#!/bin/bash
# Work on a copy of the site or an extract of the tar files

# Delete any non html. It should be located in the static/ dir
#find . -type f -not -name "*.html" -delete
#rm -r */img
#rm contents-full.html
# Initial Conversion to MarkDown
#find . -name "*.html" -exec sh -c "cat {} | sed -n '/Main content table/,/End of main content/p;/End of main content/q' | iconv -f iso8859-7 -t utf-8  | pandoc -s -r html -t markdown_github-raw_html | grep -vE '\[Περιεχόμενα\]|\[Επόμενο\]|\[Προηγούμενο\]' > {}.md_1" \;
# or
#find . -name "*.html" -exec sh -c "cat {} | sed -n '/Main content table/,/End of main content/p;/End of main content/q' | iconv -f iso8859-7 -t utf-8  | pandoc -s --columns=160 -r html -t markdown+pipe_tables-raw_html | grep -vE '\[Περιεχόμενα\]|\[Επόμενο\]|\[Προηγούμενο\]|^:::' > {}.md_1" \;

for d in `find . -mindepth 1 -maxdepth 1 -type d`; do
    echo \#Entering dir $d
        cd $d
            for chapter in $(ls [0-1]*html.md_1 | sed -e 's/\(.*\)-[0-9]*.html\.md_1/\1/' | cut -f1 -d\. | uniq); do
                issue=$(basename $d)
                # construct article header
                intro=${chapter}.html.md_1
                title=$(cat $intro | sed '/^====.*/Q' | grep -v "^$" | tr -d "\'")
                author="$(cat $intro |  sed -n '/^\[\*.*/p;/.+\*\]/Q' | sed -e "s/\ <.*>//" | sed -e "s/{.*}//" | sed -e "s/  (.*)//" | tr -d "*,[]\\\'" )"
                adate="$(cat $intro | sed -n '/^*\[/p' | sed -e 's/\*\[\(.*\)\].*/\1/' | tr -d "*\'")"
                case "$adate" in
                    *Ιανουάριος* | *Ιανουαρίου*)
                       amonth="1"
                       ;;
                    *Φεβρουάριος* | *Φεβρουαρίου*)
                       amonth="2"
                       ;;
                    *Μάρτιος* | *Μαρτίου*)
                       amonth="3"
                       ;;
                    *Απρίλιος* | *Απριλίου*)
                       amonth="4"
                       ;;
                    *Μάϊος* | *Μαΐου*)
                       amonth="5"
                       ;;
                    *Ιούνιος* | *Ιουνίου*)
                       amonth="6"
                       ;;
                    *Ιούλιος* | *Ιουλίου*)
                       amonth="7"
                       ;;
                    *Αύγουστος* | *Αυγούστου*)
                       amonth="8"
                       ;;
                    *Σεπτέμβριος* | *Σεπτεμβρίου*)
                       amonth="9"
                       ;;
                    *Οκτώβριος* | *Οκτωβρίου*)
                       amonth="10"
                       ;;
                    *Νοέμβριος* | *Νοεμβρίου*)
                       amonth="11"
                       ;;
                    *Δεκέμβριος*)
                       amonth="12"
                       ;;
                    *)
                        echo -n "failed date"
                        ;;
                esac
                ayear=$(echo $adate | sed -e "s/.*\([[:digit:]]\{4\}\).*/\1/")
                series="Magaz ${issue}"
                order="$(basename $chapter | cut -d_ -f1|bc)"
                # Start Template
                # Start Front Matter
                echo '+++' > ${chapter}.md
                echo "title = '$title'">> ${chapter}.md
                echo "date = '$(date -d $ayear-$amonth-01 +%Y-%m-%dT00:00:00Z)'" >> ${chapter}.md
                echo "description = ''" >> ${chapter}.md
                echo "author = '$author'">> ${chapter}.md
                echo "issue = ['$series']">> ${chapter}.md
                echo "issue_weight = $order">> ${chapter}.md
                echo '+++'>> ${chapter}.md
                # Append content
                #cat ${intro} | sed -e "s/(${chapter}.*)//" | sed -e "s/^-.*\[\(.*\)\]$/- \1/" | sed -e "s/\([[:digit:]]\)\. \[\(.*\)\]$/\1. \2/" >> ${chapter}.md
                cat ${intro} | sed -e "s/${chapter}.*\.html//" | sed -e 's/\[\(.*\)\]{.*} \[\(.*\)\](.*/**\1 \2**/' | tail +7 >> ${chapter}.md
                cat ${chapter}-?.html.md_1 >> ${chapter}.md 2>/dev/null
                cat ${chapter}-??.html.md_1 >> ${chapter}.md 2>/dev/null

                # Change image links to absolute paths
                sed -i "s/(img\//(\/${issue}\/img\//g" ${chapter}.md

            done
            cd ..
        done