@xapp/arachne-web-service 1.18.6 → 1.20.0

package/lib/utils/getPrompt.js

@@ -2,6 +2,7 @@
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.getTokens = getTokens;
 exports.getWebMarkdownPrompt = getWebMarkdownPrompt;
+exports.getAboutUsPrompt = getAboutUsPrompt;
 exports.getJSONSchemaPrompt = getJSONSchemaPrompt;
 /*! Copyright (c) 2023, XAPP AI */
 // TODO: REPLACE, has problem with webpack
@@ -63,6 +64,32 @@ Please print out in JSON format on one line information about the website above
     const tokens = getTokens(prompt);
     return Object.assign({ prompt: prompt.trim() }, tokens);
 }
+/**
+ * A narrow prompt for an about-us page: founding year, ownership, service areas, team size and licences, which the
+ * homepage often does not state.
+ *
+ * @param markdown The about-us page as markdown
+ * @param url URL of the about-us page
+ */
+function getAboutUsPrompt(markdown, url) {
+    const prompt = `
+Your task is to analyze the "About Us" page of a business website (provided in markdown below) found at ${url}.  Only report what the page states.  If the page does not state something, leave that key out rather than guessing.
+
+===========
+${markdown}
+===========
+
+Please print out in JSON format on one line the following keys:
+
+"founded_year" - The year the business was founded or started, as a four digit number.  Only include it if the year itself is written on the page.
+"ownership" - An array describing how the business is owned, only using these values: "family-owned", "veteran-owned", "woman-owned", "minority-owned", "locally-owned", "employee-owned", "franchise".
+"service_areas" - An array of the cities, counties or regions the business says it serves, written exactly as they appear on the page.
+"team_size" - The number of employees or team members, as a number.  Only include it if the number itself is written on the page.
+"licenses" - An array of state-issued contractor or trade license numbers found on the page.  Do not include general business registrations, ISO certifications, or membership numbers.  Use an object with keys: number (just the license number itself), type (the type of license such as HVAC, Plumbing, Electrical, General Contractor, Roofing, etc), state (two-letter state code if identifiable from context), fullText (the complete license text exactly as shown on the page).
+`;
+    const tokens = getTokens(prompt);
+    return Object.assign({ prompt: prompt.trim() }, tokens);
+}
 function getJSONSchemaPrompt(json, url) {
     const jsonString = JSON.stringify(json);
     const prompt = `

package/lib/utils/getPrompt.js.map

@@ -1 +1 @@
-{"version":3,"file":"getPrompt.js","sourceRoot":"","sources":["../../src/utils/getPrompt.ts"],"names":[],"mappings":";;AA2BA,8BAQC;AAGD,oDAiDC;AAED,kDAoBC;AA7GD,kCAAkC;AAClC,0CAA0C;AAC1C,MAAM,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,eAAe,CAAC,CAAC;AAE5C,MAAM,eAAe,GAAG,GAAG,CAAC;AAiB5B;;;;;GAKG;AACH,SAAgB,SAAS,CAAC,MAAc,EAAE,aAAqB,KAAK;IAEhE,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC;IAErC,4CAA4C;IAC5C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,UAAU,GAAG,MAAM,GAAG,eAAe,CAAC,CAAC;IAErE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,CAAC;AACjC,CAAC;AAGD,SAAgB,oBAAoB,CAAC,QAAgB,EAAE,GAAW;IAE9D,MAAM,MAAM,GAAG;0FACuE,GAAG;;;EAG3F,QAAQ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAmCT,CAAC;IAEE,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;IAEjC,uBACI,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE,IAClB,MAAM,EACZ;AACL,CAAC;AAED,SAAgB,mBAAmB,CAAC,IAAS,EAAE,GAAW;IAEtD,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;IAExC,MAAM,MAAM,GAAG;EACjB,UAAU;;kDAEsC,GAAG;;;CAGpD,CAAC;IAEE,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC;IAE9B,MAAM,MAAM,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;IAElC,uBACI,MAAM,EAAE,OAAO,IACZ,MAAM,EACZ;AACL,CAAC"}
+{"version":3,"file":"getPrompt.js","sourceRoot":"","sources":["../../src/utils/getPrompt.ts"],"names":[],"mappings":";;AA2BA,8BAQC;AAGD,oDAiDC;AASD,4CAwBC;AAED,kDAoBC;AA9ID,kCAAkC;AAClC,0CAA0C;AAC1C,MAAM,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,eAAe,CAAC,CAAC;AAE5C,MAAM,eAAe,GAAG,GAAG,CAAC;AAiB5B;;;;;GAKG;AACH,SAAgB,SAAS,CAAC,MAAc,EAAE,aAAqB,KAAK;IAEhE,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC;IAErC,4CAA4C;IAC5C,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,UAAU,GAAG,MAAM,GAAG,eAAe,CAAC,CAAC;IAErE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,CAAC;AACjC,CAAC;AAGD,SAAgB,oBAAoB,CAAC,QAAgB,EAAE,GAAW;IAE9D,MAAM,MAAM,GAAG;0FACuE,GAAG;;;EAG3F,QAAQ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CAmCT,CAAC;IAEE,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;IAEjC,uBACI,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE,IAClB,MAAM,EACZ;AACL,CAAC;AAED;;;;;;GAMG;AACH,SAAgB,gBAAgB,CAAC,QAAgB,EAAE,GAAW;IAE1D,MAAM,MAAM,GAAG;0GACuF,GAAG;;;EAG3G,QAAQ;;;;;;;;;;CAUT,CAAC;IAEE,MAAM,MAAM,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;IAEjC,uBACI,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE,IAClB,MAAM,EACZ;AACL,CAAC;AAED,SAAgB,mBAAmB,CAAC,IAAS,EAAE,GAAW;IAEtD,MAAM,UAAU,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;IAExC,MAAM,MAAM,GAAG;EACjB,UAAU;;kDAEsC,GAAG;;;CAGpD,CAAC;IAEE,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC;IAE9B,MAAM,MAAM,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;IAElC,uBACI,MAAM,EAAE,OAAO,IACZ,MAAM,EACZ;AACL,CAAC"}

package/lib/utils/getSiteMap.d.ts

@@ -42,11 +42,44 @@ export interface SiteMapData {
      * A terms of service page, typically found at /terms, /terms-of-service, /tos, /legal/terms, /terms-and-conditions
      */
     termsOfService?: string;
+    /**
+     * A page where customers book or request an appointment, typically found at /book-online, /schedule, /appointments,
+     * /make-appointment, /request-service, /estimate
+     */
+    bookingPage?: string;
+    /**
+     * Service slugs read from the sitemap URLs, e.g. "water-heater-repair" from /services/water-heater-repair
+     * or "drain-cleaning" from /drain-cleaning.  No page is loaded to find them.
+     *
+     * Only present when requested, see `dataOptions.serviceSlugsFromSitemap`.
+     */
+    serviceSlugs?: string[];
+}
+export interface SiteMapDataOptions {
+    /**
+     * Add `serviceSlugs` read from the sitemap URLs.
+     */
+    serviceSlugs?: boolean;
 }
+/** Most service slugs returned, so a sitemap full of location pages cannot bloat the result. */
+export declare const MAX_SERVICE_SLUGS = 100;
+/**
+ * Reads the services a business names in its URLs, without loading any page.
+ *
+ * Two shapes are recognised:
+ * - every segment under a services section: /services/water-heater-repair, /our-services/plumbing/leak-detection
+ * - a single top-level segment containing a service word: /drain-cleaning, /ac-repair.html, /hvac-services
+ *
+ * A bare top-level slug such as /plumbing is not taken, since it cannot be told apart from /gallery.
+ *
+ * @param urls URLs, typically every URL in the sitemap
+ * @returns Unique lowercase slugs in first-seen order, at most MAX_SERVICE_SLUGS
+ */
+export declare function extractServiceSlugs(urls: string[]): string[];
 /**
  * Returns descriptive data about a sitemap.
  *
  * @param url
  * @returns
  */
-export declare function getSiteMapData(url: string): Promise<SiteMapData | undefined>;
+export declare function getSiteMapData(url: string, options?: SiteMapDataOptions): Promise<SiteMapData | undefined>;

package/lib/utils/getSiteMap.js

@@ -42,7 +42,9 @@ var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, ge
     });
 };
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.MAX_SERVICE_SLUGS = void 0;
 exports.getSitemap = getSitemap;
+exports.extractServiceSlugs = extractServiceSlugs;
 exports.getSiteMapData = getSiteMapData;
 /*! Copyright (c) 2023, XAPP AI */
 const cheerio = __importStar(require("cheerio"));
@@ -139,14 +141,123 @@ function crawlSitemap(sitemapUrl, visited) {
         return allUrls;
     });
 }
+/** Most service slugs returned, so a sitemap full of location pages cannot bloat the result. */
+exports.MAX_SERVICE_SLUGS = 100;
+/** A path segment under which every following segment names a service. */
+const SERVICE_PARENT_SEGMENT = /^((our|residential|commercial)-)?services?$|^what-we-do$/;
+/** Words that mark a top-level slug as naming a service, e.g. /drain-cleaning, /ac-repair, /hvac-services. */
+const SERVICE_WORDS = new Set([
+    "repair",
+    "repairs",
+    "installation",
+    "installations",
+    "install",
+    "replacement",
+    "replacements",
+    "cleaning",
+    "maintenance",
+    "inspection",
+    "inspections",
+    "removal",
+    "restoration",
+    "remodeling",
+    "renovation",
+    "treatment",
+    "service",
+    "services",
+]);
+/** Words that mark a slug as something other than a service, e.g. /customer-service, /service-areas. */
+const NON_SERVICE_WORDS = new Set(["area", "areas", "customer", "terms", "privacy", "policy", "careers", "jobs"]);
+/** Slugs that never name a service on their own. */
+const NON_SERVICE_SLUGS = new Set(["page", "all", "residential", "commercial", "category", "tag"]);
+/** A path containing any of these segments is content or taxonomy, not a service page. */
+const NON_SERVICE_SECTIONS = new Set([
+    "blog",
+    "news",
+    "article",
+    "articles",
+    "post",
+    "posts",
+    "category",
+    "categories",
+    "tag",
+    "tags",
+    "author",
+    "wp-content",
+    "feed",
+    "careers",
+    "jobs",
+    "events",
+    "press",
+]);
+const SLUG = /^[a-z0-9]+(-[a-z0-9]+)*$/;
+function isServiceSlug(slug) {
+    return SLUG.test(slug) && /[a-z]/.test(slug) && slug.length <= 80 && !NON_SERVICE_SLUGS.has(slug);
+}
+/**
+ * Reads the services a business names in its URLs, without loading any page.
+ *
+ * Two shapes are recognised:
+ * - every segment under a services section: /services/water-heater-repair, /our-services/plumbing/leak-detection
+ * - a single top-level segment containing a service word: /drain-cleaning, /ac-repair.html, /hvac-services
+ *
+ * A bare top-level slug such as /plumbing is not taken, since it cannot be told apart from /gallery.
+ *
+ * @param urls URLs, typically every URL in the sitemap
+ * @returns Unique lowercase slugs in first-seen order, at most MAX_SERVICE_SLUGS
+ */
+function extractServiceSlugs(urls) {
+    const slugs = new Set();
+    for (const value of urls) {
+        if (slugs.size >= exports.MAX_SERVICE_SLUGS)
+            break;
+        let pathname;
+        try {
+            pathname = new URL(value).pathname;
+        }
+        catch (_a) {
+            continue;
+        }
+        const segments = pathname
+            .toLowerCase()
+            .split("/")
+            .filter((segment) => segment.length > 0)
+            .map((segment) => segment.replace(/\.(html?|php|aspx?)$/, ""));
+        if (segments.some((segment) => NON_SERVICE_SECTIONS.has(segment)))
+            continue;
+        const candidates = [];
+        const parent = segments.findIndex((segment) => SERVICE_PARENT_SEGMENT.test(segment));
+        if (parent >= 0) {
+            const rest = segments.slice(parent + 1);
+            // pagination of the services index, e.g. /services/page/2
+            if (rest[0] === "page")
+                continue;
+            candidates.push(...rest);
+        }
+        else if (segments.length === 1) {
+            const words = segments[0].split("-");
+            if (words.some((word) => SERVICE_WORDS.has(word)) && !words.some((word) => NON_SERVICE_WORDS.has(word))) {
+                candidates.push(segments[0]);
+            }
+        }
+        for (const candidate of candidates) {
+            if (isServiceSlug(candidate) && !SERVICE_PARENT_SEGMENT.test(candidate)) {
+                slugs.add(candidate);
+                if (slugs.size >= exports.MAX_SERVICE_SLUGS)
+                    break;
+            }
+        }
+    }
+    return [...slugs];
+}
 /**
  * Returns descriptive data about a sitemap.
  *
  * @param url
  * @returns
  */
-function getSiteMapData(url) {
-    return __awaiter(this, void 0, void 0, function* () {
+function getSiteMapData(url_1) {
+    return __awaiter(this, arguments, void 0, function* (url, options = {}) {
         let siteUrl;
         let originalPath;
         let sitemapUrl;
@@ -233,7 +344,14 @@ function getSiteMapData(url) {
         if (termsOfService) {
             (0, stentor_logger_1.log)().debug(`Found terms of service page at ${termsOfService}`);
         }
-        return {
+        const bookingPage = filtered.find((value) => {
+            const parsed = new URL(value);
+            return pagePatterns_1.BOOKING_PAGE_PATH_REGEX.test(parsed.pathname);
+        });
+        if (bookingPage) {
+            (0, stentor_logger_1.log)().debug(`Found booking page at ${bookingPage}`);
+        }
+        const data = {
             siteMapUrl: sitemapUrl.toString(),
             totalPages: filtered.length,
             aboutUs,
@@ -242,7 +360,13 @@ function getSiteMapData(url) {
             paymentPage,
             privacyPolicy,
             termsOfService,
+            bookingPage,
         };
+        if (options.serviceSlugs) {
+            data.serviceSlugs = extractServiceSlugs(filtered);
+            (0, stentor_logger_1.log)().debug(`Found ${data.serviceSlugs.length} service slugs in the sitemap`);
+        }
+        return data;
     });
 }
 //# sourceMappingURL=getSiteMap.js.map

package/lib/utils/getSiteMap.js.map

@@ -1 +1 @@
-{"version":3,"file":"getSiteMap.js","sourceRoot":"","sources":["../../src/utils/getSiteMap.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA0DA,gCAEC;AAqGD,wCAqHC;AAtRD,kCAAkC;AAClC,iDAAmC;AAEnC,2CAA6C;AAC7C,mDAAqC;AAErC,iDAOwB;AAExB,MAAM,YAAY,GAAG,CAAO,GAAW,EAAmB,EAAE;IACxD,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAC;IACzC,MAAM,EAAE,MAAM,EAAE,GAAG,UAAU,CAAC;IAE9B,MAAM,OAAO,GAAG,IAAI,CAAC;IAErB,wFAAwF;IACxF,MAAM,SAAS,GAAG,UAAU,CAAC,GAAG,EAAE;QAC9B,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,WAAW,OAAO,+BAA+B,GAAG,EAAE,CAAC,CAAC;QACpE,UAAU,CAAC,KAAK,EAAE,CAAC;IACvB,CAAC,EAAE,OAAO,CAAC,CAAC;IAEZ,aAAa;IACb,IAAI,CAAC;QACD,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,MAAM,EAAE,CAAC;aACpC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE;YACf,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;gBACf,MAAM,IAAI,KAAK,CAAC,yCAAyC,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC;YAChF,CAAC;YAED,OAAO,QAAQ,CAAC,IAAI,EAAE,CAAC;QAC3B,CAAC,CAAC;aACD,OAAO,CAAC,GAAG,EAAE;YACV,YAAY,CAAC,SAAS,CAAC,CAAC;QAC5B,CAAC,CAAC,CAAC;QAEP,OAAO,IAAI,CAAC;IAChB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,OAAO,CAAC,KAAK,CAAC,4BAA4B,KAAK,EAAE,CAAC,CAAC;QACnD,OAAO,EAAE,CAAC;IACd,CAAC;AACL,CAAC,CAAA,CAAC;AAEF;;;;;;;;;GASG;AACH,SAAsB,UAAU,CAAC,UAAkB;;QAC/C,OAAO,YAAY,CAAC,UAAU,EAAE,IAAI,GAAG,EAAU,CAAC,CAAC;IACvD,CAAC;CAAA;AAED;;;;;;;;;;;GAWG;AACH,SAAe,YAAY,CAAC,UAAkB,EAAE,OAAoB;;QAChE,MAAM,oBAAoB,GAAG,IAAA,sBAAY,EAAC,UAAU,CAAC,CAAC;QAEtD,IAAI,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,EAAE,CAAC;YACpC,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,2BAA2B,oBAAoB,aAAa,CAAC,CAAC;YAC1E,OAAO,EAAE,CAAC;QACd,CAAC;QAED,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,CAAC;QAElC,MAAM,OAAO,GAAG,MAAM,YAAY,CAAC,oBAAoB,CAAC,CAAC;QAEzD,IAAI,CAAC,OAAO,EAAE,CAAC;YACX,OAAO,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;YAC5C,OAAO,EAAE,CAAC;QACd,CAAC;QAED,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,OAAO,EAAE;YAC5B,OAAO,EAAE,IAAI;SAChB,CAAC,CAAC;QAEH,IAAI,OAAO,GAAa,EAAE,CAAC;QAE3B,mBAAmB;QACnB,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE;YACvB,MAAM,GAAG,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;YACxC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACtB,CAAC,CAAC,CAAC;QAEH,yCAAyC;QACzC,MAAM,YAAY,GAAG,CAAC,CAAC,SAAS,CAAC,CAAC,OAAO,EAAE,CAAC;QAE5C,KAAK,MAAM,WAAW,IAAI,YAAY,EAAE,CAAC;YACrC,MAAM,GAAG,GAAG,CAAC,CAAC,WAAW,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;YAE9C,IAAI,CAAC,GAAG,EAAE,CAAC;gBACP,SAAS;YACb,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,YAAY,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;YAC9C,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QACnC,CAAC;QAED,OAAO,OAAO,CAAC;IACnB,CAAC;CAAA;AAqCD;;;;;GAKG;AACH,SAAsB,cAAc,CAAC,GAAW;;QAC5C,IAAI,OAAY,CAAC;QACjB,IAAI,YAAoB,CAAC;QACzB,IAAI,UAAe,CAAC;QAEpB,IAAI,CAAC;YACD,OAAO,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;YACvB,YAAY,GAAG,OAAO,CAAC,QAAQ,CAAC;YAEhC,UAAU,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;YAC1B,UAAU,CAAC,QAAQ,GAAG,GAAG,YAAY,aAAa,CAAC;QACvD,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACT,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,oCAAoC,GAAG,GAAG,CAAC,CAAC;YACxD,OAAO,SAAS,CAAC;QACrB,CAAC;QAED,IAAI,CAAC,UAAU,EAAE,CAAC;YACd,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,oCAAoC,GAAG,GAAG,CAAC,CAAC;YACxD,OAAO,SAAS,CAAC;QACrB,CAAC;QAED,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,sBAAsB,UAAU,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QAE3D,MAAM,OAAO,GAAG,MAAM,UAAU,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;QAExD,QAAQ;QACR,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;YAClC,IAAI,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBACxB,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;gBAChC,UAAU,CAAC,QAAQ,GAAG,KAAK,CAAC;gBAC5B,OAAO,UAAU,CAAC,QAAQ,EAAE,CAAC;YACjC,CAAC;iBAAM,CAAC;gBACJ,OAAO,KAAK,CAAC;YACjB,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,wFAAwF;QACxF,MAAM,QAAQ,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE;YACtC,IAAI,CAAC;gBACD,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;gBAC9B,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC;gBACjC,OAAO,QAAQ,CAAC,UAAU,CAAC,YAAY,CAAC,CAAC;YAC7C,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACT,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,0BAA0B,KAAK,EAAE,CAAC,CAAC;YACnD,CAAC;YACD,OAAO,KAAK,CAAC;QACjB,CAAC,CAAC,CAAC;QAEH,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,eAAe,OAAO,CAAC,MAAM,uBAAuB,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC;QAEnF,qBAAqB;QACrB,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACpC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,kCAAmB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACrD,CAAC,CAAC,CAAC;QAEH,IAAI,OAAO,EAAE,CAAC;YACV,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,0BAA0B,OAAO,EAAE,CAAC,CAAC;QACrD,CAAC;QAED,MAAM,SAAS,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACtC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,oCAAqB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACvD,CAAC,CAAC,CAAC;QAEH,IAAI,SAAS,EAAE,CAAC;YACZ,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,4BAA4B,SAAS,EAAE,CAAC,CAAC;QACzD,CAAC;QAED,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACpC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,kCAAmB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACrD,CAAC,CAAC,CAAC;QAEH,IAAI,OAAO,EAAE,CAAC;YACV,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,qBAAqB,OAAO,EAAE,CAAC,CAAC;QAChD,CAAC;QAED,MAAM,WAAW,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACxC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,sCAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACzD,CAAC,CAAC,CAAC;QAEH,IAAI,WAAW,EAAE,CAAC;YACd,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,yBAAyB,WAAW,EAAE,CAAC,CAAC;QACxD,CAAC;QAED,MAAM,aAAa,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YAC1C,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,wCAAyB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QAC3D,CAAC,CAAC,CAAC;QAEH,IAAI,aAAa,EAAE,CAAC;YAChB,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,gCAAgC,aAAa,EAAE,CAAC,CAAC;QACjE,CAAC;QAED,MAAM,cAAc,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YAC3C,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,0CAA2B,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QAC7D,CAAC,CAAC,CAAC;QAEH,IAAI,cAAc,EAAE,CAAC;YACjB,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,kCAAkC,cAAc,EAAE,CAAC,CAAC;QACpE,CAAC;QAED,OAAO;YACH,UAAU,EAAE,UAAU,CAAC,QAAQ,EAAE;YAEjC,UAAU,EAAE,QAAQ,CAAC,MAAM;YAE3B,OAAO;YACP,SAAS;YACT,OAAO;YACP,WAAW;YACX,aAAa;YACb,cAAc;SACjB,CAAC;IACN,CAAC;CAAA"}
+{"version":3,"file":"getSiteMap.js","sourceRoot":"","sources":["../../src/utils/getSiteMap.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA2DA,gCAEC;AA2LD,kDA6CC;AAQD,wCAsIC;AAnbD,kCAAkC;AAClC,iDAAmC;AAEnC,2CAA6C;AAC7C,mDAAqC;AAErC,iDAQwB;AAExB,MAAM,YAAY,GAAG,CAAO,GAAW,EAAmB,EAAE;IACxD,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAC;IACzC,MAAM,EAAE,MAAM,EAAE,GAAG,UAAU,CAAC;IAE9B,MAAM,OAAO,GAAG,IAAI,CAAC;IAErB,wFAAwF;IACxF,MAAM,SAAS,GAAG,UAAU,CAAC,GAAG,EAAE;QAC9B,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,WAAW,OAAO,+BAA+B,GAAG,EAAE,CAAC,CAAC;QACpE,UAAU,CAAC,KAAK,EAAE,CAAC;IACvB,CAAC,EAAE,OAAO,CAAC,CAAC;IAEZ,aAAa;IACb,IAAI,CAAC;QACD,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,MAAM,EAAE,CAAC;aACpC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE;YACf,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;gBACf,MAAM,IAAI,KAAK,CAAC,yCAAyC,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC;YAChF,CAAC;YAED,OAAO,QAAQ,CAAC,IAAI,EAAE,CAAC;QAC3B,CAAC,CAAC;aACD,OAAO,CAAC,GAAG,EAAE;YACV,YAAY,CAAC,SAAS,CAAC,CAAC;QAC5B,CAAC,CAAC,CAAC;QAEP,OAAO,IAAI,CAAC;IAChB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,OAAO,CAAC,KAAK,CAAC,4BAA4B,KAAK,EAAE,CAAC,CAAC;QACnD,OAAO,EAAE,CAAC;IACd,CAAC;AACL,CAAC,CAAA,CAAC;AAEF;;;;;;;;;GASG;AACH,SAAsB,UAAU,CAAC,UAAkB;;QAC/C,OAAO,YAAY,CAAC,UAAU,EAAE,IAAI,GAAG,EAAU,CAAC,CAAC;IACvD,CAAC;CAAA;AAED;;;;;;;;;;;GAWG;AACH,SAAe,YAAY,CAAC,UAAkB,EAAE,OAAoB;;QAChE,MAAM,oBAAoB,GAAG,IAAA,sBAAY,EAAC,UAAU,CAAC,CAAC;QAEtD,IAAI,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,EAAE,CAAC;YACpC,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,2BAA2B,oBAAoB,aAAa,CAAC,CAAC;YAC1E,OAAO,EAAE,CAAC;QACd,CAAC;QAED,OAAO,CAAC,GAAG,CAAC,oBAAoB,CAAC,CAAC;QAElC,MAAM,OAAO,GAAG,MAAM,YAAY,CAAC,oBAAoB,CAAC,CAAC;QAEzD,IAAI,CAAC,OAAO,EAAE,CAAC;YACX,OAAO,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;YAC5C,OAAO,EAAE,CAAC;QACd,CAAC;QAED,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,OAAO,EAAE;YAC5B,OAAO,EAAE,IAAI;SAChB,CAAC,CAAC;QAEH,IAAI,OAAO,GAAa,EAAE,CAAC;QAE3B,mBAAmB;QACnB,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE;YACvB,MAAM,GAAG,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;YACxC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACtB,CAAC,CAAC,CAAC;QAEH,yCAAyC;QACzC,MAAM,YAAY,GAAG,CAAC,CAAC,SAAS,CAAC,CAAC,OAAO,EAAE,CAAC;QAE5C,KAAK,MAAM,WAAW,IAAI,YAAY,EAAE,CAAC;YACrC,MAAM,GAAG,GAAG,CAAC,CAAC,WAAW,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC;YAE9C,IAAI,CAAC,GAAG,EAAE,CAAC;gBACP,SAAS;YACb,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,YAAY,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;YAC9C,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QACnC,CAAC;QAED,OAAO,OAAO,CAAC;IACnB,CAAC;CAAA;AAwDD,gGAAgG;AACnF,QAAA,iBAAiB,GAAG,GAAG,CAAC;AAErC,0EAA0E;AAC1E,MAAM,sBAAsB,GAAG,0DAA0D,CAAC;AAE1F,8GAA8G;AAC9G,MAAM,aAAa,GAAG,IAAI,GAAG,CAAC;IAC1B,QAAQ;IACR,SAAS;IACT,cAAc;IACd,eAAe;IACf,SAAS;IACT,aAAa;IACb,cAAc;IACd,UAAU;IACV,aAAa;IACb,YAAY;IACZ,aAAa;IACb,SAAS;IACT,aAAa;IACb,YAAY;IACZ,YAAY;IACZ,WAAW;IACX,SAAS;IACT,UAAU;CACb,CAAC,CAAC;AAEH,wGAAwG;AACxG,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,OAAO,EAAE,UAAU,EAAE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,CAAC,CAAC,CAAC;AAElH,oDAAoD;AACpD,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,aAAa,EAAE,YAAY,EAAE,UAAU,EAAE,KAAK,CAAC,CAAC,CAAC;AAEnG,0FAA0F;AAC1F,MAAM,oBAAoB,GAAG,IAAI,GAAG,CAAC;IACjC,MAAM;IACN,MAAM;IACN,SAAS;IACT,UAAU;IACV,MAAM;IACN,OAAO;IACP,UAAU;IACV,YAAY;IACZ,KAAK;IACL,MAAM;IACN,QAAQ;IACR,YAAY;IACZ,MAAM;IACN,SAAS;IACT,MAAM;IACN,QAAQ;IACR,OAAO;CACV,CAAC,CAAC;AAEH,MAAM,IAAI,GAAG,0BAA0B,CAAC;AAExC,SAAS,aAAa,CAAC,IAAY;IAC/B,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM,IAAI,EAAE,IAAI,CAAC,iBAAiB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;AACtG,CAAC;AAED;;;;;;;;;;;GAWG;AACH,SAAgB,mBAAmB,CAAC,IAAc;IAC9C,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAC;IAEhC,KAAK,MAAM,KAAK,IAAI,IAAI,EAAE,CAAC;QACvB,IAAI,KAAK,CAAC,IAAI,IAAI,yBAAiB;YAAE,MAAM;QAE3C,IAAI,QAAgB,CAAC;QACrB,IAAI,CAAC;YACD,QAAQ,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC,QAAQ,CAAC;QACvC,CAAC;QAAC,WAAM,CAAC;YACL,SAAS;QACb,CAAC;QAED,MAAM,QAAQ,GAAG,QAAQ;aACpB,WAAW,EAAE;aACb,KAAK,CAAC,GAAG,CAAC;aACV,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;aACvC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,OAAO,CAAC,sBAAsB,EAAE,EAAE,CAAC,CAAC,CAAC;QAEnE,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;YAAE,SAAS;QAE5E,MAAM,UAAU,GAAa,EAAE,CAAC;QAChC,MAAM,MAAM,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,sBAAsB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;QAErF,IAAI,MAAM,IAAI,CAAC,EAAE,CAAC;YACd,MAAM,IAAI,GAAG,QAAQ,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YACxC,0DAA0D;YAC1D,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,MAAM;gBAAE,SAAS;YACjC,UAAU,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC;QAC7B,CAAC;aAAM,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC/B,MAAM,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YACrC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,iBAAiB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;gBACtG,UAAU,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;YACjC,CAAC;QACL,CAAC;QAED,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;YACjC,IAAI,aAAa,CAAC,SAAS,CAAC,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;gBACtE,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;gBACrB,IAAI,KAAK,CAAC,IAAI,IAAI,yBAAiB;oBAAE,MAAM;YAC/C,CAAC;QACL,CAAC;IACL,CAAC;IAED,OAAO,CAAC,GAAG,KAAK,CAAC,CAAC;AACtB,CAAC;AAED;;;;;GAKG;AACH,SAAsB,cAAc;yDAAC,GAAW,EAAE,UAA8B,EAAE;QAC9E,IAAI,OAAY,CAAC;QACjB,IAAI,YAAoB,CAAC;QACzB,IAAI,UAAe,CAAC;QAEpB,IAAI,CAAC;YACD,OAAO,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;YACvB,YAAY,GAAG,OAAO,CAAC,QAAQ,CAAC;YAEhC,UAAU,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;YAC1B,UAAU,CAAC,QAAQ,GAAG,GAAG,YAAY,aAAa,CAAC;QACvD,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACT,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,oCAAoC,GAAG,GAAG,CAAC,CAAC;YACxD,OAAO,SAAS,CAAC;QACrB,CAAC;QAED,IAAI,CAAC,UAAU,EAAE,CAAC;YACd,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,oCAAoC,GAAG,GAAG,CAAC,CAAC;YACxD,OAAO,SAAS,CAAC;QACrB,CAAC;QAED,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,sBAAsB,UAAU,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QAE3D,MAAM,OAAO,GAAG,MAAM,UAAU,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;QAExD,QAAQ;QACR,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;YAClC,IAAI,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBACxB,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;gBAChC,UAAU,CAAC,QAAQ,GAAG,KAAK,CAAC;gBAC5B,OAAO,UAAU,CAAC,QAAQ,EAAE,CAAC;YACjC,CAAC;iBAAM,CAAC;gBACJ,OAAO,KAAK,CAAC;YACjB,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,wFAAwF;QACxF,MAAM,QAAQ,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE;YACtC,IAAI,CAAC;gBACD,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;gBAC9B,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC;gBACjC,OAAO,QAAQ,CAAC,UAAU,CAAC,YAAY,CAAC,CAAC;YAC7C,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACT,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,0BAA0B,KAAK,EAAE,CAAC,CAAC;YACnD,CAAC;YACD,OAAO,KAAK,CAAC;QACjB,CAAC,CAAC,CAAC;QAEH,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,eAAe,OAAO,CAAC,MAAM,uBAAuB,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC;QAEnF,qBAAqB;QACrB,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACpC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,kCAAmB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACrD,CAAC,CAAC,CAAC;QAEH,IAAI,OAAO,EAAE,CAAC;YACV,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,0BAA0B,OAAO,EAAE,CAAC,CAAC;QACrD,CAAC;QAED,MAAM,SAAS,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACtC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,oCAAqB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACvD,CAAC,CAAC,CAAC;QAEH,IAAI,SAAS,EAAE,CAAC;YACZ,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,4BAA4B,SAAS,EAAE,CAAC,CAAC;QACzD,CAAC;QAED,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACpC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,kCAAmB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACrD,CAAC,CAAC,CAAC;QAEH,IAAI,OAAO,EAAE,CAAC;YACV,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,qBAAqB,OAAO,EAAE,CAAC,CAAC;QAChD,CAAC;QAED,MAAM,WAAW,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACxC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,sCAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACzD,CAAC,CAAC,CAAC;QAEH,IAAI,WAAW,EAAE,CAAC;YACd,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,yBAAyB,WAAW,EAAE,CAAC,CAAC;QACxD,CAAC;QAED,MAAM,aAAa,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YAC1C,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,wCAAyB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QAC3D,CAAC,CAAC,CAAC;QAEH,IAAI,aAAa,EAAE,CAAC;YAChB,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,gCAAgC,aAAa,EAAE,CAAC,CAAC;QACjE,CAAC;QAED,MAAM,cAAc,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YAC3C,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,0CAA2B,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QAC7D,CAAC,CAAC,CAAC;QAEH,IAAI,cAAc,EAAE,CAAC;YACjB,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,kCAAkC,cAAc,EAAE,CAAC,CAAC;QACpE,CAAC;QAED,MAAM,WAAW,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE;YACxC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9B,OAAO,sCAAuB,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;QACzD,CAAC,CAAC,CAAC;QAEH,IAAI,WAAW,EAAE,CAAC;YACd,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,yBAAyB,WAAW,EAAE,CAAC,CAAC;QACxD,CAAC;QAED,MAAM,IAAI,GAAgB;YACtB,UAAU,EAAE,UAAU,CAAC,QAAQ,EAAE;YAEjC,UAAU,EAAE,QAAQ,CAAC,MAAM;YAE3B,OAAO;YACP,SAAS;YACT,OAAO;YACP,WAAW;YACX,aAAa;YACb,cAAc;YACd,WAAW;SACd,CAAC;QAEF,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;YACvB,IAAI,CAAC,YAAY,GAAG,mBAAmB,CAAC,QAAQ,CAAC,CAAC;YAClD,IAAA,oBAAG,GAAE,CAAC,KAAK,CAAC,SAAS,IAAI,CAAC,YAAY,CAAC,MAAM,+BAA+B,CAAC,CAAC;QAClF,CAAC;QAED,OAAO,IAAI,CAAC;IAChB,CAAC;CAAA"}

package/lib/utils/pagePatterns.d.ts

@@ -10,7 +10,7 @@ export interface PagePattern {
     /** Regex to match link text */
     textRegex: RegExp;
 }
-export type PageType = "aboutUs" | "contactUs" | "faqPage" | "paymentPage" | "privacyPolicy" | "termsOfService";
+export type PageType = "aboutUs" | "contactUs" | "faqPage" | "paymentPage" | "privacyPolicy" | "termsOfService" | "bookingPage";
 /**
  * About Us page pattern
  * Matches: /about, /about-us, /company/about, /who-we-are, /our-story, /info/about, /en/about, /about/our-company
@@ -47,6 +47,21 @@ export declare const PRIVACY_POLICY_TEXT_REGEX: RegExp;
  */
 export declare const TERMS_OF_SERVICE_PATH_REGEX: RegExp;
 export declare const TERMS_OF_SERVICE_TEXT_REGEX: RegExp;
+/**
+ * Booking page pattern, where a business embeds its scheduler (Demandforce, Housecall Pro, ServiceTitan, Jobber...)
+ * Matches: /book, /book-online, /book-now, /booking, /schedule, /schedule-service, /schedule-online,
+ * /schedule-appointment, /appointment(s), /make-appointment(s), /make-an-appointment, /request-appointment,
+ * /request-an-appointment, /request-service, /estimate, /free-estimate
+ *
+ * Only the last path segment is matched, so a blog post such as /blog/how-to-book-a-plumber is not a booking page.
+ */
+export declare const BOOKING_PAGE_PATH_REGEX: RegExp;
+/**
+ * Link text for a booking page: "Book Online", "Schedule Service", "Request an Appointment", "Make an Appointment"...
+ *
+ * Anchored to the whole link text, so a headline such as "How to book a plumber" does not match.
+ */
+export declare const BOOKING_PAGE_TEXT_REGEX: RegExp;
 /**
  * All page patterns in a single array for iteration
  */

package/lib/utils/pagePatterns.js

@@ -1,7 +1,7 @@
 "use strict";
 /*! Copyright (c) 2026, XAPP AI */
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.PAGE_KEYS = exports.PAGE_PATTERNS = exports.TERMS_OF_SERVICE_TEXT_REGEX = exports.TERMS_OF_SERVICE_PATH_REGEX = exports.PRIVACY_POLICY_TEXT_REGEX = exports.PRIVACY_POLICY_PATH_REGEX = exports.PAYMENT_PAGE_TEXT_REGEX = exports.PAYMENT_PAGE_PATH_REGEX = exports.FAQ_PAGE_TEXT_REGEX = exports.FAQ_PAGE_PATH_REGEX = exports.CONTACT_US_TEXT_REGEX = exports.CONTACT_US_PATH_REGEX = exports.ABOUT_US_TEXT_REGEX = exports.ABOUT_US_PATH_REGEX = void 0;
+exports.PAGE_KEYS = exports.PAGE_PATTERNS = exports.BOOKING_PAGE_TEXT_REGEX = exports.BOOKING_PAGE_PATH_REGEX = exports.TERMS_OF_SERVICE_TEXT_REGEX = exports.TERMS_OF_SERVICE_PATH_REGEX = exports.PRIVACY_POLICY_TEXT_REGEX = exports.PRIVACY_POLICY_PATH_REGEX = exports.PAYMENT_PAGE_TEXT_REGEX = exports.PAYMENT_PAGE_PATH_REGEX = exports.FAQ_PAGE_TEXT_REGEX = exports.FAQ_PAGE_PATH_REGEX = exports.CONTACT_US_TEXT_REGEX = exports.CONTACT_US_PATH_REGEX = exports.ABOUT_US_TEXT_REGEX = exports.ABOUT_US_PATH_REGEX = void 0;
 /**
  * About Us page pattern
  * Matches: /about, /about-us, /company/about, /who-we-are, /our-story, /info/about, /en/about, /about/our-company
@@ -38,6 +38,21 @@ exports.PRIVACY_POLICY_TEXT_REGEX = /^privacy(\s+policy)?$/i;
  */
 exports.TERMS_OF_SERVICE_PATH_REGEX = /\/(terms(-of-(service|use))?|tos|legal\/terms|terms-and-conditions)(\/)?$/i;
 exports.TERMS_OF_SERVICE_TEXT_REGEX = /^(terms(\s+(of\s+(service|use)|and\s+conditions)?)?|tos)$/i;
+/**
+ * Booking page pattern, where a business embeds its scheduler (Demandforce, Housecall Pro, ServiceTitan, Jobber...)
+ * Matches: /book, /book-online, /book-now, /booking, /schedule, /schedule-service, /schedule-online,
+ * /schedule-appointment, /appointment(s), /make-appointment(s), /make-an-appointment, /request-appointment,
+ * /request-an-appointment, /request-service, /estimate, /free-estimate
+ *
+ * Only the last path segment is matched, so a blog post such as /blog/how-to-book-a-plumber is not a booking page.
+ */
+exports.BOOKING_PAGE_PATH_REGEX = /\/(book(-online|-now)?|booking|schedule(-service|-online|-appointment|-an-appointment)?|appointments?|make-(an-)?appointments?|request-(an-)?appointment|request-service|(free-)?estimate)(\/)?$/i;
+/**
+ * Link text for a booking page: "Book Online", "Schedule Service", "Request an Appointment", "Make an Appointment"...
+ *
+ * Anchored to the whole link text, so a headline such as "How to book a plumber" does not match.
+ */
+exports.BOOKING_PAGE_TEXT_REGEX = /^((book|schedule)(\s+(online|now|(an?\s+)?(appointment|service|visit|consultation)))?|request\s+(an?\s+)?(appointment|service|visit|consultation|estimate)|make\s+an?\s+appointment|free\s+estimate)(\s+(online|now|today))?$/i;
 /**
  * All page patterns in a single array for iteration
  */
@@ -72,6 +87,11 @@ exports.PAGE_PATTERNS = [
         pathRegex: exports.TERMS_OF_SERVICE_PATH_REGEX,
         textRegex: exports.TERMS_OF_SERVICE_TEXT_REGEX,
     },
+    {
+        key: "bookingPage",
+        pathRegex: exports.BOOKING_PAGE_PATH_REGEX,
+        textRegex: exports.BOOKING_PAGE_TEXT_REGEX,
+    },
 ];
 /**
  * All page type keys, derived from PAGE_PATTERNS to maintain single source of truth

package/lib/utils/pagePatterns.js.map

@@ -1 +1 @@
-{"version":3,"file":"pagePatterns.js","sourceRoot":"","sources":["../../src/utils/pagePatterns.ts"],"names":[],"mappings":";AAAA,kCAAkC;;;AAiBlC;;;GAGG;AACU,QAAA,mBAAmB,GAC5B,qGAAqG,CAAC;AAC7F,QAAA,mBAAmB,GAAG,kBAAkB,CAAC;AAEtD;;;GAGG;AACU,QAAA,qBAAqB,GAC9B,oEAAoE,CAAC;AAC5D,QAAA,qBAAqB,GAAG,oBAAoB,CAAC;AAE1D;;;GAGG;AACU,QAAA,mBAAmB,GAAG,0DAA0D,CAAC;AACjF,QAAA,mBAAmB,GAAG,+CAA+C,CAAC;AAEnF;;;GAGG;AACU,QAAA,uBAAuB,GAAG,+DAA+D,CAAC;AAC1F,QAAA,uBAAuB,GAAG,uDAAuD,CAAC;AAE/F;;;GAGG;AACU,QAAA,yBAAyB,GAAG,+DAA+D,CAAC;AAC5F,QAAA,yBAAyB,GAAG,wBAAwB,CAAC;AAElE;;;GAGG;AACU,QAAA,2BAA2B,GACpC,4EAA4E,CAAC;AACpE,QAAA,2BAA2B,GAAG,4DAA4D,CAAC;AAExG;;GAEG;AACU,QAAA,aAAa,GAAkB;IACxC;QACI,GAAG,EAAE,SAAS;QACd,SAAS,EAAE,2BAAmB;QAC9B,SAAS,EAAE,2BAAmB;KACjC;IACD;QACI,GAAG,EAAE,WAAW;QAChB,SAAS,EAAE,6BAAqB;QAChC,SAAS,EAAE,6BAAqB;KACnC;IACD;QACI,GAAG,EAAE,SAAS;QACd,SAAS,EAAE,2BAAmB;QAC9B,SAAS,EAAE,2BAAmB;KACjC;IACD;QACI,GAAG,EAAE,aAAa;QAClB,SAAS,EAAE,+BAAuB;QAClC,SAAS,EAAE,+BAAuB;KACrC;IACD;QACI,GAAG,EAAE,eAAe;QACpB,SAAS,EAAE,iCAAyB;QACpC,SAAS,EAAE,iCAAyB;KACvC;IACD;QACI,GAAG,EAAE,gBAAgB;QACrB,SAAS,EAAE,mCAA2B;QACtC,SAAS,EAAE,mCAA2B;KACzC;CACJ,CAAC;AAEF;;GAEG;AACU,QAAA,SAAS,GAAwB,qBAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC"}
+{"version":3,"file":"pagePatterns.js","sourceRoot":"","sources":["../../src/utils/pagePatterns.ts"],"names":[],"mappings":";AAAA,kCAAkC;;;AAwBlC;;;GAGG;AACU,QAAA,mBAAmB,GAC5B,qGAAqG,CAAC;AAC7F,QAAA,mBAAmB,GAAG,kBAAkB,CAAC;AAEtD;;;GAGG;AACU,QAAA,qBAAqB,GAC9B,oEAAoE,CAAC;AAC5D,QAAA,qBAAqB,GAAG,oBAAoB,CAAC;AAE1D;;;GAGG;AACU,QAAA,mBAAmB,GAAG,0DAA0D,CAAC;AACjF,QAAA,mBAAmB,GAAG,+CAA+C,CAAC;AAEnF;;;GAGG;AACU,QAAA,uBAAuB,GAAG,+DAA+D,CAAC;AAC1F,QAAA,uBAAuB,GAAG,uDAAuD,CAAC;AAE/F;;;GAGG;AACU,QAAA,yBAAyB,GAAG,+DAA+D,CAAC;AAC5F,QAAA,yBAAyB,GAAG,wBAAwB,CAAC;AAElE;;;GAGG;AACU,QAAA,2BAA2B,GACpC,4EAA4E,CAAC;AACpE,QAAA,2BAA2B,GAAG,4DAA4D,CAAC;AAExG;;;;;;;GAOG;AACU,QAAA,uBAAuB,GAChC,mMAAmM,CAAC;AACxM;;;;GAIG;AACU,QAAA,uBAAuB,GAChC,gOAAgO,CAAC;AAErO;;GAEG;AACU,QAAA,aAAa,GAAkB;IACxC;QACI,GAAG,EAAE,SAAS;QACd,SAAS,EAAE,2BAAmB;QAC9B,SAAS,EAAE,2BAAmB;KACjC;IACD;QACI,GAAG,EAAE,WAAW;QAChB,SAAS,EAAE,6BAAqB;QAChC,SAAS,EAAE,6BAAqB;KACnC;IACD;QACI,GAAG,EAAE,SAAS;QACd,SAAS,EAAE,2BAAmB;QAC9B,SAAS,EAAE,2BAAmB;KACjC;IACD;QACI,GAAG,EAAE,aAAa;QAClB,SAAS,EAAE,+BAAuB;QAClC,SAAS,EAAE,+BAAuB;KACrC;IACD;QACI,GAAG,EAAE,eAAe;QACpB,SAAS,EAAE,iCAAyB;QACpC,SAAS,EAAE,iCAAyB;KACvC;IACD;QACI,GAAG,EAAE,gBAAgB;QACrB,SAAS,EAAE,mCAA2B;QACtC,SAAS,EAAE,mCAA2B;KACzC;IACD;QACI,GAAG,EAAE,aAAa;QAClB,SAAS,EAAE,+BAAuB;QAClC,SAAS,EAAE,+BAAuB;KACrC;CACJ,CAAC;AAEF;;GAEG;AACU,QAAA,SAAS,GAAwB,qBAAa,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC"}

package/lib/utils/subpageDetection.d.ts

@@ -0,0 +1,140 @@
+import { DetectorResult, Evaluator } from "@xapp/arachne-parsers";
+import { DetectorConstructor } from "../contants";
+import { DetectionSource, SourcedDetectorResult, SubpageDetectionOptions, SubpageType } from "../models";
+/** Default budget for one subpage: navigation plus detectors. */
+export declare const DEFAULT_SUBPAGE_TIMEOUT_MS = 10000;
+/** Default cap on subpages visited in one crawl. */
+export declare const DEFAULT_MAX_SUBPAGES = 3;
+/** A subpage is not started with less than this left of the overall budget. */
+export declare const MIN_SUBPAGE_BUDGET_MS = 1000;
+/** Longest wait for a subpage tab to close before moving on. */
+export declare const TAB_CLOSE_TIMEOUT_MS = 2000;
+export interface SubpageTarget {
+    pageType: SubpageType;
+    url: string;
+}
+export interface SubpageVisit extends SubpageTarget {
+    status: "detected" | "skipped";
+    /**
+     * Why the page was skipped.
+     */
+    reason?: string;
+}
+/**
+ * True when `candidate` is an http(s) URL on the same site as `base`.  www and the apex count as the same site;
+ * any other subdomain does not.
+ */
+export declare function isSameSite(candidate: string, base: string): boolean;
+/**
+ * True when the options ask for at least one page.
+ */
+export declare function isSubpageDetectionEnabled(options: SubpageDetectionOptions | undefined): boolean;
+/**
+ * Picks the subpages to visit: the configured page types that were discovered and are on the same site as the
+ * homepage, one visit per page type and per URL, never the homepage itself, at most `maxPages`.
+ *
+ * @param pages Discovered pages, e.g. `site.pages`
+ * @param homepageUrl The URL the crawl started from
+ * @param options Subpage detection options
+ */
+export declare function resolveSubpageTargets(pages: Partial<Record<SubpageType, string>> | undefined, homepageUrl: string, options: SubpageDetectionOptions | undefined): SubpageTarget[];
+/**
+ * Merges one page's detector results into the results gathered so far.
+ *
+ * Each technology appears once.  A technology not seen yet is added with `source`.  A technology already detected
+ * keeps its first source; one only reported as not installed so far is replaced, in place, by a detection.
+ *
+ * @param target Results gathered so far, mutated
+ * @param incoming Results from the page
+ * @param source The page they came from
+ */
+export declare function mergeDetectorResults(target: SourcedDetectorResult[], incoming: DetectorResult[], source: DetectionSource): void;
+/**
+ * Navigates to a page on the same site and returns its rendered HTML.
+ *
+ * Rejects, without navigating, for an off-site URL; and rejects when the page does not load within `timeoutMs`,
+ * answers with an error status, or redirects off-site.
+ *
+ * @param page The live page to navigate
+ * @param url The page to load
+ * @param homepageUrl The URL the crawl started from
+ * @param timeoutMs Budget for the navigation and reading the content
+ */
+export declare function loadSameSitePage(page: Evaluator, url: string, homepageUrl: string, timeoutMs: number): Promise<string>;
+/**
+ * A browser tab used for subpage visits, i.e. a puppeteer Page.
+ */
+export interface SubpageTab extends Evaluator {
+    isClosed(): boolean;
+    close(options?: {
+        runBeforeUnload?: boolean;
+    }): Promise<void>;
+}
+export interface LazyTab {
+    /**
+     * The tab, opened on first use.  Rejects once the tab has been closed.
+     */
+    get(): Promise<SubpageTab>;
+    /**
+     * Closes the tab if one was opened.  Never throws, and waits at most the close timeout; a tab still opening is
+     * closed as soon as it opens.
+     */
+    close(): Promise<void>;
+}
+/**
+ * A tab that is only opened when first needed, so subpage visits never navigate the page the homepage detectors
+ * ran on: a detector that outlived its budget keeps reading the homepage's DOM, not a subpage's.
+ *
+ * Use one per page visit and `close()` it in a `finally`.  Closing is what cancels a navigation that outlived its
+ * budget; a timeout race alone leaves it pending, and a second `goto()` on that tab would fail.
+ *
+ * @param open Opens a new tab and applies the same setup as the crawler's own pages
+ * @param closeTimeoutMs Longest wait for the tab to close
+ */
+export declare function createLazyTab(open: () => Promise<SubpageTab>, closeTimeoutMs?: number): LazyTab;
+export interface DetectOnSubpagesProps {
+    /**
+     * Opens a new tab, with the crawler's page setup.  Called once per target; never returns the page the homepage
+     * detectors ran on, nor a tab used before.
+     */
+    openTab: () => Promise<SubpageTab>;
+    /**
+     * The URL the crawl started from.
+     */
+    homepageUrl: string;
+    /**
+     * Pages to visit, from `resolveSubpageTargets()`.
+     */
+    targets: SubpageTarget[];
+    detectors: DetectorConstructor[];
+    /**
+     * Results gathered so far; each page's results are merged in once that page finishes.
+     */
+    detectedTechnologies: SourcedDetectorResult[];
+    /**
+     * Budget for each page.
+     */
+    pageTimeoutMs: number;
+    /**
+     * Epoch milliseconds by which all subpages must be done.
+     */
+    deadline: number;
+    /**
+     * Receives the HTML of each page that loaded.
+     */
+    onPageContent?: (target: SubpageTarget, html: string) => void;
+    /**
+     * Longest wait for each tab to close.  Defaults to TAB_CLOSE_TIMEOUT_MS.
+     */
+    tabCloseTimeoutMs?: number;
+}
+/**
+ * Visits each target in turn, each on a fresh tab closed once that target is done, and runs the detector chain on it.
+ *
+ * A page that fails to load, errors, or does not finish its detectors within its budget is skipped and its partial
+ * results are discarded; results already gathered are never lost.  Nothing is started once less than
+ * MIN_SUBPAGE_BUDGET_MS is left before `deadline`.
+ *
+ * @returns What happened on each target
+ */
+export declare function detectOnSubpages(props: DetectOnSubpagesProps): Promise<SubpageVisit[]>;