html_to_markdown function
parent
0d37159afb
commit
1fc54a4380
20
poc.sh
20
poc.sh
|
|
@ -29,19 +29,23 @@ original() {
|
|||
for url in $(foo | grep -o 'https:..www.dndbeyond.com.compendium[^"]*' | sed 's/#.*//' | sort -u); do
|
||||
echo $url
|
||||
foo $url \
|
||||
| pandoc -f html -t markdown --ascii --toc --wrap=none --strip-comments -o - \
|
||||
| sed '/^:::.*/d' \
|
||||
| sed -e '/^# Chapter/p' -e '0,/^# Chapter/d' \
|
||||
| sed -e '/^Share$/,$d' \
|
||||
| sed 's/^\(\#\#*\) \[](#\([^)]*\))/\1 /' \
|
||||
| sed 's/{#\([^ ]*\)[^}]*}/\n\n<a name="\1"><\/a>/' \
|
||||
| sed 's/\[\([^]]*\)]\(([^)]*)\)*\({[^}]*}\)*/\1/g' \
|
||||
| sed "s/\\\\\([\"']\)/\1/g" \
|
||||
| html_to_markdown \
|
||||
| less || notes-server -root $d
|
||||
break
|
||||
done
|
||||
}
|
||||
|
||||
html_to_markdown() {
|
||||
pandoc -f html -t markdown --ascii --toc --wrap=none --strip-comments -o - \
|
||||
| sed '/^:::.*/d' \
|
||||
| sed -e '/^# Chapter/p' -e '0,/^# Chapter/d' \
|
||||
| sed -e '/^Share$/,$d' \
|
||||
| sed 's/^\(\#\#*\) \[](#\([^)]*\))/\1 /' \
|
||||
| sed 's/{#\([^ ]*\)[^}]*}/\n\n<a name="\1"><\/a>/' \
|
||||
| sed 's/\[\([^]]*\)]\(([^)]*)\)*\({[^}]*}\)*/\1/g' \
|
||||
| sed "s/\\\\\([\"']\)/\1/g"
|
||||
}
|
||||
|
||||
if [ "$0" == "$BASH_SOURCE" ]; then
|
||||
main "$@"
|
||||
fi
|
||||
|
|
|
|||
Loading…
Reference in New Issue