Check spelling using codespell#

Today I discovered codespell via this Rich commit. codespell is a really simple spell checker that can be run locally or incorporated into a CI flow.

codespell is designed to run against source code. Instead of using a dictionary of correctly spelled words it instead uses a dictionary of known common spelling mistakes, derived from English Wikipedia (defined here). This makes it less likely to be confused by variable or function names, while still being able to spot spelling mistakes in comments.

Basic usage:

1
pip install codespell
2
codespell
3
# Or point it at a folder, or files in that folder:
4
codespell docs/*.rst

This outputs any spelling errors it finds in those files. I got this the first time I ran it against the Datasette documentation:

1
docs/authentication.rst:63: perfom ==> perform
2
docs/authentication.rst:76: perfom ==> perform
3
docs/changelog.rst:429: repsonse ==> response
4
docs/changelog.rst:503: permissons ==> permissions
5
docs/changelog.rst:717: compatibilty ==> compatibility
6
docs/changelog.rst:1172: browseable ==> browsable
7
docs/deploying.rst:191: similiar ==> similar
8
docs/internals.rst:434: Respons ==> Response, respond
9
docs/internals.rst:440: Respons ==> Response, respond
10
docs/internals.rst:717: tha ==> than, that, the
11
docs/performance.rst:42: databse ==> database
12
docs/plugin_hooks.rst:667: utilites ==> utilities
13
docs/publish.rst:168: countainer ==> container
14
docs/settings.rst:352: inalid ==> invalid
15
docs/sql_queries.rst:406: preceeded ==> preceded, proceeded

You can create a file of additional words that it should ignore and pass that using the --ignore-words option:

1
codespell docs/*.rst --ignore-words docs/codespell-ignore-words.txt

Since I don’t have any words in that file yet I added one fake word, so my file looks like this:

1
AddWordsToIgnoreHere

Each ignored word should be on a separate line.

I added it to my GitHub Actions CI like this:

1
name: Check spelling in documentation
2

3
on: [push, pull_request]
4

5
jobs:
6
  spellcheck:
7
    runs-on: ubuntu-latest
8
    steps:
9
    - uses: actions/checkout@v2
10
    - name: Set up Python ${{ matrix.python-version }}
11
      uses: actions/setup-python@v2
12
      with:
13
        python-version: 3.9
14
    - uses: actions/cache@v2
15
      name: Configure pip caching
16
      with:
17
        path: ~/.cache/pip
18
        key: ${{ runner.os }}-pip-spellcheck
19
        restore-keys: |
20
          ${{ runner.os }}-pip-spellcheck
21
    - name: Install dependencies
22
      run: |
23
        pip install codespell
24
    - name: Check spelling
25
      run: codespell docs/*.rst --ignore-words docs/codespell-ignore-words.txt

Now any push or pull request will have the spell checker applied to it, and will fail if any new incorrectly spelled words are detected.

Here’s the full PR where I added this to Datasette, and the commit where I added this to sqlite-utils.