Added new demo "Wikipedia assistant"

Author: Cutwell

python-web-io/.envrc.example.sh

@@ -1,6 +1,6 @@
 # NOTE: save this file as .envrc and activate using `direnv allow` from same directory
 
-# if testing `tests/linux_gpt.py`, set your OpenAI API Key
+# if testing `examples/wikipedia_assistant.py`, set your OpenAI API Key
 export OPENAI_API_KEY=""
 
 # server env vars

python-web-io/.pythonwebio/config.toml

@@ -1,5 +1,5 @@
 [script]
-filepath = "examples/color_names.py"
+filepath = "examples/wikipedia_assistant.py"
 entrypoint = "main"
 
 [page]

python-web-io/Makefile

@@ -6,7 +6,7 @@ install: ## install dependencies
 
 publish: ## build module and publish as new version to PyPI
 	@poetry build
-	@poetry publish
+	@poetry publish --username=$(PYPI_USERNAME) --password=$(PYPI_PASSWORD)
 
 help: ## display this help screen
     @grep -h -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-30s\033[0m %s\n", $$1, $$2}'

python-web-io/README.md

@@ -1,51 +1,60 @@
-# Python Web I/O
+# <img src="./.github/logo.png" style="width:40px;padding-right:10px;margin-bottom:-8px;" alt="Sticker of a cute yellow Python snake, representing the use of the Python programming language in this project."> Python Web I/O
  Generate a webpage as a GUI for a Python script, and serve from anywhere.
 
-## Usage
-```
-$ export SECRET_KEY="someSecureSecretKey"
-$ python_web_io .\example.py
-```
-* Create a `.envrc` file, setting `SECRET_KEY` as per [`python_web_io/.envrc.example`](https://github.com/Cutwell/python-web-io/blob/main/python_web_io/.envrc.example).
-* Try running the [`example.py`](https://github.com/Cutwell/python-web-io/blob/main/python_web_io/example.py) script using `python_web_io example.py`.
+## Documentation
+Check out the [wiki](https://github.com/Cutwell/python-web-io/wiki).
 
-|Argument|||
-|:---:|:---:|:---:|
-|`"example.py"`|Required|Specify the file path for the app Python script / entrypoint.|
-|`--debug`|Optional|Run the Flask server with debug output enabled.|
+## Quickstart
 
-## Config
-### Magic
-`input()` and `print()` both support the `magic` keyword argument. For `input()`, `magic` sets the `type` of the input html element. For `print()`, `magic` sets the element type.
+Install `python-web-io` locally using:
+```bash
+pip install python-web-io
+```
 
-||Magic|Default|
-|:---:|:---:|:---:|
-|`input()`|`button`, `checkbox`, `color`, `date`, `datetime-local`, `email`, `file`, `image`, `month`, `number`, `password`, `radio`, `range`, `search`, `tel`, `text`, `time`, `url`, `week`|`text`|
-|`print()`|`style`, `img`, `address`, `footer`, `aside`, `header`, `h1..6`, `blockquote`, `p`, `b`, `abbr`, `code`, `em`, `i`, `mark`, `q`, `s`, `small`, `span`, `strong`, |`p`|
+Or via `poetry` using:
+```bash
+poetry add python_web_io
+```
 
-#### Arguments
-`input()` and `print()` both support the `magic_args` keyword argument. `magic_args` accepts a dictionary, which can be used to set attributes for the html element.
+If evaluating / testing `python-web-io`, install dependencies for the example apps using:
+```bash
+poetry add python_web_io --with examples
+```
 
-### Cache
-The user script is re-evaluated after each user interaction, to progress the script to the next `input()`, etc. This means expensive functions may be called more than once per session. To reduce latency, a cache decorator is made available through the `python_web_io` module. The `@cache_to_file()` decorator accepts a single argument: `file_path`, which indicates where the cache (a `.pkl` file) should be stored.
+After installing the project, some environment setup is required:
 
-```python3
-import python_web_io as io
+### Required setup
+Create an `app.py` file containing your script, and an `.envrc` file to store project secrets. (Note: remember to add `.envrc` to your `.gitignore`)
+Look for example apps in [`./python_web_io/examples`](https://github.com/Cutwell/python-web-io/tree/main/python_web_io/examples).
+```
+.
+├── .envrc
+├── config.toml
+└── app.py
+```
 
-@io.cache_to_file('cache.pickle')
-def expensive_function(arg):
-    # Calculate the result here
-    return result
+Add the following environment variables to your `.envrc`. (Note: remember to activate the `.envrc` in your terminal using `direnv allow`)
+```bash
+# server env vars
+export PYTHON_WEB_IO_SECRET=""
+export PYTHON_WEB_IO_CONFIG=".pythonwebio/config.toml" 	# defaults to .pythonwebio/config.toml if not set
 ```
 
-Cache is persistent across sessions, allowing multiple users to access it. Session specific data can be stored using `session` from `flask`.
+Generate a random key for `PYTHON_WEB_IO_SECRET` using this python command line snippet:
+```bash
+python -c 'import secrets; print(secrets.token_hex())'
+```
 
-```python3
-from flask import session
-session['some_var] = 'some_val'
+If testing `wikipedia_assistant.py`, an OpenAI API key will also need to be set.
+```bash
+export OPENAI_API_KEY=""
 ```
 
-Reserved keys for the `session` namespace are: `io` and `counter`. 
+### Running the webapp
+We recommend running `python_web_io` using `uvicorn`:
+```bash
+poetry run uvicorn python_web_io.main:app
+```
 
 ## License
 MIT

python-web-io/examples/README.md

@@ -20,7 +20,7 @@ Download the free [World Cities Database](https://simplemaps.com/data/world-citi
 Demo of LangChain integration, using an OpenAI LLM to imitate a linux terminal.
 
 ```bash
-poetry run python_web_io --config=examples/linux_gpt.toml
+poetry run python_web_io --config=examples/wikipedia_assistant.toml
 ```
 
 ## 🎨 Color names

python-web-io/examples/color_names.py

@@ -72,3 +72,6 @@ def main():
 
     hex_color = input("What's your favourite color?", type="color")    
     print(f"Nice! That colour looks like {get_nearest_color_name(hex_color)}!")
+
+if __name__ == "__main__":
+    main()

python-web-io/examples/linux_gpt.py

@@ -198,4 +198,7 @@ def main():
         f"Last updated ↻: {day_of_week}, {date_time_str}<br>",
         f"Temperature 🌡️: {weather['current_weather']['temperature']}°C<br>",
         f"Wind speed 💨: {weather['current_weather']['windspeed']} km/h, Wind direction 🧭: {winddirection}° ({degrees_to_direction(winddirection)})"
-    )
+    )
+
+if __name__ == "__main__":
+    main()

python-web-io/examples/weather_app.py

@@ -0,0 +1,42 @@
+import openai
+import python_web_io as io
+
+template = "I want you to act like a wikipedia expert. You have knowledge of every topic that exists on wikipedia, and can accurately give information about any query. When answering questions, provide links to every wikipedia page referenced. Format all links using markdown formatting. Do not complete the previous input, generate a new response as its own paragraph."
+
+chat_history = [
+    template
+]
+
+# cache prompt + responses so the conversation being replayed up to N steps doesn't require N API calls.
+@io.cache_to_file("cache.pickle")
+def chat_with_model(prompt):
+    chat_history.append(prompt)
+    response = openai.Completion.create(
+        engine="text-davinci-003",
+        prompt=" ".join(chat_history),
+        max_tokens=1024,  # Adjust as needed
+        n=1,
+        stop=None,
+        temperature=0.5,
+    )
+    return response.choices[0].text.strip()
+
+def main():
+    print("<h2>Wikipedia Assistant</h2>")
+
+    print("Chat with the OpenAI model (type 'exit' to quit)")
+
+    print("<small>System prompt:", template, "</small>")
+
+    while True:
+        print("<br>")
+        user_input = input("👩‍💻 You: ")
+        if user_input.lower() == 'exit':
+            print("🤖: Goodbye!")
+            break
+
+        response = chat_with_model(user_input)
+        print(f"🤖: {response}")
+
+if __name__ == "__main__":
+    main()

python-web-io/examples/wikipedia_assistant.py

@@ -1,5 +1,5 @@
 [script]
-filepath = "examples/linux_gpt.py"
+filepath = "examples/wikipedia_assistant.py"
 entrypoint = "main"
 
 [page]