Update .gitlab-ci.yml file

generate_temp_docx:

  stage: preprocessing

  before_script:

    - pip3 install python-docx #install dependances

  script:

    - python3 preprocessing.py "./API.docx"

    - docker run -u $(id -u):$(id -g) --entrypoint "/bin/sh" -v "./:/data" serafinopirronitto/test-image:v3 -c "cd data && python3 preprocessing.py './API.docx'"

  artifacts:

    paths:

      - temp.docx

images_extraction:

  stage: preprocessing

  script:

    - pandoc --extract-media ./ -f docx -t html -s API.docx -o API.html

    - docker run -u $(id -u):$(id -g) --entrypoint "/bin/sh" -v "./:/data" serafinopirronitto/test-image:v3 -c "cd data && pandoc --extract-media ./ -f docx -t html -s API.docx -o API.html"

  after_script:

    - cd media

    - libreoffice --headless --convert-to png *.emf

    - mogrify -trim *.png

    - docker run -u $(id -u):$(id -g) --entrypoint "/bin/sh" -v "./:/data" serafinopirronitto/test-image:v3 -c "cd data/media && libreoffice --headless --convert-to png *.emf && mogrify -trim *.png"

  artifacts:

    paths:

      - media

chunk_html:

  stage: chunk

  before_script:

    - docker pull pandoc/latex:3.1.13-alpine

  needs:

    - "generate_temp_docx"

  script:

    - docker run --entrypoint "/bin/sh" -v ./:/data pandoc/latex:3.1.13-alpine -c "pandoc --extract-media ./ -f docx -t chunkedhtml -L filter_1.lua -L filter_2.lua  --css=styling.css --css=API.css -s temp.docx -o ./API.zip --toc --toc-depth 4 --template=template.html --split-level=1"

    - docker run -u $(id -u):$(id -g) --entrypoint "/bin/sh" -v "./:/data" serafinopirronitto/test-image:v3 -c "cd data && pandoc --extract-media ./ -f docx -t chunkedhtml -L filter_1.lua -L filter_2.lua  --css=styling.css --css=API.css -s temp.docx -o ./API.zip --toc --toc-depth 4 --template=template.html --split-level=1"

  artifacts:

    paths:

      - ./API.zip

saving_script:

  stage: build

  stage: chunk

  needs:

    - "chunk_html"

    - "images_extraction"

  before_script:

    - mkdir API

    - unzip ./API.zip /API

    - cd docx_to_html

    - unzip ./API.zip -d ./API

    - cp -r ./dist ./API

    - cd ./API

    - npm install --save puppeteer fs

    - cp pre-save_script.js ./API

  script:

    - node pre-save_script.js

    - cd saved_files

    - for file in *_updated*; do mv "$file" "$(echo "../$file" | sed 's/_updated//g')"; done

    - docker run -u $(id -u):$(id -g) --entrypoint "/bin/sh" -v "./API:/data" node:20-alpine3.20 -c "cd data && npm init -y && npm install --save puppeteer fs &&\

      apk --update --upgrade add chromium && export ENV PUPPETEER_EXECUTABLE_PATH="/usr/bin/chromium-browser &&\

      node pre-save_script.js && rm ./*.html && cd saved_files && for file in *_updated*; do mv "$file" "$(echo "../$file" | sed 's/_updated//g')"; done"

  artifacts:

    paths:

      - "./API/saved_files/"

      - docx_to_html/API/

docx:

  stage: build

  before_script:

    - pip3 install python-docx && pip3 install cssutils && pip3 install beautifulsoup4 #install dependances

    - docker pull serafinopirronitto/test-image:v3

  needs:

    - "saving_script"

  script:

    - docker run --entrypoint "/bin/sh" -v "./:/data" serafinopirronitto/test-image:v3 -c "cd data && python3 html_to_docx.py ./API && python3 postprocessing.py html_to_docx_output.docx ./media"

    - docker run -u $(id -u):$(id -g) --entrypoint "/bin/sh" -v "./:/data" serafinopirronitto/test-image:v3 -c "cd data && python3 html_to_docx.py ./API && python3 postprocessing.py html_to_docx_output.docx ./media"

  artifacts:

    paths:

      - html_to_docx_output_fixed.docx

pages:

  stage: deploy

  needs:

    - "docx"

  script: #copy into public everything it needs to make the page work

    - cp ./API/saved_files/* public/ #html files

    - cp API.css background_highlight.css public/ #css files