Weather Project - Initial Release

Series

• Weather Project
• Weather Project - PyPi
• Weather Project - Requirements and Roadmap
• Weather Project - Initial Release

Overview

From the previous article, the Weather Project is now on PyPi and is on Github. This article will be a bit of an update and the steps to the initial working release.

Current Status

The last update has been quite awhile ago. The initial goal the project more or less hasn’t change but I have change some of the goals of the project. This program will still have the goal of pulling data from OpenWeather but for now only the One Call API since this is a free API and contains a lot of data:

• Current weather
• Minute forecast for 1 hour
• Hourly forecast for 48 hours
• Daily forecast for 8 days
• National weather alerts
• Historical weather data for 40+ years back (since January 1, 1979)

The latest release (0.1.5), openweather_report will only get all the data listed but the last bullet point.

Switch from Poetry to Flit

In the past year I used Poetry in multiple projects at work and the workflow was getting in my way. Most of my projects are relatively simple and a lot of the features that could help a large project was slowing me down. I went back to using Flit with pip-tools and pip-chill. This provides what I need while still being performant.

This release I didn’t use pip-tools and pip-chill but will going forward.

Getting 0.1.5 Released

I already have a crude private version of this program that was getting data from OpenWeather but it wasn’t generic enough to share.

pyproject.toml

Switching from Poetry to Flit was straightforward. Instead of using SQLAlchemy, I decided to use aiosql. The main reason is that the SQL required to save data is simple and I don’t want it mixed within my Python code like my internal program had. SQLAlchemy would be overkill. I also included the development dependencies but only used mypy on this release. Once I get the rest of the tools setup, I will go into more detail of the settings in my pyproject.toml.

I did figure out how to get pipx to recognize that this was a CLI program and not a library by adding the following:

[project.scripts]
openweather_report = "openweather_report.cli:main"

README.md

I wanted to make sure that the README.md had a quickstart as well as describing what is required to be able to save to PostgreSQL and SQLite. The wiki will still house the detail documentation.

CLI Arguments

The program will work using the latitude and longitude of the city you would like the weather conditions. The API Key is a required as well. An invalid API Key will cause OpenWeather to return JSON that describes the error. Optionally:

• --save: Will either save to JSON file, PostgreSQL or SQLite.
• --save_path: This is only for saving to JSON and will default to a timestamp filename.
• --db_string: This is only for the databases.
  • For PostgreSQL it will be postgres://username:password@host:port/database.
  • For SQLite it will be the filename.

The program will crash if the save location is incorrect or if a database the tables not set up ahead of time.

Save to JSON

This was straightforward since I would just dump to a file. There is an option to provide a filename but will default to a timestamp filename.

PostgreSQL

It requires a schema of weather and a table weather.raw_json_data which is described in the README. Will print to screen if there is an error but won’t stop the program from running.

SQLite

It is similar to PostgreSQL except there isn’t a schema. The class is similar to PostgreSQL and will get cleaned up in the next release.

SQL Code

By using aiosql, I have the SQL separate from the Python code. Right now there are two function save_json_data and save_json_data_no_schema. I may decide to separate openweather.sql into one from PostgreSQL and one from SQLite as the project grows.

Next Steps

I verified that the program works and was able to replace my crusty internal version. I did create a helper program so that I could use this program to loop through the cities I track the weather. I should provide at least an example of this file so others could do the same. I need to flesh out the Roadmap and get the program tested before it gets any larger.

I also would like to explain how to take this data and do some basic analysis. For example it can be slow to query JSON in PostgreSQL and I plan on showing how to take the JSON data and put it into separate indexed tables to make querying more performant.

Finally I really need to revisit Cookiecutter to help me be more consistent with my projects.