Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/jrmontag/talk-python-fastapi
OpenWeather API from TalkPython course
https://github.com/jrmontag/talk-python-fastapi
api-rest openweather-api python3
Last synced: 2 days ago
JSON representation
OpenWeather API from TalkPython course
- Host: GitHub
- URL: https://github.com/jrmontag/talk-python-fastapi
- Owner: jrmontag
- License: mit
- Created: 2022-11-10T23:25:52.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2023-10-31T02:19:53.000Z (about 1 year ago)
- Last Synced: 2024-11-18T02:45:14.962Z (about 1 month ago)
- Topics: api-rest, openweather-api, python3
- Language: Python
- Homepage: http://wx.jrmontag.xyz/
- Size: 52.7 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# notes
- [course site](https://training.talkpython.fm/courses/getting-started-with-fastapi)
- [course repo](https://github.com/talkpython/modern-apis-with-fastapi)
## Ch 3: a first api
- [HTTP status codes](https://www.webfx.com/web-development/glossary/http-status-codes/)
## Ch 4: modern language foundations
- type hints
- helpful in editor, still separate from runtime
- async/await
- mostly useful for improving scalability during i/o heavy work like db r/w
- wsgi vs asgi
- write an application in one framework e.g. flask
- put it on heroku, run under gunicorn, etc
- web frameworks plug into a general architecture: wsgi
- web service gateway interface
- expects a method e.g. request(), returns response over network
- to avoid breaking changes for async, introduce asgi e.g. uvicorn
- pydantic classes
- rich editor support
## Ch 5: a realistic api
- html uses jinja templating
- try a vs code extension? requires different file extensions e.g. `.html.j2` or `.jinja`
- bunch of layout, css, etc - seems like just lift this content for now
- perhaps use something like md -> html in the future
- interesting that api routers don't require `__init__.py` in directories
- initial `/weather` api path doesn't work as demonstrated in the lecture recording - requires exact path match (no extra params)
- managing secrets
- `.gitignore` a local json file; could also use env vars
- don't end up part of shhgit
## Ch 6: error handling and perf
- use of async/await not well motivated yet here
- cache response in memory to start; maybe not the most elegant solution, but works
- esp in production, likely have multiple instances/processes running and in-memory cache won't be shared
- alternatives might be redis, another db service, etc
- additional benefit: fewer API calls against budget
- web server app is a rare places where it's acceptable a blank exception catch (+ logging)
## Ch 7: inbound data
- to simplify, again choose an in-memory storage for reports in lieu of database
- [Postman app](https://www.postman.com/) (free + no account needed) is useful for crafting POSTs for local API
- more nuance to the 20X HTTP code -> more developer-friendly e.g. after POST request
- make a little client app with `input()` and `localhost` requests
- jinja templating is a bit unintuitive
- OpenAPI (fka Swagger) docs are neat
## Ch 8: deployment (linux x gunicorn x nginx)
- spectrum of options for "where should I deploy my python app?"
- Platform as a Service (PaaS) e.g. heroku: easy, and good python support; less control, can be expensive
- Virtual Private Servier (VPS) e.g. [DigitalOcean](https://m.do.co/c/a4ad5ae8d042), [Linode](talkpython.fm/linode): lower cost; some server maintainance
- talkpython stuff runs on Linode
- Compute / Platform as a Service e.g. AWS: complicated and expensive
- AWS Lightsail: new offering, lower cost, competing with DO and Linode
- follow along with DO setup
```
$ ssh root@
# apt update
# apt upgrade
# reboot
```
- architecture
- nginx (web server, serves html/css, talks to internet)
- gunicorn (manages app lifecycle, restart stuck apps)
- uvicorn (+ python code; many copies of application (processes), will distribute requests appropriately)
- follow along with scripts/server_setup.sh
- clone from course repo to start (TODO: come back and get my copy running)
- [] read more about the commands/programs used in server_setup.sh
- [] read more about the weather.service configuration and systemd
- edit weather.nginx IP mapping in-place
- SSL setup fails as written, but would be the approach to use after pointing an owned domain to the IP address (after DNS records have updated)
- posting a report to deployed version
```
curl --header "Content-Type: application/json" \
--request POST \
--data '{"description":"Sunshine clearing away the snow","location": {"city":"Denver"}}' \
http:///api/reports
```
Update Oct 2023: moved this app to a new host; now accessible at [http://wx.jrmontag.xyz/](http://wx.jrmontag.xyz/)
# course setup
- note that current dev machine is old so likely some brittle pieces
- may have to skip to docker or something if things fall apart
## python install
- try out [pyenv](https://github.com/pyenv/pyenv) for this project on recommendation
- set global to latest, point to that one for new venvs
```
$ pyenv global 3.11
$ # cd into each ch dir
$ python3 -m venv venv
$ . venv/bin/activate
(venv) $ pip install -r requirements.txt
(venv) $ python main.py
```
## IDE
- recommend pycharm or vs code
- install Python plugin (microsoft)