From d85513c6f24968abd35181c902819dc8e489f7e7 Mon Sep 17 00:00:00 2001
From: Iru Cai <mytbk920423@gmail.com>
Date: Sat, 25 Aug 2018 00:16:14 +0800
Subject: local store cleaner

---
 mxclean/clean_files.cpp      | 43 +++++++++++++++++++++++++++++++++++++++++++
 mxclean/clean_local_store.sh | 31 +++++++++++++++++++++++++++++++
 2 files changed, 74 insertions(+)
 create mode 100644 mxclean/clean_files.cpp
 create mode 100644 mxclean/clean_local_store.sh

diff --git a/mxclean/clean_files.cpp b/mxclean/clean_files.cpp
new file mode 100644
index 0000000..8c3b0ad
--- /dev/null
+++ b/mxclean/clean_files.cpp
@@ -0,0 +1,43 @@
+#include <set>
+#include <string>
+#include <fstream>
+#include <iostream>
+
+using namespace std;
+
+string path_to_id(string &s)
+{
+	string t = "";
+	for (auto c: s) {
+		if (isalpha(c))
+			t = t + c;
+	}
+	return t;
+}
+
+int main()
+{
+	ifstream flist("/tmp/mxclean/filelist");
+	ifstream keeplist("/tmp/mxclean/files_to_keep");
+	ofstream rm("/tmp/mxclean/remove.sh");
+
+	set<string> keep_set;
+	string s;
+	int keep = 0, remove = 0;
+
+	while (keeplist >> s) {
+		keep_set.insert(s);
+	}
+
+	while (flist >> s) {
+		if (keep_set.find(path_to_id(s)) != keep_set.end()) {
+			keep++;
+		} else {
+			remove++;
+			rm << "rm " << s << endl;
+		}
+	}
+
+	cout << "keep " << keep << " files, remove " << remove << " files." << endl;
+	return 0;
+}
diff --git a/mxclean/clean_local_store.sh b/mxclean/clean_local_store.sh
new file mode 100644
index 0000000..12402e7
--- /dev/null
+++ b/mxclean/clean_local_store.sh
@@ -0,0 +1,31 @@
+#!/bin/sh
+
+set -e
+
+purge_ts=$(($(date +%s)-$((3600*24*30))))000
+server=matrixim.cc
+content_dir=/var/lib/matrix-synapse/media_store/local_content
+tmp=/tmp/mxclean
+CLEANER="$(pwd)/clean_files"
+
+make "${CLEANER}"
+cd "${content_dir}"
+mkdir -p "${tmp}"
+find -type f > ${tmp}/filelist
+
+# for avatars in state events
+# E2EE messages doesn't have the URI in the events, so do not rely on it on all events
+echo 'select content from events' | psql -d synapse -Aqt | grep -F "mxc://${server}/" > ${tmp}/file_in_db
+jq -r .avatar_url ${tmp}/file_in_db | awk -F '/' '{print $NF}' | sed '/null/d' > ${tmp}/dbfiles
+jq -r .url ${tmp}/file_in_db | awk -F '/' '{print $NF}' | sed '/null/d' >> ${tmp}/dbfiles
+
+# profile avatars
+echo 'select avatar_url from profiles' | psql -d synapse -Aqt | sed '/^$/d' | awk -F '/' '{print $NF}' > ${tmp}/avatars
+
+echo "select media_id from local_media_repository where created_ts > ${purge_ts}" \
+	| psql -d synapse -Aqt > ${tmp}/recent_media
+
+cat ${tmp}/dbfiles ${tmp}/avatars ${tmp}/recent_media | sort -u > ${tmp}/files_to_keep
+${CLEANER}
+
+echo "Please run 'cd ${content_dir} && sh ${tmp}/remove.sh' to clean the local content."
-- 
cgit v1.2.3