marvvin

commit 67f0823fed70d95f1f206351754e2fabaa1a90bb
parent 7f04dc4e2f2cddb207ea9a914bcdfdbff3dcafe4
Author: hhvn <dev@hhvn.uk>
Date:   Mon, 28 Feb 2022 20:36:10 +0000

URL titles.

Diffstat:
M
handler.rc
 | 
63
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

1 file changed, 63 insertions(+), 0 deletions(-)
diff --git a/handler.rc b/handler.rc
@@ -48,6 +48,10 @@ fn send_msg {
 	send_raw 'PRIVMSG %s :%s\r\n' $1 $2
 }
 
+fn send_notice {
+	send_raw 'NOTICE %s :%s\r\n' $1 $2
+}
+
 fn lock {
 	echo `{date +%s} ^ ' + ' ^ $2 | bc > data/locks/$1
 }
@@ -71,6 +75,59 @@ fn randomize {
 	echo $*($n)
 }
 
+fn title {
+	url     = `$nl{echo $1 | sed 's/invidio.us/www.youtube.com/g;s/m.imdb.com/www.imdb.com/g' | tr -d '\r' | sed 's~embed/~watch?v=~g' | grep -o '[a-z]*://[^ ]*' | sed 's/[\.?>,]$//'}
+	curl -qsLD - $url | tr -d '\r' | sed '/^[^[:print:]]*$/q' | tr '[:upper:]' '[:lower:]' > data/headers
+	curl -qsL $url | tr -d '\r' > data/content
+	if (echo $url | grep 'http[s]*://[w.]*youtu[.]*be' >/dev/null) {
+		title  = `{< data/content \
+			grep '\[{\"videoPrimaryInfoRenderer\":{\"title\":{\"runs\"\:\[{\"text\":' |
+			sed 's/.*\[{\"videoPrimaryInfoRenderer\":{\"title\":{\"runs\"\:\[{\"text\":// ; s/}\]}.*// ; s/\\//g ; s/^\"// ; s/.$//'}
+		author = `{< data/content \
+			grep -o 'author":[^}]*}' |
+			sed -E 's/.*:"([^"]*).*/\1/g'}
+		date   = `{< data/content \
+			grep -o 'uploadDate":[^}]*}' |
+			sed -E 's/.*:"([^"]*).*/\1/g;s/([0-9]*)-([0-9]*)-([0-9]*)/\3-\2-\1/'}
+
+		send_notice $TARGET '05You05,16Tube ::: Title: ' ^ $title ^ ' ::: Author: ' ^ $author ^ ' ::: Uploaded: ' ^ $date
+	} else {
+		code   = `{head -n 1 < data/headers}
+		ctype  = `{awk -F: '$1 == "content-type" {print $2}' < data/headers}
+		server = `{awk -F: '$1 == "server" {print $2}' < data/headers}
+
+		if (~ $server 'Cloudflare') {
+			out = $out ^ 'Cloudflared   '
+		} else {
+			out = $out ^ 'Server: ' ^ $server ^ '   '
+		}
+
+		if (~ $ctype 'text/html'*) {
+			title       = `{< data/content grep -o '<title[^>]*>[^<]*</title>' | sed 's~.*<title[^>]*>~~g;s~</title>.*~~g' | head -n 1}
+			published   = `{< data/content grep -o '<meta[^>]*"published_time[^>]*>' | sed -E 's/.*content="([^"]*)".*/\1/;s/T.*//;s~([0-9]*)-([0-9]*)-([0-9]*)~\3/\2/\1~'}
+			modified    = `{< data/content grep -o '<meta[^>]*"modified_time[^>]*>' | sed -E 's/.*content="([^"]*)".*/\1/;s/T.*//;s~([0-9]*)-([0-9]*)-([0-9]*)~\3/\2/\1~'}
+			author      = `{< data/content grep -o '<meta[^>]*"author[^>]*>' | sed -E 's/.*content="([^"]*)".*/\1/'}
+			generator   = `{< data/content grep -io '<meta[^>]*"generator[^>]*>' | sed -E 's/.*content="([^"]*)".*/\1/'}
+
+			if (echo $title | grep '.' >/dev/null)
+				out = $out ^ 'Title: ' ^ $title ^ '   '
+			if (echo $published | grep '.' >/dev/null)
+				out = $out ^ 'Published: ' ^ $published ^ '   '
+			if (echo $modified | grep '.' >/dev/null)
+				out = $out ^ 'Modified: ' ^ $modified ^ '   '
+			if (echo $author | grep '.' >/dev/null)
+				out = $out ^ 'Author: ' ^ $author ^ '   '
+			if (echo $generator | grep '.' >/dev/null)
+				out = $out ^ 'Generator: ' ^ $generator ^ '   '
+		} else {
+			if (echo $ctype | grep '.' >/dev/null)
+				out = $out ^ 'Content type: ' ^ $ctype ^ '   '
+		}
+
+		send_notice $TARGET $out
+	}
+}
+
 if (~ $PARAMS(1) 'PING') {
 	send_raw 'PONG :%s\r\n' $PARAMS(2)
 } else if (~ $PARAMS(1) '376') {
@@ -120,4 +177,10 @@ if (~ $PARAMS(1) 'PING') {
 		'Try something else..' \
 		'I''m a fucking bot man, cut me some slack' \
 		'*yawn* RTFM'}
+} else {
+	urls = `$nl{echo $MSG | grep -Eo 'https?://[^ ]*'}
+	for (url in $urls) {
+		title $url
+		echo $url >/dev/stderr
+	}
 }