@measureoneinc/savant 1.0.0

package/dist/services/domain.service.js

@@ -0,0 +1,244 @@
+"use strict";
+/**
+ * @fileoverview Domain Service
+ *
+ * Service for managing domains, seeds, and domain validation.
+ * Ensures that seeds exist for domains before allowing learning creation.
+ *
+ * @author MeasureOne, Inc.
+ * @version 1.0.0
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.DomainService = void 0;
+exports.get_domain_service = get_domain_service;
+const logger_1 = require("../lib/utils/logger");
+const sequelize_service_1 = require("../lib/sequelize-service");
+const models_1 = require("../models");
+const sequelize_1 = require("sequelize");
+const genai_1 = require("@google/genai");
+const domain_identifier_agent_1 = require("../ai/agents/domain-identifier.agent");
+const config_1 = require("../config");
+const logger = (0, logger_1.create_default_logger)("domain-service");
+class DomainService {
+    constructor() {
+        this.sequelize = (0, sequelize_service_1.get_sequelize_service)();
+    }
+    /**
+     * Get all unique domains from Embeddings table (ALL types: SEED and LEARNING)
+     * Returns all distinct domain values that exist in the system
+     *
+     * @returns Promise resolving to array of unique domain strings
+     */
+    async get_unique_domains() {
+        try {
+            // Get unique domains from ALL embedding types (SEED and LEARNING)
+            const domains = await models_1.EmbeddingModel.findAll({
+                attributes: [
+                    [this.sequelize.fn('DISTINCT', this.sequelize.col('domain')), 'domain']
+                ],
+                where: {
+                    domain: {
+                        [sequelize_1.Op.ne]: null
+                    }
+                },
+                raw: true
+            });
+            const unique_domains = domains
+                .map((d) => d.domain)
+                .filter((d) => d !== null && d.trim().length > 0);
+            logger.debug('Retrieved unique domains from all embeddings', { count: unique_domains.length, domains: unique_domains });
+            return unique_domains;
+        }
+        catch (error) {
+            logger.error('Failed to get unique domains', { error });
+            return [];
+        }
+    }
+    /**
+     * Check if a seed exists for a given domain
+     * Seeds are SEED type embeddings with a domain field
+     *
+     * @param domain - The domain to check
+     * @returns Promise resolving to true if seed exists, false otherwise
+     */
+    async seed_exists_for_domain(domain) {
+        try {
+            if (!domain || domain.trim().length === 0) {
+                return false;
+            }
+            // Check if any SEED type embedding exists for this domain
+            const seed_count = await models_1.EmbeddingModel.count({
+                where: {
+                    type: models_1.EmbeddingType.SEED,
+                    domain: domain,
+                }
+            });
+            const exists = seed_count > 0;
+            logger.debug('Seed existence check', { domain, exists, seed_count });
+            return exists;
+        }
+        catch (error) {
+            logger.error('Failed to check seed existence', { error, domain });
+            return false;
+        }
+    }
+    /**
+     * Get seed pattern for a domain
+     * Returns the pattern/format specification from the seed
+     *
+     * @param domain - The domain to get seed pattern for
+     * @returns Promise resolving to seed pattern text or null if not found
+     */
+    async get_seed_pattern(domain) {
+        try {
+            if (!domain || domain.trim().length === 0) {
+                return null;
+            }
+            // Get ALL SEED type embeddings for this domain
+            const seed_embeddings = await models_1.EmbeddingModel.findAll({
+                where: {
+                    type: models_1.EmbeddingType.SEED,
+                    domain: domain,
+                },
+                attributes: ['knowledge_id', 'knowledge_revision_id', 'text', 'chunk_index', 'created_at'],
+                order: [['created_at', 'DESC'], ['chunk_index', 'ASC']], // Most recent first, then by chunk order
+            });
+            if (!seed_embeddings || seed_embeddings.length === 0) {
+                logger.debug('No seed embeddings found for domain', { domain });
+                return null;
+            }
+            logger.info('Found seed embeddings for domain', {
+                domain,
+                total_embeddings: seed_embeddings.length,
+                unique_knowledge_ids: [...new Set(seed_embeddings.map(e => e.knowledge_id))].length
+            });
+            // Group embeddings by knowledge_id to combine chunks from the same Knowledge
+            const knowledge_groups = new Map();
+            for (const embedding of seed_embeddings) {
+                const knowledge_id = embedding.knowledge_id;
+                if (!knowledge_groups.has(knowledge_id)) {
+                    knowledge_groups.set(knowledge_id, []);
+                }
+                knowledge_groups.get(knowledge_id).push(embedding);
+            }
+            // Try each knowledge's combined text to find LEARNING FORMAT
+            // Start with the most recent knowledge (first in the sorted list)
+            for (const [knowledge_id, embeddings] of knowledge_groups.entries()) {
+                // Get the most recent revision for this knowledge
+                const most_recent_revision_id = embeddings[0]?.knowledge_revision_id;
+                // Get all chunks for the most recent revision, sorted by chunk_index
+                const revision_chunks = embeddings
+                    .filter(e => e.knowledge_revision_id === most_recent_revision_id)
+                    .sort((a, b) => (a.chunk_index || 0) - (b.chunk_index || 0));
+                // Combine chunks in order
+                const combined_text = revision_chunks
+                    .map(e => e.text)
+                    .join('\n\n');
+                logger.debug('Combined seed chunks for knowledge', {
+                    domain,
+                    knowledge_id,
+                    revision_id: most_recent_revision_id,
+                    chunk_count: revision_chunks.length,
+                    combined_length: combined_text.length
+                });
+                // Extract the LEARNING FORMAT section from the combined text
+                // Match "** LEARNING FORMAT **" (with optional spaces) and capture everything until next section or end
+                // logger.info('Extracting LEARNING FORMAT section from combined text', combined_text);
+                const learning_format_match = combined_text.match(/\*\*\s*LEARNING FORMAT\s*\*\*([\s\S]*?)(?=\n\*\*|$)/i);
+                if (learning_format_match && learning_format_match[1]) {
+                    const learning_format = learning_format_match[1].trim();
+                    logger.info('✅ Extracted LEARNING FORMAT section from seed knowledge', {
+                        domain,
+                        knowledge_id,
+                        revision_id: most_recent_revision_id,
+                        format_length: learning_format.length,
+                        format_preview: learning_format.substring(0, 200)
+                    });
+                    return learning_format;
+                }
+            }
+            // If LEARNING FORMAT section not found in any seed, try the most recent seed's full text as fallback
+            const most_recent_seed = seed_embeddings[0];
+            logger.warn('⚠️ LEARNING FORMAT section not found in any seed knowledge, using most recent seed text as fallback', {
+                domain,
+                knowledge_id: most_recent_seed.knowledge_id,
+                seed_text_length: most_recent_seed.text.length
+            });
+            return most_recent_seed.text;
+        }
+        catch (error) {
+            logger.error('Failed to get seed pattern', { error, domain });
+            return null;
+        }
+    }
+    /**
+     * Validate that a seed exists for a domain before allowing operations
+     * Throws an error if seed doesn't exist
+     *
+     * @param domain - The domain to validate
+     * @throws Error if seed doesn't exist for domain
+     */
+    async validate_seed_exists(domain) {
+        if (!domain || domain.trim().length === 0) {
+            throw new Error('Domain is required');
+        }
+        const exists = await this.seed_exists_for_domain(domain);
+        if (!exists) {
+            throw new Error(`No seed found for domain '${domain}'. Please add a seed for this domain before creating learnings.`);
+        }
+    }
+    /**
+     * Identify domain from seed content using AI
+     * Domain is ALWAYS identified using AI - never from manual input
+     *
+     * @param content - The seed content to analyze
+     * @returns Promise resolving to identified domain string
+     * @throws Error if domain cannot be identified or if GEMINI_API_KEY is not configured
+     */
+    async identify_domain_from_content(content) {
+        if (!content || content.trim().length === 0) {
+            throw new Error('Content is required for domain identification');
+        }
+        const gemini_api_key = config_1.config_utils.get('GEMINI_API_KEY');
+        if (!gemini_api_key) {
+            throw new Error('GEMINI_API_KEY not configured');
+        }
+        logger.info('Starting domain identification from content', {
+            content_length: content.length,
+            content_preview: content.substring(0, 200)
+        });
+        const domain_agent = new domain_identifier_agent_1.DomainIdentifierAgent(new genai_1.GoogleGenAI({ apiKey: gemini_api_key }));
+        logger.debug('Calling domain agent with content', {
+            content_length: content.length,
+            content_title: content.split('\n')[0]?.substring(0, 100)
+        });
+        const identified_domain = await domain_agent.identify_domain(content);
+        logger.info('Domain identification result', {
+            identified_domain: identified_domain || 'null',
+            domain_type: typeof identified_domain,
+            domain_length: identified_domain?.length || 0
+        });
+        if (!identified_domain || identified_domain.trim().length === 0) {
+            throw new Error('Could not identify domain from seed content. Domain identification is required for seed creation.');
+        }
+        // Normalize and return domain
+        const normalized_domain = identified_domain.trim();
+        logger.info('✅ Domain identified from content', {
+            domain: normalized_domain
+        });
+        return normalized_domain;
+    }
+}
+exports.DomainService = DomainService;
+// Singleton instance
+let domain_service_instance = null;
+/**
+ * Get the singleton DomainService instance
+ */
+function get_domain_service() {
+    if (!domain_service_instance) {
+        domain_service_instance = new DomainService();
+    }
+    return domain_service_instance;
+}
+//# sourceMappingURL=domain.service.js.map

package/dist/services/domain.service.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"domain.service.js","sourceRoot":"","sources":["../../src/services/domain.service.ts"],"names":[],"mappings":";AAAA;;;;;;;;GAQG;;;AAoQH,gDAKC;AAvQD,gDAA4D;AAC5D,gEAAiE;AACjE,sCAA0D;AAC1D,yCAA+B;AAC/B,yCAA4C;AAC5C,kFAA6E;AAC7E,sCAAyC;AAEzC,MAAM,MAAM,GAAG,IAAA,8BAAqB,EAAC,gBAAgB,CAAC,CAAC;AAEvD,MAAa,aAAa;IAGtB;QACI,IAAI,CAAC,SAAS,GAAG,IAAA,yCAAqB,GAAE,CAAC;IAC7C,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,kBAAkB;QACpB,IAAI,CAAC;YACD,kEAAkE;YAClE,MAAM,OAAO,GAAG,MAAM,uBAAc,CAAC,OAAO,CAAC;gBACzC,UAAU,EAAE;oBACR,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC,UAAU,EAAE,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,EAAE,QAAQ,CAAC;iBAC1E;gBACD,KAAK,EAAE;oBACH,MAAM,EAAE;wBACJ,CAAC,cAAE,CAAC,EAAE,CAAC,EAAE,IAAW;qBACvB;iBACJ;gBACD,GAAG,EAAE,IAAI;aACZ,CAAC,CAAC;YAEH,MAAM,cAAc,GAAG,OAAO;iBACzB,GAAG,CAAC,CAAC,CAAM,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC;iBACzB,MAAM,CAAC,CAAC,CAAgB,EAAE,EAAE,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YAErE,MAAM,CAAC,KAAK,CAAC,8CAA8C,EAAE,EAAE,KAAK,EAAE,cAAc,CAAC,MAAM,EAAE,OAAO,EAAE,cAAc,EAAE,CAAC,CAAC;YACxH,OAAO,cAAc,CAAC;QAC1B,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACb,MAAM,CAAC,KAAK,CAAC,8BAA8B,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;YACxD,OAAO,EAAE,CAAC;QACd,CAAC;IACL,CAAC;IAED;;;;;;OAMG;IACH,KAAK,CAAC,sBAAsB,CAAC,MAAc;QACvC,IAAI,CAAC;YACD,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACxC,OAAO,KAAK,CAAC;YACjB,CAAC;YAED,0DAA0D;YAC1D,MAAM,UAAU,GAAG,MAAM,uBAAc,CAAC,KAAK,CAAC;gBAC1C,KAAK,EAAE;oBACH,IAAI,EAAE,sBAAa,CAAC,IAAI;oBACxB,MAAM,EAAE,MAAM;iBACjB;aACJ,CAAC,CAAC;YAEH,MAAM,MAAM,GAAG,UAAU,GAAG,CAAC,CAAC;YAC9B,MAAM,CAAC,KAAK,CAAC,sBAAsB,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC,CAAC;YACrE,OAAO,MAAM,CAAC;QAClB,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACb,MAAM,CAAC,KAAK,CAAC,gCAAgC,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC;YAClE,OAAO,KAAK,CAAC;QACjB,CAAC;IACL,CAAC;IAED;;;;;;OAMG;IACH,KAAK,CAAC,gBAAgB,CAAC,MAAc;QACjC,IAAI,CAAC;YACD,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACxC,OAAO,IAAI,CAAC;YAChB,CAAC;YAED,+CAA+C;YAC/C,MAAM,eAAe,GAAG,MAAM,uBAAc,CAAC,OAAO,CAAC;gBACjD,KAAK,EAAE;oBACH,IAAI,EAAE,sBAAa,CAAC,IAAI;oBACxB,MAAM,EAAE,MAAM;iBACjB;gBACD,UAAU,EAAE,CAAC,cAAc,EAAE,uBAAuB,EAAE,MAAM,EAAE,aAAa,EAAE,YAAY,CAAC;gBAC1F,KAAK,EAAE,CAAC,CAAC,YAAY,EAAE,MAAM,CAAC,EAAE,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC,EAAE,yCAAyC;aACrG,CAAC,CAAC;YAEH,IAAI,CAAC,eAAe,IAAI,eAAe,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACnD,MAAM,CAAC,KAAK,CAAC,qCAAqC,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;gBAChE,OAAO,IAAI,CAAC;YAChB,CAAC;YAED,MAAM,CAAC,IAAI,CAAC,kCAAkC,EAAE;gBAC5C,MAAM;gBACN,gBAAgB,EAAE,eAAe,CAAC,MAAM;gBACxC,oBAAoB,EAAE,CAAC,GAAG,IAAI,GAAG,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC,MAAM;aACtF,CAAC,CAAC;YAEH,6EAA6E;YAC7E,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAkC,CAAC;YACnE,KAAK,MAAM,SAAS,IAAI,eAAe,EAAE,CAAC;gBACtC,MAAM,YAAY,GAAG,SAAS,CAAC,YAAY,CAAC;gBAC5C,IAAI,CAAC,gBAAgB,CAAC,GAAG,CAAC,YAAY,CAAC,EAAE,CAAC;oBACtC,gBAAgB,CAAC,GAAG,CAAC,YAAY,EAAE,EAAE,CAAC,CAAC;gBAC3C,CAAC;gBACD,gBAAgB,CAAC,GAAG,CAAC,YAAY,CAAE,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;YACxD,CAAC;YAED,6DAA6D;YAC7D,kEAAkE;YAClE,KAAK,MAAM,CAAC,YAAY,EAAE,UAAU,CAAC,IAAI,gBAAgB,CAAC,OAAO,EAAE,EAAE,CAAC;gBAClE,kDAAkD;gBAClD,MAAM,uBAAuB,GAAG,UAAU,CAAC,CAAC,CAAC,EAAE,qBAAqB,CAAC;gBAErE,qEAAqE;gBACrE,MAAM,eAAe,GAAG,UAAU;qBAC7B,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,qBAAqB,KAAK,uBAAuB,CAAC;qBAChE,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,CAAC,CAAC;gBAEjE,0BAA0B;gBAC1B,MAAM,aAAa,GAAG,eAAe;qBAChC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;qBAChB,IAAI,CAAC,MAAM,CAAC,CAAC;gBAElB,MAAM,CAAC,KAAK,CAAC,oCAAoC,EAAE;oBAC/C,MAAM;oBACN,YAAY;oBACZ,WAAW,EAAE,uBAAuB;oBACpC,WAAW,EAAE,eAAe,CAAC,MAAM;oBACnC,eAAe,EAAE,aAAa,CAAC,MAAM;iBACxC,CAAC,CAAC;gBAEH,6DAA6D;gBAC7D,wGAAwG;gBACxG,uFAAuF;gBACvF,MAAM,qBAAqB,GAAG,aAAa,CAAC,KAAK,CAAC,sDAAsD,CAAC,CAAC;gBAE1G,IAAI,qBAAqB,IAAI,qBAAqB,CAAC,CAAC,CAAC,EAAE,CAAC;oBACpD,MAAM,eAAe,GAAG,qBAAqB,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;oBACxD,MAAM,CAAC,IAAI,CAAC,yDAAyD,EAAE;wBACnE,MAAM;wBACN,YAAY;wBACZ,WAAW,EAAE,uBAAuB;wBACpC,aAAa,EAAE,eAAe,CAAC,MAAM;wBACrC,cAAc,EAAE,eAAe,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC;qBACpD,CAAC,CAAC;oBACH,OAAO,eAAe,CAAC;gBAC3B,CAAC;YACL,CAAC;YAED,qGAAqG;YACrG,MAAM,gBAAgB,GAAG,eAAe,CAAC,CAAC,CAAC,CAAC;YAC5C,MAAM,CAAC,IAAI,CAAC,qGAAqG,EAAE;gBAC/G,MAAM;gBACN,YAAY,EAAE,gBAAgB,CAAC,YAAY;gBAC3C,gBAAgB,EAAE,gBAAgB,CAAC,IAAI,CAAC,MAAM;aACjD,CAAC,CAAC;YACH,OAAO,gBAAgB,CAAC,IAAI,CAAC;QACjC,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACb,MAAM,CAAC,KAAK,CAAC,4BAA4B,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC;YAC9D,OAAO,IAAI,CAAC;QAChB,CAAC;IACL,CAAC;IAED;;;;;;OAMG;IACH,KAAK,CAAC,oBAAoB,CAAC,MAAc;QACrC,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACxC,MAAM,IAAI,KAAK,CAAC,oBAAoB,CAAC,CAAC;QAC1C,CAAC;QAED,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,sBAAsB,CAAC,MAAM,CAAC,CAAC;QACzD,IAAI,CAAC,MAAM,EAAE,CAAC;YACV,MAAM,IAAI,KAAK,CAAC,6BAA6B,MAAM,iEAAiE,CAAC,CAAC;QAC1H,CAAC;IACL,CAAC;IAED;;;;;;;OAOG;IACH,KAAK,CAAC,4BAA4B,CAAC,OAAe;QAC9C,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC1C,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAC;QACrE,CAAC;QAED,MAAM,cAAc,GAAG,qBAAY,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAC;QAC1D,IAAI,CAAC,cAAc,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CAAC,+BAA+B,CAAC,CAAC;QACrD,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,6CAA6C,EAAE;YACvD,cAAc,EAAE,OAAO,CAAC,MAAM;YAC9B,eAAe,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC;SAC7C,CAAC,CAAC;QAEH,MAAM,YAAY,GAAG,IAAI,+CAAqB,CAC1C,IAAI,mBAAW,CAAC,EAAE,MAAM,EAAE,cAAwB,EAAE,CAAC,CACxD,CAAC;QAEF,MAAM,CAAC,KAAK,CAAC,mCAAmC,EAAE;YAC9C,cAAc,EAAE,OAAO,CAAC,MAAM;YAC9B,aAAa,EAAE,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC;SAC3D,CAAC,CAAC;QAEH,MAAM,iBAAiB,GAAG,MAAM,YAAY,CAAC,eAAe,CAAC,OAAO,CAAC,CAAC;QAEtE,MAAM,CAAC,IAAI,CAAC,8BAA8B,EAAE;YACxC,iBAAiB,EAAE,iBAAiB,IAAI,MAAM;YAC9C,WAAW,EAAE,OAAO,iBAAiB;YACrC,aAAa,EAAE,iBAAiB,EAAE,MAAM,IAAI,CAAC;SAChD,CAAC,CAAC;QAEH,IAAI,CAAC,iBAAiB,IAAI,iBAAiB,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC9D,MAAM,IAAI,KAAK,CAAC,mGAAmG,CAAC,CAAC;QACzH,CAAC;QAED,8BAA8B;QAC9B,MAAM,iBAAiB,GAAG,iBAAiB,CAAC,IAAI,EAAE,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,kCAAkC,EAAE;YAC5C,MAAM,EAAE,iBAAiB;SAC5B,CAAC,CAAC;QAEH,OAAO,iBAAiB,CAAC;IAC7B,CAAC;CACJ;AAhPD,sCAgPC;AAED,qBAAqB;AACrB,IAAI,uBAAuB,GAAyB,IAAI,CAAC;AAEzD;;GAEG;AACH,SAAgB,kBAAkB;IAC9B,IAAI,CAAC,uBAAuB,EAAE,CAAC;QAC3B,uBAAuB,GAAG,IAAI,aAAa,EAAE,CAAC;IAClD,CAAC;IACD,OAAO,uBAAuB,CAAC;AACnC,CAAC"}

package/dist/services/embedding.service.d.ts

@@ -0,0 +1,132 @@
+/**
+ * @fileoverview Embedding Service
+ *
+ * Service for generating embeddings using Gemini and storing them in the database.
+ *
+ * @author MeasureOne, Inc.
+ * @version 1.0.0
+ */
+import { EmbeddingType } from "../models";
+export declare class EmbeddingService {
+    private gemini;
+    private default_model;
+    private query_embedding_cache;
+    private readonly EMBEDDING_CACHE_TTL_MS;
+    private readonly MAX_EMBEDDING_CACHE_SIZE;
+    private readonly SEMANTIC_SIMILARITY_THRESHOLD;
+    private readonly MAX_SEMANTIC_CHECKS;
+    private faiss_index_cache;
+    private is_index_building;
+    constructor();
+    /**
+     * Builds or retrieves a cached Faiss index for a given model.
+     * This is the core of the search optimization.
+     * @param model - The embedding model name (e.g., "text-embedding-004")
+     */
+    private get_or_build_faiss_index;
+    /**
+     * Generate embedding with caching (Phase 1 optimization)
+     * Caches query embeddings to avoid redundant API calls
+     *
+     * Cache strategy (optimized for user-generated queries):
+     * 1. Quick exact text check (O(1) - fast path, but rare for user queries)
+     * 2. Generate embedding and check semantic similarity (primary strategy)
+     * 3. Cache by semantic similarity, not exact text (since exact matches are rare)
+     *
+     * Rationale: User queries are semantically similar but textually different.
+     * Exact text caching has low hit rate, so we prioritize semantic similarity.
+     */
+    private generate_embedding_cached;
+    /**
+     * Set cache entry with LRU eviction
+     * Stores embedding with pre-computed norm for faster similarity calculations
+     */
+    private set_cache_entry;
+    /**
+     * Find semantically similar cached embedding (synchronous version)
+     * Checks cached embeddings for high similarity (>= threshold)
+     *
+     * Optimization: Limits checks to prevent O(n) performance on large cache
+     * Uses pre-computed norms for faster similarity calculation
+     *
+     * @param query_embedding - Pre-computed query embedding vector
+     * @param query_norm - Pre-computed query embedding norm
+     * @param model - Embedding model
+     * @param similarity_threshold - Minimum similarity threshold
+     * @returns Matching embedding or null
+     */
+    private find_similar_cached_embedding_sync;
+    /**
+     * Generate embedding for a single text using Gemini
+     *
+     * @param text - Text to embed
+     * @param model - Embedding model (default: text-embedding-004)
+     * @returns Promise resolving to embedding vector
+     */
+    generate_embedding(text: string, model?: string): Promise<number[]>;
+    /**
+     * Create embeddings for a document (works for both SEED and LEARNING types)
+     * This method chunks text, generates embeddings, and stores them in the database.
+     *
+     * @param knowledge_id - Required knowledge ID (must exist)
+     * @param knowledge_revision_id - Required revision ID (must exist)
+     * @param text - Text content to create embeddings for
+     * @param source_data_id - Optional source data ID
+     * @param model - Optional embedding model (defaults to default_model)
+     * @param type - Embedding type (SEED or LEARNING)
+     * @param domain - Required domain (normalized to lowercase)
+     *
+     * Note: This does NOT search for similar documents or create knowledge/revision.
+     * The caller must ensure knowledge_id and knowledge_revision_id already exist.
+     */
+    create_embeddings_for_document(knowledge_id: string, knowledge_revision_id: string, text: string, source_data_id: string | undefined, model: string | undefined, type: EmbeddingType, domain: string): Promise<{
+        embeddings_created: number;
+        knowledge_id: string;
+        knowledge_revision_id: string;
+    }>;
+    /**
+     * Find knowledge ID for learned patterns
+     */
+    find_learned_patterns_knowledge_id(): Promise<string | null>;
+    /**
+     * Search embeddings by similarity using a two-pass top-k approach
+     *
+     * @param query_text - Query text to search for
+     * @param knowledge_id - Optional knowledge ID to filter by (learned knowledge from LEARNING embeddings is ALWAYS included)
+     * @param limit - Maximum number of results (default: 10)
+     * @param model - Embedding model (default: text-embedding-004)
+     * @returns Promise resolving to array of similar embeddings with scores
+     *
+     * This optimized version uses a pre-built in-memory Faiss index for fast vector search.
+     */
+    search_embeddings(query_text: string, knowledge_id?: string, limit?: number, model?: string, type?: EmbeddingType, domain?: string): Promise<Array<{
+        text: string;
+        score: number;
+        knowledge_id: string;
+        knowledge_revision_id: string;
+        source_data_id?: string;
+        chunk_index?: number;
+        type?: EmbeddingType;
+        domain?: string;
+    }>>;
+    /**
+     * Calculate cosine similarity between two vectors
+     *
+     * @param vec_a - First vector
+     * @param vec_b - Second vector
+     * @returns Cosine similarity score (0-1)
+     */
+    cosine_similarity(vec_a: number[], vec_b: number[]): number;
+    /**
+     * Optimized cosine similarity calculation that reuses pre-computed norms
+     * This avoids recalculating norms for every comparison
+     *
+     * @param vec_a - Query vector
+     * @param vec_b - Embedding vector to compare
+     * @param vec_a_norm - Pre-computed norm of vec_a (sqrt of sum of squares)
+     * @param vec_b_norm - Optional pre-computed norm of vec_b (if not provided, calculates it)
+     * @returns Cosine similarity score (0-1)
+     */
+    private cosine_similarity_optimized;
+}
+//# sourceMappingURL=embedding.service.d.ts.map

package/dist/services/embedding.service.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"embedding.service.d.ts","sourceRoot":"","sources":["../../src/services/embedding.service.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,OAAO,EAAkB,aAAa,EAAkB,MAAM,WAAW,CAAC;AA2B1E,qBAAa,gBAAgB;IACzB,OAAO,CAAC,MAAM,CAAc;IAC5B,OAAO,CAAC,aAAa,CAAgC;IAKrD,OAAO,CAAC,qBAAqB,CAAsC;IACnE,OAAO,CAAC,QAAQ,CAAC,sBAAsB,CAAkB;IACzD,OAAO,CAAC,QAAQ,CAAC,wBAAwB,CAAQ;IACjD,OAAO,CAAC,QAAQ,CAAC,6BAA6B,CAAQ;IACtD,OAAO,CAAC,QAAQ,CAAC,mBAAmB,CAAM;IAG1C,OAAO,CAAC,iBAAiB,CAAqC;IAC9D,OAAO,CAAC,iBAAiB,CAAS;;IAUlC;;;;OAIG;YACW,wBAAwB;IA2EtC;;;;;;;;;;;OAWG;YACW,yBAAyB;IAkDvC;;;OAGG;IACH,OAAO,CAAC,eAAe;IA4BvB;;;;;;;;;;;;OAYG;IACH,OAAO,CAAC,kCAAkC;IAkE1C;;;;;;OAMG;IACG,kBAAkB,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC;IAqCzE;;;;;;;;;;;;;;OAcG;IACU,8BAA8B,CACvC,YAAY,EAAE,MAAM,EACpB,qBAAqB,EAAE,MAAM,EAC7B,IAAI,EAAE,MAAM,EACZ,cAAc,EAAE,MAAM,GAAG,SAAS,EAClC,KAAK,EAAE,MAAM,GAAG,SAAS,EACzB,IAAI,EAAE,aAAa,EACnB,MAAM,EAAE,MAAM,GACf,OAAO,CAAC;QAAE,kBAAkB,EAAE,MAAM,CAAC;QAAC,YAAY,EAAE,MAAM,CAAC;QAAC,qBAAqB,EAAE,MAAM,CAAA;KAAE,CAAC;IA4G/F;;OAEG;IACU,kCAAkC,IAAI,OAAO,CAAC,MAAM,GAAG,IAAI,CAAC;IAyBzE;;;;;;;;;;OAUG;IACG,iBAAiB,CACnB,UAAU,EAAE,MAAM,EAClB,YAAY,CAAC,EAAE,MAAM,EACrB,KAAK,GAAE,MAAW,EAClB,KAAK,CAAC,EAAE,MAAM,EACd,IAAI,CAAC,EAAE,aAAa,EACpB,MAAM,CAAC,EAAE,MAAM,GAChB,OAAO,CAAC,KAAK,CAAC;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAC;QAAC,YAAY,EAAE,MAAM,CAAC;QAAC,qBAAqB,EAAE,MAAM,CAAC;QAAC,cAAc,CAAC,EAAE,MAAM,CAAC;QAAC,WAAW,CAAC,EAAE,MAAM,CAAC;QAAC,IAAI,CAAC,EAAE,aAAa,CAAC;QAAC,MAAM,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC,CAAC;IA8G7L;;;;;;OAMG;IACI,iBAAiB,CAAC,KAAK,EAAE,MAAM,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM;IAuBlE;;;;;;;;;OASG;IACH,OAAO,CAAC,2BAA2B;CA2BtC"}