From d26a944d951c79450cb94d8470b76c96cbad23cc Mon Sep 17 00:00:00 2001
From: ed <s@ocv.me>
Date: Wed, 17 Jul 2024 21:00:59 +0000
Subject: [PATCH] hooks: add cache-warmer

---
 bin/hooks/README.md                 |   1 +
 bin/hooks/into-the-cache-it-goes.py | 140 ++++++++++++++++++++++++++++
 2 files changed, 141 insertions(+)
 create mode 100644 bin/hooks/into-the-cache-it-goes.py

diff --git a/bin/hooks/README.md b/bin/hooks/README.md
index 9af3bca6..df9ad84f 100644
--- a/bin/hooks/README.md
+++ b/bin/hooks/README.md
@@ -13,6 +13,7 @@ run copyparty with `--help-hooks` for usage details / hook type explanations (xb
 * [image-noexif.py](image-noexif.py) removes image exif by overwriting / directly editing the uploaded file
 * [discord-announce.py](discord-announce.py) announces new uploads on discord using webhooks ([example](https://user-images.githubusercontent.com/241032/215304439-1c1cb3c8-ec6f-4c17-9f27-81f969b1811a.png))
 * [reject-mimetype.py](reject-mimetype.py) rejects uploads unless the mimetype is acceptable
+* [into-the-cache-it-goes.py](into-the-cache-it-goes.py) avoids bugs in caching proxies by immediately downloading each file that is uploaded
 
 
 # upload batches
diff --git a/bin/hooks/into-the-cache-it-goes.py b/bin/hooks/into-the-cache-it-goes.py
new file mode 100644
index 00000000..99c5b669
--- /dev/null
+++ b/bin/hooks/into-the-cache-it-goes.py
@@ -0,0 +1,140 @@
+#!/usr/bin/env python3
+
+import sys
+import json
+import shutil
+import platform
+import subprocess as sp
+from urllib.parse import quote
+
+
+_ = r"""
+try to avoid race conditions in caching proxies
+(primarily cloudflare, but probably others too)
+by means of the most obvious solution possible:
+
+just as each file has finished uploading, use
+the server's external URL to download the file
+so that it ends up in the cache, warm and snug
+
+this intentionally delays the upload response
+as it waits for the file to finish downloading
+before copyparty is allowed to return the URL
+
+NOTE: you must edit this script before use,
+  replacing https://example.com with your URL
+
+NOTE: if the files are only accessible with a
+  password and/or filekey, you must also add
+  a cromulent password in the PASSWORD field
+
+NOTE: needs either wget, curl, or "requests":
+  python3 -m pip install --user -U requests
+
+
+example usage as global config:
+    --xau j,t10,bin/hooks/into-the-cache-it-goes.py
+
+parameters explained,
+    xau = execute after upload
+    j   = this hook needs upload information as json (not just the filename)
+    t10 = abort download and continue if it takes longer than 10sec
+
+example usage as a volflag (per-volume config):
+    -v srv/inc:inc:r:rw,ed:xau=j,t10,bin/hooks/into-the-cache-it-goes.py
+                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+    (share filesystem-path srv/inc as volume /inc,
+     readable by everyone, read-write for user 'ed',
+     running this plugin on all uploads with params explained above)
+
+example usage as a volflag in a copyparty config file:
+    [/inc]
+      srv/inc
+      accs:
+        r: *
+        rw: ed
+      flags:
+        xau: j,t10,bin/hooks/into-the-cache-it-goes.py
+"""
+
+
+# replace this with your site's external URL
+# (including the :portnumber if necessary)
+SITE_URL = "https://example.com"
+
+# if downloading is protected by passwords or filekeys,
+# specify a valid password between the quotes below:
+PASSWORD = ""
+
+# if file is larger than this, skip download
+MAX_MEGABYTES = 8
+
+# =============== END OF CONFIG ===============
+
+
+WINDOWS = platform.system() == "Windows"
+
+
+def main():
+    fun = download_with_python
+    if shutil.which("curl"):
+        fun = download_with_curl
+    elif shutil.which("wget"):
+        fun = download_with_wget
+
+    inf = json.loads(sys.argv[1])
+
+    if inf["sz"] > 1024 * 1024 * MAX_MEGABYTES:
+        print("[into-the-cache] file is too large; will not download")
+        return
+
+    file_url = "/"
+    if inf["vp"]:
+        file_url += inf["vp"] + "/"
+    file_url += inf["ap"].replace("\\", "/").split("/")[-1]
+    file_url = SITE_URL.rstrip("/") + quote(file_url, safe=b"/")
+
+    print("[into-the-cache] %s(%s)" % (fun.__name__, file_url))
+    fun(file_url, PASSWORD.strip())
+
+    print("[into-the-cache] Download OK")
+
+
+def download_with_curl(url, pw):
+    cmd = ["curl"]
+
+    if pw:
+        cmd += ["-HPW:%s" % (pw,)]
+
+    nah = sp.DEVNULL
+    sp.check_call(cmd + [url], stdout=nah, stderr=nah)
+
+
+def download_with_wget(url, pw):
+    cmd = ["wget", "-O"]
+
+    cmd += ["nul" if WINDOWS else "/dev/null"]
+
+    if pw:
+        cmd += ["--header=PW:%s" % (pw,)]
+
+    nah = sp.DEVNULL
+    sp.check_call(cmd + [url], stdout=nah, stderr=nah)
+
+
+def download_with_python(url, pw):
+    import requests
+
+    headers = {}
+    if pw:
+        headers["PW"] = pw
+
+    with requests.get(url, headers=headers, stream=True) as r:
+        r.raise_for_status()
+        for _ in r.iter_content(chunk_size=1024 * 256):
+            pass
+
+
+if __name__ == "__main__":
+    main()