Preface

--I have a Github account, but I haven't published anything ――I sometimes want a general-purpose text check tool, but I didn't really want it, so let's make it and publish it! ――I heard that the next company uses CircleCI --There is also a free plan, so let's use it! ――I will write down the points that I suffered when doing the above.

I made a tool to format and validate dates in Python

GitHub：https://github.com/nir-takemi/text_verifier --Atmosphere like this (YYYY / MM / DD (w) hh: mm is defined in the correct format)

[Err file path]/Users/xxxx/Documents/test.txt
[Err]Date format does't conform to rules: ２０１６/2/29 (Monday) 23:59
[Err]Date format does't conform to rules: 2016/2/29 (Month) 23：59
[Err]Date format does't conform to rules(NG list): \n2/29 (Monday) 23:59
[Err]2016/2/30 isn't exist: 2016/2/30 (Tue) 23:59

――I wanted to check various texts in the future, so I named it text_ broadly.

Difficulties in creating tools

About Python naming conventions

I was allowed to reference! : List of Python naming conventions

Do you put the test code?

――I will put the conclusion --The test code is also included in TensorFlow published by Google in the world, so I decided that it is okay to put it in. ――As I will describe later, when you want to test with CircleCI, it seems to be troublesome if it is not in the project, so you have to put it in

What is a good library to write test?

--Select pytest. ――I was worried about unittest, but when I asked an experienced Python person, it was intuitively easy to understand, so I chose pytest. It's a simple tool, so you don't have to get lost at this point, and you can wrap it around long ones.

I was allowed to reference! : Notes on getting started with pytest

What does the Python file hierarchy look like in the first place?

――Conclusion I made it like this

tree .
.
├── tests #Test related
│   ├── conftest.py
│   └── test_text_verifier.py
├── textverifier #Main module related
│   ├── config.yaml
│   ├── pattern.yaml
│   └── text_verifier.py
└── verify_text.py #Existence that only executes

--It's a unit test, but I used it as a reference! : Directory structure when writing tests with Python 3 standard unittest

I can't run pytest because I can't load the module ...

--The conclusion is that I created conftest.py (which is executed as a pre-process for the entire test) and described the process to pass the relative path of the module directory. --Reference -Import by directly specifying the directory path -Introduction to pytest --Notes from fighting IT engineers --In the above file hierarchy, I imported the module as follows

from text_verifier import TextVerifier

But such an error

ImportError while importing test module 'xxx'.
Hint: make sure your test modules/packages have valid Python names.
Traceback:
tests/test_text_verifier.py:5: in <module>
    from text_verifier import TextVerifier
E   ModuleNotFoundError: No module named 'text_verifier'

--So, I wrote conftest.py under the test directory as follows.

`conftest.py`


import sys 
import os

sys.path.append(os.path.abspath(os.path.dirname(os.path.abspath(__file__)) + "/../textverifier/"))

About CircleCI

getting started

I was allowed to reference! : I've just started CircleCI, so I've summarized it in an easy-to-understand manner --Basically, I did the following to run the test --CircleCI registration --Link with GitHub --Create config.yml in the project on GitHub side --Install Docker & CicleCI locally and repeat the verification test of the created config.yml --Push config.yml to GitHub --If you push it, it will be executed on the CicleCI side.

--By the way, the final config.yml looks like this

version: 2
jobs:
  build:
    docker:
      - image: circleci/python:latest
    steps:
      - checkout

      - run:
          command: |
            sudo pip install pipenv
            pipenv install
            pipenv run pip install pyyaml
            pipenv run pip install pytest
            pipenv run pytest

I want to display the CircleCI batch because it's a big deal

--This is the readme on GitHub 　

I was allowed to reference! : Display CircleCI badge in README.md on github

Does the test result make an error?

--Create a state where an error occurs once in the local test --push! & Run! (Failed properly!) --The error was the same as the local error, and it became FAILED there. --Batch was also FAILED

Impressions

--As for the tool, I noticed it after making it, but there was also a convenient one called textlint ... I wanted to check the format pattern match for this tool. So it's okay, but I felt that the validity check was good with textlint ――Because I've been an uncle Jenkins for a long time, it was quite meaningful just to know CircleCI's glue! I was used to putting the configuration file (execution shell etc.) on the server side, so it was fresh to put it on the repository side (Is GitHub Actions the same?) ――Since it's still the starting line, I'd like to play with CircleCI in particular ... ――Next, I want to make something that can be published on PyPI.

The story of releasing a Python text check tool on GitHub x CircleCI for the first time