html_md.sh

#!/bin/sh

#wget --mirror --convert-links --adjust-extension --page-requisites  --no-parent https://docs.it4i.cz


if [ "$1" = "-d" ]; then
	# delete all .md files
	if [ "$2" = "-html" ]; then
		# delete all .html files
		find . -name "*.ht*" | 
		while read i; 
		do
			rm "$i"; 	
			echo "$i deleted"; 
		done
	fi
	if [ "$2" = "-md" ]; then
		# delete all .md files
		find . -name "*.md" | 
		while read i; 
		do
			rm "$i"; 		
			echo "$i deleted"; 
		done
	fi
else
	# delete all .md files
	find . -name "*.md" | 
	while read i; 
	do
		rm "$i"; 
		echo "$i deleted"; 
	done

	# create folder info
	mkdir info;

	find . -name "*.png" | 
	while read i; 
	do		
		echo "$i" >> ./info/list_png.txt; 
	done

	find . -name "*.jpg" | 
	while read i; 
	do		
		echo "$i" >> ./info/list_jpg.txt; 
	done
	

	# html -> md
	find . -name "*.ht*" | 
	while read i; 
	do 
		# filtering html
		echo "$i"; 
		echo "\t\tfiltering html files...";
		
		HEAD=$(cat "$i" | grep -n -m1 '<h1' |cut -f1 -d: | tr --delete '\n')
		END=$(cat "$i" | grep -n -m1 '<!-- <div tal:content=' |cut -f1 -d: | tr --delete '\n')
		LAST=$(cat "$i" | wc -l | tr --delete '\n')
		DOWN=$((LAST-END+2))

		cat "$i" | sed '1,'"$((HEAD-1))"'d' | sed -n -e :a -e '1,'"$DOWN"'!{P;N;D;};N;ba' > "${i%.*}TMP.html"	

		# converted .html to .md
		echo "\t\t.html -> .md"
		pandoc -f html -t markdown+pipe_tables-grid_tables "${i%.*}TMP.html" -o "${i%.*}.md"; 
		rm "${i%.*}TMP.html";

		# folder info, file strukture, list of all files and his addres into folders
		echo "${i%.*}" >> ./info/files_md.txt;

		# create filter_auto
		cat "${i%.*}.md" | grep -o -P '(?<={).*(?=})' | sort -u | sed '/{/d' | sed '/\$/d' >> filter_auto;
		sort -u filter_auto -o filter_auto; 

		# exceptions filter_auto
		cat exceptions_filter_auto | 
		while read y; 
		do 
			# search and delete according with filter_auto
			cat filter_auto | sed -e 's/'"$y"'//g' > filter_autoTMP;
			cat filter_autoTMP > filter_auto;
		done
	
		# text filtering of html, css, ...
		echo "\t\tautomatic filter..."
		cat filter_auto | 
		while read y; 
		do 
			# search and delete according with filter_auto
			cat "${i%.*}.md" | sed -e 's/{'"$y"'}//g' | sed -e 's/\\//g' | sed -e 's/: //g' | sed -e 's/<\/div>//g' | sed '/^<div/d'  | sed '/^$/d' > "${i%.*}TMP.md";
			cat "${i%.*}TMP.md" > "${i%.*}.md";
		done

		echo "\t\tother filter..."
		cat filter_other | 
		while read a; 
		do 
			# search and delete according with filter_other
			cat "${i%.*}.md" | sed -e 's/'"$a"'//g'  > "${i%.*}TMP.md";
			cat "${i%.*}TMP.md" > "${i%.*}.md";
		done

		# delete temporary files
		rm "${i%.*}TMP.md";
		
	done
	rm filter_autoTMP
	rm filter_auto
fi