|
|
@@ -47,6 +47,14 @@ do |
|
|
|
url="${line}" |
|
|
|
fi |
|
|
|
|
|
|
|
if [[ "${url}" == *' | '* ]] |
|
|
|
then |
|
|
|
suffix=" | ${url#* | }" |
|
|
|
url="${url%% | *}" |
|
|
|
else |
|
|
|
suffix="" |
|
|
|
fi |
|
|
|
|
|
|
|
if [[ "${url}" =~ ^https?://((www|m|[a-z][a-z]-[a-z][a-z]).)?facebook.com/login/.*[?\&]next=https?%3A%2F%2F((www|m|[a-z][a-z]-[a-z][a-z]).)?facebook.com%2F && "${url}" != *'%0A'* && "${url}" != *'%00'* ]] |
|
|
|
then |
|
|
|
url="${url##*\?next=}" |
|
|
@@ -67,7 +75,7 @@ do |
|
|
|
user="$(grep -Po '<div\s[^>]*(?<=\s)data-key\s*=\s*"tab_home".*?</div>' <<< "${page}" | grep -Po '<a\s[^>]*(?<=\s)href="/\K[^/]+')" |
|
|
|
if [[ "${user}" ]] |
|
|
|
then |
|
|
|
echo "${prefix}https://www.facebook.com/${user}/" |
|
|
|
echo "${prefix}https://www.facebook.com/${user}/${suffix}" |
|
|
|
continue |
|
|
|
elif grep -q 'id="pagelet_loggedout_sign_up"' <<< "${page}" |
|
|
|
then |
|
|
@@ -76,7 +84,7 @@ do |
|
|
|
user="$(grep -Po '<link rel="canonical" href="\K[^"]+' <<< "${page}")" |
|
|
|
if [[ "${user}" ]] |
|
|
|
then |
|
|
|
echo "${prefix}${user}" |
|
|
|
echo "${prefix}${user}${suffix}" |
|
|
|
continue |
|
|
|
fi |
|
|
|
fi |
|
|
@@ -97,7 +105,7 @@ do |
|
|
|
user="$(curl -sL --max-time 10 -A "Opera/9.80 (Windows NT 6.1; WOW64) Presto/2.12.388 Version/12.18" "https://twitter.com/${unnormalisedUser}" | grep -Po '<a class="([^"]*\s)?ProfileHeaderCard-screennameLink(\s[^"]*)?" href="/\K[^/"]+(?=")')" |
|
|
|
if [[ "${user}" ]] |
|
|
|
then |
|
|
|
echo "${prefix}https://twitter.com/${user}" |
|
|
|
echo "${prefix}https://twitter.com/${user}${suffix}" |
|
|
|
else |
|
|
|
echo "Failed to normalise Twitter URL: ${url}" >&2 |
|
|
|
echo "${line}" |
|
|
@@ -113,7 +121,7 @@ do |
|
|
|
verbose_echo "Normalising Instagram URL: ${url}" >&2 |
|
|
|
user="${url%/}" |
|
|
|
user="${user##*/}" |
|
|
|
echo "${prefix}https://www.instagram.com/${user,,}/" |
|
|
|
echo "${prefix}https://www.instagram.com/${user,,}/${suffix}" |
|
|
|
elif [[ "${url}" =~ ^https?://(www\.)?youtube\.com/ ]] |
|
|
|
then |
|
|
|
verbose_echo "Normalising YouTube URL: ${url}" >&2 |
|
|
@@ -127,12 +135,12 @@ do |
|
|
|
canonical="$(grep -Po '<link itemprop="url" href="http://www\.youtube\.com/\Kuser/[^"]+' <<< "${page}")" |
|
|
|
if [[ "${canonical}" ]] |
|
|
|
then |
|
|
|
echo "${prefix}https://www.youtube.com/${canonical}" |
|
|
|
echo "${prefix}https://www.youtube.com/${canonical}${suffix}" |
|
|
|
else |
|
|
|
canonical="$(grep -Po '<link itemprop="url" href="http://www\.youtube\.com/\Kchannel/[^"]+' <<< "${page}")" |
|
|
|
if [[ "${canonical}" ]] |
|
|
|
then |
|
|
|
echo "${prefix}https://www.youtube.com/${canonical}" |
|
|
|
echo "${prefix}https://www.youtube.com/${canonical}${suffix}" |
|
|
|
else |
|
|
|
echo "Failed to normalise YouTube URL: ${url}" >&2 |
|
|
|
echo "${line}" |
|
|
@@ -143,7 +151,7 @@ do |
|
|
|
canonical="$(curl -sS ${otherCurlRedirectOpt} --max-time 10 -A "${userAgent}" -o /dev/null -w '%{url_effective}' "${url}")" |
|
|
|
if [[ "${canonical}" ]] |
|
|
|
then |
|
|
|
echo "${prefix}${canonical}" |
|
|
|
echo "${prefix}${canonical}${suffix}" |
|
|
|
else |
|
|
|
echo "Failed to normalise other URL: ${url}" >&2 |
|
|
|
echo "${line}" |
|
|
|