@adobe/spacecat-shared-data-access 2.38.1 → 2.40.0

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+# [@adobe/spacecat-shared-data-access-v2.40.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.39.0...@adobe/spacecat-shared-data-access-v2.40.0) (2025-07-24)
+
+
+### Features
+
+* amend llmo config with URL patterns ([#869](https://github.com/adobe/spacecat-shared/issues/869)) ([c6cb047](https://github.com/adobe/spacecat-shared/commit/c6cb0476c6d2b3dad048855ae49481ae8fcfb5ce))
+
+# [@adobe/spacecat-shared-data-access-v2.39.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.38.1...@adobe/spacecat-shared-data-access-v2.39.0) (2025-07-22)
+
+
+### Features
+
+* add llmo specific configuration in site config ([#853](https://github.com/adobe/spacecat-shared/issues/853)) ([476b8e8](https://github.com/adobe/spacecat-shared/commit/476b8e84009602a4c9d7ef8d42b9efcae3beae9e))
+
 # [@adobe/spacecat-shared-data-access-v2.38.1](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.38.0...@adobe/spacecat-shared-data-access-v2.38.1) (2025-07-22)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/spacecat-shared-data-access",
-  "version": "2.38.1",
+  "version": "2.40.0",
   "description": "Shared modules of the Spacecat Services - Data Access",
   "type": "module",
   "engines": {

package/src/models/site/config.js

@@ -37,6 +37,32 @@ export const IMPORT_SOURCES = {
   RUM: 'rum',
 };
 
+const LLMO_TAG_PATTERN = /^(market|product|topic):\s?.+/;
+const LLMO_TAG = Joi.alternatives()
+  .try(
+    // Tag market, product, topic like this: "market: ch", "product: firefly", "topic: copyright"
+    Joi.string().pattern(LLMO_TAG_PATTERN),
+    Joi.string(),
+  );
+
+// LLMO question schema for both Human and AI questions
+const QUESTION_SCHEMA = Joi.object({
+  key: Joi.string().required(),
+  question: Joi.string().required(),
+  source: Joi.string().optional(),
+  volume: Joi.string().optional(),
+  keyword: Joi.string().optional(),
+  url: Joi.string().uri().optional(),
+  tags: Joi.array().items(LLMO_TAG).optional(),
+  importTime: Joi.string().isoDate().optional(),
+});
+
+const LLMO_URL_PATTERN_SCHEMA = {
+  urlPattern: Joi.string().uri().required(),
+  tags: Joi.array().items(LLMO_TAG).optional(),
+};
+const LLMO_URL_PATTERNS_SCHEMA = Joi.array().items(LLMO_URL_PATTERN_SCHEMA);
+
 const IMPORT_BASE_KEYS = {
   destinations: Joi.array().items(Joi.string().valid(IMPORT_DESTINATIONS.DEFAULT)).required(),
   sources: Joi.array().items(Joi.string().valid(...Object.values(IMPORT_SOURCES))).required(),
@@ -201,6 +227,15 @@ export const configSchema = Joi.object({
     headers: Joi.object().pattern(Joi.string(), Joi.string()),
     overrideBaseURL: Joi.string().uri().optional(),
   }).optional(),
+  llmo: Joi.object({
+    dataFolder: Joi.string().required(),
+    brand: Joi.string().required(),
+    questions: Joi.object({
+      Human: Joi.array().items(QUESTION_SCHEMA).optional(),
+      AI: Joi.array().items(QUESTION_SCHEMA).optional(),
+    }).optional(),
+    urlPatterns: LLMO_URL_PATTERNS_SCHEMA.optional(),
+  }).optional(),
   cdnLogsConfig: Joi.object({
     bucketName: Joi.string().required(),
     filters: Joi.array().items(
@@ -257,6 +292,20 @@ export function validateConfiguration(config) {
   return value; // Validated and sanitized configuration
 }
 
+export function extractWellKnownTags(tags) {
+  const wellKnownTags = {};
+  for (const tag of tags) {
+    if (LLMO_TAG_PATTERN.test(tag)) {
+      const colonIdx = tag.indexOf(':');
+      const value = tag.slice(colonIdx + 1).trim();
+      if (colonIdx !== -1 && value) {
+        wellKnownTags[tag.slice(0, colonIdx).trim()] = value;
+      }
+    }
+  }
+  return wellKnownTags;
+}
+
 export const Config = (data = {}) => {
   let configData;
 
@@ -274,7 +323,7 @@ export const Config = (data = {}) => {
   }
 
   const state = { ...configData };
-  const self = { state };
+  const self = { state, extractWellKnownTags };
   self.getSlackConfig = () => state.slack;
   self.isInternalCustomer = () => state?.slack?.workspace === 'internal';
   self.getSlackMentions = (type) => state?.handlers?.[type]?.mentions?.slack;
@@ -291,6 +340,12 @@ export const Config = (data = {}) => {
   self.getFetchConfig = () => state?.fetchConfig;
   self.getBrandConfig = () => state?.brandConfig;
   self.getCdnLogsConfig = () => state?.cdnLogsConfig;
+  self.getLlmoConfig = () => state?.llmo;
+  self.getLlmoDataFolder = () => state?.llmo?.dataFolder;
+  self.getLlmoBrand = () => state?.llmo?.brand;
+  self.getLlmoHumanQuestions = () => state?.llmo?.questions?.Human;
+  self.getLlmoAIQuestions = () => state?.llmo?.questions?.AI;
+  self.getLlmoUrlPatterns = () => state?.llmo?.urlPatterns;
 
   self.updateSlackConfig = (channel, workspace, invitedUserCount) => {
     state.slack = {
@@ -300,6 +355,95 @@ export const Config = (data = {}) => {
     };
   };
 
+  self.updateLlmoConfig = (dataFolder, brand, questions = {}, urlPatterns = undefined) => {
+    state.llmo = {
+      dataFolder,
+      brand,
+      questions,
+      urlPatterns,
+    };
+  };
+
+  self.updateLlmoDataFolder = (dataFolder) => {
+    state.llmo = state.llmo || {};
+    state.llmo.dataFolder = dataFolder;
+  };
+
+  self.updateLlmoBrand = (brand) => {
+    state.llmo = state.llmo || {};
+    state.llmo.brand = brand;
+  };
+
+  self.addLlmoHumanQuestions = (questions) => {
+    state.llmo = state.llmo || {};
+    state.llmo.questions = state.llmo.questions || {};
+    state.llmo.questions.Human = state.llmo.questions.Human || [];
+    state.llmo.questions.Human.push(...questions);
+  };
+
+  self.addLlmoAIQuestions = (questions) => {
+    state.llmo = state.llmo || {};
+    state.llmo.questions = state.llmo.questions || {};
+    state.llmo.questions.AI = state.llmo.questions.AI || [];
+    state.llmo.questions.AI.push(...questions);
+  };
+
+  self.removeLlmoQuestion = (key) => {
+    state.llmo = state.llmo || {};
+    state.llmo.questions = state.llmo.questions || {};
+    state.llmo.questions.Human = state.llmo.questions.Human || [];
+    state.llmo.questions.Human = state.llmo.questions.Human.filter(
+      (question) => question.key !== key,
+    );
+    state.llmo.questions.AI = state.llmo.questions.AI || [];
+    state.llmo.questions.AI = state.llmo.questions.AI.filter(
+      (question) => question.key !== key,
+    );
+  };
+
+  self.updateLlmoQuestion = (key, questionUpdate) => {
+    state.llmo = state.llmo || {};
+    state.llmo.questions = state.llmo.questions || {};
+    state.llmo.questions.Human = state.llmo.questions.Human || [];
+    state.llmo.questions.Human = state.llmo.questions.Human.map(
+      (question) => (question.key === key ? { ...question, ...questionUpdate, key } : question),
+    );
+    state.llmo.questions.AI = state.llmo.questions.AI || [];
+    state.llmo.questions.AI = state.llmo.questions.AI.map(
+      (question) => (question.key === key ? { ...question, ...questionUpdate, key } : question),
+    );
+  };
+
+  self.addLlmoUrlPatterns = (urlPatterns) => {
+    Joi.assert(urlPatterns, LLMO_URL_PATTERNS_SCHEMA, 'Invalid URL patterns');
+
+    state.llmo ??= {};
+    state.llmo.urlPatterns ??= [];
+    const byPattern = new Map(
+      state.llmo.urlPatterns.map((p) => [p.urlPattern, p]),
+    );
+    for (const p of urlPatterns) {
+      byPattern.set(p.urlPattern, p);
+    }
+
+    state.llmo.urlPatterns = [...byPattern.values()];
+  };
+
+  self.replaceLlmoUrlPatterns = (urlPatterns) => {
+    Joi.assert(urlPatterns, LLMO_URL_PATTERNS_SCHEMA, 'Invalid URL patterns');
+    state.llmo ??= {};
+    state.llmo.urlPatterns = urlPatterns;
+  };
+
+  self.removeLlmoUrlPattern = (urlPattern) => {
+    const urlPatterns = state.llmo?.urlPatterns;
+    if (!urlPatterns) return;
+
+    state.llmo.urlPatterns = urlPatterns.filter(
+      (pattern) => pattern.urlPattern !== urlPattern,
+    );
+  };
+
   self.updateImports = (imports) => {
     state.imports = imports;
   };
@@ -409,4 +553,5 @@ Config.toDynamoItem = (config) => ({
   fetchConfig: config.getFetchConfig(),
   brandConfig: config.getBrandConfig(),
   cdnLogsConfig: config.getCdnLogsConfig(),
+  llmo: config.getLlmoConfig(),
 });

package/src/models/site/index.d.ts

@@ -21,7 +21,7 @@ import type {
   Organization,
   SiteCandidate,
   SiteTopPage,
-} from '../index';
+} from '../index.js';
 
 export interface HlxConfig {
   hlxVersion: number; // helix (AEM Edge Delivery) major version
@@ -61,6 +61,25 @@ export interface ImportConfig {
   limit?: number;
 }
 
+export type WellKnownLmmoTag = 'market' | 'product' | 'topic';
+export type LmmoTag = `${WellKnownLmmoTag}:${string}` | string;
+
+export interface LlmoQuestion {
+  key: string;
+  question: string;
+  source?: string;
+  volume?: string;
+  importTime?: string;
+  keyword?: string;
+  url?: string;
+  tags?: LmmoTag[];
+}
+
+export interface LlmoUrlPattern {
+  urlPattern: string;
+  tags?: LmmoTag[];
+}
+
 export interface SiteConfig {
   state: {
     slack?: {
@@ -97,7 +116,17 @@ export interface SiteConfig {
       headers?: Record<string, string>;
       overrideBaseURL?: string;
     };
+    llmo?: {
+      dataFolder: string;
+      brand: string;
+      questions?: {
+        Human?: Array<LlmoQuestion>;
+        AI?: Array<LlmoQuestion>;
+      };
+      urlPatterns?: Array<LlmoUrlPattern>
+    };
   };
+  extractWellKnownTags(tags: Array<string>): Partial<Record<WellKnownLmmoTag, string>>;
   getSlackConfig(): { workspace?: string; channel?: string; invitedUserCount?: number };
   getImports(): ImportConfig[];
   getImportConfig(type: ImportType): ImportConfig | undefined;
@@ -116,6 +145,28 @@ export interface SiteConfig {
   getLatestMetrics(type: string):
     { pageViewsChange: number; ctrChange: number; projectedTrafficValue: number } | undefined;
   getFetchConfig(): { headers?: Record<string, string>, overrideBaseURL?: string } | undefined;
+  getLlmoConfig(): {
+    dataFolder: string;
+    brand: string;
+    questions?: { Human?: Array<LlmoQuestion>; AI?: Array<LlmoQuestion> };
+  } | undefined;
+  updateLlmoConfig(dataFolder: string, brand: string, questions?: {
+    Human?: Array<LlmoQuestion>;
+    AI?: Array<LlmoQuestion>;
+  }, urlPatterns?: Array<LlmoUrlPattern>): void;
+  updateLlmoDataFolder(dataFolder: string): void;
+  updateLlmoBrand(brand: string): void;
+  getLlmoDataFolder(): string | undefined;
+  getLlmoBrand(): string | undefined;
+  getLlmoHumanQuestions(): LlmoQuestion[] | undefined;
+  getLlmoAIQuestions(): LlmoQuestion[] | undefined;
+  getLlmoUrlPatterns(): Array<LlmoUrlPattern> | undefined;
+  addLlmoHumanQuestions(questions: LlmoQuestion[]): void;
+  addLlmoAIQuestions(questions: LlmoQuestion[]): void;
+  removeLlmoQuestion(key: string): void;
+  updateLlmoQuestion(key: string, questionUpdate: Partial<LlmoQuestion>): void;
+  addLlmoUrlPatterns(urlPatterns: Array<LlmoUrlPattern>): void;
+  removeLlmoUrlPattern(urlPattern: string): void;
 }
 
 export interface Site extends BaseModel {