#!/bin/bash

# PURPOSE
# This is not a nightly script. You use it to regenerate the model-collections
# if Greenstone has changed fundamentally, such as what HASH OIDs get assigned
# to documents or something that changes the contents of the index and 
# archives folders. This has happened now with the commits 
# http://trac.greenstone.org/changeset/28022 and 
# http://trac.greenstone.org/changeset/28021
# These commits generate new stable HASH OIDs for the existing documents.


# USAGE
# Put this file in the toplevel of the Greenstone 2 binary/SVN installation 
# that you want to generate the model collections with.
# Call with 0 args to process all model collections
# Call with n args, where each is a collection name in the model-collections
# list

# PSEUDOCODE
# This script:
# Checks out the model-collections folder from SVN 
# Makes a copy
# In the copy: gets rid of their .svn folders, and builds each collection in turn, moving building to index once done



function single_collection () {
    collection=$1    

    #escape the filename (in case of space)
    collection=`echo $collection | sed 's@ @\\\ @g'`
    
    #get just the basename
    collection=`basename $collection`
    
    import.pl -removeold $collection
    buildcol.pl -removeold $collection
    rm -rf collect/$collection/index
    mv collect/$collection/building collect/$collection/index
    
    echo
    echo "*********************"
    echo "Done processing $collection"
    echo "*********************"
    echo
}


function all_collections() {

    #for each collection, import, build, move building to index
    for collection in collect/*; do
	single_collection $collection;
    done
}


# The program starts here

# Need pdfbox for the PDFBox tutorial
if [ ! -e ext/pdf-box ]; then
    cd ext
    if [ ! -e ext/pdf-box-java.tar.gz ]; then
	wget http://trac.greenstone.org/export/head/gs2-extensions/pdf-box/trunk/pdf-box-java.tar.gz
	tar -xvzf pdf-box-java.tar.gz
    fi
    cd ..
fi


if test -e model-collect; then   
    svn up model-collect
else
    svn co http://svn.greenstone.org/other-projects/nightly-tasks/diffcol/trunk/model-collect
fi

# move the existing collect folder out of the way
if [ -e collect ] && [ ! -e collect_orig ] ; then
    mv collect collect_orig
fi

# make a copy of the model-collect named as the new collect 
# and remove the copy's .svn folders
if [ -e collect_orig ]; then
    if [ -e collect ]; then
	rm -rf collect
    fi
    cp -r model-collect collect
fi

#cd collect
#find . -name ".svn" -type d -exec rm -rf {} \;
#cd ..
find collect -name ".svn" -type d -exec rm -rf {} \;


# Set up the Greenstone environment for building
source setup.bash

# parse arguments
# http://stackoverflow.com/questions/12711786/bash-convert-command-line-arguments-into-array
# http://stackoverflow.com/questions/255898/how-to-iterate-over-arguments-in-bash-script

if [ "$1" == "" ]; then
	all_collections
else
    # Command-line args are a list of collections, 
    # process each command-line arg, after confirming such a collection exists

    for collection in collect/"$@"; do
	if test -e $collection; then   
	    single_collection $collection;
	else
	    echo "Can't find collection $collection. Skipping."
	fi
    done
fi