@graphext/cuery 0.2.0

package/script/src/schemas/sources.schema.d.ts

@@ -0,0 +1,29 @@
+export interface Source {
+    title: string;
+    url: string;
+    domain: string;
+    cited?: boolean;
+    positions?: Array<number>;
+}
+/**
+ * Extended Source with brand mentions and links.
+ * Mentions refer to the source title, and links refer to the source URL.
+ */
+export interface EnrichedSource extends Source {
+    mentionedBrands: Array<string>;
+    mentionedCompetitors: Array<string>;
+    linkedBrand: string | null;
+    linkedCompetitor: string | null;
+}
+/**
+ * Source with category information from WEB_TAXONOMY.
+ */
+export interface CategorizedSource extends EnrichedSource {
+    category: string | null;
+    subcategory: string | null;
+}
+export interface SearchSource extends Source {
+    rank: number;
+    datePublished: string | null;
+}
+//# sourceMappingURL=sources.schema.d.ts.map

package/script/src/schemas/sources.schema.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"sources.schema.d.ts","sourceRoot":"","sources":["../../../src/src/schemas/sources.schema.ts"],"names":[],"mappings":"AAAA,MAAM,WAAW,MAAM;IACtB,KAAK,EAAE,MAAM,CAAC;IACd,GAAG,EAAE,MAAM,CAAC;IACZ,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,OAAO,CAAC;IAChB,SAAS,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;CAC1B;AAED;;;GAGG;AACH,MAAM,WAAW,cAAe,SAAQ,MAAM;IAC7C,eAAe,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IAC/B,oBAAoB,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC;IACpC,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;IAC3B,gBAAgB,EAAE,MAAM,GAAG,IAAI,CAAC;CAChC;AAED;;GAEG;AACH,MAAM,WAAW,iBAAkB,SAAQ,cAAc;IACxD,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;CAC3B;AAED,MAAM,WAAW,YAAa,SAAQ,MAAM;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,aAAa,EAAE,MAAM,GAAG,IAAI,CAAC;CAC7B"}

package/script/src/schemas/sources.schema.js

@@ -0,0 +1,2 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });

package/script/src/schemas/topics.schema.d.ts

@@ -0,0 +1,33 @@
+import { z } from '../../deps/jsr.io/@zod/zod/4.1.12/src/index.js';
+/**
+ * Topic schema - a topic with its subtopics.
+ */
+export declare const TopicSchema: z.ZodObject<{
+    topic: z.ZodString;
+    subtopics: z.ZodArray<z.ZodString>;
+}, z.core.$strip>;
+/**
+ * Taxonomy schema - contains all topics and their subtopics.
+ */
+export declare const TaxonomySchema: z.ZodObject<{
+    topics: z.ZodArray<z.ZodObject<{
+        topic: z.ZodString;
+        subtopics: z.ZodArray<z.ZodString>;
+    }, z.core.$strip>>;
+}, z.core.$strip>;
+/**
+ * A topic with its subtopics.
+ */
+export type TopicType = z.infer<typeof TopicSchema>;
+/**
+ * Taxonomy structure containing all topics and their subtopics.
+ */
+export type TaxonomyType = z.infer<typeof TaxonomySchema>;
+/**
+ * Topic label assigned to a text (topic + subtopic pair).
+ */
+export interface TopicLabel {
+    topic: string;
+    subtopic: string;
+}
+//# sourceMappingURL=topics.schema.d.ts.map

package/script/src/schemas/topics.schema.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"topics.schema.d.ts","sourceRoot":"","sources":["../../../src/src/schemas/topics.schema.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,gDAAgD,CAAC;AAEnE;;GAEG;AACH,eAAO,MAAM,WAAW;;;iBAGtB,CAAC;AAEH;;GAEG;AACH,eAAO,MAAM,cAAc;;;;;iBAEzB,CAAC;AAEH;;GAEG;AACH,MAAM,MAAM,SAAS,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,WAAW,CAAC,CAAC;AAEpD;;GAEG;AACH,MAAM,MAAM,YAAY,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,cAAc,CAAC,CAAC;AAE1D;;GAEG;AACH,MAAM,WAAW,UAAU;IAC1B,KAAK,EAAE,MAAM,CAAC;IACd,QAAQ,EAAE,MAAM,CAAC;CACjB"}

package/script/src/schemas/topics.schema.js

@@ -0,0 +1,17 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.TaxonomySchema = exports.TopicSchema = void 0;
+const index_js_1 = require("../../deps/jsr.io/@zod/zod/4.1.12/src/index.js");
+/**
+ * Topic schema - a topic with its subtopics.
+ */
+exports.TopicSchema = index_js_1.z.object({
+    topic: index_js_1.z.string(),
+    subtopics: index_js_1.z.array(index_js_1.z.string())
+});
+/**
+ * Taxonomy schema - contains all topics and their subtopics.
+ */
+exports.TaxonomySchema = index_js_1.z.object({
+    topics: index_js_1.z.array(exports.TopicSchema)
+});

package/script/src/tools/brands.d.ts

@@ -0,0 +1,55 @@
+import type { CompetitorSearchOptions, BrandSearchOptions, BrandListResponse } from '../schemas/brand.schema.js';
+import { type Brand, type FlaggedBrand, type Product } from '../schemas/brand.schema.js';
+import type { Entity } from './entities.js';
+interface BrandContextOptions {
+    brand?: string | Array<string>;
+    brandDomain?: string | Array<string>;
+    sector?: string | null;
+    market?: string | null;
+    briefing?: string | null;
+}
+export declare function buildBrandContext({ brand, brandDomain, sector, market, briefing }: BrandContextOptions): string;
+export type { BrandContextOptions };
+/**
+ * Generate seed keywords for a single portfolio item using an LLM.
+ */
+export declare function generatePortfolioKeywords(product: Product, sector: string, market: string, language?: string, model?: string): Promise<Array<string>>;
+/**
+ * Generate seed keywords for all portfolio items in a brand concurrently,
+ * updating the keywordSeeds field in-place for each product.
+ */
+export declare function enrichBrandPortfolioWithKeywords(brand: Brand, language?: string, model?: string, maxConcurrency?: number, sector?: string | null, market?: string | null): Promise<Brand>;
+export declare function generateCompetitorsInfo({ brand, brandDomain, sector, market, briefing, strict, instructions, brands, language, userLanguage, model, useSearch, countryISOCode, contextSize, reasoningEffort }: CompetitorSearchOptions): Promise<BrandListResponse>;
+export declare function generateBrandInfo({ brand, brandDomain, language, userLanguage, sector, market, briefing, model, useSearch, countryISOCode, contextSize, reasoningEffort }: BrandSearchOptions): Promise<Brand>;
+export declare function concatBrands(ownBrands: Array<Brand>, competitors: Array<Brand>): Array<FlaggedBrand>;
+/**
+ * Normalizes a brand name for comparison.
+ * - Converts to lowercase
+ * - Normalizes & to " and " (with spaces to separate words)
+ * - Removes accents/diacritics
+ * - Expands CamelCase to spaces (e.g., "AcmeCorp" -> "acme corp")
+ * - Expands embedded "and" (e.g., "benandjerrys" -> "ben and jerrys")
+ * - Separates numbers from letters (e.g., "7eleven" -> "7 eleven")
+ * - Converts hyphens to spaces (e.g., "Coca-Cola" -> "coca cola")
+ * - Removes apostrophes (e.g., "Jerry's" -> "jerrys")
+ * - Removes periods/dots (e.g., "Dr. Pepper" -> "dr pepper")
+ */
+export declare function normalizeBrandName(name: string): string;
+/**
+ * Creates an ultra-normalized key for brand matching (removes all spaces and non-alphanumerics).
+ * Used to match brands like "Acme&Co" with entities like "acmeco",
+ * or "Ben & Jerry's" with "benandjerrys".
+ */
+export declare function createBrandMatchKey(brandName: string): string;
+/**
+ * Creates an ultra-normalized version of text for key-based matching.
+ * Removes all spaces and non-alphanumerics for exact key comparison.
+ */
+/**
+ * Checks which brands are mentioned in a text.
+ * Returns shortNames(!) of brands in order they're mentioned.
+ */
+export declare function rankBrandsInText(text: string, brands: Array<FlaggedBrand>, entities?: Array<Entity>): Array<string>;
+export declare function rankBrandsInTexts(texts: Array<string>, brands: Array<FlaggedBrand>, entities?: Array<Array<Entity>>): Promise<Array<Array<string>>>;
+export type * from '../schemas/brand.schema.js';
+//# sourceMappingURL=brands.d.ts.map

package/script/src/tools/brands.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"brands.d.ts","sourceRoot":"","sources":["../../../src/src/tools/brands.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,uBAAuB,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAC;AACjH,OAAO,EAAgC,KAAK,KAAK,EAAE,KAAK,YAAY,EAAE,KAAK,OAAO,EAAE,MAAM,4BAA4B,CAAC;AAIvH,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,eAAe,CAAC;AAE5C,UAAU,mBAAmB;IAC5B,KAAK,CAAC,EAAE,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC;IAC/B,WAAW,CAAC,EAAE,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC;IACrC,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACvB,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACvB,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CACzB;AAED,wBAAgB,iBAAiB,CAAC,EACjC,KAAK,EACL,WAAW,EACX,MAAM,EACN,MAAM,EACN,QAAQ,EACR,EAAE,mBAAmB,GAAG,MAAM,CAuB9B;AAED,YAAY,EAAE,mBAAmB,EAAE,CAAC;AAqHpC;;GAEG;AACH,wBAAsB,yBAAyB,CAC9C,OAAO,EAAE,OAAO,EAChB,MAAM,EAAE,MAAM,EACd,MAAM,EAAE,MAAM,EACd,QAAQ,GAAE,MAAa,EACvB,KAAK,GAAE,MAAkB,GACvB,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAaxB;AAED;;;GAGG;AACH,wBAAsB,gCAAgC,CACrD,KAAK,EAAE,KAAK,EACZ,QAAQ,GAAE,MAAa,EACvB,KAAK,GAAE,MAAkB,EACzB,cAAc,GAAE,MAAY,EAC5B,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,EACtB,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,GACpB,OAAO,CAAC,KAAK,CAAC,CAoBhB;AAGD,wBAAsB,uBAAuB,CAAC,EAC7C,KAAK,EACL,WAAW,EACX,MAAa,EACb,MAAa,EACb,QAAQ,EACR,MAAa,EACb,YAAiB,EACjB,MAAa,EACb,QAAe,EACf,YAAmB,EACnB,KAA4B,EAC5B,SAAgB,EAChB,cAAqB,EACrB,WAAsB,EACtB,eAAuB,EACvB,EAAE,uBAAuB,GAAG,OAAO,CAAC,iBAAiB,CAAC,CAoDtD;AAED,wBAAsB,iBAAiB,CAAC,EACvC,KAAK,EACL,WAAW,EACX,QAAe,EACf,YAAmB,EACnB,MAAM,EACN,MAAM,EACN,QAAQ,EACR,KAA4B,EAC5B,SAAgB,EAChB,cAAqB,EACrB,WAAsB,EACtB,eAAuB,EACvB,EAAE,kBAAkB,GAAG,OAAO,CAAC,KAAK,CAAC,CAyCrC;AAED,wBAAgB,YAAY,CAC3B,SAAS,EAAE,KAAK,CAAC,KAAK,CAAC,EACvB,WAAW,EAAE,KAAK,CAAC,KAAK,CAAC,GACvB,KAAK,CAAC,YAAY,CAAC,CAYrB;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAwBvD;AAED;;;;GAIG;AACH,wBAAgB,mBAAmB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAI7D;AAuED;;;GAGG;AAQH;;;GAGG;AACH,wBAAgB,gBAAgB,CAC/B,IAAI,EAAE,MAAM,EACZ,MAAM,EAAE,KAAK,CAAC,YAAY,CAAC,EAC3B,QAAQ,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,GACtB,KAAK,CAAC,MAAM,CAAC,CA6Ff;AAED,wBAAsB,iBAAiB,CACtC,KAAK,EAAE,KAAK,CAAC,MAAM,CAAC,EACpB,MAAM,EAAE,KAAK,CAAC,YAAY,CAAC,EAC3B,QAAQ,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,GAC7B,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAE/B;AAED,mBAAmB,4BAA4B,CAAC"}

package/script/src/tools/brands.js

@@ -0,0 +1,470 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.buildBrandContext = buildBrandContext;
+exports.generatePortfolioKeywords = generatePortfolioKeywords;
+exports.enrichBrandPortfolioWithKeywords = enrichBrandPortfolioWithKeywords;
+exports.generateCompetitorsInfo = generateCompetitorsInfo;
+exports.generateBrandInfo = generateBrandInfo;
+exports.concatBrands = concatBrands;
+exports.normalizeBrandName = normalizeBrandName;
+exports.createBrandMatchKey = createBrandMatchKey;
+exports.rankBrandsInText = rankBrandsInText;
+exports.rankBrandsInTexts = rankBrandsInTexts;
+const index_js_1 = require("../../deps/jsr.io/@zod/zod/4.1.12/src/index.js");
+const async_js_1 = require("../async.js");
+const openai_js_1 = require("../openai.js");
+const brand_schema_js_1 = require("../schemas/brand.schema.js");
+const search_js_1 = require("./search.js");
+const utils_js_1 = require("../utils.js");
+const urls_js_1 = require("../urls.js");
+function buildBrandContext({ brand, brandDomain, sector, market, briefing }) {
+    let brandContext;
+    if (brandDomain) {
+        const domainText = Array.isArray(brandDomain) ? brandDomain.join(', ') : brandDomain;
+        if (brand) {
+            const nameText = Array.isArray(brand) ? brand.join(', ') : brand;
+            brandContext = ` "${nameText}" (domain: ${domainText})`;
+        }
+        else {
+            brandContext = ` with domain "${domainText}"`;
+        }
+    }
+    else {
+        brandContext = ` "${brand}"`;
+    }
+    if (sector) {
+        brandContext += ` in the ${sector} sector`;
+    }
+    if (market) {
+        brandContext += ` operating in ${market}`;
+    }
+    if (briefing) {
+        brandContext += ` Briefing: ${briefing}`;
+    }
+    return brandContext;
+}
+const DEFAULT_SEARCH_MODEL = 'gpt-4.1';
+const PORTFOLIO_KEYWORDS_PROMPT = (0, utils_js_1.dedent)(`
+Generate 5 Google search keywords that real users would type when researching this product or service.
+
+Product/Service: {product_name}
+Category: {category}
+Sector: {sector}
+Market: {market}
+
+Context:
+- The product/service is part of a brand in the {sector} sector operating in {market}
+- Generate what people ACTUALLY search, NOT literal product names or brand mentions
+- Keywords should be broad (1-3 words) suitable for Google Keyword Planner seed expansion
+- Focus on generic search terms that potential customers would use
+- DO NOT include brand names in the keywords
+
+Examples of GOOD keywords for different contexts:
+- Running shoe product → "running shoes", "athletic footwear", "trail running"
+- CRM software → "customer management", "sales automation", "crm tools"
+- Online course → "learn programming", "coding tutorials", "web development"
+
+Requirements:
+- Natural language people actually type in search bars
+- 5 realistic search queries
+- Broad enough for keyword expansion (1-3 words typically)
+- Output in {language}
+- No brand names or specific product names
+
+What would someone in {market} naturally search when looking for products/services like {product_name}?
+`);
+const PortfolioKeywordsSchema = index_js_1.z.object({
+    keywords: index_js_1.z.array(index_js_1.z.string()).describe('List of 5 seed keywords for the product or service.')
+});
+const COMPETITORS_PROMPT = (0, utils_js_1.dedent)(`
+You are an expert in market analysis and competitive intelligence. Given the brand(s)
+information in the below section, find and return an exhaustive list of competitors. Consider
+competitors to be brands that offer similar products or services and target the same or similar
+customer segments. {strictness_clause}
+
+The output should be a JSON array with one object for each competitor. Each competitor object
+should have the following fields:
+
+- name: Name of the competitor brand.
+- shortName: A short, common/canonical name of the competitor brand, if different from the official name.
+  E.g. "Tesla" instead of "Tesla, Inc.", or "Peugeot" instead of "Automobiles Peugeot". If not applicable,
+  set same as name.
+- description: A brief description of the competitor brand's main activity.
+- domain: The official website of the competitor brand, if available.
+- sectors: A list of main industrial sectors the competitor brand operates in.
+- markets: A list of main geographical markets the competitor brand operates in.
+- marketPosition: The market position of the competitor brand, which can be one of the following values:
+  "leader", "challenger", "niche", "follower".
+- portfolio: A list of products or services offered by the competitor brand. Each product or service should
+  have a name and a category.
+- favicon: URL of the brand's favicon, if available.
+
+It's more important to be comprehensive than selective, so include more competitors rather than less.
+Try to order them by relevance, with the most direct and bigger competitors first.
+
+Return the answer (field values) in the language "{language}".
+
+Also provide an "explanation" field with a brief summary (2-3 sentences) explaining the reasoning behind
+the competitors you identified: what makes them relevant competitors, how they relate to the original brand,
+and what competitive landscape they represent.
+
+{instructions}
+
+# Brand(s) information
+
+{context}
+
+{currentCompetitorsInfo}
+`);
+const STRICT_CLAUSE = 'Only consider as competitors those that do NOT belong to the same parent company as the original brand(s) (e.g. Fanta is NOT a competitor of Coca-Cola in this sense).';
+const NON_STRICT_CLAUSE = 'Consider all relevant competitors, including those from the same parent company (e.g. Fanta IS a competitor of Coca-Cola in this sense).';
+const CURRENT_DATA_CLAUSE = (0, utils_js_1.dedent)(`
+# Current Competitors Information
+
+The following is the current information the user has about competitors for the brand(s).
+
+{currentData}
+
+IMPORTANT: Unless the instructions explicitly specify to modify, replace, add, or remove competitors
+from this list, you MUST return at least all the same competitors that are already present in the
+current data. If no modification instructions are provided, maintain the existing list and only add
+new competitors if they are relevant.
+`);
+const BRAND_PROMPT = (0, utils_js_1.dedent)(`
+You are an expert in market analysis and competitive intelligence investigating the brand{brand_context}.
+Find out the following information about this brand:
+
+- name: Brand/company name
+- shortName: A short, common/canonical name of the brand, if different from the official name.
+  E.g. "Tesla" instead of "Tesla, Inc.", or "Peugeot" instead of "Automobiles Peugeot". If not applicable,
+  set same as name.
+- description: A brief description of the brand's main activity.
+- domain: The official website of the brand, if available.
+- sectors: A list of specific industrial sectors the brand operates in, ordered by relevance. Be precise and concrete rather than generic - for example, instead of "education", use "professional training for unemployed", "corporate learning and development", or "online coding bootcamps". Adapt the specificity level to match the brand's actual focus and niche.
+- markets: A list of main geographical markets the brand operates in.
+- marketPosition: The market position of the brand, which can be one of the following values:
+  "leader", "challenger", "niche", "follower".
+- portfolio: A list of products or services offered by the competitor brand. Each product or service should
+  have a name and a category.
+- favicon: URL of the brand's favicon, if available.
+
+Return the answer as a JSON object with field values in the language "{language}".
+`);
+/**
+ * Generate seed keywords for a single portfolio item using an LLM.
+ */
+async function generatePortfolioKeywords(product, sector, market, language = 'en', model = 'gpt-4.1') {
+    const prompt = PORTFOLIO_KEYWORDS_PROMPT
+        .replace('{product_name}', product.name)
+        .replace('{category}', product.category || 'general')
+        .replaceAll('{sector}', sector)
+        .replaceAll('{market}', market)
+        .replaceAll('{language}', language);
+    const { parsed } = await (0, openai_js_1.askOpenAISafe)(prompt, model, PortfolioKeywordsSchema);
+    if (!parsed) {
+        throw new Error('Failed to parse portfolio keywords from OpenAI response');
+    }
+    return parsed.keywords;
+}
+/**
+ * Generate seed keywords for all portfolio items in a brand concurrently,
+ * updating the keywordSeeds field in-place for each product.
+ */
+async function enrichBrandPortfolioWithKeywords(brand, language = 'en', model = 'gpt-4.1', maxConcurrency = 100, sector, market) {
+    if (!brand.portfolio || brand.portfolio.length === 0) {
+        return brand;
+    }
+    sector = sector || brand.sectors[0] || 'general';
+    market = market || brand.markets[0] || 'global';
+    await (0, async_js_1.mapParallel)(brand.portfolio, maxConcurrency, async (product) => {
+        const keywords = await generatePortfolioKeywords(product, sector, market, language, model);
+        product.keywordSeeds = keywords;
+    });
+    return brand;
+}
+async function generateCompetitorsInfo({ brand, brandDomain, sector = null, market = null, briefing, strict = true, instructions = '', brands = null, language = 'en', userLanguage = null, model = DEFAULT_SEARCH_MODEL, useSearch = true, countryISOCode = null, contextSize = 'medium', reasoningEffort = 'low' }) {
+    if (!brand && !brandDomain) {
+        throw new Error('Either brand or brandDomain must be provided');
+    }
+    const strictnessClause = strict ? STRICT_CLAUSE : NON_STRICT_CLAUSE;
+    const context = buildBrandContext({
+        brand,
+        brandDomain,
+        sector,
+        market,
+        briefing
+    });
+    const currentData = brands ? CURRENT_DATA_CLAUSE.replace('{currentData}', JSON.stringify(brands, null, 2)) : '';
+    const formattedPrompt = COMPETITORS_PROMPT
+        .replace('{strictness_clause}', strictnessClause)
+        .replace('{instructions}', instructions || '')
+        .replace('{language}', userLanguage ?? language)
+        .replace('{context}', context)
+        .replace('{currentCompetitorsInfo}', currentData);
+    const response = await (0, search_js_1.searchWithFormat)({
+        prompt: formattedPrompt,
+        model: model,
+        useSearch: useSearch,
+        responseSchema: brand_schema_js_1.BrandListSchema,
+        countryISOCode,
+        contextSize,
+        reasoningEffort
+    });
+    // Make sure these are clean domains
+    response.brands.map(brand => {
+        if (brand.domain) {
+            brand.domain = (0, urls_js_1.extractDomain)(brand.domain);
+        }
+    });
+    const enrichedBrands = await (0, async_js_1.mapParallel)(response.brands, 100, async (brand) => enrichBrandPortfolioWithKeywords(brand, language, model, undefined, sector, market));
+    return {
+        brands: enrichedBrands,
+        explanation: response.explanation
+    };
+}
+async function generateBrandInfo({ brand, brandDomain, language = 'en', userLanguage = null, sector, market, briefing, model = DEFAULT_SEARCH_MODEL, useSearch = true, countryISOCode = null, contextSize = 'medium', reasoningEffort = 'low' }) {
+    if (!brand && !brandDomain) {
+        throw new Error('Either brand or brandDomain must be provided');
+    }
+    const brandContext = buildBrandContext({
+        brand,
+        brandDomain,
+        sector,
+        market,
+        briefing
+    });
+    const prompt = BRAND_PROMPT
+        .replace('{brand_context}', brandContext)
+        .replace('{language}', userLanguage ?? language);
+    const brandInfo = await (0, search_js_1.searchWithFormat)({
+        prompt,
+        model,
+        responseSchema: brand_schema_js_1.BrandSchema,
+        countryISOCode: countryISOCode,
+        contextSize: contextSize,
+        reasoningEffort: reasoningEffort,
+        useSearch
+    });
+    // Make sure this is a clean domain
+    brandInfo.domain = (0, urls_js_1.extractDomain)(brandInfo.domain);
+    const enrichedBrand = await enrichBrandPortfolioWithKeywords(brandInfo, language, model, undefined, sector, market);
+    return enrichedBrand;
+}
+function concatBrands(ownBrands, competitors) {
+    const ownBrandsWithFlag = ownBrands.map((brand) => ({
+        ...brand,
+        isCompetitor: false
+    }));
+    const competitorsWithFlag = competitors.map((brand) => ({
+        ...brand,
+        isCompetitor: true
+    }));
+    return [...ownBrandsWithFlag, ...competitorsWithFlag];
+}
+/**
+ * Normalizes a brand name for comparison.
+ * - Converts to lowercase
+ * - Normalizes & to " and " (with spaces to separate words)
+ * - Removes accents/diacritics
+ * - Expands CamelCase to spaces (e.g., "AcmeCorp" -> "acme corp")
+ * - Expands embedded "and" (e.g., "benandjerrys" -> "ben and jerrys")
+ * - Separates numbers from letters (e.g., "7eleven" -> "7 eleven")
+ * - Converts hyphens to spaces (e.g., "Coca-Cola" -> "coca cola")
+ * - Removes apostrophes (e.g., "Jerry's" -> "jerrys")
+ * - Removes periods/dots (e.g., "Dr. Pepper" -> "dr pepper")
+ */
+function normalizeBrandName(name) {
+    return name
+        // Remove apostrophes before other processing (e.g., "Jerry's" -> "Jerrys")
+        .replace(/'/g, '')
+        // Remove periods/dots (e.g., "Dr. Pepper" -> "Dr Pepper")
+        .replace(/\./g, '')
+        // Normalize & to " and " BEFORE CamelCase expansion to ensure proper word separation
+        .replace(/&/g, ' and ')
+        // Expand CamelCase to spaces (insert space before uppercase letters)
+        .replace(/([a-z])([A-Z])/g, '$1 $2')
+        .toLowerCase()
+        .trim()
+        // Convert hyphens to spaces (e.g., "coca-cola" -> "coca cola")
+        .replace(/-/g, ' ')
+        // Expand embedded "and" (e.g., "benandjerrys" -> "ben and jerrys")
+        .replace(/([a-z])and([a-z])/g, '$1 and $2')
+        // Separate numbers from letters (e.g., "7eleven" -> "7 eleven")
+        .replace(/([0-9])([a-z])/g, '$1 $2')
+        .replace(/([a-z])([0-9])/g, '$1 $2')
+        // Normalize accents/diacritics
+        .normalize('NFD')
+        .replace(/[\u0300-\u036f]/g, '')
+        // Normalize multiple spaces to single space
+        .replace(/\s+/g, ' ');
+}
+/**
+ * Creates an ultra-normalized key for brand matching (removes all spaces and non-alphanumerics).
+ * Used to match brands like "Acme&Co" with entities like "acmeco",
+ * or "Ben & Jerry's" with "benandjerrys".
+ */
+function createBrandMatchKey(brandName) {
+    return normalizeBrandName(brandName)
+        .replace(/\s+/g, '')
+        .replace(/[^a-z0-9]/g, ''); // drop any other non-alphanumerics
+}
+/**
+ * Creates a regex pattern that matches various text representations of a brand name.
+ * Handles: CamelCase, spaces, &/and variations, hyphens, apostrophes, accents.
+ */
+function createBrandMatchPattern(brandName) {
+    // First normalize the brand name to get a canonical form
+    const normalized = normalizeBrandName(brandName);
+    // Split into words
+    const words = normalized.split(' ').filter(w => w.length > 0);
+    // Build pattern where each word can have flexible separators between them
+    const wordPatterns = words.map(word => {
+        if (word === 'and') {
+            return '(?:&|and)?';
+        }
+        // For other words, escape special chars and allow accent variations
+        const escaped = word.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+        // Allow accented versions of vowels
+        return escaped
+            .replace(/a/g, '[aáàäâ]')
+            .replace(/e/g, '[eéèëê]')
+            .replace(/i/g, '[iíìïî]')
+            .replace(/o/g, '[oóòöô]')
+            .replace(/u/g, '[uúùüû]')
+            .replace(/n/g, '[nñ]');
+    });
+    // Flexible separator: allows space, hyphen, apostrophe, ampersand, "and", or nothing (CamelCase)
+    // The separator is optional to handle cases like "BenJerrys" matching "Ben Jerry's"
+    const flexibleSeparator = "(?:[\\s\\-'&]|and)*";
+    const pattern = wordPatterns.join(flexibleSeparator);
+    // Word boundary handling:
+    // - Start: beginning of string OR non-alphanumeric character
+    // - End: end of string OR non-alphanumeric character
+    // This allows matching "7-Eleven" in "I shop at 7-Eleven daily"
+    return new RegExp(`(?:^|[^a-zA-Z0-9])${pattern}(?:$|[^a-zA-Z0-9])|^${pattern}$`, 'i');
+}
+/**
+ * Creates a normalized version of text for matching purposes.
+ * Expands CamelCase, separates numbers from letters, and expands common connectors.
+ */
+function normalizeTextForMatching(text) {
+    return text
+        // Remove apostrophes
+        .replace(/'/g, '')
+        // Remove periods/dots (Dr. Pepper → Dr Pepper)
+        .replace(/\./g, '')
+        // Expand CamelCase (lowercase followed by uppercase)
+        .replace(/([a-z])([A-Z])/g, '$1 $2')
+        // Convert hyphens to spaces (Coca-Cola → Coca Cola)
+        .replace(/-/g, ' ')
+        // Separate numbers from letters (7eleven → 7 eleven, eleven7 → eleven 7)
+        .replace(/([0-9])([a-zA-Z])/g, '$1 $2')
+        .replace(/([a-zA-Z])([0-9])/g, '$1 $2')
+        // Expand "and" when embedded in words (benandjerrys → ben and jerrys)
+        // This handles cases where "and" is glued to other words
+        .replace(/([a-zA-Z])and([a-zA-Z])/gi, '$1 and $2')
+        // Normalize & to " and "
+        .replace(/&/g, ' and ')
+        // Normalize accents
+        .normalize('NFD').replace(/[\u0300-\u036f]/g, '')
+        // Normalize spaces
+        .replace(/\s+/g, ' ');
+}
+/**
+ * Creates an ultra-normalized version of text for key-based matching.
+ * Removes all spaces and non-alphanumerics for exact key comparison.
+ */
+// function createTextMatchKey(text: string): string {
+// 	return normalizeTextForMatching(text)
+// 		.toLowerCase()
+// 		.replace(/\s+/g, '')
+// 		.replace(/[^a-z0-9]/g, '');
+// }
+/**
+ * Checks which brands are mentioned in a text.
+ * Returns shortNames(!) of brands in order they're mentioned.
+ */
+function rankBrandsInText(text, brands, entities) {
+    const mentionedBrands = [];
+    // Create a normalized version of text for pattern matching (expands CamelCase, etc.)
+    const normalizedText = normalizeTextForMatching(text);
+    const brandEntities = entities?.filter(entity => entity.type.toLowerCase() === 'brand') ?? [];
+    // Build a map from ultra-normalized FlaggedBrand name to the entity text variation (if any)
+    // Using createBrandMatchKey to handle cases like "acmeco" matching "Acme&Co"
+    const entityTextByNormalizedBrand = new Map();
+    for (const entity of brandEntities) {
+        entityTextByNormalizedBrand.set(createBrandMatchKey(entity.name), entity.name);
+    }
+    // Set of ultra-normalized entity names that match a FlaggedBrand (to exclude from separate processing)
+    const matchedEntityNames = new Set();
+    // Helper function to search in both original and normalized text
+    const findPatternPosition = (pattern) => {
+        const originalMatch = text.match(pattern);
+        if (originalMatch != null) {
+            return originalMatch.index;
+        }
+        const normalizedMatch = normalizedText.match(pattern);
+        if (normalizedMatch != null) {
+            return normalizedMatch.index;
+        }
+        return -1;
+    };
+    // Process FlaggedBrands, using entity text as an additional search variant when available
+    for (const brand of brands) {
+        let earliestPosition = Infinity;
+        const normalizedShortName = createBrandMatchKey(brand.shortName);
+        // Check if there's a matching entity for this brand (using ultra-normalized comparison)
+        const entityText = entityTextByNormalizedBrand.get(normalizedShortName);
+        if (entityText != null) {
+            matchedEntityNames.add(createBrandMatchKey(entityText));
+            // Use flexible pattern matching to find the entity in text (handles CamelCase, &/and, etc.)
+            const entityPattern = createBrandMatchPattern(entityText);
+            const position = findPatternPosition(entityPattern);
+            if (position !== -1) {
+                earliestPosition = Math.min(earliestPosition, position);
+            }
+        }
+        // Check for shortName using flexible pattern matching
+        if (brand.shortName != null && brand.shortName.trim() !== '') {
+            const brandPattern = createBrandMatchPattern(brand.shortName);
+            const position = findPatternPosition(brandPattern);
+            if (position !== -1) {
+                earliestPosition = Math.min(earliestPosition, position);
+            }
+        }
+        // Check for domain occurrence
+        if (brand.domain != null && brand.domain.trim() !== '') {
+            const domainIndex = text.toLowerCase().indexOf(brand.domain.toLowerCase());
+            if (domainIndex !== -1) {
+                earliestPosition = Math.min(earliestPosition, domainIndex);
+            }
+        }
+        if (earliestPosition !== Infinity) {
+            mentionedBrands.push({
+                name: brand.shortName,
+                position: earliestPosition
+            });
+        }
+    }
+    // Process entity brands that don't match any FlaggedBrand
+    for (const entity of brandEntities) {
+        if (matchedEntityNames.has(createBrandMatchKey(entity.name))) {
+            continue;
+        }
+        // Use flexible pattern matching for unmatched entities too
+        const entityPattern = createBrandMatchPattern(entity.name);
+        const position = findPatternPosition(entityPattern);
+        if (position !== -1) {
+            mentionedBrands.push({
+                name: entity.name,
+                position: position
+            });
+        }
+    }
+    return mentionedBrands
+        .sort((a, b) => a.position - b.position)
+        .map(item => item.name);
+}
+async function rankBrandsInTexts(texts, brands, entities) {
+    return texts.map((text, index) => rankBrandsInText(text, brands, entities?.[index]));
+}

package/script/src/tools/classifier.d.ts

@@ -0,0 +1,17 @@
+/**
+ * Classifies a single data record into one of the provided categories using an LLM call.
+ */
+export declare function classify(record: Record<string, unknown> | null, labels: Record<string, string>, instructions?: string, model?: string): Promise<string | null>;
+/**
+ * Classifies multiple data records concurrently while preserving order.
+ */
+export declare function classifyBatch(records: Array<Record<string, unknown> | null>, labels: Record<string, string>, instructions?: string, model?: string, maxConcurrency?: number): Promise<Array<string | null>>;
+/**
+ * Assigns one or more labels to a single data record using an LLM call.
+ */
+export declare function label(record: Record<string, unknown> | null, labels: Record<string, string>, instructions?: string, model?: string): Promise<Array<string> | null>;
+/**
+ * Assigns labels to multiple data records concurrently while preserving order.
+ */
+export declare function labelBatch(records: Array<Record<string, unknown> | null>, labels: Record<string, string>, instructions?: string, model?: string, maxConcurrency?: number): Promise<Array<Array<string> | null>>;
+//# sourceMappingURL=classifier.d.ts.map

package/script/src/tools/classifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"classifier.d.ts","sourceRoot":"","sources":["../../../src/src/tools/classifier.ts"],"names":[],"mappings":"AAsGA;;GAEG;AACH,wBAAsB,QAAQ,CAC7B,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,EACtC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAC9B,YAAY,GAAE,MAAW,EACzB,KAAK,GAAE,MAAuB,GAC5B,OAAO,CAAC,MAAM,GAAG,IAAI,CAAC,CAiBxB;AAED;;GAEG;AACH,wBAAgB,aAAa,CAC5B,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC,EAC9C,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAC9B,YAAY,GAAE,MAAW,EACzB,KAAK,GAAE,MAAuB,EAC9B,cAAc,GAAE,MAAY,GAC1B,OAAO,CAAC,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC,CAAC,CAM/B;AAED;;GAEG;AACH,wBAAsB,KAAK,CAC1B,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,EACtC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAC9B,YAAY,GAAE,MAAW,EACzB,KAAK,GAAE,MAAuB,GAC5B,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,CAiB/B;AAED;;GAEG;AACH,wBAAgB,UAAU,CACzB,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC,EAC9C,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,EAC9B,YAAY,GAAE,MAAW,EACzB,KAAK,GAAE,MAAuB,EAC9B,cAAc,GAAE,MAAY,GAC1B,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,CAAC,CAMtC"}