#!/bin/sh
# Find dead (404) links in an html file
# Requires curl, grep, hxwls, parallel
# $Id: deadlink.sh,v 1.4 2025/02/17 10:09:32 oc45ujef Exp $
# https://wwwcip.cs.fau.de/~oc45ujef/misc/src/deadlink.sh

links(){
    if [ "$1" = "-" ]
    then cat
    else
        if [ -f "$1" ]
        then
            cat "$1"
        else
            curl -Ls "$1"
        fi
    fi | hxwls -
}

links "$1" \
    | grep -e '^http' \
    | parallel '
        case {} in
            *"youtube.com/watch"*)
                curl -ILs -o /dev/null -w "%{http_code}\t{= uq =}\n" "https://www.youtube.com/oembed?format=json&url={= uq =}"
                ;;
            *)
                curl -ILs -o /dev/null -w "%{http_code}\t%{url}\n" {}
                ;;
        esac' \
    | grep -e '^404'