Add first queries.

README.md

@@ -1,2 +1,4 @@
 # statbot-queries
-A list of queries useful for a Statbot dataset
+A list of queries useful for a [Statbot](https://github.com/strinking/statbot) dataset.
+
+Available under the terms of the [MIT License](LICENSE).

channel_messages.sql

@@ -0,0 +1,12 @@
+-- Lists all channels by number of messages.
+
+SELECT channels.name, t.*
+FROM (
+    SELECT channel_id, count(*)
+    FROM messages
+    WHERE guild_id = 181866934353133570 -- Programming
+    GROUP BY channel_id
+) AS t
+JOIN channels
+    ON channels.channel_id = t.channel_id
+ORDER BY channels.name;

channel_percentage.sql

@@ -0,0 +1,34 @@
+-- This query measures the "meme" ratio of our guild.
+-- Really this is just what percentage of a person's
+-- posts are in #general vs. the rest of the guild.
+-- Since percentages for people with low post counts
+-- can be skewed, this query will ignore those with
+-- fewer than 400 messages total.
+
+SELECT
+    users.name,
+    totals.user_id,
+    100.0 * general.count / totals.count as percentage,
+    totals.count as total_messages
+FROM (
+    SELECT
+        user_id,
+        COUNT(user_id)
+    FROM messages
+    WHERE guild_id = 181866934353133570 -- Programming
+    GROUP BY user_id
+) as totals
+JOIN (
+    SELECT
+        user_id,
+        COUNT(user_id)
+    FROM messages
+    WHERE channel_id = 181866934353133570 -- #general
+    GROUP BY user_id
+) as general
+    ON totals.user_id = general.user_id
+JOIN users
+    ON totals.user_id = users.user_id
+WHERE totals.count > 400 -- Limit to people 400 posts or more
+GROUP BY totals.user_id, general.count, totals.count, users.name
+ORDER BY percentage DESC;

music_links.sql

@@ -0,0 +1,13 @@
+-- This query extracts YouTube URLs from #music.
+-- It works on messages with multiple links, and
+-- doesn't need any regex or parsing since it
+-- just queries the list of embeds.
+
+SELECT DISTINCT json_array_elements(embeds::json)->>'url' AS url
+FROM messages
+WHERE channel_id = 280778849615216640 -- #music
+AND (
+    content LIKE '%youtube.com/watch%'
+    OR content LIKE '%youtube.com/playlist%'
+    OR content LIKE '%youtu.be/%'
+);

selfbot_embeds.sql

@@ -0,0 +1,11 @@
+-- This query counts how many messages have embeds,
+-- but were not made by bot accounts.
+-- This means that they are running selfbots.
+
+SELECT count(*)
+FROM messages
+JOIN users
+    ON messages.user_id = users.user_id
+WHERE messages.guild_id = 181866934353133570 -- Programming
+AND messages.embeds != '[]'
+AND NOT users.is_bot;