Filter out broken-completely links

This commit is contained in:
Michael Raskin 2014-07-18 17:41:51 +04:00
parent 5ef6745be3
commit c38a11316d

View File

@ -9,6 +9,6 @@ relpath="${path#$server}"
echo "URL: $url" >&2
curl -A 'text/html; text/xhtml; text/xml; */*' -L -k "$url" | sed -re 's/^/-/;s/[hH][rR][eE][fF]=("([^"]*)"|'\''([^'\'']*)'\''|([^"'\'' <>&]+)[ <>&])/\n+\2\3\4\n-/g' | \
curl -A 'text/html; text/xhtml; text/xml; */*' -L -k "$url" | sed -re 's/^/-/;s/[^a-zA-Z][hH][rR][eE][fF]=("([^"]*)"|'\''([^'\'']*)'\''|([^"'\'' <>&]+)[ <>&])/\n+\2\3\4\n-/g' | \
sed -e '/^-/d; s/^[+]//; /^#/d;'"s/^\\//$protocol:\\/\\/$server\\//g" | \
sed -re 's`^[^:]*$`'"$protocol://$basepath/&\`"