added fastapi-server instrucs

fastapi-server/.dockerignore

@@ -2,4 +2,5 @@
 .venv
 poetry.lock
 pyproject.toml
+README.md
 

fastapi-server/Dockerfile

@@ -1,4 +1,4 @@
-FROM --platform=linux/amd64 python:3.11
+FROM python:3.11
 
 
 # Set the working directory in the container

fastapi-server/Makefile

@@ -1,14 +1,19 @@
+LOCAL_ARCH=$(shell uname -m)
+GCLOUD_ARCH=amd64
+IMAGE_NAME=langroid-server
+TAG=latest
+
 server:
-	docker build -t server .
+	docker build --platform linux/$(LOCAL_ARCH) -t $(IMAGE_NAME):$(TAG) .
 
 run:
-	docker run -d -p 80:80 server
+	docker run --env-file .env -d -p 80:80 $(IMAGE_NAME):$(TAG)
 
 stop:
 	docker stop $(shell docker ps -q)
 
-gbuild:
-	docker build -t gcr.io/langroid/langroid-server:v1 .
+gserver:
+	docker build --platform=linux/$(GCLOUD_ARCH) -t gcr.io/langroid/$(IMAGE_NAME):$(TAG) .
 
 gpush:
-	docker push gcr.io/langroid/langroid-server:v1
+	docker push gcr.io/langroid/$(IMAGE_NAME):$(TAG)

fastapi-server/README.md

@@ -1,8 +1,16 @@
 # Setup a REST API server on GCP for a Langroid script
 
+Ensure that your env secrets are in this folder as a `.env` file.
+We will of course *not* include it in the docker image
+(which is why it's in the `.dockerignore` file) but we will use it
+for local testing by passing this file in as an env file to the uvicorn server.
+(See the defn of `make run` in the Makefile.)
+
 ## local testing
 
-Build server using `make server`, run it with `make run`
+Build server using `make server`, run it with `make run`.
+See the definitions of these in the Makefile. Notice that 
+we are passing in the env variables
 
 ## Curl examples to test the server locally
 
@@ -25,6 +33,13 @@ curl -X POST \
 ```
 (If you omit the `-o out.txt` part, the response will be printed to the terminal.)
 
+`process_file_and_data` endpoint:
+
+```bash
+curl -X POST "http://localhost:8000/process_file_and_data/" \
+     -F "file=@/tmp/query.txt" \
+     -F 'complex_data={"id": 1, "item": {"name": "Item Name", "description": "A description", "quantity": 10, "tags": ["tag1", "tag2"]}, "related_items": [{"name": "Related Item 1", "description": "Description 1", "quantity": 5, "tags": ["tag3", "tag4"]}, {"name": "Related Item 2", "description": "Description 2", "quantity": 3, "tags": ["tag5", "tag6"]}]}'
+```
 
 ## Deploy to google cloud run
 
@@ -36,20 +51,12 @@ See other details here:
 https://chat.openai.com/share/c34583c8-b88e-4a70-bf24-83229700c020
 
 
-Run these from within the dir where the Dockerfile is located:
-
-```bash
-gcloud auth configure-docker
-docker build -t gcr.io/langroid/langroid-server:v1 .
-docker push  gcr.io/langroid/langroid-server:v1
-```
-
-The `build` and `push` cmds are also available via the Makefile 
-as `make gbuild` and `make gpush` respectively.
+Run `make gserver`, `make gpush` from within the dir where the Dockerfile is 
+located.
 
 Go to Google Cloud Run Service and create a new service, 
-selecting the latest version of the pushed docker image above: 
-`gcr.io/langroid/langroid-server:v1`
+selecting the latest version of the pushed docker image above, e.g.: 
+`gcr.io/langroid/langroid-server:latest`
 
 When setting up the service:
 - ensure you select the same port number as in the Dockerfile, e.g. 80.
@@ -61,6 +68,33 @@ If the service fails to start due to an error like `uvicorn: exec format error`,
 then you may be able to fix it by explicitly choosing an architecture 
 in the Dockerfile, e.g. `linux/amd64` (which we chose in the Dockerfile).
 
+### Creating secrets in google cloud
+Some commands for quick reference:
+
+```bash
+gcloud secrets create openai-api-key --replication-policy="automatic"
+echo -n "your-openai-api-key" | gcloud secrets versions add openai-api-key --data-file=-
+```
+
+After creating your secret and adding its value, you may need to set appropriate 
+permissions for the secret. Use gcloud secrets add-iam-policy-binding to grant access 
+to the secret:
+
+```bash
+gcloud secrets add-iam-policy-binding openai-api-key \
+  --member="serviceAccount:[email protected]" \
+  --role="roles/secretmanager.secretAccessor"
+```
+
+To expose one of these as environment var named `OPENAI_API_KEY` in the cloud run 
+service:
+
+```bash
+gcloud run services update langroid-server \
+ --update-secrets OPENAI_API_KEY=openai-api-key:latest \
+ --region=us-east4
+```
+
 ## Test GCP endpoints
 
 Same curl cmds as above, but use the endpoint url from the GCP Cloud Run service, 

fastapi-server/app.py

@@ -1,7 +1,7 @@
-from typing import Any
-from fastapi import FastAPI, File, UploadFile
+from typing import Any, Dict, List
+from fastapi import FastAPI, File, UploadFile, Form
 from fastapi.responses import FileResponse
-from pydantic import BaseModel
+from pydantic import BaseModel, Json
 import uvicorn
 import os
 import langroid as lr
@@ -10,6 +10,19 @@
 class TextInput(BaseModel):
     text: str
 
+# Define your Pydantic models for the complex payload
+class Item(BaseModel):
+    name: str
+    description: str
+    quantity: int
+    tags: List[str]
+
+class ComplexItem(BaseModel):
+    id: int
+    item: Item
+    related_items: List[Item]
+
+
 class Server:
     def __init__(self):
         self.setup()
@@ -60,6 +73,22 @@ def process_text(text_input: TextInput) -> Any:
     result = server.serve_text(text_input.text)
     return {"message": result, "status": "ok"}
 
+@app.post("/process_file_and_data/")
+async def process_file_and_data(
+    file: UploadFile = File(...),
+    complex_data: Json[ComplexItem] = Form(...)
+) -> Dict[str, Any]:
+    # You can now access the file and the complex data
+    # For example, save the file and process the complex data
+    # Here, just return a message indicating success for demonstration
+
+    return {
+        "message": "Received file and complex data successfully!",
+        "file_name": file.filename,
+        "complex_data_id": complex_data.id,
+        "complex_data_item_name": complex_data.item.name
+    }
+
 # This port number must match the port number in the Dockerfile
 if __name__ == "__main__":
     uvicorn.run(app, host="0.0.0.0", port=80)