feat(readme): add readme with edge fn template

Author: Daniele Briggi

.github/workflows/test.yaml

@@ -17,6 +17,5 @@ jobs:
           connection_string: ${{ secrets.CONNECTION_STRING }}
           base_url: https://docs.sqlitecloud.io/docs/
           database_name: aisearch-action-test.sqlite
-          database_path: sqliterag.sqlite
           # only few files for testing
           source_files: docs/sqlite-cloud/sdks/php

README.md

@@ -0,0 +1,103 @@
+[<img src="https://github.com/user-attachments/assets/0d406c41-ff61-41d7-a8de-249e9e652946" alt="https://sqlite.ai" width="110"/>](https://sqlite.ai)
+
+# SQLite AI Search Action
+    
+Index your documentation files with SQLite AI-powered embeddings for intelligent search capabilities.
+
+[![Test](https://github.com/sqliteai/sqlite-aisearch-action/actions/workflows/test.yaml/badge.svg?branch=main)](https://github.com/sqliteai/sqlite-aisearch-action/actions/workflows/test.yaml)
+[![GitHub Release](https://img.shields.io/github/v/release/sqliteai/sqlite-aisearch-action?label=Version)](https://github.com/sqliteai/sqlite-aisearch-action/releases/latest)
+
+## Overview
+
+The SQLite AI Search Action automatically processes your documentation files and creates an AI-powered searchable database on SQLite Cloud. This action uses both embeddings to understand document content semantically and traditional full-text search, enabling more intelligent and contextual search results.
+
+This action uses [sqlite-rag](https://github.com/sqliteai/sqlite-rag).
+
+## How it works
+
+1. **Downloads AI Model**: Retrieves the specified Hugging Face model to generate embeddings for semantic search
+2. **Processes Documents**: Recursively scans and parses all documentation files in your repository  
+3. **Creates Dual Index**: Generates both AI embeddings and traditional full-text search indices for comprehensive search capabilities
+4. **Builds Database**: Creates a SQLite database containing documents, embeddings, and search indices
+5. **Uploads to SQLite Cloud**: Transfers the complete database to your SQLite Cloud project
+6. **Integrates with Edge Function**: Works out-of-the-box with the predefined SQLite Cloud Edge Function template ([aisearch-docs.js](search_edge_function_template/aisearch-docs.js)) for instant search functionality
+
+> **Note**: the SQLite Cloud Search Edge Function supports only the default model. Do not change the model settings if you want to use our search edge function.
+
+## Usage
+
+### Setup your workflow
+
+1. **Get your Connection String**: Ensure you have a project on [SQLite Cloud dashboard](https://dashboard.sqlitecloud.io). If not, sign up to [SQLite AI](https://sqlite.ai) to create one for free.
+
+2. **Set GitHub Secret**: Add your connection string as `SQLITECLOUD_CONNECTION_STRING` in your repository secrets.
+
+3. **Add to Workflow**: Create or update your GitHub workflow:
+
+```yaml
+name: AI Search Index
+
+on:
+  push:
+    branches: [main]
+  workflow_dispatch:
+
+jobs:
+  build-search:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      
+      - name: Build AI Search Database
+        uses: sqliteai/sqlite-aisearch-action@v1
+        with:
+          connection_string: ${{ secrets.SQLITECLOUD_CONNECTION_STRING }}
+          base_url: https://docs.yoursite.com
+          database_name: aidocs_search.db
+          source_files: ./path/to/documents
+```
+
+### Create the Search Edge Function
+
+To enable search functionality on your indexed database, you need to create an Edge Function using the provided template:
+
+1. Access your dashboard on https://dashboard.sqlitecloud.io
+2. Enter the same project where the created database has been uploaded
+3. Go to the **Edge Functions** section
+4. Create a new `Javascript Function` and copy and paste the code from [aisearch-docs.js](search_edge_function_template/aisearch-docs.js) in the editor
+5. Deploy and test
+
+#### How to perform a search
+
+1. Go in the **Detail** in the Edge Function panel and copy the **Function URL**
+2. Execute a GET request and send a URL-econded query as `query` parameter. 
+    
+    Example: `GET
+	    https://myproject.cloud/v2/functions/aisearch-docs?query=what%27s+Offsync%3F`
+
+Response example:
+```json
+{
+  "data": {
+    "search": [
+      {
+        "id": "c41a6c2e-34e9-4e8e-92b9-41b8065047c7",
+        "uri": "docs/sqlite-cloud/sdks/php/methods.mdx",
+        "metadata": "{\"base_url\": \"https://docs.sqlitecloud.io/docs/\"}",
+        "snippet": "---\ntitle: OffSync\ndescription: Enable local-first applications with automatic data synchronization between edge devices and SQLite Cloud...",
+        "vec_rank": 1,
+        "fts_rank": null,
+        "combined_rank": 0.0163934426229508,
+        "vec_distance": 0.581515073776245,
+        "fts_score": null
+      },
+      ...
+    ]
+  }
+}
+```
+
+## Support
+
+- 📖 [SQLite Cloud Documentation](https://docs.sqlitecloud.io)
+- 💬 [Community Support](https://github.com/orgs/sqlitecloud/discussions)

action.yaml

@@ -1,5 +1,5 @@
 name: "SQLite AI - AI search for documents"
-description: "Parses document files to create a database on your SQLite Cloud project."
+description: "Out-of-the-box database for AI search for documents powered by SQLite AI."
 author: "SQLite AI Team"
 
 inputs:
@@ -12,9 +12,6 @@ inputs:
   database_name:
     description: The name of the database to use on SQLite Cloud
     required: true
-  database_path:
-    description: The local path of the database file to upload to SQLite Cloud.
-    required: true
   source_files:
     description: The path of the files, by default it will parse every file recursively starting from the working directory.
     required: false
@@ -81,14 +78,16 @@ runs:
       run: |
         sqlite-rag add \
           --recursive "${{ inputs.source_files }}" \
-          --metadata '{"base_url": "${{ inputs.base_url }}"}'
+          --metadata '{"base_url": "${{ inputs.base_url }}"}' \
+          --relative-paths
       shell: bash
 
     - name: Upload the database to SQLite Cloud
       run: bash $GITHUB_ACTION_PATH/upload.sh
       env:
         GITHUB_ACTION_PATH: ${{ github.action_path }}
         CONNECTION_STRING: ${{ inputs.connection_string }}
-        DATABASE_PATH: ${{ inputs.database_path }}
+        # default sqlite-rag database path
+        DATABASE_PATH: "./sqliterag.sqlite"
         DATABASE: ${{ inputs.database_name }}
       shell: bash

search_edge_function_template/aisearch-docs.js

@@ -0,0 +1,111 @@
+/**
+ * Edge Function template for performing AI-powered search on databases created by the GitHub Action.
+ * 
+ * This template provides a ready-to-use Edge Function that users can copy and paste 
+ * into their SQLite Cloud project to enable semantic search capabilities on the 
+ * documentation database generated by the sqlite-aisearch-action GitHub Action.
+ * 
+ * The function combines vector embeddings with full-text search using Reciprocal 
+ * Rank Fusion (RRF) to deliver accurate and contextually relevant search results.
+ * 
+ * See README.md for deployment instructions.
+ */
+
+//---- CONFIGURATION ----
+const sqliteAIBaseUrl = "https://aiserver.vital-rhino.eks.euc1.ryujaz.sqlite.cloud";
+const sqliteAIAPI = "/v1/ai/embeddings"
+//-----------------------
+
+const requestid = request.params.requestid;
+const query = request.params.query;
+
+// get embedding from sqlite-ai-server
+const data = {"text": query };
+const response = await fetch(sqliteAIBaseUrl + sqliteAIAPI, {
+    method: "POST",
+    headers: {
+        "Content-Type": "application/json"
+    },
+    body: JSON.stringify(data)
+});
+
+if (!response.ok) {
+    throw new Error("Failed to generate embedding. " + response.statusText);
+}
+
+const result = await response.json();
+const query_embedding = result.data.embedding;
+
+// clean query for full-text search
+const query_fts = (query.toLowerCase().match(/\b\w+\b/g) || []).join(" ") + "*";
+
+// --- TEST ---
+//const test_embedding = await connection.sql('SELECT embedding FROM chunks LIMIT 1;');
+//const query_embedding = test_embedding[0].embedding;
+// ------------
+
+// Vector configuration must match the embedding parameters used during database generation
+await connection.sql("SELECT vector_init('chunks', 'embedding', 'type=INT8,dimension=768,distance=cosine')");
+
+const res = await connection.sql(
+    `
+    -- sqlite-vector KNN vector search results
+    WITH vec_matches AS (
+        SELECT
+            v.rowid AS chunk_id,
+            row_number() OVER (ORDER BY v.distance) AS rank_number,
+            v.distance
+        FROM vector_quantize_scan('chunks', 'embedding', ?, 10) AS v
+    ),
+    -- Full-text search results
+    fts_matches AS (
+        SELECT
+            chunks_fts.rowid AS chunk_id,
+            row_number() OVER (ORDER BY rank) AS rank_number,
+            rank AS score
+        FROM chunks_fts
+        WHERE chunks_fts MATCH ?
+        LIMIT 10
+    ),
+    -- combine FTS5 + vector search results with RRF
+    matches AS (
+        SELECT
+            COALESCE(vec_matches.chunk_id, fts_matches.chunk_id) AS chunk_id,
+            vec_matches.rank_number AS vec_rank,
+            fts_matches.rank_number AS fts_rank,
+            -- Reciprocal Rank Fusion score
+            (
+                COALESCE(1.0 / (60 + vec_matches.rank_number), 0.0) * 1.0 +
+                COALESCE(1.0 / (60 + fts_matches.rank_number), 0.0) * 1.0
+            ) AS combined_rank,
+            vec_matches.distance AS vec_distance,
+            fts_matches.score AS fts_score
+        FROM vec_matches
+            FULL OUTER JOIN fts_matches
+                ON vec_matches.chunk_id = fts_matches.chunk_id
+    )
+    SELECT
+        documents.id,
+        documents.uri,
+        documents.content as document_content,
+        documents.metadata,
+        chunks.content AS snippet,
+        vec_rank,
+        fts_rank,
+        combined_rank,
+        vec_distance,
+        fts_score
+    FROM matches
+        JOIN chunks ON chunks.id = matches.chunk_id
+        JOIN documents ON documents.id = chunks.document_id
+    ORDER BY combined_rank DESC
+    ;
+    `, query_embedding, query_fts)
+
+
+return {
+    data: {
+        search: res,
+        requestid: requestid
+    }
+}