Add memory upsert and redis functionality to local api

.well-known/ai-plugin.json

@@ -1,19 +1,19 @@
 {
-  "schema_version": "v1",
-  "name_for_model": "retrieval",
-  "name_for_human": "Retrieval Plugin",
-  "description_for_model": "Plugin for searching through the user's documents (such as files, emails, and more) to find answers to questions and retrieve relevant information. Use it whenever a user asks something that might be found in their personal information.",
-  "description_for_human": "Search through your documents.",
-  "auth": {
-    "type": "user_http",
-    "authorization_type": "bearer"
-  },
-  "api": {
-    "type": "openapi",
-    "url": "https://your-app-url.com/.well-known/openapi.yaml",
-    "has_user_authentication": false
-  },
-  "logo_url": "https://your-app-url.com/.well-known/logo.png",
-  "contact_email": "[email protected]", 
-  "legal_info_url": "http://example.com/legal-info"
+    "schema_version": "v1",
+    "name_for_model": "retrieval",
+    "name_for_human": "Retrieval Plugin",
+    "description_for_model": "Plugin for searching through the user's documents (such as files, emails, and more) to find answers to questions and retrieve relevant information. Use it whenever a user asks something that might be found in their personal information, or asks you to save information for later.",
+    "description_for_human": "Search through your documents.",
+    "auth": {
+        "type": "user_http",
+        "authorization_type": "bearer"
+    },
+    "api": {
+        "type": "openapi",
+        "url": "https://localhost:3333/.well-known/openapi.yaml",
+        "has_user_authentication": false
+    },
+    "logo_url": "https://localhost:3333/.well-known/logo.png",
+    "contact_email": "[email protected]",
+    "legal_info_url": "hello@legal.com"
 }

.well-known/openapi.yaml

@@ -3,9 +3,35 @@ info:
   title: Retrieval Plugin API
   description: A retrieval API for querying and filtering documents based on natural language queries and metadata
   version: 1.0.0
-  servers:
-    - url: https://your-app-url.com
+servers:
+  - url: https://localhost:3333
 paths:
+  /upsert:
+    post:
+      summary: Upsert
+      description: Save chat information. Accepts an array of documents with text (potential questions + conversation text), metadata (source 'chat' and timestamp, no ID as this will be generated). Confirm with the user before saving, ask for more details/context.
+      operationId: upsert_upsert_post
+      requestBody:
+        content:
+          application/json:
+            schema:
+              $ref: "#/components/schemas/UpsertRequest"
+        required: true
+      responses:
+        "200":
+          description: Successful Response
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/UpsertResponse"
+        "422":
+          description: Validation Error
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/HTTPValidationError"
+      security:
+        - HTTPBearer: []
   /query:
     post:
       summary: Query
@@ -34,6 +60,20 @@ paths:
         - HTTPBearer: []
 components:
   schemas:
+    Document:
+      title: Document
+      required:
+        - text
+      type: object
+      properties:
+        id:
+          title: Id
+          type: string
+        text:
+          title: Text
+          type: string
+        metadata:
+          $ref: "#/components/schemas/DocumentMetadata"
     DocumentChunkMetadata:
       title: DocumentChunkMetadata
       type: object
@@ -79,6 +119,24 @@ components:
         score:
           title: Score
           type: number
+    DocumentMetadata:
+      title: DocumentMetadata
+      type: object
+      properties:
+        source:
+          $ref: "#/components/schemas/Source"
+        source_id:
+          title: Source Id
+          type: string
+        url:
+          title: Url
+          type: string
+        created_at:
+          title: Created At
+          type: string
+        author:
+          title: Author
+          type: string
     DocumentMetadataFilter:
       title: DocumentMetadataFilter
       type: object
@@ -169,6 +227,28 @@ components:
         - chat
       type: string
       description: An enumeration.
+    UpsertRequest:
+      title: UpsertRequest
+      required:
+        - documents
+      type: object
+      properties:
+        documents:
+          title: Documents
+          type: array
+          items:
+            $ref: "#/components/schemas/Document"
+    UpsertResponse:
+      title: UpsertResponse
+      required:
+        - ids
+      type: object
+      properties:
+        ids:
+          title: Ids
+          type: array
+          items:
+            type: string
     ValidationError:
       title: ValidationError
       required:

PROMPTS.md

@@ -0,0 +1,19 @@
+# Prompts
+
+Use the following prompt in ChatGPT at chat.openai.com to test the plugin:
+
+```text
+As an AI language model, you have access to a variety of plugins that enhance your capabilities. I want you to utilize these plugins effectively to provide the most accurate and comprehensive responses. Here's how you should proceed:
+
+1. **Data Retrieval Plugin**: Whenever a question is asked or a task is given, your first step should be to use the data retrieval plugin to check your memory storage. This plugin allows you to search through stored documents, files, emails, and more to find answers to questions and retrieve relevant information. See if a similar question has been asked before, or if the data from a previous information gathering task is stored.
+
+2. **WebPilot Plugin**: If no relevant information is found in memory, use the web browsing plugins like WebPilot to search the internet. The WebPilot plugin allows you to visit web pages, retrieve content from them, and interact with the content if necessary. Use this tool to download the relevant information.
+
+3. **Data Storage**: Any new information gathered from the internet or other sources should be saved into the retrieval system for future reference. Use the 'Retrieval Plugin (local)' `/upsert` endpoint to upload this information into the plugin's vector datastore. This will allow you to access this information in the future without needing to search the internet again.
+
+4. **Clarification**: If there's any uncertainty about the provided information or the task at hand, don't hesitate to ask follow-up questions for clarification. It's better to ask and be sure than to assume and make mistakes.
+
+5. **Prompt Understanding**: Read each prompt carefully and ask if there are any questions about the instructions. This will ensure that you fully understand the task at hand and can provide the most accurate response.
+
+Remember, these plugins are tools to enhance your capabilities. Use them effectively to provide the best possible service.
+```

examples/docker/redis/docker-compose.yml

@@ -3,10 +3,11 @@ version: "3.9"
 services:
   redis:
     image: redis/redis-stack-server:latest
+    restart: unless-stopped
     ports:
       - "6379:6379"
     volumes:
-        - redis_data:/data
+      - redis_data:/data
     healthcheck:
       test: ["CMD", "redis-cli", "-h", "localhost", "-p", "6379", "ping"]
       interval: 2s
@@ -15,4 +16,4 @@ services:
       start_period: 5s
 
 volumes:
-  redis_data:
+  redis_data:

examples/memory/ai-plugin.json

@@ -5,15 +5,15 @@
     "description_for_model": "Plugin for searching through the user's documents (such as files, emails, and more) to find answers to questions and retrieve relevant information. Use it whenever a user asks something that might be found in their personal information, or asks you to save information for later.",
     "description_for_human": "Search through your documents.",
     "auth": {
-      "type": "user_http",
-      "authorization_type": "bearer"
+        "type": "user_http",
+        "authorization_type": "bearer"
     },
     "api": {
-      "type": "openapi",
-      "url": "https://your-app-url.com/.well-known/openapi.yaml",
-      "has_user_authentication": false
+        "type": "openapi",
+        "url": "https://your-app-url.com/.well-known/openapi.yaml",
+        "has_user_authentication": false
     },
     "logo_url": "https://your-app-url.com/.well-known/logo.png",
-    "contact_email": "[email protected]", 
+    "contact_email": "[email protected]",
     "legal_info_url": "[email protected]"
-  }
+}

examples/providers/redis/semantic-search-and-filter.ipynb

@@ -498,7 +498,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.11.1"
+   "version": "3.10.6"
   },
   "vscode": {
    "interpreter": {

local_server/ai-plugin-local.json

@@ -0,0 +1,17 @@
+{
+    "schema_version": "v1",
+    "name_for_model": "retrieval",
+    "name_for_human": "Retrieval Plugin (localhost)",
+    "description_for_model": "Plugin for searching through the user's documents (such as files, emails, and more) to find answers to questions and retrieve relevant information. Use it whenever a user asks something that might be found in their personal information.",
+    "description_for_human": "Search through your documents.",
+    "auth": {
+        "type": "none"
+    },
+    "api": {
+        "type": "openapi",
+        "url": "http://localhost:3333/.well-known/openapi.yaml"
+    },
+    "logo_url": "http://localhost:3333/.well-known/logo.png",
+    "contact_email": "[email protected]",
+    "legal_info_url": "[email protected]"
+}

local_server/ai-plugin.json

@@ -1,18 +1,17 @@
 {
-  "schema_version": "v1",
-  "name_for_model": "retrieval",
-  "name_for_human": "Retrieval Plugin",
-  "description_for_model": "Plugin for searching through the user's documents (such as files, emails, and more) to find answers to questions and retrieve relevant information. Use it whenever a user asks something that might be found in their personal information.",
-  "description_for_human": "Search through your documents.",
-  "auth": {
-    "type": "none"
-  },
-  "api": {
-    "type": "openapi",
-    "url": "http://localhost:3333/.well-known/openapi.yaml"
-  },
-  "logo_url": "http://localhost:3333/.well-known/logo.png",
-  "contact_email": "[email protected]", 
-  "legal_info_url": "[email protected]"
+    "schema_version": "v1",
+    "name_for_model": "retrieval",
+    "name_for_human": "Retrieval Plugin (local)",
+    "description_for_model": "Plugin for searching through the user's documents (such as files, emails, and more) to find answers to questions and retrieve relevant information. Use it whenever a user asks something that might be found in their personal information, or asks you to save information for later.",
+    "description_for_human": "Search through your documents.",
+    "auth": {
+        "type": "none"
+    },
+    "api": {
+        "type": "openapi",
+        "url": "http://localhost:3333/.well-known/openapi.yaml"
+    },
+    "logo_url": "http://localhost:3333/.well-known/logo.png",
+    "contact_email": "[email protected]",
+    "legal_info_url": "[email protected]"
 }
-

local_server/main.py

@@ -50,7 +50,7 @@ async def get_manifest(request):
 
 @app.route("/.well-known/logo.png")
 async def get_logo(request):
-    file_path = "./local_server/logo.png"
+    file_path = "./local_server/logo-local.png"
     return FileResponse(file_path, media_type="text/json")
 
 
@@ -59,6 +59,14 @@ async def get_openapi(request):
     file_path = "./local_server/openapi.yaml"
     return FileResponse(file_path, media_type="text/json")
 
+# Create a sub-application, in order to access just the upsert and query endpoints in the OpenAPI schema, found at http://0.0.0.0:3333/sub/openapi.json when the app is running locally
+sub_app = FastAPI(
+    title="Retrieval Plugin API",
+    description="A retrieval API for querying and filtering documents based on natural language queries and metadata",
+    version="1.0.0",
+    servers=[{"url": "http://localhost:{PORT}"}],
+)
+app.mount("/sub", sub_app)
 
 @app.post(
     "/upsert-file",
@@ -91,6 +99,23 @@ async def upsert_file(
     "/upsert",
     response_model=UpsertResponse,
 )
+async def upsert_main(
+    request: UpsertRequest = Body(...),
+):
+    try:
+        ids = await datastore.upsert(request.documents)
+        return UpsertResponse(ids=ids)
+    except Exception as e:
+        logger.error(e)
+        raise HTTPException(status_code=500, detail="Internal Service Error")
+
+
+@sub_app.post(
+    "/upsert",
+    response_model=UpsertResponse,
+    # NOTE: We are describing the shape of the API endpoint input due to a current limitation in parsing arrays of objects from OpenAPI schemas. This will not be necessary in the future.
+    description="Save chat information. Accepts an array of documents with text (potential questions + conversation text), metadata (source 'chat' and timestamp, no ID as this will be generated). Confirm with the user before saving, ask for more details/context.",
+)
 async def upsert(
     request: UpsertRequest = Body(...),
 ):
@@ -102,8 +127,32 @@ async def upsert(
         raise HTTPException(status_code=500, detail="Internal Service Error")
 
 
-@app.post("/query", response_model=QueryResponse)
-async def query_main(request: QueryRequest = Body(...)):
+@app.post(
+    "/query",
+    response_model=QueryResponse,
+)
+async def query_main(
+    request: QueryRequest = Body(...),
+):
+    try:
+        results = await datastore.query(
+            request.queries,
+        )
+        return QueryResponse(results=results)
+    except Exception as e:
+        logger.error(e)
+        raise HTTPException(status_code=500, detail="Internal Service Error")
+
+
+@sub_app.post(
+    "/query",
+    response_model=QueryResponse,
+    # NOTE: We are describing the shape of the API endpoint input due to a current limitation in parsing arrays of objects from OpenAPI schemas. This will not be necessary in the future.
+    description="Accepts search query objects array each with query and optional filter. Break down complex questions into sub-questions. Refine results by criteria, e.g. time / source, don't do this often. Split queries if ResponseTooLargeError occurs.",
+)
+async def query(
+    request: QueryRequest = Body(...),
+):
     try:
         results = await datastore.query(
             request.queries,