nixpkgs/pkgs/build-support/upstream-updater/urls-from-page.sh

15 lines
468 B
Bash
Raw Permalink Normal View History

#! /bin/sh
url="$1"
protocol="${url%%:*}"
path="${url#$protocol://}"
server="${path%%/*}"
basepath="${path%/*}"
relpath="${path#$server}"
echo "URL: $url" >&2
2014-07-18 13:41:51 +00:00
curl -A 'text/html; text/xhtml; text/xml; */*' -L -k "$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" | \
2012-12-08 20:36:38 +00:00
sed -re 's`^[^:]*$`'"$protocol://$basepath/&\`"