@akotliar/sitemap-qa 1.0.0-alpha.0

package/dist/index.cjs

@@ -0,0 +1,484 @@
+#!/usr/bin/env node
+"use strict";var ue=Object.create;var F=Object.defineProperty;var de=Object.getOwnPropertyDescriptor;var pe=Object.getOwnPropertyNames;var ge=Object.getPrototypeOf,fe=Object.prototype.hasOwnProperty;var he=(e,t,r,s)=>{if(t&&typeof t=="object"||typeof t=="function")for(let n of pe(t))!fe.call(e,n)&&n!==r&&F(e,n,{get:()=>t[n],enumerable:!(s=de(t,n))||s.enumerable});return e};var M=(e,t,r)=>(r=e!=null?ue(ge(e)):{},he(t||!e||!e.__esModule?F(r,"default",{value:e,enumerable:!0}):r,e));var tr=require("dotenv/config"),me=require("commander");var ce=require("commander"),D=require("fs"),k=M(require("ora"),1),x=M(require("chalk"),1);var E=require("fs/promises"),_=require("fs"),A=require("path"),O=require("os");var z={timeout:30,concurrency:10,outputFormat:"html",outputDir:"./sitemap-qa/report",verbose:!1,baseUrl:"https://example.com",acceptedPatterns:[]};async function G(e){let t={...z},r=(0,A.join)((0,O.homedir)(),".sitemap-qa","config.json");if((0,_.existsSync)(r))try{let a=JSON.parse(await(0,E.readFile)(r,"utf-8"));t={...t,...a}}catch(a){console.warn(`Warning: Failed to load global config: ${a}`)}let s=(0,A.join)(process.cwd(),".sitemap-qa.config.json");if((0,_.existsSync)(s))try{let a=JSON.parse(await(0,E.readFile)(s,"utf-8"));t={...t,...a}}catch(a){console.warn(`Warning: Failed to load project config: ${a}`)}let n=ye();return t={...t,...n},t=ve(t,e),e.baseUrl&&(t.baseUrl=e.baseUrl),be(t),t}function ye(){let e={};return process.env.SITEMAP_VERIFY_TIMEOUT&&(e.timeout=parseInt(process.env.SITEMAP_VERIFY_TIMEOUT,10)),e}function ve(e,t){let r={...e};return t.timeout&&t.timeout!=="30"&&(r.timeout=parseInt(t.timeout,10)),t.output&&(r.outputFormat=t.output),t.outputDir&&(r.outputDir=t.outputDir),t.verbose===!0&&(r.verbose=!0),t.acceptedPatterns&&(r.acceptedPatterns=t.acceptedPatterns.split(",").map(s=>s.trim()).filter(Boolean)),r}function be(e){if(e.timeout<1||e.timeout>300)throw new Error("Timeout must be between 1 and 300 seconds");if(!["json","html"].includes(e.outputFormat))throw new Error("Output format must be json or html")}var U=class extends Error{constructor(r,s){super(`Network request failed for ${r}: ${s.message}`);this.url=r;this.originalError=s;this.name="NetworkError"}code="NETWORK_ERROR"},w=class extends Error{constructor(r,s,n){let a=`HTTP ${s} error for ${r}`;s===403&&(a+=`
+   Note: 403 Forbidden often indicates bot protection (Cloudflare, etc.) or access restrictions`);super(a);this.url=r;this.statusCode=s;this.statusText=n;this.name="HttpError"}code="HTTP_ERROR"};var q=require("playwright");async function we(e,t){let r;try{r=await q.chromium.launch({headless:!0,args:["--disable-blink-features=AutomationControlled","--disable-dev-shm-usage","--no-sandbox"]});let n=await(await r.newContext({userAgent:"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",viewport:{width:1920,height:1080},locale:"en-US",timezoneId:"America/New_York",extraHTTPHeaders:{Accept:"text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8","Accept-Language":"en-US,en;q=0.9","Accept-Encoding":"gzip, deflate, br",DNT:"1",Connection:"keep-alive","Upgrade-Insecure-Requests":"1"}})).newPage();await n.addInitScript(()=>{Object.defineProperty(navigator,"webdriver",{get:()=>!1}),window.chrome={runtime:{}};let m=window.navigator.permissions.query;window.navigator.permissions.query=l=>l.name==="notifications"?Promise.resolve({state:Notification.permission}):m(l)}),n.setDefaultTimeout(t*1e3);let a=await n.goto(e,{waitUntil:"domcontentloaded",timeout:t*1e3});if(!a)throw new Error("No response received from page");let c=a.status(),o=await n.content(),i=n.url();if(await r.close(),c>=200&&c<300)return{content:o,statusCode:c,url:i};throw new w(i,c)}catch(s){throw r&&await r.close(),s.code==="HTTP_ERROR"?s:new U(e,s)}}async function S(e,t={}){let{timeout:r=30,maxRetries:s=3,retryDelay:n=1e3,useBrowser:a=!1}=t;new URL(e);let c=[408,429,500,502,503,504],o=null,i=!1;for(let m=0;m<=s;m++){try{if(a||i)return await we(e,r);let l=new AbortController,d=setTimeout(()=>l.abort(),r*1e3),p=await fetch(e,{method:"GET",headers:{"User-Agent":"sitemap-qa/1.0.0",Accept:"text/xml,application/xml,text/plain,*/*"},signal:l.signal,redirect:"follow"});clearTimeout(d);let u=p.status,g=await p.text();if(u>=200&&u<300)return{content:g,statusCode:u,url:p.url};if(u===403&&!i){i=!0;continue}if(!c.includes(u))throw new w(p.url,u);o=new w(p.url,u)}catch(l){if(l.code==="HTTP_ERROR"){let d=l;if(!c.includes(d.statusCode))throw l;o=l}else o=new U(e,l);if(m===s)break}if(m<s){let l=n*Math.pow(2,m);await new Promise(d=>setTimeout(d,l))}}throw o}function xe(e){return new URL(e).origin}async function Re(e,t){let r=new URL(e),n=r.hostname.startsWith("www.")?r.hostname.substring(4):`www.${r.hostname}`,a=`${r.protocol}//${n}/robots.txt`;try{let c=await S(a,{timeout:t.timeout,maxRetries:1});return c.statusCode===200||c.statusCode===404?n:r.hostname}catch(c){return c instanceof w&&c.statusCode===301,r.hostname}}async function j(e,t){let r=new URL(e).origin,s=[],n=["/sitemap.xml","/sitemap_index.xml","/sitemap-index.xml"],a=await Promise.allSettled(n.map(async c=>{let o=`${r}${c}`;try{return(await S(o,{timeout:t.timeout,maxRetries:1})).statusCode===200?(t.verbose&&console.log(`\u2713 Found sitemap at: ${o}`),{found:!0,url:o}):{found:!1}}catch(i){return i instanceof w?i.statusCode===401||i.statusCode===403?(s.push({url:o,statusCode:i.statusCode,error:i.statusCode===401?"Unauthorized":"Access Denied"}),t.verbose&&console.log(`\u26A0 Access denied: ${o} (${i.statusCode})`)):t.verbose&&console.log(`\u2717 Not found: ${o} (${i.statusCode})`):t.verbose&&console.log(`\u2717 Not found: ${o}`),{found:!1}}}));for(let c of a)if(c.status==="fulfilled"&&c.value.found)return{sitemaps:[c.value.url],issues:s};return t.verbose&&console.log("No sitemap found at standard paths"),{sitemaps:[],issues:s}}async function H(e,t){let r=`${new URL(e).origin}/robots.txt`;try{let n=(await S(r,{timeout:t.timeout,maxRetries:1})).content.split(`
+`),a=[];for(let c of n){let o=c.match(/^Sitemap:\s*(.+)$/i);if(o){let i=o[1].trim();try{new URL(i),a.push(i)}catch{t.verbose&&console.warn(`Invalid sitemap URL in robots.txt: ${i}`)}}}return t.verbose&&a.length>0&&console.log(`Found ${a.length} sitemap(s) in robots.txt`),a}catch{return t.verbose&&console.log(`No robots.txt found at ${r}`),[]}}function Se(e){if(e.includes("<sitemapindex"))return!0;if(e.includes("<urlset")){let t=/<url[^>]*>.*?<loc>([^<]+)<\/loc>.*?<\/url>/gs,r=Array.from(e.matchAll(t)),s=Math.min(5,r.length),n=0;for(let a=0;a<s;a++){let c=r[a][1].trim().toLowerCase();(c.includes("sitemap")||c.endsWith(".xml"))&&n++}return n>s/2}return!1}function ke(e){let t=[];if(e.includes("<sitemapindex")){let r=/<sitemap[^>]*>(.*?)<\/sitemap>/gs,s;for(;(s=r.exec(e))!==null;){let n=/<loc>([^<]+)<\/loc>/i.exec(s[1]);if(n){let a=n[1].trim();try{new URL(a),t.push(a)}catch{}}}}else{let r=/<url[^>]*>(.*?)<\/url>/gs,s;for(;(s=r.exec(e))!==null;){let n=/<loc>([^<]+)<\/loc>/i.exec(s[1]);if(n){let a=n[1].trim();if(a.toLowerCase().includes("sitemap")||a.toLowerCase().endsWith(".xml"))try{new URL(a),t.push(a)}catch{}}}}return t}async function C(e,t,r,s,n=10){let a=[],c=[...e],o=new Set,i=new Set,m=new Set,l=s,d=5;for(;c.length>0;){let b=c.splice(0,Math.min(d,c.length));if(await Promise.all(b.map(async h=>{if(o.has(h)){t.verbose&&console.warn(`Skipping duplicate sitemap: ${h}`);return}o.add(h);try{let y=await S(h,{timeout:t.timeout,maxRetries:2});if(Se(y.content)){t.verbose&&console.log(`Found sitemap index: ${h}`);let R=ke(y.content);c.push(...R),t.verbose&&console.log(`  \u2514\u2500 Contains ${R.length} child sitemap(s)`)}else a.push(h),t.verbose&&console.log(`\u2713 Discovered sitemap: ${h}`)}catch(y){if(y instanceof w&&y.statusCode===301?m.add(h):i.add(h),t.verbose){let R=y instanceof Error?y.message:String(y);if(y instanceof w&&y.statusCode===301){l||(l=await Re(r,t),t.verbose&&console.log(`Canonical domain detected: ${l}`));try{let T=new URL(h);T.hostname!==l?(console.warn(`\u26A0\uFE0F  Sitemap URL redirects (301): ${h}`),console.warn("   Problem: The sitemap index contains a URL that redirects."),console.warn(`   Likely issue: Domain mismatch - expected "${l}" but got "${T.hostname}"`),console.warn(`   Fix: Update sitemap index to use "https://${l}${T.pathname}"`)):(console.warn(`\u26A0\uFE0F  Sitemap URL redirects (301): ${h}`),console.warn("   Fix: Update the sitemap index to reference the final URL after redirect."))}catch{console.warn(`Failed to fetch sitemap ${h}: ${R}`)}}else console.warn(`Failed to fetch sitemap ${h}: ${R}`)}}})),o.size>1e3){console.warn("\u26A0\uFE0F  Processed over 1000 sitemap URLs. Stopping to prevent excessive requests.");break}}let p=o.size,u=i.size,g=m.size,f=p-a.length-u-g;return a.length===0&&p>0&&(console.warn(`
+\u26A0\uFE0F  SITEMAP DISCOVERY ISSUE`),f>0&&(u>0||g>0)?(console.warn(`Found ${f} sitemap index(es) containing ${u+g} child sitemap(s):`),g>0&&console.warn(`  - ${g} sitemap(s) return 301 redirects (content not accessible without following redirect)`),u>0&&console.warn(`  - ${u} sitemap(s) returned errors (404, 403, 500, or network issues)`)):g>0?console.warn(`All ${g} sitemap(s) return 301 redirects.`):u>0?(console.warn(`All ${u} sitemap(s) returned errors.`),console.warn(`
+Common causes:`),console.warn("  - 403 Forbidden: Bot protection (Cloudflare, etc.) or IP blocking"),console.warn("  - 404 Not Found: Sitemaps don't exist at these URLs"),console.warn("  - 500/502/503: Server errors or maintenance"),console.warn(`
+If sitemaps work in your browser but not here, the site likely has bot protection.`),console.warn("Try: Check if sitemaps load without JavaScript, or contact site administrator.")):console.warn(`Processed ${p} URL(s) but found no accessible sitemaps.`),console.warn(`
+Note: This tool does not follow redirects for sitemap URLs.`),g>0&&(console.warn(`
+Possible causes of redirects:`),console.warn("  - Sitemap index uses non-canonical domain (e.g., missing 'www' or vice versa)"),console.warn("  - Sitemap URLs redirect from HTTP to HTTPS"),console.warn("  - Intentional redirects in your site configuration"),console.warn(`
+Recommendation: Update sitemap index URLs to match the final destination (no redirects).`)),console.warn("")),{sitemaps:a,canonicalDomain:l}}async function V(e,t){let r=xe(e),s=[],n;t.verbose&&console.log("Strategy 1: Checking robots.txt for sitemap directives...");let a=await H(r,t);if(a.length>0){let{sitemaps:m,canonicalDomain:l}=await C(a,t,r,n);return n=l,{sitemaps:m,source:"robots-txt",accessIssues:[],canonicalDomain:n}}t.verbose&&console.log("Strategy 2: Trying standard sitemap paths...");let{sitemaps:c,issues:o,redirectedToCanonical:i}=await j(r,t);if(s=o,c.length>0){let{sitemaps:m,canonicalDomain:l}=await C(c,t,r,n);return n=l,{sitemaps:m,source:"standard-path",accessIssues:[],canonicalDomain:n}}if(i){let m=`https://${i}`;console.log(`
+\u{1F4A1} All requests redirected. Retrying with canonical domain: ${i}
+`);let l=await H(m,t);if(l.length>0){let{sitemaps:p,canonicalDomain:u}=await C(l,t,m,i);return{sitemaps:p,source:"robots-txt",accessIssues:[],canonicalDomain:u||i}}let{sitemaps:d}=await j(m,t);if(d.length>0){let{sitemaps:p,canonicalDomain:u}=await C(d,t,m,i);return{sitemaps:p,source:"standard-path",accessIssues:[],canonicalDomain:u||i}}}return{sitemaps:[],source:"none",accessIssues:s,canonicalDomain:n}}var $=require("fast-xml-parser"),Ue=new $.XMLParser({ignoreAttributes:!1,attributeNamePrefix:"@_",textNodeName:"_text",parseAttributeValue:!0,trimValues:!0,allowBooleanAttributes:!0,parseTagValue:!1});function Ce(e,t){let r=[];if(e.urlset){let s=Array.isArray(e.urlset.url)?e.urlset.url:[e.urlset.url];for(let n of s)!n||!n.loc||r.push({loc:n.loc,lastmod:n.lastmod,changefreq:n.changefreq,priority:n.priority?parseFloat(n.priority):void 0,source:t})}return r}async function B(e,t){let r=[];try{let s=$.XMLValidator.validate(e);if(s!==!0){let o=typeof s=="object"?s.err.msg:"Invalid XML";return{urls:[],errors:[`[${t}] XML parsing failed: ${o}`],totalCount:0,sitemapUrl:t}}let n=Ue.parse(e),a=Ce(n,t),c=[];for(let o of a)try{new URL(o.loc),o.priority!==void 0&&(o.priority<0||o.priority>1)&&(r.push(`Invalid priority ${o.priority} for ${o.loc} - clamping to 0-1`),o.priority=Math.max(0,Math.min(1,o.priority))),o.changefreq&&(["always","hourly","daily","weekly","monthly","yearly","never"].includes(o.changefreq.toLowerCase())||(r.push(`Invalid changefreq "${o.changefreq}" for ${o.loc}`),o.changefreq=void 0)),c.push(o)}catch{r.push(`Invalid URL format: ${o.loc}`)}return{urls:c,errors:r,totalCount:c.length,sitemapUrl:t}}catch(s){let n=s instanceof Error?s.message:String(s);return{urls:[],errors:[`[${t}] XML parsing failed: ${n}`],totalCount:0,sitemapUrl:t}}}async function W(e,t){let r=[],s=[],n=0,a=0;t.verbose&&console.log(`
+Extracting URLs from ${e.length} sitemap(s)...`);let o=await $e(e,10,async i=>{try{t.verbose&&console.log(`Extracting URLs from: ${i}`);let m=await S(i,{timeout:t.timeout,maxRetries:2}),l=await B(m.content,i),d=l.urls.map(p=>({...p,extractedAt:new Date().toISOString()}));return t.verbose&&console.log(`  \u2713 Extracted ${l.urls.length} URLs from ${i}`),{success:!0,urls:d,errors:l.errors}}catch(m){let l=`Failed to process ${i}: ${m instanceof Error?m.message:String(m)}`;return t.verbose&&console.error(`  \u2717 ${l}`),{success:!1,urls:[],errors:[l]}}});for(let i of o)i.success?(n++,r.push(...i.urls)):a++,s.push(...i.errors);return t.verbose&&(console.log(`
+Extraction complete:`),console.log(`  - Sitemaps processed: ${n}`),console.log(`  - Sitemaps failed: ${a}`),console.log(`  - Total URLs: ${r.length}`),console.log(`  - Errors: ${s.length}`)),{allUrls:r,sitemapsProcessed:n,sitemapsFailed:a,totalUrls:r.length,errors:s}}async function $e(e,t,r){let s=[];for(let n=0;n<e.length;n+=t){let a=e.slice(n,n+t),c=await Promise.all(a.map(r));s.push(...c)}return s}function Pe(e){try{let t=new URL(e),r=t.pathname;r.endsWith("/")&&r!=="/"&&(r=r.slice(0,-1));let s=Array.from(t.searchParams.entries()).sort(([a],[c])=>a.localeCompare(c)),n=new URLSearchParams(s);return`${t.protocol}//${t.host}${r}${n.toString()?"?"+n.toString():""}${t.hash}`}catch{return e}}function Te(e){if(e.length===1)return e[0];let t={...e[0]},r=e.map(o=>o.source);t.source=r.join(", ");let s=e.map(o=>o.lastmod).filter(o=>!!o).map(o=>new Date(o).getTime()).sort((o,i)=>i-o);s.length>0&&(t.lastmod=new Date(s[0]).toISOString());let n=e.map(o=>o.priority).filter(o=>o!==void 0);n.length>0&&(t.priority=Math.max(...n));let a=e.map(o=>o.changefreq).filter(o=>!!o);if(a.length>0){let o=new Map;for(let m of a)o.set(m,(o.get(m)||0)+1);let i=Array.from(o.entries()).sort((m,l)=>l[1]-m[1]);t.changefreq=i[0][0]}let c=e.map(o=>o.extractedAt).filter(o=>!!o).map(o=>new Date(o).getTime()).sort((o,i)=>i-o);return c.length>0&&(t.extractedAt=new Date(c[0]).toISOString()),t}function J(e,t=!1){let r=e.length;t&&console.log(`
+Consolidating ${e.length} URL(s)...`);let s=new Map;for(let c of e){let o=Pe(c.loc);s.has(o)||s.set(o,[]),s.get(o).push(c)}let n=[],a=[];for(let[c,o]of s.entries()){let i=Te(o);n.push(i),o.length>1&&a.push({url:c,count:o.length,sources:o.map(m=>m.source)})}if(t&&(console.log("Consolidation complete:"),console.log(`  - Input URLs: ${r}`),console.log(`  - Unique URLs: ${n.length}`),console.log(`  - Duplicates removed: ${r-n.length}`),a.length>0)){console.log(`
+Top duplicates:`);let c=a.sort((o,i)=>i.count-o.count).slice(0,5);for(let o of c)console.log(`  - ${o.url} (${o.count} times)`)}return{uniqueUrls:n,totalInputUrls:r,duplicatesRemoved:r-n.length,duplicateGroups:a}}var Q=[{name:"Authentication Parameter",category:"sensitive_params",severity:"high",regex:/[?&](token|auth|key|password|secret|apikey|session|credentials)=/i,description:"Query parameter may contain sensitive authentication data"},{name:"Debug Parameter",category:"sensitive_params",severity:"medium",regex:/[?&](debug|trace|verbose|test_mode)=/i,description:"Query parameter may contain debug or diagnostic flag"},{name:"HTTP in HTTPS Site",category:"protocol_inconsistency",severity:"medium",regex:/^http:\/\//,description:"HTTP URL in HTTPS sitemap (potential mixed content)"},{name:"Test Content Path",category:"test_content",severity:"medium",regex:/\/(?:test-|demo-|sample-|temp-|temporary-|placeholder-)|\/(test|demo|sample|temp|temporary|placeholder)(?:\/|$)/i,description:"URL path suggests test, demo, or unfinished content that may not be intended for indexing"}];function L(e){return e.replace(/[.*+?^${}()|[\]\\]/g,"\\$&")}function Ee(e){let t=e.split(".");return t.length>=2?t.slice(-2).join("."):e}function K(e,t){let r=new URL(e).hostname,s=Ee(r);if(t?.allowedSubdomains&&t.allowedSubdomains.length>0){let c=L(s),i=`^https?://(?!(?:(?:${t.allowedSubdomains.map(L).join("|")})\\.)?${c}(?:/|$))`;return{name:"Domain Mismatch",category:"domain_mismatch",severity:"high",regex:new RegExp(i),description:"URL does not match expected domain or allowed subdomains"}}let a=`^https?://(?!(?:www\\.)?${L(s)}(?:/|$))`;return{name:"Domain Mismatch",category:"domain_mismatch",severity:"high",regex:new RegExp(a),description:`URL does not match expected domain: ${s} (including www variant)`}}var Y=[{name:"Staging Subdomain",category:"environment_leakage",severity:"high",regex:/^https?:\/\/(staging|stg)\./i,description:"URL uses staging subdomain"},{name:"Development Subdomain",category:"environment_leakage",severity:"high",regex:/^https?:\/\/(dev|development)\./i,description:"URL uses development subdomain"},{name:"QA/Test Subdomain",category:"environment_leakage",severity:"high",regex:/^https?:\/\/(qa|test|uat|preprod)\./i,description:"URL uses test environment subdomain"},{name:"Localhost URL",category:"environment_leakage",severity:"high",regex:/^https?:\/\/(localhost|127\.0\.0\.1|0\.0\.0\.0)/,description:"URL points to localhost (development environment)"},{name:"Environment in Path",category:"environment_leakage",severity:"high",regex:/^https?:\/\/[^/]+\/(staging|dev|qa|uat|preprod)\//i,description:"URL path contains environment identifier at root level"}];var X=[{name:"Admin Path",category:"admin_paths",severity:"high",regex:/\/(admin|administrator)(?:\/|$|\?)/i,description:"URL contains /admin or /administrator as a path segment"},{name:"Dashboard Path",category:"admin_paths",severity:"high",regex:/\/dashboard(?:\/|$|\?)/i,description:"URL contains /dashboard as a path segment"},{name:"Config Path",category:"admin_paths",severity:"high",regex:/\/(config|configuration)(?:\/|$|\?)/i,description:"URL contains /config or /configuration as a path segment"},{name:"Console Path",category:"admin_paths",severity:"high",regex:/\/console(?:\/|$|\?)/i,description:"URL contains /console as a path segment"},{name:"Control Panel Path",category:"admin_paths",severity:"high",regex:/\/(cpanel|control-panel)(?:\/|$|\?)/i,description:"URL contains control panel as a path segment"}],Z=[{name:"Internal Content Path",category:"internal_content",severity:"medium",regex:/\/internal\b/i,description:"URL contains /internal path segment - may be internal-only content not intended for public indexing"}],ee=[{name:"Authentication Token Parameter",category:"sensitive_params",severity:"high",regex:/[?&](token|auth_token|access_token|api_token)=/i,description:"Query parameter may contain authentication token"},{name:"API Key Parameter",category:"sensitive_params",severity:"high",regex:/[?&](apikey|api_key|key)=/i,description:"Query parameter may contain API key"},{name:"Password Parameter",category:"sensitive_params",severity:"high",regex:/[?&](password|passwd|pwd)=/i,description:"Query parameter may contain password"},{name:"Secret Parameter",category:"sensitive_params",severity:"high",regex:/[?&](secret|client_secret)=/i,description:"Query parameter may contain secret value"},{name:"Session Parameter",category:"sensitive_params",severity:"high",regex:/[?&](session|sessionid|sid)=/i,description:"Query parameter may contain session identifier"},{name:"Credentials Parameter",category:"sensitive_params",severity:"high",regex:/[?&]credentials=/i,description:"Query parameter may contain credentials"},{name:"Debug Parameter",category:"sensitive_params",severity:"medium",regex:/[?&](debug|trace|verbose)=/i,description:"Query parameter contains debug or diagnostic flag"},{name:"Test Mode Parameter",category:"sensitive_params",severity:"medium",regex:/[?&](test_mode|test|testing)=/i,description:"Query parameter indicates test mode"}];function te(e){try{let t=new URL(e),r=["token","auth","auth_token","access_token","api_token","apikey","api_key","key","password","passwd","pwd","secret","client_secret","session","sessionid","sid","credentials"];for(let s of r)t.searchParams.has(s)&&t.searchParams.set(s,"[REDACTED]");return t.toString()}catch{return e}}function _e(e,t,r){switch(e){case"environment_leakage":return{rationale:`Production sitemap contains ${r} URL(s) from non-production environments (staging, dev, QA, test). This indicates configuration errors or environment leakage.`,recommendedAction:"Verify sitemap generation excludes non-production environments. Review deployment configuration and environment filtering rules."};case"admin_paths":return{rationale:`${r} administrative path(s) detected in public sitemap (admin, dashboard, config). These paths may expose privileged access points.`,recommendedAction:"Confirm if admin paths should be publicly indexed. Consider excluding via robots.txt or removing from sitemap. Verify access controls."};case"internal_content":return{rationale:`${r} URL(s) contain "internal" in the path. These may be internal-facing content not intended for public indexing.`,recommendedAction:"Review URLs to determine if they should be publicly accessible. Consider excluding internal content from sitemap or adding noindex meta tags."};case"test_content":return{rationale:`${r} URL(s) contain test/demo/sample identifiers. These may be placeholder or unfinished content not intended for indexing.`,recommendedAction:"Review and remove test content from production sitemaps. Verify content is production-ready before including in sitemap."};case"sensitive_params":return{rationale:`${r} URL(s) contain sensitive query parameters (token, auth, key, password, session). This may expose authentication credentials or debugging flags.`,recommendedAction:"Review why sensitive parameters are in sitemap URLs. Remove authentication tokens from URLs. Consider POST requests for sensitive data."};case"protocol_inconsistency":return{rationale:`${r} URL(s) use HTTP protocol in HTTPS sitemap. This creates mixed content warnings and potential security issues.`,recommendedAction:"Update URLs to use HTTPS consistently. Verify SSL certificate coverage. Check for hardcoded HTTP URLs in content."};case"domain_mismatch":return{rationale:`${r} URL(s) do not match expected base domain. This may indicate external links, CDN URLs, or configuration errors.`,recommendedAction:"Verify if external domains are intentional. Review sitemap generation logic. Confirm CDN or subdomain configuration is correct."};default:return{rationale:`${r} URL(s) flagged in category: ${e}`,recommendedAction:"Review flagged URLs and determine appropriate action."}}}function P(e,t=5){let r=new Map;for(let i of e)r.has(i.category)||r.set(i.category,[]),r.get(i.category).push(i);let s=[];for(let[i,m]of r.entries()){let l=Array.from(new Set(m.map(f=>f.url))),d=m.reduce((f,b)=>{let h=["low","medium","high"];return h.indexOf(b.severity)>h.indexOf(f)?b.severity:f},"low"),p=l.slice(0,t),{rationale:u,recommendedAction:g}=_e(i,d,l.length);s.push({category:i,severity:d,count:l.length,rationale:u,sampleUrls:p,recommendedAction:g,allUrls:l})}s.sort((i,m)=>{let l=["high","medium","low"];return l.indexOf(i.severity)-l.indexOf(m.severity)});let n=new Set(e.map(i=>i.url)).size,a=s.filter(i=>i.severity==="high").reduce((i,m)=>i+m.count,0),c=s.filter(i=>i.severity==="medium").reduce((i,m)=>i+m.count,0),o=s.filter(i=>i.severity==="low").reduce((i,m)=>i+m.count,0);return{groups:s,totalRiskUrls:n,highSeverityCount:a,mediumSeverityCount:c,lowSeverityCount:o}}async function re(e,t,r){let s=Date.now(),n=[],a=K(t),c=[...Q,...Y,...X,...ee,...Z,a],o=[];if(r.acceptedPatterns&&r.acceptedPatterns.length>0)for(let p of r.acceptedPatterns)try{let u=p.replace(/[.+?^${}()|[\]\\]/g,"\\$&").replace(/\*/g,"[^/]*");!u.endsWith("$")&&!u.includes("(?:")&&(u=u+"(?:/|$|\\?|#)"),o.push(new RegExp(u,"i"))}catch{r.verbose&&console.warn(`Invalid accepted pattern: ${p}`)}if(r.verbose){console.log(`
+Analyzing ${e.length} URLs for risk patterns...`);try{console.log(`Base domain: ${new URL(t).hostname}`)}catch{console.log(`Base URL: ${t}`)}o.length>0&&console.log(`Accepted patterns: ${o.length}`)}let i;try{i=new URL(t).protocol}catch{r.verbose&&console.warn(`Invalid base URL: ${t}, defaulting to https:`),i="https:"}let m=0;for(let p of e){let u=p.loc;m++,(m%1e4===0||m===e.length)&&process.stdout.write(`\r\x1B[K  Analyzing: ${m.toLocaleString()}/${e.length.toLocaleString()} URLs...`);let g=!1;for(let f of o)if(f.test(u)){g=!0;break}if(!g)for(let f of c)if(f.category==="protocol_inconsistency")try{let b=new URL(u).protocol;i==="https:"&&b==="http:"&&n.push({url:u,category:f.category,severity:f.severity,pattern:f.name,rationale:f.description,matchedValue:"http://"})}catch{r.verbose&&console.warn(`Skipping invalid URL: ${u}`);continue}else try{let b=u.match(f.regex);b&&n.push({url:f.category==="sensitive_params"?te(u):u,category:f.category,severity:f.severity,pattern:f.name,rationale:f.description,matchedValue:b[0]})}catch(b){r.verbose&&console.error(`Pattern matching failed for ${f.name}: ${b instanceof Error?b.message:String(b)}`);continue}}e.length>=1e4&&process.stdout.write("\r\x1B[K");let l=P(n),d=Date.now()-s;if(r.verbose&&(console.log(`
+Risk Summary:`),console.log(`  - Total URLs analyzed: ${e.length}`),console.log(`  - Risk URLs found: ${l.totalRiskUrls}`),console.log(`  - HIGH severity: ${l.highSeverityCount}`),console.log(`  - MEDIUM severity: ${l.mediumSeverityCount}`),console.log(`  - LOW severity: ${l.lowSeverityCount}`),console.log(`  - Processing time: ${d}ms`),l.groups.length>0)){console.log(`
+Risk Categories Found:`);for(let p of l.groups)console.log(`  - ${p.category}: ${p.count} URLs (${p.severity.toUpperCase()})`)}return{findings:n,groups:l.groups,totalUrlsAnalyzed:e.length,riskUrlCount:l.totalRiskUrls,cleanUrlCount:e.length-l.totalRiskUrls,highSeverityCount:l.highSeverityCount,mediumSeverityCount:l.mediumSeverityCount,lowSeverityCount:l.lowSeverityCount,processingTimeMs:d}}function oe(e){let t={high:0,medium:0,low:0},r=e.riskGroups.map(c=>{t[c.severity]+=c.count;let o=c.allUrls||c.sampleUrls;return{category:c.category,count:c.count,severity:c.severity,summary:c.rationale,examples:o.slice(0,3),allUrls:o}}),s=e.riskGroups.reduce((c,o)=>c+o.count,0),n=s>0?`Found ${s} potentially risky URLs across ${e.riskGroups.length} categories in ${e.totalUrls} total URLs.`:`Analyzed ${e.totalUrls} URLs. No suspicious patterns detected.`,a=[];return t.high>0&&a.push(`${t.high} high-severity issues require immediate attention`),t.medium>0&&a.push(`${t.medium} medium-severity issues should be reviewed`),t.low>0&&a.push(`${t.low} low-severity items flagged for awareness`),{overview:n,keyFindings:a,categoryInsights:r,severityBreakdown:t,recommendations:[],generatedBy:"rule-based analysis",metadata:{tokensUsed:0,processingTime:e.processingTime||0,model:"pattern-matching"}}}var Ae="1.0.0-alpha.0";function se(e,t,r,s,n,a,c={}){let{pretty:o=!0,indent:i=2}=c,m=Le(e,t,r,s,n,a),l=Ne(m);return o?JSON.stringify(l,null,i):JSON.stringify(l)}function Le(e,t,r,s,n,a){let c=Ie(n.baseUrl||"unknown",a,e),o=s.map(d=>({category:d.category,severity:d.severity,count:d.count,pattern:d.category,rationale:d.rationale,sampleUrls:d.sampleUrls.slice(0,5),recommendedAction:d.recommendedAction})),i={highSeverityCount:e.severityBreakdown.high,mediumSeverityCount:e.severityBreakdown.medium,lowSeverityCount:e.severityBreakdown.low,totalRiskyUrls:s.reduce((d,p)=>d+p.count,0),overallStatus:De(e.severityBreakdown,r.errors)},m={overview:e.overview,keyFindings:e.keyFindings,recommendations:e.recommendations},l=r.errors.map(Ge);return{analysisMetadata:c,sitemapsDiscovered:t.sitemaps,totalUrlCount:r.totalCount,urlsAnalyzed:r.totalCount,suspiciousGroups:o,riskSummary:m,summary:i,errors:l}}function Ie(e,t,r){return{baseUrl:e,analysisTimestamp:new Date().toISOString(),toolVersion:Ae,executionTimeMs:Date.now()-t,analysisType:r.generatedBy}}function De(e,t){return t.length>0?"errors":e.high+e.medium+e.low>0?"issues_found":"clean"}function Ne(e){return{analysis_metadata:Fe(e.analysisMetadata),sitemaps_discovered:e.sitemapsDiscovered,total_url_count:e.totalUrlCount,urls_analyzed:e.urlsAnalyzed,suspicious_groups:e.suspiciousGroups.map(Me),risk_summary:ze(e.riskSummary),summary:Oe(e.summary),errors:e.errors}}function Fe(e){return{base_url:e.baseUrl,analysis_timestamp:e.analysisTimestamp,tool_version:e.toolVersion,execution_time_ms:e.executionTimeMs,analysis_type:e.analysisType}}function Me(e){return{category:e.category,severity:e.severity,count:e.count,pattern:e.pattern,rationale:e.rationale,sample_urls:e.sampleUrls,recommended_action:e.recommendedAction}}function ze(e){return{overview:e.overview,key_findings:e.keyFindings,recommendations:e.recommendations}}function Oe(e){return{high_severity_count:e.highSeverityCount,medium_severity_count:e.mediumSeverityCount,low_severity_count:e.lowSeverityCount,total_risky_urls:e.totalRiskyUrls,overall_status:e.overallStatus}}function Ge(e){if("code"in e){let t=e,r={code:t.code||"UNKNOWN_ERROR",message:e.message};return"attemptedPaths"in t?r.context={attempted_paths:t.attemptedPaths}:"sitemapUrl"in t&&"lineNumber"in t?r.context={sitemap_url:t.sitemapUrl,line_number:t.lineNumber}:"url"in t&&(r.context={url:t.url}),r}return{code:"UNKNOWN_ERROR",message:e.message}}var ie=require("fs"),qe="1.0.0-alpha.0";function je(e,t,r,s,n,a={}){let c=a.maxUrlsPerGroup??10,o=new Date().toISOString(),i=e.categoryInsights.reduce((u,g)=>u+g.count,0),m=e.categoryInsights.filter(u=>u.severity==="high"),l=e.categoryInsights.filter(u=>u.severity==="medium"),d=e.categoryInsights.filter(u=>u.severity==="low");return`<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1.0">
+  <title>Sitemap QA Report - ${s.baseUrl}</title>
+  <style>
+    * { margin: 0; padding: 0; box-sizing: border-box; }
+    body {
+      font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, 'Helvetica Neue', Arial, sans-serif;
+      line-height: 1.6;
+      color: #1f2937;
+      background: #ffffff;
+      padding: 24px;
+    }
+    .container {
+      max-width: 1400px;
+      margin: 0 auto;
+      background: white;
+      box-shadow: 0 1px 3px rgba(0,0,0,0.05);
+      border-radius: 12px;
+      overflow: hidden;
+      border: 1px solid #e5e7eb;
+    }
+    .header {
+      background: #0f172a;
+      color: white;
+      padding: 48px 40px;
+      border-bottom: 3px solid #3b82f6;
+    }
+    .header h1 { 
+      font-size: 1.875rem; 
+      font-weight: 700;
+      margin-bottom: 12px;
+      letter-spacing: -0.025em;
+    }
+    .header .meta { 
+      opacity: 0.75; 
+      font-size: 0.875rem;
+      font-weight: 400;
+    }
+    .summary {
+      display: grid;
+      grid-template-columns: repeat(auto-fit, minmax(240px, 1fr));
+      gap: 1px;
+      background: #e5e7eb;
+      border-bottom: 1px solid #e5e7eb;
+    }
+    .summary-card {
+      background: white;
+      padding: 28px 32px;
+      text-align: center;
+    }
+    .summary-card .label { 
+      font-size: 0.75rem; 
+      color: #6b7280;
+      text-transform: uppercase; 
+      letter-spacing: 0.05em;
+      font-weight: 600;
+      margin-bottom: 8px;
+    }
+    .summary-card .value { 
+      font-size: 2.25rem; 
+      font-weight: 700;
+      color: #0f172a;
+      font-variant-numeric: tabular-nums;
+    }
+    .content { padding: 40px; }
+    .status-clean {
+      text-align: center;
+      padding: 80px 32px;
+      background: #f0fdf4;
+      border-radius: 8px;
+      border: 1px solid #86efac;
+    }
+    .status-clean h2 { 
+      font-size: 1.875rem;
+      margin-bottom: 12px;
+      color: #166534;
+      font-weight: 700;
+    }
+    .status-clean p { 
+      font-size: 1rem;
+      color: #65a30d;
+    }
+    .severity-section { margin-bottom: 32px; }
+    .severity-section h2 {
+      font-size: 1.125rem;
+      font-weight: 600;
+      padding: 16px 20px;
+      margin-bottom: 16px;
+      border-radius: 8px;
+      display: flex;
+      align-items: center;
+      gap: 12px;
+      cursor: pointer;
+      user-select: none;
+      transition: all 0.2s;
+    }
+    .severity-section h2:hover {
+      opacity: 0.85;
+      transform: translateY(-1px);
+    }
+    .severity-section h2::after {
+      content: '\u25BC';
+      margin-left: auto;
+      font-size: 0.8em;
+      transition: transform 0.3s ease;
+      opacity: 0.7;
+    }
+    .severity-section h2.collapsed::after {
+      transform: rotate(-90deg);
+    }
+    .severity-section h2.collapsed {
+      margin-bottom: 0;
+    }
+    .severity-content {
+      max-height: none;
+      overflow: visible;
+      transition: max-height 0.4s ease-out, opacity 0.3s ease-out;
+      opacity: 1;
+    }
+    .severity-content.collapsed {
+      max-height: 0;
+      overflow: hidden;
+      opacity: 0;
+    }
+    .severity-high { background: #fef2f2; color: #dc2626; border: 1px solid #fecaca; }
+    .severity-medium { background: #fffbeb; color: #d97706; border: 1px solid #fde68a; }
+    .severity-low { background: #eff6ff; color: #2563eb; border: 1px solid #dbeafe; }
+    .risk-group {
+      background: white;
+      border: 1px solid #e5e7eb;
+      border-radius: 8px;
+      padding: 24px;
+      margin-bottom: 16px;
+    }
+    .risk-group h3 {
+      font-size: 1rem;
+      margin-bottom: 12px;
+      color: #0f172a;
+      font-weight: 600;
+    }
+    .risk-group .count {
+      display: inline-block;
+      background: #3b82f6;
+      color: white;
+      padding: 2px 10px;
+      border-radius: 9999px;
+      font-size: 0.75rem;
+      font-weight: 600;
+      margin-left: 8px;
+    }
+    .risk-group .impact {
+      color: #64748b;
+      margin-bottom: 16px;
+      font-size: 0.875rem;
+      line-height: 1.6;
+    }
+    .risk-group .urls {
+      background: #f8fafc;
+      border: 1px solid #e2e8f0;
+      border-radius: 6px;
+      padding: 16px;
+    }
+    .risk-group .urls h4 {
+      font-size: 0.75rem;
+      color: #64748b;
+      margin-bottom: 12px;
+      text-transform: uppercase;
+      letter-spacing: 0.05em;
+      font-weight: 600;
+    }
+    .risk-group .urls ul { list-style: none; }
+    .risk-group .urls li {
+      padding: 10px 12px;
+      border-bottom: 1px solid #e2e8f0;
+      font-family: 'SF Mono', 'Monaco', 'Cascadia Code', 'Consolas', monospace;
+      font-size: 0.8125rem;
+      color: #334155;
+      background: white;
+      margin-bottom: 4px;
+      border-radius: 4px;
+      word-break: break-all;
+      line-height: 1.6;
+    }
+    .risk-group .urls li:last-child { border-bottom: none; margin-bottom: 0; }
+    .risk-group .more {
+      color: #3b82f6;
+      font-style: italic;
+      margin-top: 8px;
+      font-size: 0.8125rem;
+    }
+    .download-btn {
+      display: inline-block;
+      background: #3b82f6;
+      color: white;
+      padding: 8px 16px;
+      border-radius: 6px;
+      text-decoration: none;
+      font-size: 0.8125rem;
+      font-weight: 500;
+      margin-top: 12px;
+      cursor: pointer;
+      border: none;
+      transition: all 0.15s;
+    }
+    .download-btn:hover {
+      background: #2563eb;
+      transform: translateY(-1px);
+      box-shadow: 0 4px 6px -1px rgba(0,0,0,0.1);
+    }
+    .footer {
+      background: #f8fafc;
+      padding: 24px 40px;
+      border-top: 1px solid #e5e7eb;
+      text-align: center;
+      color: #64748b;
+      font-size: 0.8125rem;
+    }
+    .sitemaps {
+      background: white;
+      border: 1px solid #e5e7eb;
+      border-radius: 8px;
+      margin-bottom: 24px;
+      overflow: hidden;
+    }
+    .sitemaps h3 {
+      font-size: 1.125rem;
+      font-weight: 600;
+      padding: 16px 20px;
+      margin: 0;
+      color: #0f172a;
+      background: #f8fafc;
+      cursor: pointer;
+      user-select: none;
+      transition: all 0.15s;
+      display: flex;
+      align-items: center;
+      gap: 10px;
+    }
+    .sitemaps h3:hover {
+      background: #f1f5f9;
+    }
+    .sitemaps h3::after {
+      content: '\u25BC';
+      margin-left: auto;
+      font-size: 0.8em;
+      transition: transform 0.3s ease;
+      opacity: 0.7;
+    }
+    .sitemaps h3.collapsed::after {
+      transform: rotate(-90deg);
+    }
+    .sitemaps-content {
+      max-height: none;
+      overflow: visible;
+      transition: max-height 0.4s ease-out, opacity 0.3s ease-out;
+      opacity: 1;
+      padding: 20px;
+    }
+    .sitemaps-content.collapsed {
+      max-height: 0;
+      overflow: hidden;
+      opacity: 0;
+      padding: 0 20px;
+    }
+    .sitemaps ul { list-style: none; }
+    .sitemaps li {
+      padding: 10px 12px;
+      font-family: 'SF Mono', 'Monaco', 'Cascadia Code', 'Consolas', monospace;
+      font-size: 0.8125rem;
+      color: #475569;
+      word-break: break-all;
+      line-height: 1.6;
+      background: #f8fafc;
+      margin-bottom: 4px;
+      border-radius: 4px;
+    }
+    .sitemaps li:last-child { margin-bottom: 0; }
+    .errors-section {
+      background: #fffbeb;
+      border-left: 4px solid #f59e0b;
+      padding: 20px;
+      margin-bottom: 24px;
+      border-radius: 8px;
+      border: 1px solid #fde68a;
+    }
+    .errors-section h3 {
+      color: #92400e;
+      margin-bottom: 16px;
+      font-size: 1.125rem;
+      font-weight: 600;
+      display: flex;
+      align-items: center;
+      gap: 8px;
+    }
+    .errors-section ul {
+      list-style: none;
+      padding: 0;
+    }
+    .errors-section li {
+      padding: 12px;
+      background: white;
+      margin-bottom: 8px;
+      border-radius: 6px;
+      font-family: 'SF Mono', 'Monaco', 'Cascadia Code', 'Consolas', monospace;
+      font-size: 0.8125rem;
+      color: #78350f;
+      word-break: break-all;
+      line-height: 1.6;
+      border: 1px solid #fde68a;
+    }
+    .errors-section li:last-child {
+      margin-bottom: 0;
+    }
+  </style>
+</head>
+<body>
+  <div class="container">
+    <div class="header">
+      <h1>Sitemap Analysis</h1>
+      <div class="meta">
+        <div>${s.baseUrl}</div>
+        <div>${new Date(o).toLocaleString()}</div>
+      </div>
+    </div>
+
+    <div class="summary">
+      <div class="summary-card">
+        <div class="label">Sitemaps</div>
+        <div class="value">${t.sitemaps.length}</div>
+      </div>
+      <div class="summary-card">
+        <div class="label">URLs Analyzed</div>
+        <div class="value">${r.toLocaleString()}</div>
+      </div>
+      <div class="summary-card">
+        <div class="label">Issues Found</div>
+        <div class="value" style="color: ${i>0?"#dc2626":"#059669"}">${i}</div>
+      </div>
+      <div class="summary-card">
+        <div class="label">Scan Time</div>
+        <div class="value">${(e.metadata.processingTime/1e3).toFixed(1)}s</div>
+      </div>
+    </div>
+
+    <div class="content">
+      ${n.length>0?`
+      <div class="errors-section">
+        <h3>Parsing Errors & Warnings (${n.length})</h3>
+        <ul>
+          ${n.map(u=>`<li>${u.message}</li>`).join(`
+          `)}
+        </ul>
+      </div>
+      `:""}
+
+      ${t.sitemaps.length>0?`
+      <div class="sitemaps">
+        <h3 class="collapsed" onclick="toggleSection(this)">Sitemaps Discovered (${t.sitemaps.length})</h3>
+        <div class="sitemaps-content collapsed">
+          <ul>
+            ${t.sitemaps.map(u=>`<li>\u2022 ${u}</li>`).join(`
+            `)}
+          </ul>
+        </div>
+      </div>
+      `:""}
+
+      ${i===0?`
+      <div class="status-clean">
+        <h2>No Issues Found</h2>
+        <p>All URLs in the sitemap passed validation checks.</p>
+      </div>
+      `:""}
+
+      ${m.length>0?`
+      <div class="severity-section">
+        <h2 class="severity-high" onclick="toggleSection(this)">High Severity (${m.reduce((u,g)=>u+g.count,0)} URLs)</h2>
+        <div class="severity-content">
+          ${m.map(u=>I(u,c)).join(`
+          `)}
+        </div>
+      </div>
+      `:""}
+
+      ${l.length>0?`
+      <div class="severity-section">
+        <h2 class="severity-medium" onclick="toggleSection(this)">Medium Severity (${l.reduce((u,g)=>u+g.count,0)} URLs)</h2>
+        <div class="severity-content">
+          ${l.map(u=>I(u,c)).join(`
+          `)}
+        </div>
+      </div>
+      `:""}
+
+      ${d.length>0?`
+      <div class="severity-section">
+        <h2 class="severity-low" onclick="toggleSection(this)">Low Severity (${d.reduce((u,g)=>u+g.count,0)} URLs)</h2>
+        <div class="severity-content">
+          ${d.map(u=>I(u,c)).join(`
+          `)}
+        </div>
+      </div>
+      `:""}
+    </div>
+
+    <div class="footer">
+      Generated by <strong>sitemap-qa</strong> v${qe}
+    </div>
+  </div>
+  
+  <script>
+    function toggleSection(header) {
+      header.classList.toggle('collapsed');
+      const content = header.nextElementSibling;
+      content.classList.toggle('collapsed');
+    }
+    
+    function downloadUrls(categorySlug, encodedUrls) {
+      // Decode HTML entities and parse JSON
+      const textarea = document.createElement('textarea');
+      textarea.innerHTML = encodedUrls;
+      const urls = JSON.parse(textarea.value);
+      
+      // Create text content (one URL per line)
+      const textContent = urls.join('\\n');
+      
+      // Create blob and download
+      const blob = new Blob([textContent], { type: 'text/plain' });
+      const url = URL.createObjectURL(blob);
+      const a = document.createElement('a');
+      a.href = url;
+      a.download = categorySlug + '_urls.txt';
+      document.body.appendChild(a);
+      a.click();
+      document.body.removeChild(a);
+      URL.revokeObjectURL(url);
+    }
+  </script>
+</body>
+</html>`}function I(e,t){let r=e.category.split("_").map(i=>i.charAt(0).toUpperCase()+i.slice(1)).join(" "),s=e.examples.slice(0,t),n=e.count-s.length,a=e.category.toLowerCase(),c=JSON.stringify(e.allUrls),o=ne(c);return`<div class="risk-group">
+        <h3>${r} <span class="count">${e.count} URLs</span></h3>
+        <div class="impact">${e.summary}</div>
+        <div class="urls">
+          <h4>Sample URLs</h4>
+          <ul>
+            ${s.map(i=>`<li>${ne(i)}</li>`).join(`
+            `)}
+          </ul>
+          ${n>0?`<div class="more">... and ${n} more</div>`:""}
+          <button class="download-btn" onclick="downloadUrls('${a}', '${o}')">\u{1F4E5} Download All ${e.count} URLs</button>
+        </div>
+      </div>`}function ne(e){return e.replace(/&/g,"&amp;").replace(/</g,"&lt;").replace(/>/g,"&gt;").replace(/"/g,"&quot;").replace(/'/g,"&#039;")}async function ae(e,t,r,s,n,a,c={}){let o=je(e,t,r,s,a,c);await ie.promises.writeFile(n,o,"utf-8")}var le=new ce.Command("analyze").description("Analyze sitemap for QA issues").argument("<url>","Base URL to analyze").option("--timeout <seconds>","HTTP timeout in seconds","30").option("--no-progress","Disable progress bar").option("--output <format>","Output format: html or json","html").option("--output-dir <path>","Output directory for reports").option("--output-file <path>","Custom output filename").option("--accepted-patterns <patterns>","Comma-separated regex patterns to exclude from risk detection").option("--no-color","Disable ANSI color codes in CLI output").option("--verbose","Enable verbose logging",!1).action(async(e,t)=>{let r;try{He(t),r=await G({...t,baseUrl:e,outputFormat:t.output}),console.log(`
+\u{1F50D} Analyzing ${e}...
+`);let n=await Be(e,r);Ve(n),await D.promises.mkdir(r.outputDir,{recursive:!0});let a=t.outputFile||`sitemap-qa-report-${Date.now()}.html`,c=`${r.outputDir}/${a}`;if(await ae(n.summary,n.discoveryResult,n.totalUrls,r,c,n.errors,{maxUrlsPerGroup:10}),console.log(`
+\u{1F4C4} Full report saved to: ${x.default.cyan(c)}`),t.output==="json"){let i=a.replace(/\.html$/,".json"),m=`${r.outputDir}/${i}`,l=se(n.summary,n.discoveryResult,{totalCount:n.totalUrls,uniqueUrls:[],errors:[]},n.riskGroups,r,n.executionTime,{pretty:!0,indent:2});await D.promises.writeFile(m,l,"utf-8"),console.log(`\u{1F4C4} JSON report saved to: ${x.default.cyan(m)}`)}let o=We(n);process.exit(o)}catch(s){Je(s,r),process.exit(2)}});function He(e){let t=["json","html"];if(!t.includes(e.output))throw new Error(`Invalid output format: ${e.output}. Must be one of: ${t.join(", ")}`);let r=parseInt(e.timeout);if(isNaN(r)||r<=0)throw new Error(`Invalid timeout: ${e.timeout}. Must be a positive number.`)}function Ve(e){console.log("");let t=e.summary.categoryInsights.reduce((r,s)=>r+s.count,0);if(t===0)console.log(x.default.green("\u2705 No issues found - sitemap looks clean!"));else{console.log(x.default.yellow(`\u26A0\uFE0F  Found ${t} potentially risky URL(s)`)),console.log("");let{high:r,medium:s,low:n}=e.summary.severityBreakdown;r>0&&console.log(x.default.red(`   \u{1F6A8} High severity:   ${r} URLs`)),s>0&&console.log(x.default.yellow(`   \u26A0\uFE0F  Medium severity: ${s} URLs`)),n>0&&console.log(x.default.blue(`   \u2139\uFE0F  Low severity:    ${n} URLs`))}console.log("")}async function Be(e,t){let r=Date.now(),s=[],n=(0,k.default)("Discovering sitemaps...").start(),a=await V(e,t);if(n.succeed(`Found ${a.sitemaps.length} sitemap(s)`),a.accessIssues.length>0){console.warn(`\u26A0\uFE0F  Warning: ${a.accessIssues.length} sitemap(s) are access-blocked`);for(let y of a.accessIssues)s.push(new Error(`Access blocked: ${y.url} (${y.statusCode})`))}if(a.sitemaps.length===0)throw new Error(`No sitemaps found at ${e}. Tried: /sitemap.xml, /sitemap_index.xml, /robots.txt`);let c=(0,k.default)("Parsing sitemaps...").start(),o=await W(a.sitemaps,t);if(c.succeed(`Extracted ${o.allUrls.length.toLocaleString()} URLs`),o.errors.length>0)for(let y of o.errors)typeof y=="string"?s.push(new Error(y)):s.push(y);if(o.allUrls.length===0)throw new Error("No URLs extracted from sitemaps");let i=(0,k.default)("Removing duplicates...").start(),m=J(o.allUrls),l=o.allUrls.length-m.uniqueUrls.length;l>0?i.succeed(`${m.uniqueUrls.length.toLocaleString()} unique URLs (removed ${l.toLocaleString()} duplicates)`):i.succeed(`${m.uniqueUrls.length.toLocaleString()} unique URLs`);let d=(0,k.default)("Analyzing for risks...").start(),p=await re(m.uniqueUrls,e,t),u=P(p.findings),g=u.groups.reduce((y,R)=>y+R.count,0);g>0?d.warn(`Found ${g} risky URL(s)`):d.succeed("No risks detected");let f=Date.now()-r,b=(0,k.default)("Generating report...").start(),h=oe({riskGroups:u.groups,totalUrls:m.uniqueUrls.length,sitemapUrl:e,processingTime:f});return b.succeed("Analysis complete"),{discoveryResult:a,totalUrls:m.uniqueUrls.length,riskGroups:u.groups,summary:h,errors:s,executionTime:f}}function We(e){return e.summary.severityBreakdown.high>0?1:0}function Je(e,t){console.error(`
+\u274C Analysis failed
+`),e instanceof Error?(console.error(`Error: ${e.message}`),t?.verbose&&e.stack&&(console.error(`
+Stack trace:`),console.error(e.stack)),e.message.includes("No sitemaps found")?(console.error(`
+Suggestions:`),console.error("  \u2022 Verify the base URL is correct"),console.error("  \u2022 Check if the site has a sitemap"),console.error("  \u2022 Ensure the sitemap is publicly accessible")):(e.message.includes("Network")||e.message.includes("timeout"))&&(console.error(`
+Suggestions:`),console.error("  \u2022 Check your internet connection"),console.error("  \u2022 Verify the URL is accessible"),console.error("  \u2022 Try increasing the timeout with --timeout option"))):(console.error("Unknown error occurred"),console.error(String(e)))}var N=new me.Command;N.name("sitemap-qa").version("1.0.0").description("sitemap analysis for QA teams");N.addCommand(le);process.on("unhandledRejection",(e,t)=>{console.error("Unhandled Rejection at:",t,"reason:",e),process.exit(1)});process.on("SIGINT",()=>{console.log(`
+Gracefully shutting down...`),process.exit(0)});process.on("SIGTERM",()=>{console.log(`
+Gracefully shutting down...`),process.exit(0)});N.parse();
+//# sourceMappingURL=index.cjs.map