strapi-content-embeddings 0.1.4 → 0.1.5

package/dist/server/src/services/embeddings.d.ts

@@ -1,4 +1,5 @@
 import type { Core } from "@strapi/strapi";
+import type { PluginConfigSchema } from "../config";
 export interface CreateEmbeddingData {
     data: {
         title: string;
@@ -10,6 +11,8 @@ export interface CreateEmbeddingData {
             __type: string;
             id: number;
         };
+        /** Enable chunking for large content (overrides config.autoChunk) */
+        autoChunk?: boolean;
     };
 }
 export interface UpdateEmbeddingData {
@@ -17,17 +20,55 @@ export interface UpdateEmbeddingData {
         title?: string;
         content?: string;
         metadata?: Record<string, any>;
+        /** Enable chunking for large content on update (overrides config.autoChunk) */
+        autoChunk?: boolean;
     };
 }
+export interface ChunkedEmbeddingResult {
+    /** The parent/first embedding entity */
+    entity: any;
+    /** All chunk entities created */
+    chunks: any[];
+    /** Total number of chunks created */
+    totalChunks: number;
+    /** Whether content was chunked */
+    wasChunked: boolean;
+}
 declare const embeddings: ({ strapi }: {
     strapi: Core.Strapi;
 }) => {
-    createEmbedding(data: CreateEmbeddingData): Promise<import("@strapi/types/dist/modules/documents").AnyDocument>;
+    /**
+     * Get plugin config with defaults
+     */
+    getConfig(): PluginConfigSchema;
+    /**
+     * Create a single embedding (no chunking)
+     */
+    createEmbedding(data: CreateEmbeddingData): Promise<any>;
+    /**
+     * Create embeddings with automatic chunking for large content
+     * Creates multiple embedding entities, one per chunk
+     */
+    createChunkedEmbedding(data: CreateEmbeddingData): Promise<ChunkedEmbeddingResult>;
     deleteEmbedding(id: number | string): Promise<{
         documentId: string;
         entries: import("@strapi/types/dist/modules/documents").AnyDocument[];
     }>;
-    updateEmbedding(id: string, data: UpdateEmbeddingData): Promise<import("@strapi/types/dist/modules/documents").AnyDocument>;
+    /**
+     * Find all chunks related to a parent document
+     * Returns chunks including the parent itself
+     */
+    findRelatedChunks(documentId: string): Promise<any[]>;
+    /**
+     * Delete all chunks related to a parent document
+     */
+    deleteRelatedChunks(documentId: string): Promise<number>;
+    /**
+     * Update embeddings with automatic chunking support
+     * Handles re-chunking when content changes and exceeds chunk size
+     */
+    updateChunkedEmbedding(id: string, data: UpdateEmbeddingData): Promise<ChunkedEmbeddingResult>;
+    updateEmbedding(id: string, data: UpdateEmbeddingData): Promise<any>;
     queryEmbeddings(query: string): Promise<import("../plugin-manager").QueryResponse | {
         error: string;
     }>;

package/dist/server/src/services/index.d.ts

@@ -2,12 +2,17 @@ declare const _default: {
     embeddings: ({ strapi }: {
         strapi: import("@strapi/types/dist/core").Strapi;
     }) => {
-        createEmbedding(data: import("./embeddings").CreateEmbeddingData): Promise<import("@strapi/types/dist/modules/documents").AnyDocument>;
+        getConfig(): import("../config").PluginConfigSchema;
+        createEmbedding(data: import("./embeddings").CreateEmbeddingData): Promise<any>;
+        createChunkedEmbedding(data: import("./embeddings").CreateEmbeddingData): Promise<import("./embeddings").ChunkedEmbeddingResult>;
         deleteEmbedding(id: string | number): Promise<{
             documentId: string;
             entries: import("@strapi/types/dist/modules/documents").AnyDocument[];
         }>;
-        updateEmbedding(id: string, data: import("./embeddings").UpdateEmbeddingData): Promise<import("@strapi/types/dist/modules/documents").AnyDocument>;
+        findRelatedChunks(documentId: string): Promise<any[]>;
+        deleteRelatedChunks(documentId: string): Promise<number>;
+        updateChunkedEmbedding(id: string, data: import("./embeddings").UpdateEmbeddingData): Promise<import("./embeddings").ChunkedEmbeddingResult>;
+        updateEmbedding(id: string, data: import("./embeddings").UpdateEmbeddingData): Promise<any>;
         queryEmbeddings(query: string): Promise<import("../plugin-manager").QueryResponse | {
             error: string;
         }>;
@@ -22,5 +27,21 @@ declare const _default: {
             totalCount: number;
         }>;
     };
+    sync: ({ strapi }: {
+        strapi: import("@strapi/types/dist/core").Strapi;
+    }) => {
+        syncFromNeon(options?: {
+            removeOrphans?: boolean;
+            dryRun?: boolean;
+        }): Promise<import("./sync").SyncResult>;
+        getSyncStatus(): Promise<{
+            neonCount: number;
+            strapiCount: number;
+            inSync: boolean;
+            missingInStrapi: number;
+            missingInNeon: number;
+            contentDifferences: number;
+        }>;
+    };
 };
 export default _default;

package/dist/server/src/services/sync.d.ts

@@ -0,0 +1,48 @@
+import type { Core } from "@strapi/strapi";
+export interface SyncResult {
+    success: boolean;
+    timestamp: string;
+    neonCount: number;
+    strapiCount: number;
+    actions: {
+        created: number;
+        updated: number;
+        orphansRemoved: number;
+    };
+    details: {
+        created: string[];
+        updated: string[];
+        orphansRemoved: string[];
+    };
+    errors: string[];
+}
+declare const sync: ({ strapi }: {
+    strapi: Core.Strapi;
+}) => {
+    /**
+     * Sync embeddings from Neon DB to Strapi DB
+     *
+     * This performs the following operations:
+     * 1. Fetches all embeddings from Neon DB (source of truth)
+     * 2. Fetches all embeddings from Strapi DB
+     * 3. Creates missing entries in Strapi that exist in Neon
+     * 4. Updates Strapi entries where content differs from Neon
+     * 5. Optionally removes orphaned Strapi entries (no matching Neon record)
+     */
+    syncFromNeon(options?: {
+        removeOrphans?: boolean;
+        dryRun?: boolean;
+    }): Promise<SyncResult>;
+    /**
+     * Get sync status - compare Neon and Strapi without making changes
+     */
+    getSyncStatus(): Promise<{
+        neonCount: number;
+        strapiCount: number;
+        inSync: boolean;
+        missingInStrapi: number;
+        missingInNeon: number;
+        contentDifferences: number;
+    }>;
+};
+export default sync;

package/dist/server/src/utils/chunking.d.ts

@@ -0,0 +1,44 @@
+/**
+ * Text chunking utilities for splitting large content into embeddable chunks
+ */
+export interface ChunkOptions {
+    /** Maximum characters per chunk (default: 4000, roughly ~1000 tokens) */
+    chunkSize?: number;
+    /** Number of characters to overlap between chunks (default: 200) */
+    chunkOverlap?: number;
+    /** Separator to use when splitting (default: splits on paragraphs, sentences, then words) */
+    separators?: string[];
+}
+export interface TextChunk {
+    /** The chunk text content */
+    text: string;
+    /** Zero-based chunk index */
+    chunkIndex: number;
+    /** Total number of chunks */
+    totalChunks: number;
+    /** Character offset in original text */
+    startOffset: number;
+    /** Character end offset in original text */
+    endOffset: number;
+}
+/**
+ * Estimate token count from character count
+ * OpenAI models average ~4 characters per token for English text
+ */
+export declare function estimateTokens(text: string): number;
+/**
+ * Check if content exceeds the recommended chunk size
+ */
+export declare function needsChunking(content: string, maxChars?: number): boolean;
+/**
+ * Split content into chunks suitable for embedding
+ *
+ * @param content - The text content to split
+ * @param options - Chunking options
+ * @returns Array of TextChunk objects
+ */
+export declare function chunkContent(content: string, options?: ChunkOptions): TextChunk[];
+/**
+ * Format chunk title with index information
+ */
+export declare function formatChunkTitle(baseTitle: string, chunkIndex: number, totalChunks: number): string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "strapi-content-embeddings",
-  "version": "0.1.4",
+  "version": "0.1.5",
   "description": "Strapi v5 plugin for vector embeddings with OpenAI and Neon PostgreSQL. Enables semantic search, RAG chat, and MCP (Model Context Protocol) integration.",
   "keywords": [
     "strapi",