Add scripts to publish ServiceInfo metadata

2 years ago · ed22cb9147
3 changed files with 164 additions and 0 deletions
--- a/serviceinfo/README.md
+++ b/serviceinfo/README.md
@ -0,0 +1,4 @@
+# ServiceInfo generator
+
+ServiceInfo is a JSON format for publishing meta-data about fediverse-related
+services available on a DNS domain.
--- a/serviceinfo/template-serviceinfo.json
+++ b/serviceinfo/template-serviceinfo.json
@ -0,0 +1,82 @@
+{
+    "version": "1.0",
+    "id": "https://xmpp.chapril.org/",
+    "name": "Messagerie instantanée Chapril",
+    "classifiers": ["xmpp", "instant-messaging"],
+    "software": {
+        "name": "ejabberd",
+        "version": "__EJABBERD_VERSION__",
+        "repository": "https://github.com/processone/ejabberd/"
+    },
+    "organization": {
+        "name": "Chapril",
+        "contact": "https://www.chapril.org/contact.html"
+    },
+    "protocols": [
+        {   
+            "name": "xmpp",
+            "capabilities": [
+                {   
+                    "presence": true,
+                    "mam": true,
+                    "in-band-registration": true,
+                    "web-registration": false,
+                    "storage-quota": "__USER_QUOTA__",
+                    "upload-autovaccum": "__UPLOAD_RETENTION__",
+                    "irc-gateway": false
+                }
+            ]
+        }
+    ],
+    "otherServices": {
+        "outbound": [
+            "irc"
+        ],
+        "inbound": []
+    },
+    "signups": "open",
+    "visibilityPreferences": {
+        "allowCrawling": true,
+        "preferUnlisted": false
+    },
+    "metrics": [
+        {    
+            "type": "totalUsers",
+            "value": __TOTAL_USER_ACCOUNTS__
+        },
+        {    
+            "type": "activeUsers",
+            "value": __ACTIVE_USER_WEEK__,
+            "periodLength": 604800,
+            "startTime": "__STARTTIME_WEEK__"
+        },
+        {    
+            "type": "activeUsers",
+            "value": __ACTIVE_USER_MONTH__,
+            "periodLength": 2592000,
+            "startTime": "__STARTTIME_MONTH__"
+        },
+        {    
+            "type": "activeUsers",
+            "value": __ACTIVE_USER_HALFYEAR__,
+            "periodLength": 15552000,
+            "startTime": "__STARTTIME_HALFYEAR__"
+        },
+        {    
+            "type": "localMessages",
+            "value": 60640
+        }
+    ],
+    "features": [
+        {    
+            "name": "Movim",
+            "type": "UI",
+            "description": "Web-based XMPP client"
+        },
+        {   
+            "name": "biboumi",
+            "type": "gateway",
+            "description": "XMPP->IRC gateway"
+        }
+    ]
+}
--- a/serviceinfo/update_serviceinfo
+++ b/serviceinfo/update_serviceinfo
@ -0,0 +1,78 @@
+#! /bin/bash
+
+TEMPLATE=/srv/xmpp.chapril.org/tools/serviceinfo/template-serviceinfo.json
+TARGET_FILE=/srv/serviceinfo.json
+
+main() {
+    cat "${TEMPLATE}" \
+        | sed "s/__USER_QUOTA__/$(get_hard_quota)/" \
+        | sed "s/__UPLOAD_RETENTION__/1y/" \
+        | sed "s/__TOTAL_USER_ACCOUNTS__/$(count_user_accounts)/" \
+        | sed "s/__EJABBERD_VERSION__/$(get_ejabberd_version)/" \
+        | sed "s/__STARTTIME_WEEK__/$(starttime 7)/" \
+        | sed "s/__STARTTIME_MONTH__/$(starttime 30)/" \
+        | sed "s/__STARTTIME_HALFYEAR__/$(starttime 180)/" \
+        | sed "s/__ACTIVE_USER_WEEK__/$(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)/" \
+        > ${TARGET_FILE}
+#  | sed "s/__AVG_DISK_USED_PER_USER__/$(TODO)/" \
+}
+
+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: ${avg}M"
+}
+
+starttime() {
+    days_back=${1:-30}
+    date -u +%Y-%m-%dT%H:%M:%SZ --date="08:42 ${days_back} days ago"
+}
+
+user_activity() {
+    days_back=${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