====
To build and run this site:
====

1. Check-out a base Greenstone3 system and compile up
  
    svn co https://svn.greenstone.org/main/greenstone3 greenstone3-svn-mars
    cd greenstone3-svn-mars


    # Optionally, check-out out a 'local' folder for your OS, e.g. for MacOs:
    svn co https://svn.greenstone.org/local/greenstone3/darwin-64bit local

    source ./gs3-devel.sh

    ant

    # Again, optionally activate the gnome and imagemagick
    emacs build.properties   

    ant prepare

    ant install

2. Check-out the 'mars' extension

    cd ext
    svn co https://svn.greenstone.org/gs3-extensions/mars-src/trunk mars

    Follow the details in the Mar extensions README.txt file
    In essence (for first-time setup):
      a. Get setup with a Python3 under your own control
      b. Source its activate script
      c. Add in (pip install) the specified packages
      d. Setup up (GSLDL top-level) some symbolic links to make things run more smoothly in the future
      e. Source the extension's gs3-setup.bash file
      f. Run the extention's ./CASCADE-MAKE.sh
      g. Wait a while (NodeJS in particular takes a long time to compile from source)

3. Check-out the 'mars' site and interface

    Already covered off in the Mars extension README.txt, but if not already done:

      cd web/sites
      svn co https://svn.greenstone.org/main/trunk/model-sites-dev/mars
      cd ../interfaces/
      svn co https://svn.greenstone.org/main/trunk/model-interfaces-dev/mars
      cd ../..

    Now Update:
        web/WEB-INF/servlet.xsml

    So the main 'library' servlet specifies the 'mars' site and interface:

        <init-param>
                <param-name>site_name</param-name>
                <param-value>mars</param-value>
        </init-param>
        <init-param>
                <param-name>interface_name</param-name>
                <param-value>mars</param-value>
        </init-param>

4. Build the Weka Arousal-Valence model from the DEAM data

    Follow the instructions in the 'deam' collection README.txt
    In essence:
      a. download the files
      b. process the files
      c. move up to the right level to train the models

    Note: there is no need to build this collection

5. Generate the Arousal-Valence prediction values for the 'amc-essentia' collection

   At the 'mars site' level:

   ./NOHUP-FOREACH-CSV-APPLY-AV-MODELS.sh