restructure examples, add nextjs and fastapi starters (#7)

Author: aaronvg

README.md

@@ -6,21 +6,17 @@ Clone the repo to see syntax highlights using the BAML VSCode extension!
 
 ### Requirements
 
-1. BAML CLI, BAML VSCode extension
-2. This repository uses ⚠️⚠️⚠️ Poetry ⚠️⚠️⚠️ as the python environment manager. If you want to use Conda, pip or another dependency mgmt, run `baml init` to get yourself setup properly.
-3. Python >=3.9 (Contact us if you have an older version).
+1. BAML CLI,
+2. BAML VSCode extension
+3. OPENAI_API_KEY is set in your .env file. See .env.example at the root.
 
 **Contact us on Discord if you need help running the examples using Conda, pip or another dependency mgmt**.
 
 ### Setup
 
-Note: You can always just copy the `.baml` files you want into your own project that you have initialized with `baml init`.
+We recommend running `baml init` in your own project (unless you just want to clone the NextJS or FastAPI starter projects). Then just copy the .baml files and functions you want.
 
-1. Clone the repo
-2. Install [poetry](https://python-poetry.org/docs/)
-3. Run `poetry shell` in the root
-4. Run `poetry install`
-5. Make sure you can ctrl + s one of the .baml files after you install the BAML VSCode extension to generate a baml_client dir.
+Make sure you can ctrl + s one of the .baml files after you install the BAML VSCode extension to generate a baml_client dir.
 
 ## Troubleshooting
 
@@ -30,6 +26,7 @@ Some common steps that help fix things:
 
 1. Make sure you also add `baml` pip package to the project dependencies if you are not using poetry (see pyproject.toml for dependency list).
 1. Make sure you're in a poetry shell when you run the python main.py files.
+1. Enable Python > Analysis: Type Checking Mode - Basic or greater.
 1. Make sure environment variables are set (some baml files use an env.OPEN_AI_KEY). Add a .env file to the root dir with the appropriate api key set to fix this.
 1. Restart VScode if the playground isn't working
 1. Restart VSCode if you're not getting error highlights for baml-generated code, or ensure the right Python interpreter is set (Command + Shift + P -> Select interpreter)

fastapi-starter/README.md

@@ -0,0 +1,20 @@
+## Setup
+
+Open this project up in VSCode (we recommend only opening this folder, and not at the root, or VSCode may not detect the python environment and you may not get type completions for BAML functions).
+
+Ensure your `settings.json` has:
+
+```
+{
+  "python.analysis.typeCheckingMode": "basic"
+}
+```
+
+1. Run `poetry install`
+2. Run `poetry shell`
+3. Open up vscode command palette (command + shift + p, and select the .venv folder that was created in this directory as the interpreter)
+4. Run `uvicorn fastapi_starter.app:app --reload`
+5. Curl the streaming endpoint:
+   ```
+   curl -X GET -H "Content-Type: application/json" http://localhost:8000/extract_resume
+   ```

fastapi-starter/baml-README.md

@@ -0,0 +1,46 @@
+# Getting started with BAML
+
+## Installations
+
+Make sure to download the [VSCode Playground](https://marketplace.visualstudio.com/items?itemName=gloo.baml).
+
+To use BAML with either python or typescript, you should run:
+
+```shell
+$ baml update-client
+```
+
+This will keep client side libraries in sync. It also prints the commands being run, which you can run manually if they fail.
+
+## Running tests
+
+You can run tests via:
+
+```shell
+# To run tests
+$ baml test run
+
+# To list tests
+$ baml test
+
+# For more help
+$ baml test --help
+```
+
+## Integrating BAML with python / ts
+
+You can run:
+
+```shell
+$ python -m baml_example_app
+```
+
+The `baml_example_app.py` file shows how to import from the code BAML generates.
+
+## Deploying
+
+You don't need the BAML compiler when you deploy / release. Your `baml_client` folder contains everything you may need.
+
+## Reporting bugs
+
+Report any issues on our [Github](https://www.github.com/boundaryml/baml) or [Discord](https://discord.gg/BTNBeXGuaS)

fastapi-starter/baml_example_app.py

@@ -0,0 +1,100 @@
+"""
+Run this script to see how the BAML client can be used in Python.
+
+python -m example_baml_app
+"""
+
+import asyncio
+from baml_client import baml as b
+from datetime import datetime
+from typing import List
+from typing_extensions import TypedDict
+
+async def extract_resume(resume: str) -> None:
+    """
+    Extracts the resume and prints the extracted data.
+    """
+    print("Parsing resume...")
+    print(resume[:100] + "..." if len(resume) > 100 else resume)
+    parsed_resume = await b.ExtractResume(resume)
+    print(parsed_resume.model_dump_json(indent=2))
+
+    await asyncio.sleep(1)
+    print("\n\nNow extracting using streaming")
+    async with b.ExtractResume.stream(resume) as stream:
+        async for x in stream.parsed_stream:
+            if x.is_parseable:
+                print(f"streaming: {x.parsed.model_dump_json()}")
+        response = await stream.get_final_response()
+        if response.has_value:
+            print(f"\n final: {response.value.model_dump_json(indent=2)}")
+        else:
+            print("No final response")
+
+
+class ChatMessage(TypedDict):
+    sender: str
+    message: str
+
+
+async def classify_chat(messages: List[ChatMessage]) -> None:
+    """
+    Classifies the chat and prints the classification.
+    """
+    print("Classifying chat...")
+    chat = "\n".join(map(lambda m: f'{m["sender"]}: {m["message"]}', messages))
+    print(chat[:100] + "..." if len(chat) > 100 else chat)
+
+    classification = await b.ClassifyMessage(
+        message=chat, message_date=datetime.now().strftime("%Y-%m-%d")
+    )
+    print("Got categories: ", classification)
+
+
+async def main():
+    resume = """
+    John Doe
+    1234 Elm Street
+    Springfield, IL 62701
+    (123) 456-7890
+
+    Objective: To obtain a position as a software engineer.
+
+    Education:
+    Bachelor of Science in Computer Science
+    University of Illinois at Urbana-Champaign
+    May 2020 - May 2024
+
+    Experience:
+    Software Engineer Intern
+    Google
+    May 2022 - August 2022
+    - Worked on the Google Search team
+    - Developed new features for the search engine
+    - Wrote code in Python and C++
+
+    Software Engineer Intern
+    Facebook
+    May 2021 - August 2021
+    - Worked on the Facebook Messenger team
+    - Developed new features for the messenger app
+    - Wrote code in Python and Java
+    """
+    await extract_resume(resume)
+
+    messages = [
+        {"sender": "Alice", "message": "I'm having issues with my computer."},
+        {
+            "sender": "Assistant",
+            "message": "I'm sorry to hear that. What seems to be the problem?",
+        },
+        {
+            "sender": "Alice",
+            "message": "It's running really slow. I need to return it. Can I get a refund?",
+        },
+    ]
+    await classify_chat(messages)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

fastapi-starter/baml_src/__tests__/ClassifyMessage/blue_dolphin.json

@@ -0,0 +1,6 @@
+{
+  "input": {
+    "message": "This is so frustrating, i bought a laptop and it's not working properly. I want to return it and get my money back. I'm so disappointed",
+    "message_date": "2019-01-01T00:00:00Z"
+  }
+}

fastapi-starter/baml_src/__tests__/ClassifyMessage/case1.json

@@ -0,0 +1,6 @@
+{
+  "input": {
+    "message": "Hi! I'm having an issue with my account. Can you help me?",
+    "message_date": "2019-01-01T00:00:00Z"
+  }
+}

fastapi-starter/baml_src/__tests__/ExtractResume/jason.json

@@ -0,0 +1,3 @@
+{
+  "input": "Jason Doe\nPython, Rust\nUniversity of California, Berkeley, B.S.\nin Computer Science, 2020\nAlso an expert in Tableau, SQL, and C++\n"
+}

fastapi-starter/baml_src/__tests__/ExtractResume/sarah.json

@@ -0,0 +1,3 @@
+{
+  "input": "Sarah Montez\nHarvard University\nMay 2015-2019\n3.92 GPA\nGoogle\nSoftware Engineer\nJune 2019-Present\n- Backend engineer\n- Rewrote search and uplifted metrics by 120%\n- Used C++ and Python\nMicrosoft\nSoftware Intern\nJune 2018-August 2018\n- Worked on the Windows team\n- Updated the UI\n- Used C++\n"
+}

fastapi-starter/baml_src/clients.baml

@@ -0,0 +1,23 @@
+client<llm> GPT4 {
+  provider baml-openai-chat
+  options {
+    model gpt-4
+    api_key env.OPENAI_API_KEY
+  }
+}
+
+client<llm> GPT4Turbo {
+  provider baml-openai-chat
+  options {
+    model gpt-4-1106-preview
+    api_key env.OPENAI_API_KEY
+  }
+}
+
+client<llm> GPT3 {
+  provider baml-openai-chat
+  options {
+    model gpt-3.5-turbo
+    api_key env.OPENAI_API_KEY
+  }
+} 

fastapi-starter/baml_src/example_1.baml

@@ -0,0 +1,33 @@
+class Resume {
+  name string
+  education Education[]
+  skills string[]
+}
+
+class Education {
+  school string
+  degree string
+  year int
+}
+
+function ExtractResume {
+  input string
+  output Resume
+}
+
+impl<llm, ExtractResume> version1 {
+  client GPT4
+  prompt #"
+    Parse the following resume and return a structured representation of the data in the schema below.
+
+    Resume:
+    ---
+    {#input}
+    ---
+
+    Output JSON format (only include these fields, and no others):
+    {#print_type(output)}
+
+    Output JSON:
+  "#
+}