validate-links

#!/bin/bash -e

if [[ $# -lt 2 ]]; then
    echo "Builds html from the yelp pages and then validates the page links through the html."
    echo "You need to have w3c-linkchecker installed to run this script"
    echo "Usage: ./validate-links in out"
    echo "   in: any directory with a complete set of .page files, e.g. /usr/share/help/C/gnome-help"
    echo "  out: directory to build html for checking, e.g. html"
    exit 1
fi

indir=$1
htmldir=$2
mkdir -p $htmldir

echo "Building the html"
yelp-build html -o $htmldir $1/*.page

echo "Validating the html links"
pushd $htmldir > /dev/null
# Since we do some hacky parsing of checklink output, make sure we and in the C
# locale
export LANG=C
for file in *.html; do
    # Ignore special uri schemes our OS knows how to handle
    checklink -X '^[help|endless|install].*' $file | tee out >> aggout
    # Lines starting with file:// are broken links
    broken="$(sed -n '/^file:\/\//p' out)"
    if [ -n "$broken" ]; then
        echo -e "In \e[31m$file\e[0m, broken links to"
        echo "$broken" | sed "s/^/    /"
    fi
done
# We should have a line in output starting with "Checking link" for every link we
# have investigated
checked="$(sed -n 's/^Checking link file:\/\///p' aggout | sort -u)"
for file in *.html; do
    if [[ ! $checked == *"$file"* ]]; then
        echo -e "Orphan page \e[33m$file\e[0m"
    fi
done
popd > /dev/null

echo "Done!"