From 4eb548d8bb98548efb7bcac5a73c5e0fda406c76 Mon Sep 17 00:00:00 2001
From: pitchum <pitchum@users.noreply.chapril.org>
Date: Sat, 16 May 2020 18:33:24 +0200
Subject: [PATCH] Add scripts for publishing to https://the-federation.info/

---
 nodeinfo2/README.md                 |  6 +++
 nodeinfo2/template-x-nodeinfo2.json | 35 +++++++++++++++
 nodeinfo2/update_nodeinfo           | 67 +++++++++++++++++++++++++++++
 3 files changed, 108 insertions(+)
 create mode 100644 nodeinfo2/README.md
 create mode 100644 nodeinfo2/template-x-nodeinfo2.json
 create mode 100755 nodeinfo2/update_nodeinfo

diff --git a/nodeinfo2/README.md b/nodeinfo2/README.md
new file mode 100644
index 0000000..bfe1b48
--- /dev/null
+++ b/nodeinfo2/README.md
@@ -0,0 +1,6 @@
+# Nodeinfo2 generator
+
+Nodeinfo2 is a JSON format for publishing meta-data about fediverse-related
+services available on a DNS domain.
+Among other things it makes it possible to populate and automatically
+update information on https://the-federation.info/
diff --git a/nodeinfo2/template-x-nodeinfo2.json b/nodeinfo2/template-x-nodeinfo2.json
new file mode 100644
index 0000000..6b3a0da
--- /dev/null
+++ b/nodeinfo2/template-x-nodeinfo2.json
@@ -0,0 +1,35 @@
+{
+    "version": "1.0",
+    "openRegistrations": true,
+    "organization": {
+        "name": "Chapril",
+        "contact": "https://www.chapril.org/contact.html"
+    },
+    "server": {
+        "name": "Chapril's XMPP server",
+        "baseUrl": "https://xmpp.chapril.org/",
+        "software": "ejabberd",
+        "version": "__EJABBERD_VERSION__"
+    },
+    "services": {
+        "outbound": [
+            "xmpp"
+        ],
+        "inbound": [
+            "xmpp"
+        ]
+    },
+    "protocols": [
+        "xmpp"
+    ],
+    "usage": {
+        "users": {
+            "total": __TOTAL_USER_ACCOUNTS__,
+            "activeWeek": __ACTIVE_USER_WEEK__,
+            "activeMonth": __ACTIVE_USER_MONTH__,
+            "activeHalfyear": __ACTIVE_USER_HALFYEAR__
+        },
+        "localPosts": __POSTS_COUNT__,
+        "localComments": 0
+    }
+}
diff --git a/nodeinfo2/update_nodeinfo b/nodeinfo2/update_nodeinfo
new file mode 100755
index 0000000..03381d3
--- /dev/null
+++ b/nodeinfo2/update_nodeinfo
@@ -0,0 +1,67 @@
+#! /bin/bash
+
+TEMPLATE=/srv/xmpp.chapril.org/tools/nodeinfo2/template-x-nodeinfo2.json
+TARGET_FILE=/srv/x-nodeinfo2.json
+
+main() {
+    cat "${TEMPLATE}" \
+        | sed "s/__TOTAL_USER_ACCOUNTS__/$(count_user_accounts)/" \
+        | sed "s/__EJABBERD_VERSION__/$(get_ejabberd_version)/" \
+        | sed "s/__ACTIVE_USER_WEEK__/$(user_activity 7)/" \
+        | sed "s/__ACTIVE_USER_MONTH__/$(user_activity 30)/" \
+        | sed "s/__ACTIVE_USER_HALFYEAR__/$(user_activity 180)/" \
+        | sed "s/__POSTS_COUNT__/$(count_archived_messages)/" \
+        > ${TARGET_FILE}
+}
+
+get_ejabberd_version() {
+    dpkg -l ejabberd | grep ejabberd | awk '{print $3}'
+}
+
+count_user_accounts() {
+    # exclude special accounts like r.giskard (for compliance tester) an bart and lisa for testing.
+    sudo -u ejabberd ejabberdctl registered_users chapril.org | egrep -v -e 'r\.giskard' -e '(lisa|bart)\.simpson' | wc -l
+}
+
+get_hard_quota() {
+    grep hard_quota /etc/ejabberd/ejabberd.yml -A2 | grep all | sed -r 's/^\s*-\s([0-9]+).*/\1/'
+}
+
+per_account_storage_usage() {
+    STORAGE_DIR="/srv/ejabberd/chapril.org/upload/"
+    # TODO Directories to ignore because they are used for tests accounts (bart and lisa simpsons)
+    IGNORE_DIRS="35eae3ddc83d72e4d3fad3160a400db688d7d936 9be859c77649f33d207cbe0d74a126b2fd31cb33"
+
+    echo "Per user disk usage:"
+    tmp="$(du -sh ${STORAGE_DIR}* | sort -hr | awk '{print $1}')"
+    echo "**  Max: "$(echo "${tmp}" | head -n 1)
+#    echo "  Min: "$(echo "${tmp}" | tail -n 1)
+
+    subfolders_count=$(find ${STORAGE_DIR} -mindepth 1 -maxdepth 1 -type d | wc -l)
+    total_size=$(du -s ${STORAGE_DIR} | awk '{print $1}')
+    avg=$(((${total_size} / 1024) / ${subfolders_count}))
+    echo "**  Avg: ${avg}M"
+}
+
+user_activity() {
+    days_back=${1:-30}
+    sql="select count(distinct username)
+     from archive
+     where peer not like 'irc%'
+     and kind = 'chat'
+     and created_at > current_timestamp - interval '${days_back} days'
+     and username not in ('r.giskard', 'lisa.simpson', 'bart.simpson')"
+
+    psql -U ejabberd -h localhost ejabberd -c "${sql}" -t | grep -v '^$' | awk '{print $1}'
+}
+
+count_archived_messages() {
+    sql="select count(*)
+     from archive
+     where peer not like 'irc%'
+     and username not in ('r.giskard', 'lisa.simpson', 'bart.simpson')"
+
+    psql -U ejabberd -h localhost ejabberd -c "${sql}" -t | grep -v '^$' | awk '{print $1}'
+}
+
+main