changeset 4597:c858c76d0845

mod_tweet_data: New module that fetches and sends tweet data based on tweet URLs in MUC messages.
author JC Brand <jc@opkode.com>
date Tue, 22 Jun 2021 11:41:16 +0200
parents c406e4bf7ee5
children 09f0911c735d
files mod_ogp/README.markdown mod_tweet_data/README.md mod_tweet_data/mod_tweet_data.lua
diffstat 3 files changed, 162 insertions(+), 1 deletions(-) [+]
line wrap: on
line diff
--- a/mod_ogp/README.markdown	Sun Jun 20 13:54:23 2021 +0200
+++ b/mod_ogp/README.markdown	Tue Jun 22 11:41:16 2021 +0200
@@ -6,7 +6,7 @@
 If it finds any, it sends a [XEP-0422 fastening](https://xmpp.org/extensions/xep-0422.html) applied to the original message that looks like:
 
 ```xml
-<message id="example" from="chatroom@muc.example.org" to="chatroom@muc.example.org">
+<message id="example" from="chatroom@muc.example.org" to="user@chat.example.org/resource">
 <apply-to xmlns="urn:xmpp:fasten:0" id="origin-id-X">
 <meta xmlns="http://www.w3.org/1999/xhtml" property="og:title" content="The Rock"/>
 <meta xmlns="http://www.w3.org/1999/xhtml" property="og:url" content="https://www.imdb.com/title/tt0117500/"/>
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/mod_tweet_data/README.md	Tue Jun 22 11:41:16 2021 +0200
@@ -0,0 +1,34 @@
+# mod_tweet_data
+
+This module adds [Open Graph Protocol](https://ogp.me) metadata to Twitter.com tweet URLs sent inside a MUC.
+
+It's similar to [mod_ogp](https://modules.prosody.im/mod_ogp.html) but is adapted specifically to Twitter.com, which doesn't support the [Open Graph Protocol](https://ogp.me).
+
+When a user sends a tweet URL in a MUC (where the message has its `id` equal to its `origin-id`), this module calls that URL to get the tweet data.
+If it finds any, it sends a [XEP-0422 fastening](https://xmpp.org/extensions/xep-0422.html) applied to the original message that looks as follows (note, I haven't used real data here):
+
+```xml
+    <message xmlns="jabber:client" to="user@chat.example.org/resource" from="chatroom@muc.example.org" type="groupchat">
+        <apply-to xmlns="urn:xmpp:fasten:0" id="82dbc94c-c18a-4e51-a0d5-9fd3a7bfd267">
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:article:author" content="TwitterCritter" />
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:article:published_time" content="2021-06-22T06:44:20.000Z" />
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:description" content="I'm in ur twitterz" />
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:image" content="https://pbs.twimg.com/profile_images/984325764849045505/Ty3F93Ln_normal.jpg" />
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:title" content="TwitterCritter" />
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:type" content="tweet" />
+            <meta xmlns="http://www.w3.org/1999/xhtml" property="og:url" content="https://twitter.com/TwitterCritter/status/1407227938391707648" />
+        </apply-to>
+        <stanza-id xmlns="urn:xmpp:sid:0" by="chatroom@muc.example.org" id="90e8818d-390a-4c69-a2d8-0fd463fb3366"/>
+    </message>
+```
+
+Configuration
+-------------
+
+You'll need to provide a Twitter APIv2 bearer token.
+
+```lua
+Component "muc.example.org" "muc"
+  modules_enabled = { "tweet_data" }
+  twitter_apiv2_bearer_token  = { "some-very-long-string" }
+```
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/mod_tweet_data/mod_tweet_data.lua	Tue Jun 22 11:41:16 2021 +0200
@@ -0,0 +1,127 @@
+local mod_muc = module:depends("muc")
+local http = require "net.http"
+local st = require "util.stanza"
+local json = require "util.json"
+local url_pattern = [[https://twitter.com/%S+/status/%S+]]
+local xmlns_fasten = "urn:xmpp:fasten:0"
+local xmlns_xhtml = "http://www.w3.org/1999/xhtml"
+local twitter_apiv2_bearer_token = module:get_option_string("twitter_apiv2_bearer_token");
+
+local function fetch_tweet_data(room, url, tweet_id, origin_id)
+	if not url then return; end
+	local options = {
+		method = "GET";
+		headers = { Authorization = "Bearer "..twitter_apiv2_bearer_token; };
+	};
+
+	http.request(
+		'https://api.twitter.com/2/tweets/'..tweet_id..'?expansions=author_id&tweet.fields=created_at,text&user.fields=id,name,username,profile_image_url',
+		options,
+		function(response_body, response_code, _)
+			if response_code ~= 200 then
+				module:log("debug", "Call to %s returned code %s and body %s", url, response_code, response_body)
+				return;
+			end
+
+			local response = json.decode(response_body);
+			if not response then return; end
+
+			local tweet = response['data'];
+			local author = response['includes']['users'][1];
+
+			local to = room.jid
+			local from = room and room.jid or module.host
+			local fastening = st.message({to = to, from = from, type = 'groupchat'}):tag("apply-to", {xmlns = xmlns_fasten, id = origin_id})
+
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:article:author',
+					content = author['username']
+				}
+			):up()
+
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:article:published_time',
+					content = tweet['created_at']
+				}
+			):up()
+
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:description',
+					content = tweet['text']
+				}
+			):up()
+
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:image',
+					content = author['profile_image_url']
+				}
+			):up()
+
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:title',
+					content = author['username']
+				}
+			):up()
+
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:type',
+					content = 'tweet'
+				}
+			):up()
+			fastening:tag(
+				"meta",
+				{
+					xmlns = xmlns_xhtml,
+					property = 'og:url',
+					content = 'https://twitter.com/'..author['username']..'/status/'..tweet['id']
+				}
+			):up()
+
+			mod_muc.get_room_from_jid(room.jid):broadcast_message(fastening)
+			module:log("debug", tostring(fastening))
+		end
+	)
+end
+
+local function tweet_handler(event)
+	local room, stanza = event.room, st.clone(event.stanza)
+	local body = stanza:get_child_text("body")
+
+	if not body then return; end
+
+	local origin_id = stanza:find("{urn:xmpp:sid:0}origin-id@id")
+	if not origin_id then return; end
+
+	for url in body:gmatch(url_pattern) do
+		local _, _, _, tweet_id = string.find(url, "https://twitter.com/(%S+)/status/(%S+)");
+		fetch_tweet_data(room, url, tweet_id, origin_id);
+	end
+end
+
+module:hook("muc-occupant-groupchat", tweet_handler)
+
+
+module:hook("muc-message-is-historic", function (event)
+	local fastening = event.stanza:get_child('apply-to', xmlns_fasten)
+	if fastening and fastening:get_child('meta', xmlns_xhtml) then
+		return true
+	end
+end);