This PR adds a new script/pre-commit-hook, named remove-improper-eol-in-cn-docs.

The script aims at resolving extra whitespaces in Chinese docs, which is a long-standing HTML issue as discussed here.

To solve the issue, this script finds and removes end_of_line characters which split natural Chinese paragraphs. For example,

这是一个， 像诗一样的 测试

will be changed to

这是一个，像诗一样的测试

However, the following cases stay unchanged:

• Docs written in English

This is, a poem-like test

• Natural paragraphs (split by 2+ eol characters in Markdown)

这是一个

测试

Modify .pre-commit-config.yaml with the following config to test this hook.

  - repo: https://github.com/zhouzaida/pre-commit-hooks-1
    rev: 8dd4395  # Use the ref you want to point at
    hooks:
    -   id: check-copyright
        args: ["dir_to_check", "not-found-dir", "--ignore-file-not-found-error"] 

Motivation

https://github.com/open-mmlab/mmflow/pull/87

Upgrade the version of isort.

Modification

• Remove seed-isort-config in pre-commit hook
• Upgrade isort to v5.10.1 (latest version)
• Replace known_standard_library with extra_standard_library

Modify .pre-commit-config.yaml with the following config to test this hook.

  - repo: https://github.com/zhouzaida/pre-commit-hooks-1
    rev: 0672dd0  # Use the ref you want to point at
    hooks:
      - id: check-model-index
        args: ["model-index.yml"]  # replace the dir_to_check with your expected directory to check
        additional_dependencies:
          - cerberus

In addition, since YAML does not support tuple type, the resolution field in a metafile is recognized as a string, so fix this type to be a list. You can use the following regular expression global replacement in vs code

• resolution: \(([0-9]*), ([0-9]*)\)
• resolution: [$1, $2]