https://relative.invalid":t).pathname}catch{return e.startsWith("/")?e:`/${e.replace(/^\.\//,"")}`}}function Ie(e,t,r){if(!t)return r;let s=f.dirname(f.resolve(t)),n=f.resolve(e,r.replace(/^\/+/,"")),i=f.relative(s,n).replace(/\\/g,"/");return i?(i.startsWith(".")||(i=`./${i}`),i):"."}var kt=["wp-content/","wp-includes/","wp-admin/","wp-json/","_next/"],Mt=new Set([".html",".htm"]),Oe="WPSuite.io Static Publisher",Tt=Oe.replace(/[.*+?^${}()|[\]\\]/g,"\\$&");function Lt(e,t,r){if(e.wpsuite?.subscriber===true||!r)return false;let s=f.extname(r).toLowerCase();return Mt.has(s)?/<head\b|<html\b|<!doctype html/i.test(t):false}function Dt(e){if(new RegExp(`<meta\\b(?=[^>]*\\bname=(["'])generator\\1)(?=[^>]*\\bcontent=(["'])${Tt}\\2)[^>]*\\/?>`,"i").test(e))return e;let t=`<meta name="generator" content="${Oe}" />`;if(e.match(/(\r?\n)([ \t]*)<\/head>/i))return e.replace(/(\r?\n)([ \t]*)<\/head>/i,`$1$2${t}$1$2</head>`);let s=e.includes(`\r
|
|
2
2
|
`)?`\r
|
|
3
3
|
`:`
|
|
4
|
-
`;return e.replace(/<head\b[^>]*>/i,n=>`${n}${s} ${t}`)}function
|
|
4
|
+
`;return e.replace(/<head\b[^>]*>/i,n=>`${n}${s} ${t}`)}function It(e){let t=F(e).trim();return t.startsWith("{")||t.startsWith("[")||t.includes("\\/")||/"@context"|"@type"/.test(t)}function Et(e,t){let r=[...new Set(t)].filter(i=>i.includes("/")&&!/\\+\//.test(i)).map(i=>[i,G(i)]).filter(([i,o])=>i!==o).sort((i,o)=>o[0].length-i[0].length);if(!r.length)return e;let s=i=>{let o=i;for(let[u,a]of r)o=o.split(u).join(a);return o},n=e.replace(/(<script\b[^>]*\btype=["']application\/(?:ld\+)?json["'][^>]*>)([\s\S]*?)(<\/script>)/gi,(i,o,u,a)=>`${o}${s(u)}${a}`);return n=n.replace(/(<meta\b[^>]*\bcontent=(['"]))([\s\S]*?)(\2[^>]*>)/gi,(i,o,u,a,c)=>It(a)?`${o}${s(a)}${c}`:i),n}function Ot(e,t){if(!t)return null;let r=f.dirname(f.resolve(t)),s=f.resolve(e.outputDir),n=f.relative(r,s).replace(/\\/g,"/");return n?(n.startsWith(".")||(n=`./${n}`),n.endsWith("/")?n:`${n}/`):"./"}function Ft(e,t,r){let s=Ot(t,r);if(!s)return e;let n=e;for(let i of kt){let o=i.replace(/[.*+?^${}()|[\]\\]/g,"\\$&"),u=new RegExp(`(?:\\.{1,}/)+${o}`,"g"),a=new RegExp(`(?<!\\.)/${o}`,"g");n=n.replace(u,`${s}${i}`).replace(a,`${s}${i}`);let c=i.replace(/\//g,"\\/"),l=s.replace(/\//g,"\\/"),p=c.replace(/[.*+?^${}()|[\]\\]/g,"\\$&"),d=new RegExp(`(?:\\.{1,}\\\\/)+${p}`,"g"),g=new RegExp(`(?:\\.{1,}(?:\\\\/|\\\\))+${p}`,"g"),m=new RegExp(`(?<![\\.\\\\])\\/${p}`,"g");n=n.replace(d,`${l}${c}`).replace(g,`${l}${c}`).replace(m,`${l}${c}`);}return n}function ge(e,t,r){if(e.urlRewriteMode==="absolute")return t;let s=(()=>{try{return new URL(t,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin)}catch{return null}})(),n=s?s.pathname:De(t,e.targetOrigin),i=s?`${s.search}${s.hash}`:"";return e.urlRewriteMode==="root-relative"?`${n}${i}`:`${Ie(e.outputDir,r,n)}${i}`}function pe(e,t,r,s,n){K(e,r,ge(t,s,n));}function jt(e,t,r,s,n){if(!r||r===s)return;K(e,r,s);let i=ge(t,r,n),o=ge(t,s,n);K(e,i,o);}function he(e,t,r,s,n={}){let i=e,o={};pe(o,t,t.sourceOrigin,t.targetOrigin,s);for(let[a,c]of Object.entries(t.extraReplacements))pe(o,t,a,c,s);for(let[a,c]of Object.entries(r))pe(o,t,a,c,s);for(let[a,c]of Object.entries(n)){let l=r[a];!l||l===c||jt(o,t,c,l,s);}let u=Object.entries(o).sort((a,c)=>c[0].length-a[0].length);for(let[a,c]of u)i=i.split(a).join(c);return t.urlRewriteMode!=="absolute"&&(i=Ft(i,t,s)),i=Et(i,u.map(([,a])=>a)),Lt(t,i,s)&&(i=Dt(i)),i}async function $t(e,t,r,s){let n=f.join(e.outputDir,s),i=f.extname(n).toLowerCase(),o;try{o=await S.readFile(n,"utf8");}catch(c){if(c.code!=="ENOENT")throw c;return {changed:false}}let a=he(o,e,t,i===".js"||i===".mjs"?void 0:n,r);return a!==o?(await S.writeFile(n,a,"utf8"),{changed:true}):{changed:false}}async function Ut(e,t){return await new Promise((r,s)=>{let n=a=>{if(u(),a?.error){s(new Error(a.error));return}r({changed:!!a?.changed});},i=a=>{u(),s(a);},o=a=>{u(),a!==0&&s(new Error(`Rewrite worker exited with code ${a}.`));},u=()=>{e.off("message",n),e.off("error",i),e.off("exit",o);};e.on("message",n),e.on("error",i),e.on("exit",o),e.postMessage({file:t});})}async function Fe(e,t,r={}){let n=(r.files?[...new Set(r.files)].map(d=>f.resolve(d)).filter(d=>d.startsWith(f.resolve(e.outputDir))).map(d=>f.relative(e.outputDir,d).replace(/\\/g,"/")).filter(d=>d.length>0):await Ct(["**/*"],{cwd:e.outputDir,onlyFiles:true,dot:true})).filter(d=>d!=="asset-map.json"&&$(d)),i=0;if(n.length===0)return 0;let o=Math.max(1,Number(e.rewriteConcurrency||e.assetDownloadConcurrency||1)),u=Math.max(1,Math.min(o,n.length,availableParallelism())),a=0,c=0,l=null;if(u===1){for(let d of n){let g=await $t(e,t,r.previousAssetMap??{},d);g.changed&&(i+=1),c+=1,await r.onProgress?.({index:c,totalFiles:n.length,changedTextFiles:i,file:d,changed:g.changed});}return i}let p=Array.from({length:u},async()=>{let d=new Worker(new URL("./rewrite-worker.js",import.meta.url),{workerData:{config:e,assetMap:t,previousAssetMap:r.previousAssetMap??{}}});try{for(;!l;){let g=a;if(g>=n.length)return;a+=1;let m=n[g],h=await Ut(d,m);h.changed&&(i+=1),c+=1,await r.onProgress?.({index:c,totalFiles:n.length,changedTextFiles:i,file:m,changed:h.changed});}}catch(g){throw l=g instanceof Error?g:new Error(String(g)),l}finally{await d.terminate().catch(()=>{});}});return await Promise.all(p),i}var N={error:0,warn:1,info:2,debug:3},je=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"",X=je?f.join(je,"current-progress.json"):"",$e=Promise.resolve();function Ue(e){return e.includes("crawl")?"crawl":e.includes("deploy")?"deploy":e.includes("invalidate")?"invalidate":e}function Qt(e,t){let r={...e},s=new Set(["pagesQueued","pagesRendered","assetsQueued","sitemapsQueued","pagesDiscovered","assetsDiscovered","sitemapsDiscovered","pagesSaved","assetsSaved","donePages","doneAssets","doneSitemaps","changedTextFiles","uploaded","failed","index","totalFiles","pageQueue","assetQueue","sitemapQueue"]);for(let[n,i]of Object.entries(t)){let o=r[n];if(typeof o=="number"&&typeof i=="number"&&s.has(n)){r[n]=Math.max(o,i);continue}r[n]=i;}return r}function qt(e,t,r){X&&($e=$e.then(async()=>{let s=new Date,n=s.toISOString();await S.mkdir(f.dirname(X),{recursive:true});let i=await S.readFile(X,"utf8").then(h=>JSON.parse(h)).catch(()=>null),o=i&&typeof i.details=="object"&&i.details?i.details:{},u=Ue(e),a=i?.currentStep||Ue(i?.source||""),c=i?.startedAt||n,l={...i?.stepDurationsSec??{}},p=i?.stepStartedAt||n;if(a&&a!==u&&i?.stepStartedAt){let h=Math.max(0,Math.round((s.getTime()-new Date(i.stepStartedAt).getTime())/1e3));l[a]=(l[a]??0)+h,p=n;}let d=Math.max(0,Math.round((s.getTime()-new Date(p).getTime())/1e3)),g=Math.max(0,Math.round((s.getTime()-new Date(c).getTime())/1e3)),m={checkedAt:n,source:e,message:t,details:Qt(o,r??{}),startedAt:c,currentStep:u,stepStartedAt:p,stepElapsedSec:d,totalElapsedSec:g,stepDurationsSec:l};await S.writeFile(X,JSON.stringify(m,null,2),"utf8");}).catch(()=>{}));}async function Nt(e,t){await S.mkdir(f.dirname(e),{recursive:true}),await S.appendFile(e,`${JSON.stringify(t)}
|
|
5
5
|
`,"utf8");}function _t(e,t){let r=typeof t?.phase=="string"?t.phase.trim():"";return r||(e==="page"?"render-pages":e==="asset"?"download-assets":e==="sitemap"?"discovery":e==="timing"&&typeof t?.name=="string"&&t.name.trim()||"crawl")}function Wt(e,t,r){return {checkedAt:new Date().toISOString(),currentStep:_t(e,r),level:e,message:t,details:r??{}}}var _=class{constructor(t,r="info"){this.logDir=t;this.level=typeof r=="boolean"?r?"debug":"info":r,this.ensureLogFileReady();}logDir;startedAt=Date.now();marks=new Map;initPromise=null;writeQueue=Promise.resolve();writeError=null;level;get logPath(){return f.join(this.logDir,"crawl.log.jsonl")}get currentEventPath(){return f.join(this.logDir,"current-crawl-event.json")}get rejectedPath(){return f.join(this.logDir,"rejected.jsonl")}get ignoredPath(){return f.join(this.logDir,"ignored.jsonl")}get skippedPath(){return f.join(this.logDir,"skipped-http.jsonl")}get errorsPath(){return f.join(this.logDir,"errors.jsonl")}get timingsPath(){return f.join(this.logDir,"timings.jsonl")}ensureLogFileReady(){return this.initPromise||(this.initPromise=(async()=>{await S.mkdir(this.logDir,{recursive:true}),await S.writeFile(this.logPath,"","utf8");})()),this.initPromise}enqueueTask(t){this.writeQueue=this.writeQueue.then(t).catch(r=>{this.writeError=r instanceof Error?r:new Error(String(r));});}enqueueLine(t){this.enqueueTask(async()=>{await this.ensureLogFileReady(),await S.appendFile(this.logPath,`${t}
|
|
6
|
-
`,"utf8");});}enqueueJsonLine(t,r){this.enqueueTask(()=>Nt(t,r));}updateCurrentEvent(t,r,s){let n=Wt(t,r,s);this.enqueueTask(async()=>{await S.mkdir(this.logDir,{recursive:true}),await S.writeFile(this.currentEventPath,JSON.stringify(n,null,2),"utf8");});}push(t,r,s){let n=JSON.stringify({time:new Date().toISOString(),level:t,message:r,...s||{}});this.enqueueLine(n),this.updateCurrentEvent(t,r,s),t==="error"?console.error(r,s||""):t==="warn"?console.warn(r,s||""):(N[this.level]>=N.debug||["summary","page","sitemap","asset","timing"].includes(t))&&console.log(r);}info(t,r){this.push("info",t,r);}progress(t,r){qt("crawl.log.jsonl",t,r),this.updateCurrentEvent("progress",t,r);}checkpoint(t,r){this.updateCurrentEvent("checkpoint",t,r);}page(t,r){this.push("page",t,r);}sitemap(t,r){this.push("sitemap",t,r);}asset(t,r){this.push("asset",t,r);}warn(t,r){this.push("warn",t,r);}error(t,r){this.enqueueJsonLine(this.errorsPath,{message:t,...r||{}}),this.push("error",t,r);}summary(t,r){this.push("summary",t,r);}mark(t){this.marks.set(t,Date.now());}endMark(t,r){let s=this.marks.get(t);if(!s)return;let n=Date.now()-s,i={name:t,ms:n,seconds:Number((n/1e3).toFixed(2)),...r||{}};this.enqueueJsonLine(this.timingsPath,i),this.push("timing",`Timing ${t}: ${i.seconds}s`,i);}reject(t,r,s,n){let i={kind:t,url:r,reason:s,source:n};this.enqueueJsonLine(this.rejectedPath,i),N[this.level]>=N.debug&&this.push("reject",`Rejected ${t} ${r}: ${s}`,i);}ignore(t,r,s,n){let i={kind:t,url:r,reason:s,source:n};this.enqueueJsonLine(this.ignoredPath,i),N[this.level]>=N.debug&&this.push("ignore",`Ignored ${t} ${r}: ${s}`,i);}skip(t,r,s,n){let i={kind:t,url:r,status:s,source:n};this.enqueueJsonLine(this.skippedPath,i),this.push("warn",`Skipped ${t} ${r}: HTTP ${s}`,i);}async flush(){let t=Date.now()-this.startedAt,r={name:"total",ms:t,seconds:Number((t/1e3).toFixed(2))};if(this.enqueueJsonLine(this.timingsPath,r),await this.writeQueue,this.writeError)throw this.writeError}};var Jt=["text/css","javascript","image/","font/","application/font","application/json","application/xml","text/xml","application/rss+xml","application/atom+xml","application/xslt+xml","text/xsl","image/svg+xml","application/octet-stream"],zt=new Set([".css",".js",".mjs",".json",".map",".xml",".xsl",".rss",".atom",".txt",".enc",".jws",".png",".jpg",".jpeg",".gif",".webp",".avif",".svg",".ico",".woff",".woff2",".ttf",".otf",".eot",".pdf",".mp4",".webm"]),Bt=new Set(["",".html",".htm"]),Vt=new Set([".css",".xml",".xsl",".svg",".json",".html",".htm",".txt",".enc",".jws"]),Gt=["/wp-content/","/wp-includes/","/wp-admin/","/static/","/assets/","/build/","/_next/"];function Kt(e){let t=[],r="full",s="full",n=false,i=false;for(let o=0;o<e.length;o++){let u=e[o];if(u==="--retry-timeouts")r="retry-timeouts",n=true;else if(u==="--resume")n=true;else if(u==="--resume-rewrite")n=true,i=true;else if(u==="--crawl-mode"){let a=e[++o];if(!a)throw new Error("--crawl-mode requires a value");s=a==="incremental"?"incremental":"full",n||=s==="incremental";}else if(u.startsWith("--crawl-mode="))s=u.slice(13)==="incremental"?"incremental":"full",n||=s==="incremental";else if(u==="--incremental")s="incremental",n=true;else if(u==="--url"){let a=e[++o];if(!a)throw new Error("--url requires a value");r="single-url",n=true,t.push(a);}else if(u.startsWith("--url="))r="single-url",n=true,t.push(u.slice(6));else if(u==="--urls"){let a=e[++o];if(!a)throw new Error("--urls requires a file path");r="single-url",n=true,t.push(`@${a}`);}}return {mode:r,crawlMode:s,urls:t,preserveOutput:n,resumeRewrite:i}}function Se(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.resolve(e):""}function Xt(){let e=Se();return e?f.resolve(e,".."):""}function Zt(){let e=process.env.STATIC_PUBLISHER_WP_ROOT||process.env.WPSUITE_STATIC_PUBLISHER_WP_ROOT||"";return e.trim()?f.resolve(e):""}function Yt(e){let t=e.trim(),r=[{alias:"@storage-root",root:Xt(),requiredEnv:"STATIC_PUBLISHER_RUNTIME_DIR or WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR"},{alias:"@runtime",root:Se(),requiredEnv:"STATIC_PUBLISHER_RUNTIME_DIR or WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR"},{alias:"@wp-root",root:Zt(),requiredEnv:"STATIC_PUBLISHER_WP_ROOT or WPSUITE_STATIC_PUBLISHER_WP_ROOT"}];for(let s of r){if(t!==s.alias&&!t.startsWith(`${s.alias}/`))continue;if(!s.root)return {resolvedPath:null,alias:s.alias,requiredEnv:s.requiredEnv};let n=t.slice(s.alias.length).replace(/^\/+/,"");return {resolvedPath:n?f.resolve(s.root,n):s.root,alias:s.alias}}return {resolvedPath:f.resolve(t)}}function et(e){let t=Se();return t?f.join(t,"crawl-manifest.json"):f.join(e.outputDir,".crawl-manifest.json")}function tt(){return {schemaVersion:2,updatedAt:"",pages:{},assets:{}}}function er(){return {rewriteTargets:new Set,changedFiles:new Set,deletedFiles:new Set}}function $(e,t,r){let s=V(t.outputDir,r);s&&(e.changedFiles.add(s),e.deletedFiles.delete(s));}function qe(e,t,r){let s=V(t.outputDir,r);s&&(e.deletedFiles.add(s),e.changedFiles.delete(s));}function ee(e,t){e.rewriteTargets.add(f.resolve(t));}async function tr(e){try{let t=await S.readFile(et(e),"utf8"),r=JSON.parse(t);if(r&&typeof r=="object"&&r.pages&&typeof r.pages=="object"){if(r.schemaVersion===2&&r.assets&&typeof r.assets=="object")return r;if(r.schemaVersion===1)return {schemaVersion:2,updatedAt:String(r.updatedAt||""),pages:r.pages,assets:{}}}}catch{}return tt()}function Ne(e){return JSON.parse(JSON.stringify(e))}async function rr(e,t){let r=et(e);await S.mkdir(f.dirname(r),{recursive:true}),await S.writeFile(r,JSON.stringify(t,null,2),"utf8");}function te(e){if(!e)return null;let t=e.trim();return t||null}function sr(e,t){let r=new Set,s=new Set([...Object.keys(e),...Object.keys(t)]);for(let n of s){let i=e[n],o=t[n];!i||!o||i===o||r.add(n);}return r}async function nr(e,t,r,s){let n=[...new Set(r.map(u=>u.trim()).filter(Boolean))],i=String(e.wpsuite?.siteKey||"").trim();if(n.length===0||!i)return new Map;let o=new URL("/wp-json/smartcloud-static-publisher/v1/change-tokens",e.sourceOrigin).toString();try{let u=await t.post(o,{timeout:3e4,failOnStatusCode:!1,headers:{"content-type":"application/json","x-site-key":i},data:JSON.stringify({urls:n})});if(!u.ok())return s.warn(`Change token lookup failed with HTTP ${u.status()}`,{endpoint:o,status:u.status()}),new Map;let a=await u.json().catch(()=>null);if(!a||!Array.isArray(a.items))return new Map;let c=new Map;for(let l of a.items)!l||typeof l.url!="string"||c.set(l.url,l);return c}catch(u){return s.warn("Change token lookup failed; falling back to sitemap metadata",{endpoint:o,error:String(u)}),new Map}}async function he(e,t,r,s,n){if(!r.enabled&&!r.trackManifest)return null;if(r.changeTokenCache.has(s))return r.changeTokenCache.get(s)??null;let o=(await nr(e,t,[s],n)).get(s)??null;return r.changeTokenCache.set(s,o),o}async function ir(e,t,r,s,n,i){if(at(e,n))return {action:"render",changeToken:null};if(!s.enabled)return {action:"render",changeToken:s.trackManifest?await he(e,t,s,n,i):null};let o=s.manifest.pages[n];if(!o)return {action:"render",changeToken:await he(e,t,s,n,i)};let u=await he(e,t,s,n,i);if(u?.supported&&u.token)return o.changeToken===u.token?{action:"reuse",changeToken:u}:{action:"render",changeToken:u};let a=te(r.sitemapLastmodByPage[n]);return a&&o.sitemapLastmod&&o.sitemapLastmod===a?{action:"reuse",changeToken:u}:{action:"render",changeToken:u}}async function ar(e,t,r,s,n){let i=_e(r.previousManifest),o=0;for(let[d,g]of Object.entries(r.manifest.pages))if(!r.seenPages.has(d)){qe(n,e,g.outputPath);try{await S.unlink(g.outputPath);}catch(m){(m.code||"")!=="ENOENT"&&s.warn(`Failed to remove stale incremental page output for ${d}`,{url:d,outputPath:g.outputPath,error:String(m)});}delete r.manifest.pages[d],o++;}let u=_e(r.manifest,r.runId),a=new Set,c=new Set;for(let d of u){for(let h of H(e,d))a.add(h);let g=r.manifest.assets[d]?.outputPath;g&&c.add(g);let m=re(e,t.assetMap,d);m&&c.add(m);}for(let d of Object.keys(r.manifest.assets))u.has(d)||delete r.manifest.assets[d];let l=0,p=0;for(let d of i){let g=new Set,m=r.previousManifest.assets[d]?.outputPath;m&&g.add(m);let h=re(e,t.assetMap,d);if(!u.has(d)&&h)g.add(h);else if(!u.has(d)){let y=be(e,d);g.add(y.originalFilePath),y.hashedFilePath&&g.add(y.hashedFilePath);}for(let y of g)if(!c.has(y)){qe(n,e,y);try{await S.unlink(y),l++;}catch(v){(v.code||"")!=="ENOENT"&&s.warn(`Failed to remove stale incremental asset output for ${d}`,{url:d,outputPath:y,error:String(v)});}}for(let y of H(e,d))a.has(y)||Object.prototype.hasOwnProperty.call(t.assetMap,y)&&(delete t.assetMap[y],p++);}(o>0||l>0||p>0)&&s.info(`Incremental cleanup removed ${o} stale pages, ${l} stale assets, ${p} stale asset mappings`,{removedPages:o,removedAssets:l,removedAssetMappings:p,phase:"incremental-cleanup"});}function _e(e,t){let r=new Set;for(let s of Object.values(e.pages))for(let n of s.discoveredAssets||[]){let i=n.trim();i&&r.add(i);}for(let[s,n]of Object.entries(e.assets||{})){if(t&&n.lastSeenRunId!==t)continue;let i=s.trim();i&&r.add(i);}return r}function We(e,t,r){for(let s of t)for(let n of H(e,s))if(r.has(n))return true;return false}function or(e,t,r,s,n){let i=new Set(r.rewriteTargets),o=sr(s,n);if(o.size===0)return [...i];for(let u of Object.values(t.manifest.pages))We(e,u.discoveredAssets,o)&&i.add(f.resolve(u.outputPath));for(let[u,a]of Object.entries(t.manifest.assets)){a.isText&&We(e,a.discoveredAssets,o)&&i.add(f.resolve(a.outputPath));for(let c of H(e,u))if(o.has(c)){$(r,e,a.outputPath),a.isText&&i.add(f.resolve(a.outputPath));break}}return [...i]}function H(e,t){let r=new Set([t]);try{let s=new URL(t),i=[new URL(e.sourceOrigin).origin,...Object.keys(e.extraReplacements||{})];if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{i.push(new URL(e.targetOrigin).origin);}catch{}for(let o of i)try{r.add(new URL(o).origin+s.pathname+s.search);}catch{}r.add(s.pathname+s.search),r.add(s.pathname);}catch{}return [...r]}function ur(e,t){let r=new Set([t]);try{let s=new URL(t),i=[new URL(e.sourceOrigin).origin,...Object.keys(e.extraReplacements||{})];if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{i.push(new URL(e.targetOrigin).origin);}catch{}for(let o of i)try{r.add(new URL(o).origin+s.pathname+s.search);}catch{}r.add(s.pathname+s.search);}catch{}return [...r]}function rt(e,t){let r=e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/"?e.targetOrigin:"https://relative.invalid/";try{let s=new URL(t,r);return f.join(e.outputDir,decodeURIComponent(s.pathname))}catch{return null}}function lr(e,t,r){for(let s of ur(e,r)){let n=t[s];if(!n)continue;let i=rt(e,n);if(i)return {outputPath:i,publicUrl:n}}return null}function cr(e,t,r){for(let s of H(e,r)){let n=t[s];if(!n)continue;let i=rt(e,n);if(i)return i}return null}function re(e,t,r){let s=cr(e,t,r);return s||be(e,r).originalFilePath}function dr(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.join(f.resolve(e),"queue-runner-heartbeat.json"):""}async function He(e){let t=dr();if(t)try{let r=await S.readFile(t,"utf8"),s=JSON.parse(r);if(!s||typeof s!="object")return;await S.writeFile(t,JSON.stringify({...s,checkedAt:new Date().toISOString(),status:"running",currentStep:"rewrite",message:e},null,2),"utf8");}catch{}}function pr(e,t){let r=new Set;for(let s of t.donePages)r.add(f.resolve(W(e,s)));for(let s of t.doneAssets){let n=re(e,t.assetMap,s);!n||!O(n)||r.add(f.resolve(n));}return [...r]}var se="re:",Y=new Map;function gr(e){if(!e.startsWith(se))return null;let t=Y.get(e);if(t!==void 0)return t||null;let r=e.slice(se.length).trim();if(!r)return Y.set(e,false),null;try{let s=new RegExp(r);return Y.set(e,s),s}catch{return console.warn(`[crawl] Ignoring invalid path matcher regex: ${e}`),Y.set(e,false),null}}function ve(e,t){let r=gr(t);return r?r.test(e):t.startsWith(se)?false:e.startsWith(t)}function hr(e,t){return t.find(r=>ve(e,r))}function st(e,t){let r=hr(t.pathname,e.blockedPathPrefixes);if(r)return r.startsWith(se)?`blocked path regex: ${r}`:`blocked path prefix: ${r}`;let s=e.blockedSearchFragments.find(n=>t.search.includes(n));return s?`blocked search fragment: ${s}`:null}function fr(e){let t=new URL(e.sourceOrigin),r=new Set([t.hostname,...e.allowedAssetHosts]);if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{r.add(new URL(e.targetOrigin).hostname);}catch{}for(let s of Object.keys(e.extraReplacements||{}))try{r.add(new URL(s).hostname);}catch{}return r}function mr(e,t){return fr(e).has(t.hostname)}function wr(e){let t=j(e.trim()).replace(/\\\//g,"/").replace(/^['"]|['"]$/g,"").trim();return t=t.replace(/[)]+$/g,"").trim(),t=t.replace(/;.*$/g,"").trim(),t}function nt(e){let t=j(e).replace(/\\\//g,"/");return /[{}]|\bwindow\.|\blocation\.|\bincludes\(|\?\?null|\+|%7B|%7D|%22|<|>|\s/.test(t)}function ne(e){return Gt.some(t=>e.pathname.startsWith(t))}function x(e,t,r=e.sourceOrigin,s,n="url"){let i=wr(t);if(!i||i.startsWith("data:")||i.startsWith("blob:")||i.startsWith("mailto:")||i.startsWith("tel:")||i.startsWith("#"))return s?.ignore(n,t,"empty or unsupported scheme",r),null;if(nt(i))return s?.ignore(n,t,"looks like JavaScript/code fragment",r),null;try{let o=new URL(i,r);if(!["http:","https:"].includes(o.protocol))return s?.ignore(n,t,"unsupported protocol: ",r),null;if(!mr(e,o))return s?.ignore(n,o.toString(),"host not allowed",r),null;let u=st(e,o);if(u)return s?.reject(n,o.toString(),u,r),null;let a=new URL(e.sourceOrigin);return o.protocol=a.protocol,o.host=a.host,o.hash="",o}catch{return s?.reject(n,t,"invalid URL",r),null}}function it(e){return zt.has(f.extname(e.pathname).toLowerCase())}function B(e){return ne(e)?false:Bt.has(f.extname(e.pathname).toLowerCase())}function ie(e){let t=e.pathname.toLowerCase();return t.endsWith(".xml")&&(t.includes("sitemap")||t.endsWith("/sitemap.xml"))}function R(e,t){return ie(t)||t.pathname==="/robots.txt"||t.pathname==="/llms.txt"||t.pathname.toLowerCase().endsWith(".xsl")?true:it(t)?e.assetPathPrefixes.length===0?true:e.assetPathPrefixes.some(r=>ve(t.pathname,r)):false}function W(e,t){let r=new URL(t),s=decodeURIComponent(r.pathname);return s.endsWith("/")?s+="index.html":f.extname(s)||(s+="/index.html"),f.join(e.outputDir,s)}function Pr(e){if(!e.generated404RequestPath)return null;try{return new URL(e.generated404RequestPath,e.sourceOrigin).toString()}catch{return null}}function at(e,t){let r=Pr(e);if(!r)return false;try{return new URL(t,e.sourceOrigin).toString()===r}catch{return false}}function be(e,t){let r=new URL(t),s=decodeURIComponent(r.pathname);s.endsWith("/")&&(s+="index.html");let n=!!r.search,i=s,o=f.join(e.outputDir,i),u=de(e.targetOrigin,i),a=`${u}${r.search}`;if(!n)return {originalPathname:i,originalFilePath:o,originalPublicUrl:u,originalPublicUrlWithSearch:a,preferQueryHashed:n};let c=f.extname(s),p=`${c?s.slice(0,-c.length):s}.${Ie(r.search)}${c||".bin"}`;return {originalPathname:i,originalFilePath:o,originalPublicUrl:u,originalPublicUrlWithSearch:a,preferQueryHashed:n,hashedPathname:p,hashedFilePath:f.join(e.outputDir,p),hashedPublicUrl:de(e.targetOrigin,p)}}function yr(e,t,r,s,n){t[r]=s;try{let i=new URL(r),u=[new URL(e.sourceOrigin).origin,...Object.keys(e.extraReplacements||{})];if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{u.push(new URL(e.targetOrigin).origin);}catch{}for(let a of u)try{let c=new URL(a);t[c.origin+i.pathname+i.search]=s;}catch{}try{let a=(()=>{let c=new URL(s,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin);return `${c.pathname}${c.search}`})();t[i.pathname+i.search]=a;}catch{}try{let a=new URL(s,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin).pathname;t[i.pathname]=a;}catch{}try{let a=new URL(n,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin).pathname;t[i.pathname]=a;}catch{}}catch{}}async function Pe(e){try{return await S.readFile(e)}catch(t){if(t.code==="ENOENT")return null;throw t}}async function ot(e){try{return await S.stat(e)}catch(t){if(t.code==="ENOENT")return null;throw t}}function Sr(e){let t=String(e["content-length"]||"").trim();if(!t)return null;let r=Number.parseInt(t,10);return Number.isFinite(r)&&r>=0?r:null}function ut(e){let t=String(e["last-modified"]||"").trim();if(!t)return null;let r=Date.parse(t);return Number.isFinite(r)?r:null}function vr(e,t){let r=ut(t),s=Sr(t),n=e?.mtime.getTime();return r!==null&&Math.abs((n??0)-r)>=1e3||s!==null&&e?.size!==s?false:r!==null||s!==null}async function br(e,t){if(!t)return;let r=ut(t);if(r===null)return;let s=await ot(e);if(!s)return;let n=new Date(r);await S.utimes(e,s.atime,n).catch(()=>{});}function fe(e,t){let r=f.extname(e.pathname).toLowerCase(),s=t["content-type"]||"";return Vt.has(r)||Jt.some(n=>s.includes(n))&&!s.startsWith("image/")&&!s.includes("font")}function me(e,t){(e.doneAssets.size%25===0||e.doneAssets.size===1||e.doneAssets.size===e.stats.assetsQueued)&&t.progress(`Asset download progress: downloaded ${e.doneAssets.size}, discovered ${e.stats.assetsQueued}.`,{doneAssets:e.doneAssets.size,assetsQueued:e.stats.assetsQueued,phase:"download-assets"});}async function ye(e,t,r,s,n,i,o){let u=be(e,r),a=u.originalFilePath,c=u.originalPublicUrlWithSearch;if(u.preferQueryHashed&&u.hashedFilePath&&u.hashedPublicUrl){let g=lr(e,t,r);if(g)a=g.outputPath,c=g.publicUrl;else {let m=await Pe(u.originalFilePath);if(m!==null&&!m.equals(s)){let h=O(u.originalFilePath)?Buffer.from(X(s.toString("utf8"),e,t,u.originalFilePath),"utf8"):null;(h===null||!m.equals(h))&&(a=u.hashedFilePath,c=u.hashedPublicUrl);}}}yr(e,t,r,c,u.originalPublicUrl);let l=O(a)?Buffer.from(X(s.toString("utf8"),e,t,a),"utf8"):s,p=await Pe(a),d=false;return (p===null||!p.equals(l))&&(await ce(a),await S.writeFile(a,l),d=true),await br(a,o),n.stats.assetsSaved++,i&&(n.stats.assetsSaved%50===0||n.stats.assetsSaved===1)&&i.progress(`Asset progress: saved ${n.stats.assetsSaved}, discovered ${n.stats.assetsQueued}, pending ${n.assetQueue.length}.`,{assetsSaved:n.stats.assetsSaved,assetsQueued:n.stats.assetsQueued,assetQueue:n.assetQueue.length,phase:"download-assets"}),{outputPath:a,publicUrl:c,wroteFile:d}}function J(e,t,r,s,n){let i=n&&/^https?:\/\//i.test(n)?n:e.sourceOrigin,o=x(e,r,i,s,"page");if(!o)return;if(!B(o)||it(o)){if(R(e,o)){s.info(`Seeded/non-page URL queued as asset: ${o.toString()}`,{url:o.toString(),source:n}),A(e,t,o.toString(),s,n);return}s.reject("page",o.toString(),ne(o)?"asset/internal path cannot be page":"not page-like",n);return}let u=o.toString();!t.donePages.has(u)&&!t.queuedPages.has(u)&&(t.queuedPages.add(u),t.pageQueue.push(u),t.stats.pagesQueued++,(t.stats.pagesQueued%25===0||t.stats.pagesQueued===1)&&s.progress(`Discovery progress: pages ${t.stats.pagesQueued}, assets ${t.stats.assetsQueued}, sitemaps ${t.stats.sitemapsQueued}.`,{pagesQueued:t.stats.pagesQueued,assetsQueued:t.stats.assetsQueued,sitemapsQueued:t.stats.sitemapsQueued,pageQueue:t.pageQueue.length,phase:"discovery"}));}function A(e,t,r,s,n){let i=n&&/^https?:\/\//i.test(n)?n:e.sourceOrigin,o=x(e,r,i,s,"asset");if(!o)return;if(!R(e,o)){s.reject("asset",o.toString(),"not a safe asset path/prefix",n);return}let u=o.toString();!t.doneAssets.has(u)&&!t.queuedAssets.has(u)&&(t.queuedAssets.add(u),t.assetQueue.push(u),t.stats.assetsQueued++,(t.stats.assetsQueued%100===0||t.stats.assetsQueued===1)&&s.progress(`Discovery progress: pages ${t.stats.pagesQueued}, assets ${t.stats.assetsQueued}, sitemaps ${t.stats.sitemapsQueued}.`,{pagesQueued:t.stats.pagesQueued,assetsQueued:t.stats.assetsQueued,sitemapsQueued:t.stats.sitemapsQueued,assetQueue:t.assetQueue.length,phase:"discovery"}));}function Re(e,t,r,s,n){let i=n&&/^https?:\/\//i.test(n)?n:e.sourceOrigin,o=x(e,r,i,s,"sitemap");if(!o)return;if(!ie(o)){s.reject("sitemap",o.toString(),"not sitemap-like",n);return}let u=o.toString();!t.doneSitemaps.has(u)&&!t.queuedSitemaps.has(u)&&(t.queuedSitemaps.add(u),t.sitemapQueue.push(u),t.stats.sitemapsQueued++,(t.stats.sitemapsQueued%10===0||t.stats.sitemapsQueued===1)&&s.progress(`Discovery progress: pages ${t.stats.pagesQueued}, assets ${t.stats.assetsQueued}, sitemaps ${t.stats.sitemapsQueued}.`,{pagesQueued:t.stats.pagesQueued,assetsQueued:t.stats.assetsQueued,sitemapsQueued:t.stats.sitemapsQueued,sitemapQueue:t.sitemapQueue.length,phase:"discovery"}));}function Rr(e){let t=j(e),r=[...t.matchAll(/<url>\s*([\s\S]*?)\s*<\/url>/gi)].map(s=>s[1]);return r.length>0?r.map(s=>{let n=s.match(/<loc>\s*([^<]+?)\s*<\/loc>/i);if(!n?.[1])return null;let i=s.match(/<lastmod>\s*([^<]+?)\s*<\/lastmod>/i);return {loc:n[1].trim(),lastmod:te(i?.[1]??null)}}).filter(s=>!!s):[...t.matchAll(/<loc>\s*([^<]+?)\s*<\/loc>/gi)].map(s=>({loc:s[1].trim(),lastmod:null})).filter(s=>s.loc)}function Cr(e){return e.split(",").map(t=>t.trim().split(/\s+/)[0]).filter(Boolean)}function xr(e){let t=[],r=/url\(\s*(?:"([^"]+)"|'([^']+)'|([^)]*?))\s*\)/gi;for(let s of e.matchAll(r)){let n=(s[1]||s[2]||s[3]||"").trim();n&&t.push(n);}return t}function we(e){return e.trim().replace(/\\/g,"/").replace(/^https?:\/\/[^/]+/i,"").replace(/^\/+/,"")}async function lt(e,t){let r=[];await S.mkdir(t,{recursive:true});let s=await S.readdir(e,{withFileTypes:true});for(let n of s){let i=f.join(e,n.name),o=f.join(t,n.name);if(n.isDirectory()){r.push(...await lt(i,o));continue}await S.copyFile(i,o),r.push(o);}return r}async function Ar(e,t){let r=Object.entries(e.postCrawlCopyMap||{}).map(([i,o])=>({sourcePath:i.trim(),prefix:String(o||"").trim()}));if(r.length===0)return [];let s=0,n=[];t.mark("copy-extra-paths");for(let i of r){let o=i.sourcePath,u=i.prefix;if(!o||!u){t.warn("Skipped post-crawl copy mapping with empty key/value",{sourcePath:o,prefixPath:u});continue}let a=Yt(o);if(!a.resolvedPath){t.warn("Skipped post-crawl copy source because alias root is not configured",{sourcePath:o,alias:a.alias||"",requiredEnv:a.requiredEnv||"",prefixPath:u});continue}let c=a.resolvedPath,l=we(u);if(!l){t.warn("Skipped post-crawl copy mapping with invalid export prefix",{sourcePath:o,prefixPath:u});continue}let p;try{p=await S.stat(c);}catch{t.warn("Skipped post-crawl copy source because it does not exist",{sourcePath:o,sourceAbs:c,prefixPath:u});continue}let d=f.resolve(e.outputDir,l);if(p.isDirectory()){n.push(...await lt(c,d)),s++,t.progress(`Copied static directory to export: ${c} -> /${l}`,{phase:"copy-extra-paths",sourcePath:c,targetPath:`/${l}`,copiedItems:s});continue}let g=f.basename(c),m=u.endsWith("/")?f.join(d,g):d;await S.mkdir(f.dirname(m),{recursive:true}),await S.copyFile(c,m),n.push(m),s++,t.progress(`Copied static file to export: ${c} -> /${we(u.endsWith("/")?`${u}${g}`:u)}`,{phase:"copy-extra-paths",sourcePath:c,targetPath:`/${we(u.endsWith("/")?`${u}${g}`:u)}`,copiedItems:s});}return t.endMark("copy-extra-paths",{mappedSources:r.length,copiedItems:s}),n}function z(e,t,r,s,n){let i=new Set,o=j(r).replace(/\\\//g,"/");for(let c of xr(o)){let l=x(e,c,t,s,`${n}:css-url`);l&&R(e,l)&&i.add(l.toString());}for(let c of o.matchAll(/@import\s+(?:url\()?\s*['"]?([^'"\s;]+)['"]?\s*\)?/gi)){let l=x(e,c[1],t,s,`${n}:css-import`);l&&R(e,l)&&i.add(l.toString());}for(let c of o.matchAll(/<\?xml-stylesheet[^>]+href=["']([^"']+)["'][^>]*\?>/gi)){let l=x(e,c[1],t,s,`${n}:xml-stylesheet`);l&&R(e,l)&&i.add(l.toString());}let u="(?:css|js|mjs|json|map|xml|xsl|rss|atom|txt|enc|jws|png|jpe?g|gif|webp|avif|svg|ico|woff2?|ttf|otf|eot|pdf|mp4|webm)",a=new RegExp(`(?:https?:)?//[^\\s'"<>\\);]+\\.${u}(?:\\?[^\\s'"<>\\);]*)?|(?<!\\.)/[^\\s'"<>\\);]+\\.${u}(?:\\?[^\\s'"<>\\);]*)?`,"gi");for(let c of o.matchAll(a)){let l=c[0].startsWith("//")?`${new URL(e.sourceOrigin).protocol}${c[0]}`:c[0];if(nt(l)){s.ignore(`${n}:serialized-url`,l,"looks like JavaScript/code fragment",t);continue}let p=x(e,l,t,s,`${n}:serialized-url`);p&&R(e,p)?i.add(p.toString()):p&&s.ignore("asset",p.toString(),"serialized URL did not pass safe asset path filter",t);}return [...i]}function kr(e,t,r,s){let n=new Set,i=j(r);for(let o of i.matchAll(/href=["']([^"'#\s][^"']*?)["']/gi)){let u=o[1],a=x(e,u,t,s,"page-link");a&&(st(e,a)||B(a)&&!ne(a)&&!R(e,a)&&n.add(a.toString()));}return [...n]}async function Je(e,t,r,s){for(;r.sitemapQueue.length>0;){let n=r.sitemapQueue.shift();if(r.queuedSitemaps.delete(n),!r.doneSitemaps.has(n)){r.doneSitemaps.add(n),s.sitemap(`Fetching sitemap ${n}`,{url:n});try{let i=await t.get(n,{timeout:6e4});if(!i.ok()){s.skip("sitemap",n,i.status());continue}let o=await i.body(),u=i.url()||n;await ye(e,r.assetMap,u,o,r,s);let a=o.toString("utf8");for(let c of z(e,u,a,s,"sitemap"))A(e,r,c,s,u);for(let c of Rr(a)){let l=x(e,c.loc,u,s,"sitemap-loc");l&&(ie(l)?Re(e,r,l.toString(),s,u):R(e,l)?A(e,r,l.toString(),s,u):B(l)?(c.lastmod&&(r.sitemapLastmodByPage[l.toString()]=c.lastmod),J(e,r,l.toString(),s,u)):s.reject("sitemap-loc",l.toString(),"not page/sitemap/asset-like",u));}}catch(i){s.error(`Failed sitemap ${n}`,{url:n,error:String(i)});}}}}async function Mr(e,t,r,s,n,i,o){let u=x(e,s,e.sourceOrigin,n,"asset-fetch");if(!u||!R(e,u))return;let a=u.toString();if(r.doneAssets.has(a))return;let c=i.enabled?i.previousManifest.assets[a]:void 0,l=(p,d={})=>{if(!c)return false;i.manifest.assets[a]={...c,discoveredAssets:[...c.discoveredAssets||[]],lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId},r.doneAssets.add(a);for(let g of c.discoveredAssets||[])A(e,r,g,n,a);return n.warn(`Keeping previous incremental asset ${a} after fetch issue.`,{url:a,reason:p,preservedFromPreviousManifest:true,...d,phase:"download-assets"}),me(r,n),true};n.asset(`Fetching asset ${a}`,{url:a});try{let p=re(e,r.assetMap,a),d=p?await ot(p):null;if(p&&d)try{let w=await t.head(a,{timeout:6e4}),b=w.headers();if(w.ok()&&vr(d,b)){let L=await Pe(p);if(L!==null){let I=w.url()||a,E=await ye(e,r.assetMap,I,L,r,n,b);E.wroteFile&&($(o,e,E.outputPath),O(E.outputPath)&&ee(o,E.outputPath)),r.doneAssets.add(a),n.info(`Reused existing asset ${a} based on response headers.`,{url:a,filePath:p,lastModified:b["last-modified"]||"",contentLength:b["content-length"]||"",phase:"download-assets"}),me(r,n);let U=fe(u,b),C=c?.discoveredAssets?[...c.discoveredAssets]:[];if(U||c?.isText){let dt=L.toString("utf8"),Ce=z(e,I,dt,n,"asset:cached");(Ce.length>0||!c)&&(C=Ce);for(let pt of C)A(e,r,pt,n,I);}i.trackManifest&&(i.manifest.assets[a]={url:a,outputPath:E.outputPath,isText:U||!!c?.isText,discoveredAssets:C,lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId});return}}}catch{}let g=await t.get(a,{timeout:6e4});if(!g.ok()){if(l(`http-${g.status()}`,{status:g.status()}))return;n.skip("asset",a,g.status());return}let m=await g.body(),h=g.url()||a,y=g.headers(),v=await ye(e,r.assetMap,h,m,r,n,y);v.wroteFile&&($(o,e,v.outputPath),O(v.outputPath)&&ee(o,v.outputPath)),r.doneAssets.add(a),me(r,n);let P=[];if(fe(u,y)){let w=m.toString("utf8");P=z(e,h,w,n,`asset:${f.extname(u.pathname).toLowerCase()||y["content-type"]||"unknown"}`);for(let b of P)A(e,r,b,n,h);}i.trackManifest&&(i.manifest.assets[a]={url:a,outputPath:v.outputPath,isText:fe(u,y),discoveredAssets:P,lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId});}catch(p){if(l("fetch-error",{error:String(p)}))return;n.error(`Failed asset ${a}`,{url:a,error:String(p)});}}async function Tr(e,t,r,s,n,i){let o=Math.max(1,Number(e.assetDownloadConcurrency||e.concurrency||1)),u=Array.from({length:o},async()=>{for(;r.assetQueue.length>0;){let a=r.assetQueue.shift();a&&(r.queuedAssets.delete(a),await Mr(e,t,r,a,s,n,i));}});await Promise.all(u);}async function Lr(e){await e.evaluate(async()=>{await new Promise(t=>{let r=0,s=700,n=window.setInterval(()=>{window.scrollBy(0,s),r+=s,r>=document.body.scrollHeight+window.innerHeight&&(window.clearInterval(n),window.scrollTo(0,0),t());},120);});});}async function Ir(e,t){await t.waitForLoadState("domcontentloaded",{timeout:e.readiness.timeoutMs}).catch(()=>{}),e.readiness.waitForSelector&&await t.waitForSelector(e.readiness.waitForSelector,{timeout:e.readiness.timeoutMs}).catch(()=>{}),e.readiness.waitForFunction&&await t.waitForFunction(e.readiness.waitForFunction,void 0,{timeout:e.readiness.timeoutMs}).catch(()=>{}),await Lr(t),await t.waitForTimeout(e.readiness.fallbackWaitMs);}async function Dr(e,t){await t.waitForLoadState("domcontentloaded",{timeout:e.readiness.timeoutMs}).catch(()=>{}),await t.waitForLoadState("load",{timeout:e.readiness.timeoutMs}).catch(()=>{}),e.readiness.waitForSelector&&await t.waitForSelector(e.readiness.waitForSelector,{timeout:e.readiness.timeoutMs}).catch(()=>{}),await t.waitForTimeout(e.readiness.fallbackWaitMs);}function Er(e,t){return e.noJavaScriptRenderPathPrefixes.some(r=>ve(t,r))}async function Fr(e,t,r,s,n=true){let i=new Set;if(n){let u=await t.evaluate(()=>{let a=new Set,c=["href","src","poster","data-src","data-lazy-src","data-original","data-bg","data-background","data-href"],l=new Set,p=["srcset","data-srcset","data-lazy-srcset"];return document.querySelectorAll("*").forEach(d=>{for(let m of c){let h=d.getAttribute(m);h&&a.add(h);}for(let m of p){let h=d.getAttribute(m);h&&h.split(",").forEach(y=>{let v=y.trim().split(/\s+/)[0];v&&a.add(v);});}let g=d.getAttribute("style");g&&l.add(g);}),{attrs:[...a],styles:[...l]}});for(let a of u.attrs)for(let c of Cr(a)){let l=x(e,c,r,s,"dom-attr");l&&R(e,l)&&i.add(l.toString());}for(let a of u.styles)for(let c of z(e,r,a,s,"dom-style"))i.add(c);}let o=await t.content();for(let u of z(e,r,o,s,"page-html"))i.add(u);return [...i]}async function ze(e,t,r,s,n){let i=W(e,t);await ce(i),await S.writeFile(i,r,"utf8"),s.stats.pagesSaved++,n&&(s.stats.pagesSaved%10===0||s.stats.pagesSaved===1)&&n.progress(`Page progress: processed ${s.donePages.size}, rendered ${s.stats.pagesRendered}, saved ${s.stats.pagesSaved}, discovered ${s.stats.pagesQueued}.`,{donePages:s.donePages.size,pagesRendered:s.stats.pagesRendered,pagesSaved:s.stats.pagesSaved,pagesQueued:s.stats.pagesQueued,pageQueue:s.pageQueue.length,phase:"save-pages"});}function Be(e){return e.donePages.size%5===0||e.donePages.size===1||e.donePages.size===e.stats.pagesQueued}function Ve(e,t){t.progress(`Render progress: processed ${e.donePages.size}, rendered ${e.stats.pagesRendered}, saved ${e.stats.pagesSaved}, discovered ${e.stats.pagesQueued}.`,{donePages:e.donePages.size,pagesRendered:e.stats.pagesRendered,pagesSaved:e.stats.pagesSaved,pagesQueued:e.stats.pagesQueued,phase:"render-pages"});}async function Or(e){await e.addInitScript(()=>{Object.defineProperty(window,"__WPSUITE_STATIC_EXPORT__",{value:true,writable:false,configurable:true});let t=r=>{if(r==null||String(r)==="")return true;try{let s=new URL(String(r),window.location.href),n=new URL(window.location.href);return s.hash="",n.hash="",s.href===n.href}catch{return false}};try{let r=window.location.assign.bind(window.location),s=window.location.replace.bind(window.location),n=window.location.reload.bind(window.location);Object.defineProperty(window.location,"assign",{configurable:!0,value:i=>{if(t(i)){console.warn("[smartcloud-static-publisher] blocked same-page location.assign",i??"");return}return r(i)}}),Object.defineProperty(window.location,"replace",{configurable:!0,value:i=>{if(t(i)){console.warn("[smartcloud-static-publisher] blocked same-page location.replace",i??"");return}return s(i)}}),Object.defineProperty(window.location,"reload",{configurable:!0,value:()=>{console.warn("[smartcloud-static-publisher] blocked location.reload");}});}catch{}});}async function jr(e,t,r,s,n,i,o){let u=await chromium.launch({headless:true}),a=await u.newContext({viewport:e.viewport,userAgent:"WPSuiteStaticPublisher/0.8 Playwright SitemapOnly",ignoreHTTPSErrors:e.ignoreHttpsErrors}),c=await u.newContext({viewport:e.viewport,userAgent:"WPSuiteStaticPublisher/0.8 Playwright SitemapOnly",javaScriptEnabled:false,ignoreHTTPSErrors:e.ignoreHttpsErrors});await Or(a);try{for(;t.pageQueue.length>0&&!(e.maxPages>0&&t.donePages.size>=e.maxPages);){let l=t.pageQueue.shift();if(!l)break;if(t.queuedPages.delete(l),t.donePages.has(l))continue;t.donePages.add(l),i.enabled&&i.seenPages.add(l);let p=new URL(l),d=at(e,l),g=Er(e,p.pathname);if(!B(p)||R(e,p)||ne(p)){if(R(e,p)){r.info(`Worker redirected non-page URL to asset queue: ${l}`,{url:l,source:"worker-guard"}),A(e,t,l,r,"worker-guard");continue}r.reject("page",l,"guard rejected non-page URL before rendering","worker");continue}let m=await ir(e,n,t,i,l,r);if(i.enabled&&s!=="single-url"){let P=i.manifest.pages[l];if(m.action==="reuse"&&P){for(let w of P.discoveredAssets)A(e,t,w,r,l);for(let w of P.discoveredPages)J(e,t,w,r,l);P.lastSeenRunId=i.runId,P.sitemapLastmod=te(t.sitemapLastmodByPage[l])??P.sitemapLastmod,m.changeToken?.supported&&(P.changeToken=m.changeToken.token,P.tokenSource=m.changeToken.tokenSource??P.tokenSource),r.info(`Incremental reuse skipped unchanged page ${l}`,{url:l,mode:"incremental",reason:m.changeToken?.supported===!0?"change-token-match":"sitemap-lastmod-match"}),Be(t)&&Ve(t,r);continue}}let h=await(g?c.newPage():a.newPage()),y=!1,v=new Set;await h.route("**/*",async P=>{let w=P.request();if(w.isNavigationRequest()&&w.frame()===h.mainFrame())try{let b=new URL(w.url()),L=new URL(l);if(b.hash="",L.hash="",b.href===L.href){if(y){r.warn(`Blocked same-page navigation/reload for ${l}`,{url:l,requestUrl:w.url()}),await P.abort("aborted");return}y=!0;}}catch{}await P.continue();}),h.on("response",P=>{try{let w=x(e,P.url(),e.sourceOrigin,void 0,"network-response");w&&R(e,w)&&(v.add(w.toString()),A(e,t,w.toString(),r,l));}catch{}}),r.page(`Rendering ${l}`,{url:l});try{let P=null,w=null;try{if(P=await h.goto(l,{waitUntil:"domcontentloaded",timeout:e.navigationTimeoutMs}),P&&P.ok())try{w=await P.text();}catch{w=null;}}catch(C){r.warn(`Navigation issue for ${l}; saving current DOM if available`,{url:l,error:String(C)});}if(d){let C=P?.status()??null;if(C!==404){r.error(`Generated 404 request path returned unexpected status for ${l}`,{url:l,expectedStatus:404,status:C,requestPath:e.generated404RequestPath});continue}r.info(`Capturing rendered 404 page for ${l}`,{url:l,status:C,outputPath:W(e,l)});}else if(P&&!P.ok()){r.skip("page",l,P.status()),await h.close();continue}g?(r.info(`Rendering without JS execution for ${l}`,{url:l,mode:"no-js"}),await Dr(e,h)):await Ir(e,h),t.stats.pagesRendered++,Be(t)&&Ve(t,r);let b=await Fr(e,h,l,r,!g);for(let C of b)A(e,t,C,r,l);let L=[...new Set([...b,...v])],I=w??await h.content(),E=s!=="single-url"&&!d?kr(e,l,I,r):[];if(s!=="single-url"&&!d)for(let C of E)J(e,t,C,r,l);await ze(e,l,I,t,r);let U=W(e,l);$(o,e,U),ee(o,U),i.trackManifest&&(i.manifest.pages[l]={url:l,outputPath:U,changeToken:m.changeToken?.supported?m.changeToken.token:null,tokenSource:m.changeToken?.supported?m.changeToken.tokenSource??null:null,sitemapLastmod:te(t.sitemapLastmodByPage[l])??null,discoveredPages:E,discoveredAssets:L,lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId});}catch(P){try{let w=await h.content();if(w&&w.trim()){await ze(e,l,w,t,r);let b=W(e,l);$(o,e,b),ee(o,b),r.warn(`Saved partial DOM for ${l}`,{url:l,error:String(P)});}}catch(w){r.error(`Could not save partial DOM for ${l}`,{url:l,error:String(w)});}r.error(`Failed page ${l}`,{url:l,error:String(P)});}finally{await h.close();}}}finally{await c.close(),await a.close(),await u.close();}}async function $r(e){try{let t=await S.readFile(e),s=(e.endsWith(".gz")?gunzipSync(t):t).toString("utf8").trim();if(!s)return [];try{let n=JSON.parse(s);if(Array.isArray(n))return n;if(n&&typeof n=="object")return [n]}catch{return s.split(/\r?\n/).map(n=>n.trim()).filter(Boolean).map(n=>{try{let i=JSON.parse(n);return i&&typeof i=="object"?i:null}catch{return null}}).filter(n=>n!==null)}return []}catch{return []}}function Ge(e){let t=String(e||"").toLowerCase();return t.includes("timeout")||t.includes("timed out")||t.includes("navigation timeout")}function Ur(e){let t=String(e?.job?.command||"").trim(),r=String(e?.job?.crawlMode||"full").trim();return t!=="publish"&&t!=="crawl"?false:r!=="incremental"}function Qr(e){let t=[e?.archivedAt,e?.job?.endedAt,e?.job?.startedAt];for(let r of t){let s=Date.parse(String(r||""));if(Number.isFinite(s))return s}return 0}async function ct(e){try{let t=await S.readFile(e,"utf8"),r=JSON.parse(t);return r&&typeof r=="object"&&!Array.isArray(r)?r:null}catch{return null}}function Ke(e){let t=String(e.originalFileName||"").toLowerCase();return t.endsWith("errors.jsonl")?0:t.endsWith("errors.json")?1:2}function qr(e,t){let r=Array.isArray(t?.artifacts)?t.artifacts.filter(s=>String(s?.role||"").trim()==="errors"&&String(s?.storedFileName||"").trim()!=="").sort((s,n)=>Ke(s)-Ke(n)).map(s=>f.join(e,String(s.storedFileName))):[];return [...new Set([...r,f.join(e,"errors.jsonl"),f.join(e,"errors.json"),f.join(e,"errors.jsonl.gz"),f.join(e,"errors.json.gz")])]}async function Xe(e){let t=await ct(f.join(e,"job.json"));for(let r of qr(e,t)){let s=await $r(r);if(s.length>0)return s}return []}async function Nr(e){let t=f.join(e.logDir,"archive"),r;try{r=await S.readdir(t);}catch{return ""}let s="",n=0;for(let i of r){let o=f.join(t,i);if(!(await S.stat(o).catch(()=>null))?.isDirectory())continue;let a=await ct(f.join(o,"job.json"));if(!Ur(a))continue;let c=Qr(a);c>=n&&(n=c,s=o);}return s}async function _r(e){let t=await Nr(e),r=t!==""?await Xe(t):await Xe(e.logDir),s=new Set;for(let n of r)if(Ge(n.error)||Ge(n.message)){let i=n.url;typeof i=="string"&&i&&s.add(i);}return [...s]}async function Wr(e){let t=[];for(let r of e)if(r.startsWith("@")){let s=r.slice(1),n=await S.readFile(s,"utf8");t.push(...n.split(/\r?\n/).map(i=>i.trim()).filter(i=>i&&!i.startsWith("#")));}else t.push(r);return t}async function Hr(e){try{let t=await S.readFile(f.join(e.outputDir,"asset-map.json"),"utf8");return JSON.parse(t)}catch{return {}}}async function Jr(e){let t=["crawl.log.jsonl","current-crawl-event.json","rejected.jsonl","ignored.jsonl","skipped-http.jsonl","errors.jsonl","timings.jsonl","rejected.json","ignored.json","skipped-http.json","errors.json","timings.json"];await S.mkdir(e,{recursive:true}),await Promise.all(t.map(async r=>{try{await S.unlink(f.join(e,r));}catch(s){if(s.code!=="ENOENT")throw s}}));}function Ze(e,t,r,s,n){let i=x(e,r,e.sourceOrigin,s,"manual-url");i&&(ie(i)?Re(e,t,i.toString(),s,n):R(e,i)?A(e,t,i.toString(),s,n):B(i)?J(e,t,i.toString(),s,n):s.reject("manual-url",i.toString(),"not page/sitemap/asset-like",n));}async function zr(){let e=await le(),t=Kt(process.argv.slice(2));if(t.resumeRewrite&&t.mode!=="full")throw new Error("--resume-rewrite is only supported for full crawl/publish jobs.");let r=[];t.mode==="retry-timeouts"&&(r=await _r(e)),t.mode==="single-url"&&(r=await Wr(t.urls));let s=t.crawlMode==="incremental"&&t.mode==="full",n=t.mode==="full"&&!t.resumeRewrite,i=t.preserveOutput?await Hr(e):{},o=s?await tr(e):tt(),u={enabled:s,trackManifest:n,manifest:Ne(o),previousManifest:Ne(o),runId:`${Date.now()}`,seenPages:new Set,changeTokenCache:new Map};await Jr(e.logDir);let a={pageQueue:[],queuedPages:new Set,donePages:new Set,assetQueue:[],queuedAssets:new Set,doneAssets:new Set,sitemapQueue:[],queuedSitemaps:new Set,doneSitemaps:new Set,sitemapLastmodByPage:{},assetMap:{...i},stats:{pagesQueued:0,pagesRendered:0,assetsQueued:0,sitemapsQueued:0,assetsSaved:0,pagesSaved:0}},c=er();t.preserveOutput||await S.rm(e.outputDir,{recursive:true,force:true}),await S.mkdir(e.outputDir,{recursive:true});let l=new _(e.logDir,e.logLevel);if(t.resumeRewrite)l.info("Resuming final rewrite from existing output.",{phase:"rewrite-text",mode:t.mode,crawlMode:t.crawlMode});else {let h=await chromium.launch({headless:true}),y=await h.newContext({ignoreHTTPSErrors:e.ignoreHttpsErrors}),v=y.request;try{if(l.mark("discovery"),t.mode==="full"?(e.sitemapPaths.forEach(w=>Re(e,a,w,l)),await Je(e,v,a,l),e.seedPaths.forEach(w=>Ze(e,a,w,l,"seed-path")),e.generated404RequestPath&&J(e,a,e.generated404RequestPath,l,"generated-404-request-path")):(r.forEach(w=>Ze(e,a,w,l,"cli")),await Je(e,v,a,l)),l.endMark("discovery",{pages:a.pageQueue.length,assets:a.assetQueue.length,sitemaps:a.sitemapQueue.length}),l.summary(`Queued ${a.pageQueue.length} pages, ${a.assetQueue.length} assets, ${a.sitemapQueue.length} sitemaps.`,{mode:t.mode,crawlMode:t.crawlMode,queuedPages:a.pageQueue.length,queuedAssets:a.assetQueue.length,queuedSitemaps:a.sitemapQueue.length}),t.mode==="retry-timeouts"&&a.pageQueue.length===0&&a.assetQueue.length===0&&a.sitemapQueue.length===0){l.summary("No timed-out URLs were queued. Skipping retry crawl.",{mode:t.mode,crawlMode:t.crawlMode,queuedPages:0,queuedAssets:0,queuedSitemaps:0}),await l.flush();return}l.mark("render-pages");let P=Array.from({length:Math.max(1,e.concurrency)},()=>jr(e,a,l,t.mode,v,u,c));if(await Promise.all(P),l.endMark("render-pages",{pages:a.donePages.size}),l.mark("download-assets"),await Tr(e,v,a,l,u,c),l.endMark("download-assets",{assets:a.doneAssets.size}),u.enabled&&e.maxPages===0&&await ar(e,a,u,l,c),t.mode==="full"){let w=await Ar(e,l);for(let b of w)$(c,e,b);}}finally{await y.close(),await h.close();}}u.trackManifest&&(u.manifest.updatedAt=new Date().toISOString(),await rr(e,u.manifest)),await S.writeFile(f.join(e.outputDir,"asset-map.json"),JSON.stringify(a.assetMap,null,2),"utf8"),JSON.stringify(i,null,2)!==JSON.stringify(a.assetMap,null,2)&&$(c,e,f.join(e.outputDir,"asset-map.json"));let p=t.mode==="single-url"?pr(e,a):u.enabled&&!t.resumeRewrite?or(e,u,c,i,a.assetMap):void 0;l.mark("rewrite-text");let d=t.mode==="single-url"?`Rewriting text files touched by URL crawl: 0/${p?.length??0}`:"Rewriting text files...";l.progress(d,{phase:"rewrite-text",index:0,totalFiles:p?.length,changedTextFiles:0}),await He(d);let g=Date.now(),m=await je(e,a.assetMap,{files:p,previousAssetMap:i,onProgress:async({index:h,totalFiles:y,changedTextFiles:v,file:P,changed:w})=>{w&&$(c,e,f.join(e.outputDir,P)),l.checkpoint(`Rewriting text file ${h}/${y}`,{phase:"rewrite-text",index:h,totalFiles:y,changedTextFiles:v,file:P});let b=Date.now();if(!(h===1||h===y||b-g>=5e3))return;g=b;let I=`Rewriting text files: ${h}/${y}`;l.progress(I,{phase:"rewrite-text",index:h,totalFiles:y,changedTextFiles:v,file:P}),await He(I);}});l.endMark("rewrite-text",{changedTextFiles:m}),await ke(e,{generatedAt:new Date().toISOString(),outputDir:f.resolve(e.outputDir),runMode:t.mode,crawlMode:t.crawlMode,fullSyncRequired:!(u.enabled&&!t.resumeRewrite),changedFiles:u.enabled&&!t.resumeRewrite?[...c.changedFiles].sort():[],deletedFiles:u.enabled&&!t.resumeRewrite?[...c.deletedFiles].sort():[],rewriteTargets:u.enabled&&!t.resumeRewrite?[...new Set((p||[]).map(h=>V(e.outputDir,f.isAbsolute(h)?h:f.join(e.outputDir,h))).filter(h=>h!==null))].sort():[]}),l.summary(t.resumeRewrite?`Done. Resumed final rewrite over existing output and changed ${m} text files.`:`Done. Rendered ${a.stats.pagesRendered} pages, processed ${a.doneSitemaps.size} sitemaps, downloaded ${a.doneAssets.size} assets.`,{mode:t.mode,crawlMode:t.crawlMode,resumeRewrite:t.resumeRewrite,...a.stats,pagesRendered:a.stats.pagesRendered,donePages:a.donePages.size,doneSitemaps:a.doneSitemaps.size,doneAssets:a.doneAssets.size,changedTextFiles:m}),await l.flush();}zr().catch(async e=>{console.error(e);try{let r=(await le().catch(()=>null))?.logDir??"logs",s=new _(r,"debug");s.error(`Unhandled error: ${e instanceof Error?e.message:String(e)}`,e instanceof Error?{stack:e.stack}:void 0),await s.flush();}catch{}process.exit(1);});
|
|
6
|
+
`,"utf8");});}enqueueJsonLine(t,r){this.enqueueTask(()=>Nt(t,r));}updateCurrentEvent(t,r,s){let n=Wt(t,r,s);this.enqueueTask(async()=>{await S.mkdir(this.logDir,{recursive:true}),await S.writeFile(this.currentEventPath,JSON.stringify(n,null,2),"utf8");});}push(t,r,s){let n=JSON.stringify({time:new Date().toISOString(),level:t,message:r,...s||{}});this.enqueueLine(n),this.updateCurrentEvent(t,r,s),t==="error"?console.error(r,s||""):t==="warn"?console.warn(r,s||""):(N[this.level]>=N.debug||["summary","page","sitemap","asset","timing"].includes(t))&&console.log(r);}info(t,r){this.push("info",t,r);}progress(t,r){qt("crawl.log.jsonl",t,r),this.updateCurrentEvent("progress",t,r);}checkpoint(t,r){this.updateCurrentEvent("checkpoint",t,r);}page(t,r){this.push("page",t,r);}sitemap(t,r){this.push("sitemap",t,r);}asset(t,r){this.push("asset",t,r);}warn(t,r){this.push("warn",t,r);}error(t,r){this.enqueueJsonLine(this.errorsPath,{message:t,...r||{}}),this.push("error",t,r);}summary(t,r){this.push("summary",t,r);}mark(t){this.marks.set(t,Date.now());}endMark(t,r){let s=this.marks.get(t);if(!s)return;let n=Date.now()-s,i={name:t,ms:n,seconds:Number((n/1e3).toFixed(2)),...r||{}};this.enqueueJsonLine(this.timingsPath,i),this.push("timing",`Timing ${t}: ${i.seconds}s`,i);}reject(t,r,s,n){let i={kind:t,url:r,reason:s,source:n};this.enqueueJsonLine(this.rejectedPath,i),N[this.level]>=N.debug&&this.push("reject",`Rejected ${t} ${r}: ${s}`,i);}ignore(t,r,s,n){let i={kind:t,url:r,reason:s,source:n};this.enqueueJsonLine(this.ignoredPath,i),N[this.level]>=N.debug&&this.push("ignore",`Ignored ${t} ${r}: ${s}`,i);}skip(t,r,s,n){let i={kind:t,url:r,status:s,source:n};this.enqueueJsonLine(this.skippedPath,i),this.push("warn",`Skipped ${t} ${r}: HTTP ${s}`,i);}async flush(){let t=Date.now()-this.startedAt,r={name:"total",ms:t,seconds:Number((t/1e3).toFixed(2))};if(this.enqueueJsonLine(this.timingsPath,r),await this.writeQueue,this.writeError)throw this.writeError}};var Bt=["text/css","javascript","image/","font/","application/font","application/json","application/xml","text/xml","application/rss+xml","application/atom+xml","application/xslt+xml","text/xsl","image/svg+xml","application/octet-stream"],Jt=new Set([".css",".js",".mjs",".json",".map",".xml",".xsl",".rss",".atom",".txt",".enc",".jws",".png",".jpg",".jpeg",".gif",".webp",".avif",".svg",".ico",".woff",".woff2",".ttf",".otf",".eot",".pdf",".mp4",".webm"]),zt=new Set(["",".html",".htm"]),Vt=new Set([".css",".xml",".xsl",".svg",".json",".html",".htm",".txt",".enc",".jws"]),Gt=["/wp-content/","/wp-includes/","/wp-admin/","/static/","/assets/","/build/","/_next/"];function Kt(e){let t=[],r="full",s="full",n=false,i=false;for(let o=0;o<e.length;o++){let u=e[o];if(u==="--retry-timeouts")r="retry-timeouts",n=true;else if(u==="--resume")n=true;else if(u==="--resume-rewrite")n=true,i=true;else if(u==="--crawl-mode"){let a=e[++o];if(!a)throw new Error("--crawl-mode requires a value");s=a==="incremental"?"incremental":"full",n||=s==="incremental";}else if(u.startsWith("--crawl-mode="))s=u.slice(13)==="incremental"?"incremental":"full",n||=s==="incremental";else if(u==="--incremental")s="incremental",n=true;else if(u==="--url"){let a=e[++o];if(!a)throw new Error("--url requires a value");r="single-url",n=true,t.push(a);}else if(u.startsWith("--url="))r="single-url",n=true,t.push(u.slice(6));else if(u==="--urls"){let a=e[++o];if(!a)throw new Error("--urls requires a file path");r="single-url",n=true,t.push(`@${a}`);}}return {mode:r,crawlMode:s,urls:t,preserveOutput:n,resumeRewrite:i}}function Se(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.resolve(e):""}function Xt(){let e=Se();return e?f.resolve(e,".."):""}function Zt(){let e=process.env.STATIC_PUBLISHER_WP_ROOT||process.env.WPSUITE_STATIC_PUBLISHER_WP_ROOT||"";return e.trim()?f.resolve(e):""}function Yt(e){let t=e.trim(),r=[{alias:"@storage-root",root:Xt(),requiredEnv:"STATIC_PUBLISHER_RUNTIME_DIR or WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR"},{alias:"@runtime",root:Se(),requiredEnv:"STATIC_PUBLISHER_RUNTIME_DIR or WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR"},{alias:"@wp-root",root:Zt(),requiredEnv:"STATIC_PUBLISHER_WP_ROOT or WPSUITE_STATIC_PUBLISHER_WP_ROOT"}];for(let s of r){if(t!==s.alias&&!t.startsWith(`${s.alias}/`))continue;if(!s.root)return {resolvedPath:null,alias:s.alias,requiredEnv:s.requiredEnv};let n=t.slice(s.alias.length).replace(/^\/+/,"");return {resolvedPath:n?f.resolve(s.root,n):s.root,alias:s.alias}}return {resolvedPath:f.resolve(t)}}function Ye(e){let t=Se();return t?f.join(t,"crawl-manifest.json"):f.join(e.outputDir,".crawl-manifest.json")}function et(){return {schemaVersion:2,updatedAt:"",pages:{},assets:{}}}function er(){return {rewriteTargets:new Set,changedFiles:new Set,deletedFiles:new Set}}function j(e,t,r){let s=V(t.outputDir,r);s&&(e.changedFiles.add(s),e.deletedFiles.delete(s));}function Qe(e,t,r){let s=V(t.outputDir,r);s&&(e.deletedFiles.add(s),e.changedFiles.delete(s));}function Y(e,t){e.rewriteTargets.add(f.resolve(t));}async function tr(e){try{let t=await S.readFile(Ye(e),"utf8"),r=JSON.parse(t);if(r&&typeof r=="object"&&r.pages&&typeof r.pages=="object"){if(r.schemaVersion===2&&r.assets&&typeof r.assets=="object")return r;if(r.schemaVersion===1)return {schemaVersion:2,updatedAt:String(r.updatedAt||""),pages:r.pages,assets:{}}}}catch{}return et()}function qe(e){return JSON.parse(JSON.stringify(e))}async function rr(e,t){let r=Ye(e);await S.mkdir(f.dirname(r),{recursive:true}),await S.writeFile(r,JSON.stringify(t,null,2),"utf8");}function ee(e){if(!e)return null;let t=e.trim();return t||null}function sr(e,t){let r=new Set,s=new Set([...Object.keys(e),...Object.keys(t)]);for(let n of s){let i=e[n],o=t[n];!i||!o||i===o||r.add(n);}return r}async function nr(e,t,r,s){let n=[...new Set(r.map(u=>u.trim()).filter(Boolean))],i=String(e.wpsuite?.siteKey||"").trim();if(n.length===0||!i)return new Map;let o=new URL("/wp-json/smartcloud-static-publisher/v1/change-tokens",e.sourceOrigin).toString();try{let u=await t.post(o,{timeout:3e4,failOnStatusCode:!1,headers:{"content-type":"application/json","x-site-key":i},data:JSON.stringify({urls:n})});if(!u.ok())return s.warn(`Change token lookup failed with HTTP ${u.status()}`,{endpoint:o,status:u.status()}),new Map;let a=await u.json().catch(()=>null);if(!a||!Array.isArray(a.items))return new Map;let c=new Map;for(let l of a.items)!l||typeof l.url!="string"||c.set(l.url,l);return c}catch(u){return s.warn("Change token lookup failed; falling back to sitemap metadata",{endpoint:o,error:String(u)}),new Map}}async function fe(e,t,r,s,n){if(!r.enabled&&!r.trackManifest)return null;if(r.changeTokenCache.has(s))return r.changeTokenCache.get(s)??null;let o=(await nr(e,t,[s],n)).get(s)??null;return r.changeTokenCache.set(s,o),o}async function ir(e,t,r,s,n,i){if(it(e,n))return {action:"render",changeToken:null};if(!s.enabled)return {action:"render",changeToken:s.trackManifest?await fe(e,t,s,n,i):null};let o=s.manifest.pages[n];if(!o)return {action:"render",changeToken:await fe(e,t,s,n,i)};let u=await fe(e,t,s,n,i);if(u?.supported&&u.token)return o.changeToken===u.token?{action:"reuse",changeToken:u}:{action:"render",changeToken:u};let a=ee(r.sitemapLastmodByPage[n]);return a&&o.sitemapLastmod&&o.sitemapLastmod===a?{action:"reuse",changeToken:u}:{action:"render",changeToken:u}}async function ar(e,t,r,s,n){let i=Ne(r.previousManifest),o=0;for(let[d,g]of Object.entries(r.manifest.pages))if(!r.seenPages.has(d)){Qe(n,e,g.outputPath);try{await S.unlink(g.outputPath);}catch(m){(m.code||"")!=="ENOENT"&&s.warn(`Failed to remove stale incremental page output for ${d}`,{url:d,outputPath:g.outputPath,error:String(m)});}delete r.manifest.pages[d],o++;}let u=Ne(r.manifest,r.runId),a=new Set,c=new Set;for(let d of u){for(let h of H(e,d))a.add(h);let g=r.manifest.assets[d]?.outputPath;g&&c.add(g);let m=te(e,t.assetMap,d);m&&c.add(m);}for(let d of Object.keys(r.manifest.assets))u.has(d)||delete r.manifest.assets[d];let l=0,p=0;for(let d of i){let g=new Set,m=r.previousManifest.assets[d]?.outputPath;m&&g.add(m);let h=te(e,t.assetMap,d);if(!u.has(d)&&h)g.add(h);else if(!u.has(d)){let y=be(e,d);g.add(y.originalFilePath),y.hashedFilePath&&g.add(y.hashedFilePath);}for(let y of g)if(!c.has(y)){Qe(n,e,y);try{await S.unlink(y),l++;}catch(v){(v.code||"")!=="ENOENT"&&s.warn(`Failed to remove stale incremental asset output for ${d}`,{url:d,outputPath:y,error:String(v)});}}for(let y of H(e,d))a.has(y)||Object.prototype.hasOwnProperty.call(t.assetMap,y)&&(delete t.assetMap[y],p++);}(o>0||l>0||p>0)&&s.info(`Incremental cleanup removed ${o} stale pages, ${l} stale assets, ${p} stale asset mappings`,{removedPages:o,removedAssets:l,removedAssetMappings:p,phase:"incremental-cleanup"});}function Ne(e,t){let r=new Set;for(let s of Object.values(e.pages))for(let n of s.discoveredAssets||[]){let i=n.trim();i&&r.add(i);}for(let[s,n]of Object.entries(e.assets||{})){if(t&&n.lastSeenRunId!==t)continue;let i=s.trim();i&&r.add(i);}return r}function _e(e,t,r){for(let s of t)for(let n of H(e,s))if(r.has(n))return true;return false}function or(e,t,r,s,n){let i=new Set(r.rewriteTargets),o=sr(s,n);if(o.size===0)return [...i];for(let u of Object.values(t.manifest.pages))_e(e,u.discoveredAssets,o)&&i.add(f.resolve(u.outputPath));for(let[u,a]of Object.entries(t.manifest.assets)){a.isText&&_e(e,a.discoveredAssets,o)&&i.add(f.resolve(a.outputPath));for(let c of H(e,u))if(o.has(c)){j(r,e,a.outputPath),a.isText&&i.add(f.resolve(a.outputPath));break}}return [...i]}function H(e,t){let r=new Set([t]);try{let s=new URL(t),i=[new URL(e.sourceOrigin).origin,...Object.keys(e.extraReplacements||{})];if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{i.push(new URL(e.targetOrigin).origin);}catch{}for(let o of i)try{r.add(new URL(o).origin+s.pathname+s.search);}catch{}r.add(s.pathname+s.search),r.add(s.pathname);}catch{}return [...r]}function ur(e,t){let r=new Set([t]);try{let s=new URL(t),i=[new URL(e.sourceOrigin).origin,...Object.keys(e.extraReplacements||{})];if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{i.push(new URL(e.targetOrigin).origin);}catch{}for(let o of i)try{r.add(new URL(o).origin+s.pathname+s.search);}catch{}r.add(s.pathname+s.search);}catch{}return [...r]}function tt(e,t){let r=e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/"?e.targetOrigin:"https://relative.invalid/";try{let s=new URL(t,r);return f.join(e.outputDir,decodeURIComponent(s.pathname))}catch{return null}}function lr(e,t,r){for(let s of ur(e,r)){let n=t[s];if(!n)continue;let i=tt(e,n);if(i)return {outputPath:i,publicUrl:n}}return null}function cr(e,t,r){for(let s of H(e,r)){let n=t[s];if(!n)continue;let i=tt(e,n);if(i)return i}return null}function te(e,t,r){let s=cr(e,t,r);return s||be(e,r).originalFilePath}function dr(){let e=process.env.STATIC_PUBLISHER_RUNTIME_DIR||process.env.WPSUITE_STATIC_PUBLISHER_RUNTIME_DIR||"";return e.trim()?f.join(f.resolve(e),"queue-runner-heartbeat.json"):""}async function We(e){let t=dr();if(t)try{let r=await S.readFile(t,"utf8"),s=JSON.parse(r);if(!s||typeof s!="object")return;await S.writeFile(t,JSON.stringify({...s,checkedAt:new Date().toISOString(),status:"running",currentStep:"rewrite",message:e},null,2),"utf8");}catch{}}function pr(e,t){let r=new Set;for(let s of t.donePages)r.add(f.resolve(W(e,s)));for(let s of t.doneAssets){let n=te(e,t.assetMap,s);!n||!$(n)||r.add(f.resolve(n));}return [...r]}var re="re:",Z=new Map;function gr(e){if(!e.startsWith(re))return null;let t=Z.get(e);if(t!==void 0)return t||null;let r=e.slice(re.length).trim();if(!r)return Z.set(e,false),null;try{let s=new RegExp(r);return Z.set(e,s),s}catch{return console.warn(`[crawl] Ignoring invalid path matcher regex: ${e}`),Z.set(e,false),null}}function ve(e,t){let r=gr(t);return r?r.test(e):t.startsWith(re)?false:e.startsWith(t)}function hr(e,t){return t.find(r=>ve(e,r))}function rt(e,t){let r=hr(t.pathname,e.blockedPathPrefixes);if(r)return r.startsWith(re)?`blocked path regex: ${r}`:`blocked path prefix: ${r}`;let s=e.blockedSearchFragments.find(n=>t.search.includes(n));return s?`blocked search fragment: ${s}`:null}function fr(e){let t=new URL(e.sourceOrigin),r=new Set([t.hostname,...e.allowedAssetHosts]);if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{r.add(new URL(e.targetOrigin).hostname);}catch{}for(let s of Object.keys(e.extraReplacements||{}))try{r.add(new URL(s).hostname);}catch{}return r}function mr(e,t){return fr(e).has(t.hostname)}function wr(e){let t=F(e.trim()).replace(/\\\//g,"/").replace(/^['"]|['"]$/g,"").trim();return t=t.replace(/[)]+$/g,"").trim(),t=t.replace(/;.*$/g,"").trim(),t}function st(e){let t=F(e).replace(/\\\//g,"/");return /[{}]|\bwindow\.|\blocation\.|\bincludes\(|\?\?null|\+|%7B|%7D|%22|<|>|\s/.test(t)}function se(e){return Gt.some(t=>e.pathname.startsWith(t))}function x(e,t,r=e.sourceOrigin,s,n="url"){let i=wr(t);if(!i||i.startsWith("data:")||i.startsWith("blob:")||i.startsWith("mailto:")||i.startsWith("tel:")||i.startsWith("#"))return s?.ignore(n,t,"empty or unsupported scheme",r),null;if(st(i))return s?.ignore(n,t,"looks like JavaScript/code fragment",r),null;try{let o=new URL(i,r);if(!["http:","https:"].includes(o.protocol))return s?.ignore(n,t,"unsupported protocol: ",r),null;if(!mr(e,o))return s?.ignore(n,o.toString(),"host not allowed",r),null;let u=rt(e,o);if(u)return s?.reject(n,o.toString(),u,r),null;let a=new URL(e.sourceOrigin);return o.protocol=a.protocol,o.host=a.host,o.hash="",o}catch{return s?.reject(n,t,"invalid URL",r),null}}function nt(e){return Jt.has(f.extname(e.pathname).toLowerCase())}function z(e){return se(e)?false:zt.has(f.extname(e.pathname).toLowerCase())}function ne(e){let t=e.pathname.toLowerCase();return t.endsWith(".xml")&&(t.includes("sitemap")||t.endsWith("/sitemap.xml"))}function R(e,t){return ne(t)||t.pathname==="/robots.txt"||t.pathname==="/llms.txt"||t.pathname.toLowerCase().endsWith(".xsl")?true:nt(t)?e.assetPathPrefixes.length===0?true:e.assetPathPrefixes.some(r=>ve(t.pathname,r)):false}function W(e,t){let r=new URL(t),s=decodeURIComponent(r.pathname);return s.endsWith("/")?s+="index.html":f.extname(s)||(s+="/index.html"),f.join(e.outputDir,s)}function Pr(e){if(!e.generated404RequestPath)return null;try{return new URL(e.generated404RequestPath,e.sourceOrigin).toString()}catch{return null}}function it(e,t){let r=Pr(e);if(!r)return false;try{return new URL(t,e.sourceOrigin).toString()===r}catch{return false}}function be(e,t){let r=new URL(t),s=decodeURIComponent(r.pathname);s.endsWith("/")&&(s+="index.html");let n=!!r.search,i=s,o=f.join(e.outputDir,i),u=de(e.targetOrigin,i),a=`${u}${r.search}`;if(!n)return {originalPathname:i,originalFilePath:o,originalPublicUrl:u,originalPublicUrlWithSearch:a,preferQueryHashed:n};let c=f.extname(s),p=`${c?s.slice(0,-c.length):s}.${le(r.search)}${c||".bin"}`;return {originalPathname:i,originalFilePath:o,originalPublicUrl:u,originalPublicUrlWithSearch:a,preferQueryHashed:n,hashedPathname:p,hashedFilePath:f.join(e.outputDir,p),hashedPublicUrl:de(e.targetOrigin,p)}}function yr(e,t,r,s,n){t[r]=s;try{let i=new URL(r),u=[new URL(e.sourceOrigin).origin,...Object.keys(e.extraReplacements||{})];if(e.targetOrigin&&e.targetOrigin!=="."&&e.targetOrigin!=="/")try{u.push(new URL(e.targetOrigin).origin);}catch{}for(let a of u)try{let c=new URL(a);t[c.origin+i.pathname+i.search]=s;}catch{}try{let a=(()=>{let c=new URL(s,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin);return `${c.pathname}${c.search}`})();t[i.pathname+i.search]=a;}catch{}try{let a=new URL(s,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin).pathname;t[i.pathname]=a;}catch{}try{let a=new URL(n,e.targetOrigin==="."?"https://relative.invalid":e.targetOrigin).pathname;t[i.pathname]=a;}catch{}}catch{}}async function at(e){try{return await S.readFile(e)}catch(t){if(t.code==="ENOENT")return null;throw t}}async function ot(e){try{return await S.stat(e)}catch(t){if(t.code==="ENOENT")return null;throw t}}function Sr(e){let t=String(e["content-length"]||"").trim();if(!t)return null;let r=Number.parseInt(t,10);return Number.isFinite(r)&&r>=0?r:null}function ut(e){let t=String(e["last-modified"]||"").trim();if(!t)return null;let r=Date.parse(t);return Number.isFinite(r)?r:null}function vr(e,t){let r=ut(t),s=Sr(t),n=e?.mtime.getTime();return r!==null&&Math.abs((n??0)-r)>=1e3||s!==null&&e?.size!==s?false:r!==null||s!==null}async function br(e,t){if(!t)return;let r=ut(t);if(r===null)return;let s=await ot(e);if(!s)return;let n=new Date(r);await S.utimes(e,s.atime,n).catch(()=>{});}function me(e,t){let r=f.extname(e.pathname).toLowerCase(),s=t["content-type"]||"";return Vt.has(r)||Bt.some(n=>s.includes(n))&&!s.startsWith("image/")&&!s.includes("font")}function we(e,t){(e.doneAssets.size%25===0||e.doneAssets.size===1||e.doneAssets.size===e.stats.assetsQueued)&&t.progress(`Asset download progress: downloaded ${e.doneAssets.size}, discovered ${e.stats.assetsQueued}.`,{doneAssets:e.doneAssets.size,assetsQueued:e.stats.assetsQueued,phase:"download-assets"});}function Rr(e){return le(e.toString("base64"),40)}function Cr(e,t,r){let s=e.assetVariantDigestsByOriginalPath.get(t);if(!s||s.size===0)return false;for(let n of s)if(n!==r)return true;return false}function xr(e,t,r){let s=e.assetVariantDigestsByOriginalPath.get(t);s||(s=new Set,e.assetVariantDigestsByOriginalPath.set(t,s)),s.add(r);}async function ye(e,t,r,s,n,i,o){let u=be(e,r),a=Rr(s),c=u.originalFilePath,l=u.originalPublicUrlWithSearch;if(u.preferQueryHashed&&u.hashedFilePath&&u.hashedPublicUrl&&Cr(n,u.originalFilePath,a)){let m=lr(e,t,r);m?(c=m.outputPath,l=m.publicUrl):(c=u.hashedFilePath,l=u.hashedPublicUrl);}yr(e,t,r,l,u.originalPublicUrl);let p=$(c)?Buffer.from(he(s.toString("utf8"),e,t,c),"utf8"):s,d=await at(c),g=false;return (d===null||!d.equals(p))&&(await ce(c),await S.writeFile(c,p),g=true),xr(n,u.originalFilePath,a),await br(c,o),n.stats.assetsSaved++,i&&(n.stats.assetsSaved%50===0||n.stats.assetsSaved===1)&&i.progress(`Asset progress: saved ${n.stats.assetsSaved}, discovered ${n.stats.assetsQueued}, pending ${n.assetQueue.length}.`,{assetsSaved:n.stats.assetsSaved,assetsQueued:n.stats.assetsQueued,assetQueue:n.assetQueue.length,phase:"download-assets"}),{outputPath:c,publicUrl:l,wroteFile:g}}function B(e,t,r,s,n){let i=n&&/^https?:\/\//i.test(n)?n:e.sourceOrigin,o=x(e,r,i,s,"page");if(!o)return;if(!z(o)||nt(o)){if(R(e,o)){s.info(`Seeded/non-page URL queued as asset: ${o.toString()}`,{url:o.toString(),source:n}),A(e,t,o.toString(),s,n);return}s.reject("page",o.toString(),se(o)?"asset/internal path cannot be page":"not page-like",n);return}let u=o.toString();!t.donePages.has(u)&&!t.queuedPages.has(u)&&(t.queuedPages.add(u),t.pageQueue.push(u),t.stats.pagesQueued++,(t.stats.pagesQueued%25===0||t.stats.pagesQueued===1)&&s.progress(`Discovery progress: pages ${t.stats.pagesQueued}, assets ${t.stats.assetsQueued}, sitemaps ${t.stats.sitemapsQueued}.`,{pagesQueued:t.stats.pagesQueued,assetsQueued:t.stats.assetsQueued,sitemapsQueued:t.stats.sitemapsQueued,pageQueue:t.pageQueue.length,phase:"discovery"}));}function A(e,t,r,s,n){let i=n&&/^https?:\/\//i.test(n)?n:e.sourceOrigin,o=x(e,r,i,s,"asset");if(!o)return;if(!R(e,o)){s.reject("asset",o.toString(),"not a safe asset path/prefix",n);return}let u=o.toString();!t.doneAssets.has(u)&&!t.queuedAssets.has(u)&&(t.queuedAssets.add(u),t.assetQueue.push(u),t.stats.assetsQueued++,(t.stats.assetsQueued%100===0||t.stats.assetsQueued===1)&&s.progress(`Discovery progress: pages ${t.stats.pagesQueued}, assets ${t.stats.assetsQueued}, sitemaps ${t.stats.sitemapsQueued}.`,{pagesQueued:t.stats.pagesQueued,assetsQueued:t.stats.assetsQueued,sitemapsQueued:t.stats.sitemapsQueued,assetQueue:t.assetQueue.length,phase:"discovery"}));}function Re(e,t,r,s,n){let i=n&&/^https?:\/\//i.test(n)?n:e.sourceOrigin,o=x(e,r,i,s,"sitemap");if(!o)return;if(!ne(o)){s.reject("sitemap",o.toString(),"not sitemap-like",n);return}let u=o.toString();!t.doneSitemaps.has(u)&&!t.queuedSitemaps.has(u)&&(t.queuedSitemaps.add(u),t.sitemapQueue.push(u),t.stats.sitemapsQueued++,(t.stats.sitemapsQueued%10===0||t.stats.sitemapsQueued===1)&&s.progress(`Discovery progress: pages ${t.stats.pagesQueued}, assets ${t.stats.assetsQueued}, sitemaps ${t.stats.sitemapsQueued}.`,{pagesQueued:t.stats.pagesQueued,assetsQueued:t.stats.assetsQueued,sitemapsQueued:t.stats.sitemapsQueued,sitemapQueue:t.sitemapQueue.length,phase:"discovery"}));}function Ar(e){let t=F(e),r=[...t.matchAll(/<url>\s*([\s\S]*?)\s*<\/url>/gi)].map(s=>s[1]);return r.length>0?r.map(s=>{let n=s.match(/<loc>\s*([^<]+?)\s*<\/loc>/i);if(!n?.[1])return null;let i=s.match(/<lastmod>\s*([^<]+?)\s*<\/lastmod>/i);return {loc:n[1].trim(),lastmod:ee(i?.[1]??null)}}).filter(s=>!!s):[...t.matchAll(/<loc>\s*([^<]+?)\s*<\/loc>/gi)].map(s=>({loc:s[1].trim(),lastmod:null})).filter(s=>s.loc)}function kr(e){return e.split(",").map(t=>t.trim().split(/\s+/)[0]).filter(Boolean)}function Mr(e){let t=[],r=/url\(\s*(?:"([^"]+)"|'([^']+)'|([^)]*?))\s*\)/gi;for(let s of e.matchAll(r)){let n=(s[1]||s[2]||s[3]||"").trim();n&&t.push(n);}return t}function Pe(e){return e.trim().replace(/\\/g,"/").replace(/^https?:\/\/[^/]+/i,"").replace(/^\/+/,"")}async function lt(e,t){let r=[];await S.mkdir(t,{recursive:true});let s=await S.readdir(e,{withFileTypes:true});for(let n of s){let i=f.join(e,n.name),o=f.join(t,n.name);if(n.isDirectory()){r.push(...await lt(i,o));continue}await S.copyFile(i,o),r.push(o);}return r}async function Tr(e,t){let r=Object.entries(e.postCrawlCopyMap||{}).map(([i,o])=>({sourcePath:i.trim(),prefix:String(o||"").trim()}));if(r.length===0)return [];let s=0,n=[];t.mark("copy-extra-paths");for(let i of r){let o=i.sourcePath,u=i.prefix;if(!o||!u){t.warn("Skipped post-crawl copy mapping with empty key/value",{sourcePath:o,prefixPath:u});continue}let a=Yt(o);if(!a.resolvedPath){t.warn("Skipped post-crawl copy source because alias root is not configured",{sourcePath:o,alias:a.alias||"",requiredEnv:a.requiredEnv||"",prefixPath:u});continue}let c=a.resolvedPath,l=Pe(u);if(!l){t.warn("Skipped post-crawl copy mapping with invalid export prefix",{sourcePath:o,prefixPath:u});continue}let p;try{p=await S.stat(c);}catch{t.warn("Skipped post-crawl copy source because it does not exist",{sourcePath:o,sourceAbs:c,prefixPath:u});continue}let d=f.resolve(e.outputDir,l);if(p.isDirectory()){n.push(...await lt(c,d)),s++,t.progress(`Copied static directory to export: ${c} -> /${l}`,{phase:"copy-extra-paths",sourcePath:c,targetPath:`/${l}`,copiedItems:s});continue}let g=f.basename(c),m=u.endsWith("/")?f.join(d,g):d;await S.mkdir(f.dirname(m),{recursive:true}),await S.copyFile(c,m),n.push(m),s++,t.progress(`Copied static file to export: ${c} -> /${Pe(u.endsWith("/")?`${u}${g}`:u)}`,{phase:"copy-extra-paths",sourcePath:c,targetPath:`/${Pe(u.endsWith("/")?`${u}${g}`:u)}`,copiedItems:s});}return t.endMark("copy-extra-paths",{mappedSources:r.length,copiedItems:s}),n}function J(e,t,r,s,n){let i=new Set,o=F(r).replace(/\\\//g,"/");for(let c of Mr(o)){let l=x(e,c,t,s,`${n}:css-url`);l&&R(e,l)&&i.add(l.toString());}for(let c of o.matchAll(/@import\s+(?:url\()?\s*['"]?([^'"\s;]+)['"]?\s*\)?/gi)){let l=x(e,c[1],t,s,`${n}:css-import`);l&&R(e,l)&&i.add(l.toString());}for(let c of o.matchAll(/<\?xml-stylesheet[^>]+href=["']([^"']+)["'][^>]*\?>/gi)){let l=x(e,c[1],t,s,`${n}:xml-stylesheet`);l&&R(e,l)&&i.add(l.toString());}let u="(?:css|js|mjs|json|map|xml|xsl|rss|atom|txt|enc|jws|png|jpe?g|gif|webp|avif|svg|ico|woff2?|ttf|otf|eot|pdf|mp4|webm)",a=new RegExp(`(?:https?:)?//[^\\s'"<>\\);]+\\.${u}(?:\\?[^\\s'"<>\\);]*)?|(?<!\\.)/[^\\s'"<>\\);]+\\.${u}(?:\\?[^\\s'"<>\\);]*)?`,"gi");for(let c of o.matchAll(a)){let l=c[0].startsWith("//")?`${new URL(e.sourceOrigin).protocol}${c[0]}`:c[0];if(st(l)){s.ignore(`${n}:serialized-url`,l,"looks like JavaScript/code fragment",t);continue}let p=x(e,l,t,s,`${n}:serialized-url`);p&&R(e,p)?i.add(p.toString()):p&&s.ignore("asset",p.toString(),"serialized URL did not pass safe asset path filter",t);}return [...i]}function Lr(e,t,r,s){let n=new Set,i=F(r);for(let o of i.matchAll(/href=["']([^"'#\s][^"']*?)["']/gi)){let u=o[1],a=x(e,u,t,s,"page-link");a&&(rt(e,a)||z(a)&&!se(a)&&!R(e,a)&&n.add(a.toString()));}return [...n]}async function He(e,t,r,s){for(;r.sitemapQueue.length>0;){let n=r.sitemapQueue.shift();if(r.queuedSitemaps.delete(n),!r.doneSitemaps.has(n)){r.doneSitemaps.add(n),s.sitemap(`Fetching sitemap ${n}`,{url:n});try{let i=await t.get(n,{timeout:6e4});if(!i.ok()){s.skip("sitemap",n,i.status());continue}let o=await i.body(),u=i.url()||n;await ye(e,r.assetMap,u,o,r,s);let a=o.toString("utf8");for(let c of J(e,u,a,s,"sitemap"))A(e,r,c,s,u);for(let c of Ar(a)){let l=x(e,c.loc,u,s,"sitemap-loc");l&&(ne(l)?Re(e,r,l.toString(),s,u):R(e,l)?A(e,r,l.toString(),s,u):z(l)?(c.lastmod&&(r.sitemapLastmodByPage[l.toString()]=c.lastmod),B(e,r,l.toString(),s,u)):s.reject("sitemap-loc",l.toString(),"not page/sitemap/asset-like",u));}}catch(i){s.error(`Failed sitemap ${n}`,{url:n,error:String(i)});}}}}async function Dr(e,t,r,s,n,i,o){let u=x(e,s,e.sourceOrigin,n,"asset-fetch");if(!u||!R(e,u))return;let a=u.toString();if(r.doneAssets.has(a))return;let c=i.enabled?i.previousManifest.assets[a]:void 0,l=(p,d={})=>{if(!c)return false;i.manifest.assets[a]={...c,discoveredAssets:[...c.discoveredAssets||[]],lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId},r.doneAssets.add(a);for(let g of c.discoveredAssets||[])A(e,r,g,n,a);return n.warn(`Keeping previous incremental asset ${a} after fetch issue.`,{url:a,reason:p,preservedFromPreviousManifest:true,...d,phase:"download-assets"}),we(r,n),true};n.asset(`Fetching asset ${a}`,{url:a});try{let p=te(e,r.assetMap,a),d=p?await ot(p):null;if(p&&d)try{let w=await t.head(a,{timeout:6e4}),b=w.headers();if(w.ok()&&vr(d,b)){let L=await at(p);if(L!==null){let D=w.url()||a,E=await ye(e,r.assetMap,D,L,r,n,b);E.wroteFile&&(j(o,e,E.outputPath),$(E.outputPath)&&Y(o,E.outputPath)),r.doneAssets.add(a),n.info(`Reused existing asset ${a} based on response headers.`,{url:a,filePath:p,lastModified:b["last-modified"]||"",contentLength:b["content-length"]||"",phase:"download-assets"}),we(r,n);let U=me(u,b),C=c?.discoveredAssets?[...c.discoveredAssets]:[];if(U||c?.isText){let dt=L.toString("utf8"),Ce=J(e,D,dt,n,"asset:cached");(Ce.length>0||!c)&&(C=Ce);for(let pt of C)A(e,r,pt,n,D);}i.trackManifest&&(i.manifest.assets[a]={url:a,outputPath:E.outputPath,isText:U||!!c?.isText,discoveredAssets:C,lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId});return}}}catch{}let g=await t.get(a,{timeout:6e4});if(!g.ok()){if(l(`http-${g.status()}`,{status:g.status()}))return;n.skip("asset",a,g.status());return}let m=await g.body(),h=g.url()||a,y=g.headers(),v=await ye(e,r.assetMap,h,m,r,n,y);v.wroteFile&&(j(o,e,v.outputPath),$(v.outputPath)&&Y(o,v.outputPath)),r.doneAssets.add(a),we(r,n);let P=[];if(me(u,y)){let w=m.toString("utf8");P=J(e,h,w,n,`asset:${f.extname(u.pathname).toLowerCase()||y["content-type"]||"unknown"}`);for(let b of P)A(e,r,b,n,h);}i.trackManifest&&(i.manifest.assets[a]={url:a,outputPath:v.outputPath,isText:me(u,y),discoveredAssets:P,lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId});}catch(p){if(l("fetch-error",{error:String(p)}))return;n.error(`Failed asset ${a}`,{url:a,error:String(p)});}}async function Ir(e,t,r,s,n,i){let o=Math.max(1,Number(e.assetDownloadConcurrency||e.concurrency||1)),u=Array.from({length:o},async()=>{for(;r.assetQueue.length>0;){let a=r.assetQueue.shift();a&&(r.queuedAssets.delete(a),await Dr(e,t,r,a,s,n,i));}});await Promise.all(u);}async function Er(e){await e.evaluate(async()=>{await new Promise(t=>{let r=0,s=700,n=window.setInterval(()=>{window.scrollBy(0,s),r+=s,r>=document.body.scrollHeight+window.innerHeight&&(window.clearInterval(n),window.scrollTo(0,0),t());},120);});});}async function Or(e,t){await t.waitForLoadState("domcontentloaded",{timeout:e.readiness.timeoutMs}).catch(()=>{}),e.readiness.waitForSelector&&await t.waitForSelector(e.readiness.waitForSelector,{timeout:e.readiness.timeoutMs}).catch(()=>{}),e.readiness.waitForFunction&&await t.waitForFunction(e.readiness.waitForFunction,void 0,{timeout:e.readiness.timeoutMs}).catch(()=>{}),await Er(t),await t.waitForTimeout(e.readiness.fallbackWaitMs);}async function Fr(e,t){await t.waitForLoadState("domcontentloaded",{timeout:e.readiness.timeoutMs}).catch(()=>{}),await t.waitForLoadState("load",{timeout:e.readiness.timeoutMs}).catch(()=>{}),e.readiness.waitForSelector&&await t.waitForSelector(e.readiness.waitForSelector,{timeout:e.readiness.timeoutMs}).catch(()=>{}),await t.waitForTimeout(e.readiness.fallbackWaitMs);}function jr(e,t){return e.noJavaScriptRenderPathPrefixes.some(r=>ve(t,r))}async function $r(e,t,r,s,n=true){let i=new Set;if(n){let u=await t.evaluate(()=>{let a=new Set,c=["href","src","poster","data-src","data-lazy-src","data-original","data-bg","data-background","data-href"],l=new Set,p=["srcset","data-srcset","data-lazy-srcset"];return document.querySelectorAll("*").forEach(d=>{for(let m of c){let h=d.getAttribute(m);h&&a.add(h);}for(let m of p){let h=d.getAttribute(m);h&&h.split(",").forEach(y=>{let v=y.trim().split(/\s+/)[0];v&&a.add(v);});}let g=d.getAttribute("style");g&&l.add(g);}),{attrs:[...a],styles:[...l]}});for(let a of u.attrs)for(let c of kr(a)){let l=x(e,c,r,s,"dom-attr");l&&R(e,l)&&i.add(l.toString());}for(let a of u.styles)for(let c of J(e,r,a,s,"dom-style"))i.add(c);}let o=await t.content();for(let u of J(e,r,o,s,"page-html"))i.add(u);return [...i]}async function Be(e,t,r,s,n){let i=W(e,t);await ce(i),await S.writeFile(i,r,"utf8"),s.stats.pagesSaved++,n&&(s.stats.pagesSaved%10===0||s.stats.pagesSaved===1)&&n.progress(`Page progress: processed ${s.donePages.size}, rendered ${s.stats.pagesRendered}, saved ${s.stats.pagesSaved}, discovered ${s.stats.pagesQueued}.`,{donePages:s.donePages.size,pagesRendered:s.stats.pagesRendered,pagesSaved:s.stats.pagesSaved,pagesQueued:s.stats.pagesQueued,pageQueue:s.pageQueue.length,phase:"save-pages"});}function Je(e){return e.donePages.size%5===0||e.donePages.size===1||e.donePages.size===e.stats.pagesQueued}function ze(e,t){t.progress(`Render progress: processed ${e.donePages.size}, rendered ${e.stats.pagesRendered}, saved ${e.stats.pagesSaved}, discovered ${e.stats.pagesQueued}.`,{donePages:e.donePages.size,pagesRendered:e.stats.pagesRendered,pagesSaved:e.stats.pagesSaved,pagesQueued:e.stats.pagesQueued,phase:"render-pages"});}async function Ur(e){await e.addInitScript(()=>{Object.defineProperty(window,"__WPSUITE_STATIC_EXPORT__",{value:true,writable:false,configurable:true});let t=r=>{if(r==null||String(r)==="")return true;try{let s=new URL(String(r),window.location.href),n=new URL(window.location.href);return s.hash="",n.hash="",s.href===n.href}catch{return false}};try{let r=window.location.assign.bind(window.location),s=window.location.replace.bind(window.location),n=window.location.reload.bind(window.location);Object.defineProperty(window.location,"assign",{configurable:!0,value:i=>{if(t(i)){console.warn("[smartcloud-static-publisher] blocked same-page location.assign",i??"");return}return r(i)}}),Object.defineProperty(window.location,"replace",{configurable:!0,value:i=>{if(t(i)){console.warn("[smartcloud-static-publisher] blocked same-page location.replace",i??"");return}return s(i)}}),Object.defineProperty(window.location,"reload",{configurable:!0,value:()=>{console.warn("[smartcloud-static-publisher] blocked location.reload");}});}catch{}});}async function Qr(e,t,r,s,n,i,o){let u=await chromium.launch({headless:true}),a=await u.newContext({viewport:e.viewport,userAgent:"WPSuiteStaticPublisher/0.8 Playwright SitemapOnly",ignoreHTTPSErrors:e.ignoreHttpsErrors}),c=await u.newContext({viewport:e.viewport,userAgent:"WPSuiteStaticPublisher/0.8 Playwright SitemapOnly",javaScriptEnabled:false,ignoreHTTPSErrors:e.ignoreHttpsErrors});await Ur(a);try{for(;t.pageQueue.length>0&&!(e.maxPages>0&&t.donePages.size>=e.maxPages);){let l=t.pageQueue.shift();if(!l)break;if(t.queuedPages.delete(l),t.donePages.has(l))continue;t.donePages.add(l),i.enabled&&i.seenPages.add(l);let p=new URL(l),d=it(e,l),g=jr(e,p.pathname);if(!z(p)||R(e,p)||se(p)){if(R(e,p)){r.info(`Worker redirected non-page URL to asset queue: ${l}`,{url:l,source:"worker-guard"}),A(e,t,l,r,"worker-guard");continue}r.reject("page",l,"guard rejected non-page URL before rendering","worker");continue}let m=await ir(e,n,t,i,l,r);if(i.enabled&&s!=="single-url"){let P=i.manifest.pages[l];if(m.action==="reuse"&&P){for(let w of P.discoveredAssets)A(e,t,w,r,l);for(let w of P.discoveredPages)B(e,t,w,r,l);P.lastSeenRunId=i.runId,P.sitemapLastmod=ee(t.sitemapLastmodByPage[l])??P.sitemapLastmod,m.changeToken?.supported&&(P.changeToken=m.changeToken.token,P.tokenSource=m.changeToken.tokenSource??P.tokenSource),r.info(`Incremental reuse skipped unchanged page ${l}`,{url:l,mode:"incremental",reason:m.changeToken?.supported===!0?"change-token-match":"sitemap-lastmod-match"}),Je(t)&&ze(t,r);continue}}let h=await(g?c.newPage():a.newPage()),y=!1,v=new Set;await h.route("**/*",async P=>{let w=P.request();if(w.isNavigationRequest()&&w.frame()===h.mainFrame())try{let b=new URL(w.url()),L=new URL(l);if(b.hash="",L.hash="",b.href===L.href){if(y){r.warn(`Blocked same-page navigation/reload for ${l}`,{url:l,requestUrl:w.url()}),await P.abort("aborted");return}y=!0;}}catch{}await P.continue();}),h.on("response",P=>{try{let w=x(e,P.url(),e.sourceOrigin,void 0,"network-response");w&&R(e,w)&&(v.add(w.toString()),A(e,t,w.toString(),r,l));}catch{}}),r.page(`Rendering ${l}`,{url:l});try{let P=null,w=null;try{if(P=await h.goto(l,{waitUntil:"domcontentloaded",timeout:e.navigationTimeoutMs}),P&&P.ok())try{w=await P.text();}catch{w=null;}}catch(C){r.warn(`Navigation issue for ${l}; saving current DOM if available`,{url:l,error:String(C)});}if(d){let C=P?.status()??null;if(C!==404){r.error(`Generated 404 request path returned unexpected status for ${l}`,{url:l,expectedStatus:404,status:C,requestPath:e.generated404RequestPath});continue}r.info(`Capturing rendered 404 page for ${l}`,{url:l,status:C,outputPath:W(e,l)});}else if(P&&!P.ok()){r.skip("page",l,P.status()),await h.close();continue}g?(r.info(`Rendering without JS execution for ${l}`,{url:l,mode:"no-js"}),await Fr(e,h)):await Or(e,h),t.stats.pagesRendered++,Je(t)&&ze(t,r);let b=await $r(e,h,l,r,!g);for(let C of b)A(e,t,C,r,l);let L=[...new Set([...b,...v])],D=w??await h.content(),E=s!=="single-url"&&!d?Lr(e,l,D,r):[];if(s!=="single-url"&&!d)for(let C of E)B(e,t,C,r,l);await Be(e,l,D,t,r);let U=W(e,l);j(o,e,U),Y(o,U),i.trackManifest&&(i.manifest.pages[l]={url:l,outputPath:U,changeToken:m.changeToken?.supported?m.changeToken.token:null,tokenSource:m.changeToken?.supported?m.changeToken.tokenSource??null:null,sitemapLastmod:ee(t.sitemapLastmodByPage[l])??null,discoveredPages:E,discoveredAssets:L,lastCrawledAt:new Date().toISOString(),lastSeenRunId:i.runId});}catch(P){try{let w=await h.content();if(w&&w.trim()){await Be(e,l,w,t,r);let b=W(e,l);j(o,e,b),Y(o,b),r.warn(`Saved partial DOM for ${l}`,{url:l,error:String(P)});}}catch(w){r.error(`Could not save partial DOM for ${l}`,{url:l,error:String(w)});}r.error(`Failed page ${l}`,{url:l,error:String(P)});}finally{await h.close();}}}finally{await c.close(),await a.close(),await u.close();}}async function qr(e){try{let t=await S.readFile(e),s=(e.endsWith(".gz")?gunzipSync(t):t).toString("utf8").trim();if(!s)return [];try{let n=JSON.parse(s);if(Array.isArray(n))return n;if(n&&typeof n=="object")return [n]}catch{return s.split(/\r?\n/).map(n=>n.trim()).filter(Boolean).map(n=>{try{let i=JSON.parse(n);return i&&typeof i=="object"?i:null}catch{return null}}).filter(n=>n!==null)}return []}catch{return []}}function Ve(e){let t=String(e||"").toLowerCase();return t.includes("timeout")||t.includes("timed out")||t.includes("navigation timeout")}function Nr(e){let t=String(e?.job?.command||"").trim(),r=String(e?.job?.crawlMode||"full").trim();return t!=="publish"&&t!=="crawl"?false:r!=="incremental"}function _r(e){let t=[e?.archivedAt,e?.job?.endedAt,e?.job?.startedAt];for(let r of t){let s=Date.parse(String(r||""));if(Number.isFinite(s))return s}return 0}async function ct(e){try{let t=await S.readFile(e,"utf8"),r=JSON.parse(t);return r&&typeof r=="object"&&!Array.isArray(r)?r:null}catch{return null}}function Ge(e){let t=String(e.originalFileName||"").toLowerCase();return t.endsWith("errors.jsonl")?0:t.endsWith("errors.json")?1:2}function Wr(e,t){let r=Array.isArray(t?.artifacts)?t.artifacts.filter(s=>String(s?.role||"").trim()==="errors"&&String(s?.storedFileName||"").trim()!=="").sort((s,n)=>Ge(s)-Ge(n)).map(s=>f.join(e,String(s.storedFileName))):[];return [...new Set([...r,f.join(e,"errors.jsonl"),f.join(e,"errors.json"),f.join(e,"errors.jsonl.gz"),f.join(e,"errors.json.gz")])]}async function Ke(e){let t=await ct(f.join(e,"job.json"));for(let r of Wr(e,t)){let s=await qr(r);if(s.length>0)return s}return []}async function Hr(e){let t=f.join(e.logDir,"archive"),r;try{r=await S.readdir(t);}catch{return ""}let s="",n=0;for(let i of r){let o=f.join(t,i);if(!(await S.stat(o).catch(()=>null))?.isDirectory())continue;let a=await ct(f.join(o,"job.json"));if(!Nr(a))continue;let c=_r(a);c>=n&&(n=c,s=o);}return s}async function Br(e){let t=await Hr(e),r=t!==""?await Ke(t):await Ke(e.logDir),s=new Set;for(let n of r)if(Ve(n.error)||Ve(n.message)){let i=n.url;typeof i=="string"&&i&&s.add(i);}return [...s]}async function Jr(e){let t=[];for(let r of e)if(r.startsWith("@")){let s=r.slice(1),n=await S.readFile(s,"utf8");t.push(...n.split(/\r?\n/).map(i=>i.trim()).filter(i=>i&&!i.startsWith("#")));}else t.push(r);return t}async function zr(e){try{let t=await S.readFile(f.join(e.outputDir,"asset-map.json"),"utf8");return JSON.parse(t)}catch{return {}}}async function Vr(e){let t=["crawl.log.jsonl","current-crawl-event.json","rejected.jsonl","ignored.jsonl","skipped-http.jsonl","errors.jsonl","timings.jsonl","rejected.json","ignored.json","skipped-http.json","errors.json","timings.json"];await S.mkdir(e,{recursive:true}),await Promise.all(t.map(async r=>{try{await S.unlink(f.join(e,r));}catch(s){if(s.code!=="ENOENT")throw s}}));}function Xe(e,t,r,s,n){let i=x(e,r,e.sourceOrigin,s,"manual-url");i&&(ne(i)?Re(e,t,i.toString(),s,n):R(e,i)?A(e,t,i.toString(),s,n):z(i)?B(e,t,i.toString(),s,n):s.reject("manual-url",i.toString(),"not page/sitemap/asset-like",n));}async function Gr(){let e=await ue(),t=Kt(process.argv.slice(2));if(t.resumeRewrite&&t.mode!=="full")throw new Error("--resume-rewrite is only supported for full crawl/publish jobs.");let r=[];t.mode==="retry-timeouts"&&(r=await Br(e)),t.mode==="single-url"&&(r=await Jr(t.urls));let s=t.crawlMode==="incremental"&&t.mode==="full",n=t.mode==="full"&&!t.resumeRewrite,i=t.preserveOutput?await zr(e):{},o=s?await tr(e):et(),u={enabled:s,trackManifest:n,manifest:qe(o),previousManifest:qe(o),runId:`${Date.now()}`,seenPages:new Set,changeTokenCache:new Map};await Vr(e.logDir);let a={pageQueue:[],queuedPages:new Set,donePages:new Set,assetQueue:[],queuedAssets:new Set,doneAssets:new Set,sitemapQueue:[],queuedSitemaps:new Set,doneSitemaps:new Set,sitemapLastmodByPage:{},assetMap:{...i},assetVariantDigestsByOriginalPath:new Map,stats:{pagesQueued:0,pagesRendered:0,assetsQueued:0,sitemapsQueued:0,assetsSaved:0,pagesSaved:0}},c=er();t.preserveOutput||await S.rm(e.outputDir,{recursive:true,force:true}),await S.mkdir(e.outputDir,{recursive:true});let l=new _(e.logDir,e.logLevel);if(t.resumeRewrite)l.info("Resuming final rewrite from existing output.",{phase:"rewrite-text",mode:t.mode,crawlMode:t.crawlMode});else {let h=await chromium.launch({headless:true}),y=await h.newContext({ignoreHTTPSErrors:e.ignoreHttpsErrors}),v=y.request;try{if(l.mark("discovery"),t.mode==="full"?(e.sitemapPaths.forEach(w=>Re(e,a,w,l)),await He(e,v,a,l),e.seedPaths.forEach(w=>Xe(e,a,w,l,"seed-path")),e.generated404RequestPath&&B(e,a,e.generated404RequestPath,l,"generated-404-request-path")):(r.forEach(w=>Xe(e,a,w,l,"cli")),await He(e,v,a,l)),l.endMark("discovery",{pages:a.pageQueue.length,assets:a.assetQueue.length,sitemaps:a.sitemapQueue.length}),l.summary(`Queued ${a.pageQueue.length} pages, ${a.assetQueue.length} assets, ${a.sitemapQueue.length} sitemaps.`,{mode:t.mode,crawlMode:t.crawlMode,queuedPages:a.pageQueue.length,queuedAssets:a.assetQueue.length,queuedSitemaps:a.sitemapQueue.length}),t.mode==="retry-timeouts"&&a.pageQueue.length===0&&a.assetQueue.length===0&&a.sitemapQueue.length===0){l.summary("No timed-out URLs were queued. Skipping retry crawl.",{mode:t.mode,crawlMode:t.crawlMode,queuedPages:0,queuedAssets:0,queuedSitemaps:0}),await l.flush();return}l.mark("render-pages");let P=Array.from({length:Math.max(1,e.concurrency)},()=>Qr(e,a,l,t.mode,v,u,c));if(await Promise.all(P),l.endMark("render-pages",{pages:a.donePages.size}),l.mark("download-assets"),await Ir(e,v,a,l,u,c),l.endMark("download-assets",{assets:a.doneAssets.size}),u.enabled&&e.maxPages===0&&await ar(e,a,u,l,c),t.mode==="full"){let w=await Tr(e,l);for(let b of w)j(c,e,b);}}finally{await y.close(),await h.close();}}u.trackManifest&&(u.manifest.updatedAt=new Date().toISOString(),await rr(e,u.manifest)),await S.writeFile(f.join(e.outputDir,"asset-map.json"),JSON.stringify(a.assetMap,null,2),"utf8"),JSON.stringify(i,null,2)!==JSON.stringify(a.assetMap,null,2)&&j(c,e,f.join(e.outputDir,"asset-map.json"));let p=t.mode==="single-url"?pr(e,a):u.enabled&&!t.resumeRewrite?or(e,u,c,i,a.assetMap):void 0;l.mark("rewrite-text");let d=t.mode==="single-url"?`Rewriting text files touched by URL crawl: 0/${p?.length??0}`:"Rewriting text files...";l.progress(d,{phase:"rewrite-text",index:0,totalFiles:p?.length,changedTextFiles:0}),await We(d);let g=Date.now(),m=await Fe(e,a.assetMap,{files:p,previousAssetMap:i,onProgress:async({index:h,totalFiles:y,changedTextFiles:v,file:P,changed:w})=>{w&&j(c,e,f.join(e.outputDir,P)),l.checkpoint(`Rewriting text file ${h}/${y}`,{phase:"rewrite-text",index:h,totalFiles:y,changedTextFiles:v,file:P});let b=Date.now();if(!(h===1||h===y||b-g>=5e3))return;g=b;let D=`Rewriting text files: ${h}/${y}`;l.progress(D,{phase:"rewrite-text",index:h,totalFiles:y,changedTextFiles:v,file:P}),await We(D);}});l.endMark("rewrite-text",{changedTextFiles:m}),await ke(e,{generatedAt:new Date().toISOString(),outputDir:f.resolve(e.outputDir),runMode:t.mode,crawlMode:t.crawlMode,fullSyncRequired:!(u.enabled&&!t.resumeRewrite),changedFiles:u.enabled&&!t.resumeRewrite?[...c.changedFiles].sort():[],deletedFiles:u.enabled&&!t.resumeRewrite?[...c.deletedFiles].sort():[],rewriteTargets:u.enabled&&!t.resumeRewrite?[...new Set((p||[]).map(h=>V(e.outputDir,f.isAbsolute(h)?h:f.join(e.outputDir,h))).filter(h=>h!==null))].sort():[]}),l.summary(t.resumeRewrite?`Done. Resumed final rewrite over existing output and changed ${m} text files.`:`Done. Rendered ${a.stats.pagesRendered} pages, processed ${a.doneSitemaps.size} sitemaps, downloaded ${a.doneAssets.size} assets.`,{mode:t.mode,crawlMode:t.crawlMode,resumeRewrite:t.resumeRewrite,...a.stats,pagesRendered:a.stats.pagesRendered,donePages:a.donePages.size,doneSitemaps:a.doneSitemaps.size,doneAssets:a.doneAssets.size,changedTextFiles:m}),await l.flush();}Gr().catch(async e=>{console.error(e);try{let r=(await ue().catch(()=>null))?.logDir??"logs",s=new _(r,"debug");s.error(`Unhandled error: ${e instanceof Error?e.message:String(e)}`,e instanceof Error?{stack:e.stack}:void 0),await s.flush();}catch{}process.exit(1);});
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@smart-cloud/publisher-exporter",
|
|
3
|
-
"version": "1.1.
|
|
3
|
+
"version": "1.1.2",
|
|
4
4
|
"license": "MIT",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"description": "Headless Playwright static publisher for WordPress/Elementor sites with sitemap-only page discovery, strict asset capture, escaped URL rewrite, structured logs, and targeted retry modes.",
|