From a937f2cbfbca97610488db45e0ea406a4deb0167 Mon Sep 17 00:00:00 2001
From: Chris Xiong <chirs241097@gmail.com>
Date: Mon, 12 Aug 2024 01:32:59 -0400
Subject: Implement the media cleanup command.

---
 utils/notectl.py | 30 +++++++++++++++++++++++++++---
 1 file changed, 27 insertions(+), 3 deletions(-)

(limited to 'utils/notectl.py')

diff --git a/utils/notectl.py b/utils/notectl.py
index 12e804e..7117e58 100644
--- a/utils/notectl.py
+++ b/utils/notectl.py
@@ -11,7 +11,7 @@ from datetime import datetime, timezone
 
 import postutil
 from atomgen import gen_atom
-from monolith import Monolith
+from monolith import Monolith, MediaType
 from config import conf
 
 '''
@@ -142,11 +142,32 @@ def init_instance():
 
 '''
 Clean up any media file that isn't used in the monolith file.
-TODO.
 '''
 def media_cleanup():
     conf.require()
-    pass
+    origs = [os.path.relpath(os.path.join(a, x), conf.LOCAL_DATA_ROOT) for (a, b, c) in os.walk(os.path.join(conf.LOCAL_DATA_ROOT, "media_orig")) for x in c]
+    thmbs = [os.path.relpath(os.path.join(a, x), conf.LOCAL_DATA_ROOT) for (a, b, c) in os.walk(os.path.join(conf.LOCAL_DATA_ROOT, "media_thmb")) for x in c]
+    referee = set()
+    m = Monolith(os.path.join(conf.LOCAL_DATA_ROOT, "posts.monolith"))
+    m.load_index()
+    for d in m.get_all_dates():
+        p = m.get_post(d)
+        for media in p.media:
+            if media.type == MediaType.IMAGE:
+                referee.add(media.thumbnail)
+                referee.add(media.original)
+    pdel = []
+    for f in origs + thmbs:
+        if f not in referee:
+            pdel.append(f)
+    if len(pdel) > 0:
+        for p in pdel: print(p)
+        print(f"Found {len(pdel)} unsed media file(s). Delete these files? [y/n]")
+        if input() not in ['Y', 'y']:
+            return
+        for p in pdel: os.remove(os.path.join(conf.LOCAL_DATA_ROOT, p))
+    else:
+        print("No unused media files found.")
 
 def main():
     if len(sys.argv) < 2:
@@ -157,6 +178,7 @@ def main():
         print("regen  Regenerate the entire monolith file.")
         print("sync   Sync data to remote for hosting.")
         print("init   Initialize a new Notekins instance. Requires path to the instance.")
+        print("clean  Clean up unused media files.")
         print("dump   Dump the content of the monolith file.")
         return
     match sys.argv[1]:
@@ -172,6 +194,8 @@ def main():
             sync_remote()
         case "init":
             init_instance()
+        case "clean":
+            media_cleanup()
         case "dump":
             m = Monolith(os.path.join(conf.LOCAL_DATA_ROOT, "posts.monolith"))
             m.load_index()
-- 
cgit v1.2.3