check-outdated-content.yml

# Contribution by Cloud Native Guatemala folks :)
# This workflow will check if a localized content is outdated or not
# by comparing English content in the old branch and the latest branch.
---
name: Check outdated content

on:
  pull_request:
    branches:
      - main
  workflow_dispatch:

permissions:
  contents: read
  issues: write 

jobs:
  check-outdated-content:
    name: Check outdated content
    runs-on: ubuntu-latest
    steps:
    - name: Checkout 
      uses: actions/checkout@v4
      with:
        fetch-depth: 0 # fetch all history for all tags and branches
    - name: Detecting pending translations
      shell: bash
      run: |
        #####################################################
        # Set variables to detect pending translations
        OUTPUT_DIR="./outdated"
        # Directory to compare for pending translations
        CONTENT_DIR="website/content"
        # Here we should add a new language every time a new one is added e.g: ("es" "zh" "de" "etc")
        languages=("es" "zh")
        # This is the base language to use, in this case English will be the base language
        base_lang="en"

        cd $CONTENT_DIR
        # The following lines create the necessary files to compare 
        # pending translations compared to the base_lang defined
        # This files will contain the list of pending files to translate for each language
        for lang in ${languages[@]}; do
            touch $lang.txt
        done
        # This line create the base language file where the pending
        touch $base_lang.txt


        # Make an output directory where the pending translated files are detected
        if [[ ! -e $OUTPUT_DIR ]]; then
            mkdir $OUTPUT_DIR
        elif [[ ! -d $OUTPUT_DIR ]]; then
            echo "$OUTPUT_DIR already exists but is not a directory" 1>&2
        fi

        # This part perform the comparition between the main branch and the PR branch to compare
        # and detect pending translations for languages different than English (EN)
        LATEST_BRANCH=${GITHUB_REF#refs/}
        echo "(DEBUG) LATEST_BRANCH: ${LATEST_BRANCH}"

        # Get the old branch from 'github.base_ref' 
        # The old branch can be 'upstream/dev-ko'
        OLD_BRANCH="origin/${{github.base_ref}}"
        echo "(DEBUG) OLD_BRANCH: ${OLD_BRANCH}"
            
        # This function output a file inside the OUTPUT_DIR that is pending to translate for
        # the current supported languages
        compare () {
            FILE_PATH=$1
            # Actually compare between the old and latest English content and log diff in the file
            # This detect
            if [[ -f "${FILE_PATH}" ]]; then
                # File exists
                # Check changes
                git diff ${OLD_BRANCH}..${LATEST_BRANCH} -- ${FILE_PATH} > temp.diff
                # This detect if something was changed in English (Base Language)
                if [[ -s "temp.diff" ]]; then
                    echo "(DEBUG) ${FILE_PATH} is outdated."
                    mkdir -p ${OUTPUT_DIR}/${FILE_PATH%/*}
                    mv temp.diff ${OUTPUT_DIR}/${FILE_PATH}
                else 
                # Detect if the file doesn't exists for the supported languages. This means that the file is 
                # pending to translate
                    echo "check if ${FILE_PATH} exist in other languages"
                    for lang in ${languages[@]}; do
                        NEW_FILE_PATH=$(echo "${FILE_PATH}" | sed -e "s/${base_lang}\//${lang}\//g")
                        if [[ ! -e "${NEW_FILE_PATH}" ]]; then
                            echo "The file ${FILE_PATH} needs to be translated to $lang"
                            echo ${NEW_FILE_PATH} >> $lang.txt
                        fi
                    done
                fi
            else
                echo "(DEBUG) ${FILE_PATH} does not exist."
            fi
        }

        # Get the list of files in the website for the base language (EN) and output this in files.txt
        find $base_lang -iname "*.md" > files.txt
        # Sort this file by name the final output is in files.txt
        sort files.txt > files_temp.txt;mv files_temp.txt files.txt
        # INFILE contains the name of this temporary file
        INFILE=files.txt

        # This loop call the function compare and send the file path as a parameter to check if already exists
        # in the directory for the supported languages (e.g: website/es, website/zh ) different than English (EN)
        while IFS= read -r line
        do
          compare "$line"
        done < "$INFILE"

        # This can be used to support changes for the base language English
        find outdated -iname "*.md" | sed -e "s/outdated/website\/content/g" > $base_lang.txt
            
    - name: Generate content for translation issues
      id: to_translate
      shell: bash
      run: |
        # This step generates the md file per language used to create the issue for the
        # pending files to translate using the action JasonEtco/create-an-issue@v2
        # also this creates the file for each language defined in the languages variable
        CONTENT_DIR="website/content"
        cd $CONTENT_DIR
        echo "(DEBUG) Current folder: "$(pwd)
        languages=("es" "zh")
        for lang in ${languages[@]}; do
            FILES=""

            # This part autoasign the different groups depending on the language to translate to the
            # issue to be created
            if [ "$lang" == "es" ]; then
                TEAM="Dianmz"
            elif [ "$lang" == "zh" ]; then
                TEAM="sergioarmgpl"
            fi

            # This lines output the file to be used to create the issue for pending translations per language
            echo "---" > $lang"_files.md"
            echo "title: Pending pages to translate into $lang" >> $lang"_files.md"
            echo "assignees: $TEAM" >> $lang"_files.md"
            echo 'labels:' >> $lang"_files.md"
            echo '  - help wanted' >> $lang"_files.md"
            echo '  - translations' >> $lang"_files.md"
            echo '  - good first issue' >> $lang"_files.md"
            echo '  - issue/tracking' >> $lang"_files.md"
            echo "---" >> $lang"_files.md"
            echo "Last change by: {{ payload.sender.login }}." >> $lang"_files.md"
            echo -e "*Docs to translate for "$lang" language* <br />\n" >> $lang"_files.md"
            while IFS= read -r line
            do
              FILES+="- $line <br />\n"
            done < "$lang.txt"
            echo -e $FILES >> $lang"_files.md"
        done
    # Creates the issue of pending files to translate for Spanish language
    - uses: JasonEtco/create-an-issue@v2
      env:
        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
      with:
        update_existing: true
        filename: website/content/es_files.md
    # Creates the issue of pending files to translate for Chinese language
    - uses: JasonEtco/create-an-issue@v2
      env:
        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
      with:
        update_existing: true
        filename: website/content/zh_files.md