action.yml

name: 'Vectara Documentation Indexer'
branding:
  icon: 'search'
  color: 'purple'
description: 'Send markdown files to Vectara for indexing'
inputs:
  vectaraAccountNumber:
    description: 'Vectara Account Number'
    required: true
  vectaraCorpusID:
    description: 'Vectara Corpus ID'
    required: true
  vectaraClientID:
    description: 'Vectara OAuth 2.0 Client ID'
    required: true
  vectaraClientSecret:
    description: 'Vectara OAuth 2.0 Client Secret'
    required: true
  vectaraIndexingEndpoint:
    description: 'Vectara Indexing Endpoint'
    default: 'api.vectara.io'
    required: false
  vectaraAuthEndpoint:
    description: 'Vectara Authentication Endpoint'
    default: ''
    required: false
  incrementalIndexing:
    description: 'Whether to just do an incremental index.  Set to false to index all content the first time'
    default: 'true'
    required: false
  filesPattern:
    description: 'Vectara Authentication Endpoint'
    default: '**'
    required: false
runs:
  using: "composite"
  steps:
    - uses: actions/checkout@v3
      with:
        fetch-depth: 0

    - name: Get JWT token
      id: jwt-token
      shell: bash
      run: |
        bash $GITHUB_ACTION_PATH/get_jwt_token.sh
      env:
        INPUT_VECTARA_AUTH_ENDPOINT: ${{ inputs.vectaraAuthEndpoint }}
        INPUT_VECTARA_ACCOUNT_NUMBER: ${{ inputs.vectaraAccountNumber }}
        INPUT_VECTARA_CLIENT_ID: ${{ inputs.vectaraClientID }}
        INPUT_VECTARA_CLIENT_SECRET: ${{ inputs.vectaraClientSecret }}
        
    - name: Get changed files
      id: changed-files
      uses: eskibars/changed-files@v0.1.0
      if: inputs.incrementalIndexing == 'true'
      with:
          files: |
            ${{ inputs.filesPattern }}

    - name: Get all files
      id: all-files
      uses: eskibars/changed-files@v0.1.0
      if: inputs.incrementalIndexing == 'false'
      with:
          base_sha: 4b825dc642cb6eb9a060e54bf8d69288fbee4904
          files: |
            ${{ inputs.filesPattern }}

    - name: Delete any documentation that has changed
      shell: bash
      id: delete-docs
      run: |
        for file in ${{ steps.changed-files.outputs.deleted_files }} ${{ steps.all-files.outputs.deleted_files }}; do
          PAYLOAD={"customer-id":"${{ inputs.vectaraAccountNumber }}","corpus-id":"${{ inputs.vectaraCorpusID }}","document-id":"$file"}
          echo "::debug::Delete-Payload: $PAYLOAD"
          curl -XPOST "https://${{ inputs.vectaraIndexingEndpoint }}/v1/delete-doc" -d $PAYLOAD \
           -H "Content-type: application/json" \
           -H "Authorization: Bearer $JWT_TOKEN" \
           -H "customer-id: ${{ inputs.vectaraAccountNumber }}"
        done
    
    - name: Index any documentation that has been created
      shell: bash
      id: index-docs
      run: |
        for file in ${{ steps.changed-files.outputs.added_files }} ${{ steps.all-files.outputs.added_files }}; do
          FULL_FILE_PATH=$GITHUB_WORKSPACE/$file
          echo "::debug::Full file path: $FULL_FILE_PATH"
          echo "::debug::File stats: $(ls -al $FULL_FILE_PATH)"
          curl -XPOST "https://${{ inputs.vectaraIndexingEndpoint }}/v1/upload?o=${{ inputs.vectaraCorpusID }}&c=${{ inputs.vectaraAccountNumber }}" \
           -H "Authorization: Bearer $JWT_TOKEN" -F file=@$FULL_FILE_PATH
        done

    - name: Index any documentation that has been modified
      shell: bash
      id: update-docs
      run: |
        for file in ${{ steps.changed-files.outputs.modified_files }} ${{ steps.all-files.outputs.modified_files }}; do
          #delete then add
          PAYLOAD={"customer-id":"${{ inputs.vectaraAccountNumber }}","corpus-id":"${{ inputs.vectaraCorpusID }}","document-id":"$file"}
          echo "::debug::Update-Payload: $PAYLOAD"
          curl -XPOST "https://${{ inputs.vectaraIndexingEndpoint }}/v1/delete-doc" -d $PAYLOAD \
           -H "Content-type: application/json" \
           -H "Authorization: Bearer $JWT_TOKEN" \
           -H "customer-id: ${{ inputs.vectaraAccountNumber }}"

          FULL_FILE_PATH=$GITHUB_WORKSPACE/$file
          echo "::debug::Full file path: $FULL_FILE_PATH"
          echo "::debug::File stats: $(ls -al $FULL_FILE_PATH)"
          curl -XPOST "https://${{ inputs.vectaraIndexingEndpoint }}/v1/upload?o=${{ inputs.vectaraCorpusID }}&c=${{ inputs.vectaraAccountNumber }}" \
           -H "Authorization: Bearer $JWT_TOKEN" -F file=@$FULL_FILE_PATH
        done