From 43dc70c0136ddb3cfe02fb91a0d3ae8095859f08 Mon Sep 17 00:00:00 2001
From: id628 <ingo.dean@gmail.com>
Date: Mon, 26 Feb 2024 17:37:38 -0500
Subject: [PATCH] Create mergerfs.findfile

This script tells you which drive holds the file you're looking for. It also warns you if there are multiple copies in different drives.
---
 src/mergerfs.findfile | 105 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 105 insertions(+)
 create mode 100644 src/mergerfs.findfile

diff --git a/src/mergerfs.findfile b/src/mergerfs.findfile
new file mode 100644
index 0000000..75d8ec8
--- /dev/null
+++ b/src/mergerfs.findfile
@@ -0,0 +1,105 @@
+#!/usr/bin/env python3
+import argparse
+import ctypes
+import errno
+import glob
+import os
+import sys
+import stat
+
+_libc = ctypes.CDLL("libc.so.6", use_errno=True)
+_lgetxattr = _libc.lgetxattr
+_lgetxattr.argtypes = [ctypes.c_char_p, ctypes.c_char_p, ctypes.c_void_p, ctypes.c_size_t]
+
+def lgetxattr(path, name):
+    if type(path) == str:
+        path = path.encode(errors='backslashreplace')
+    if type(name) == str:
+        name = name.encode(errors='backslashreplace')
+    length = 64
+    while True:
+        buf = ctypes.create_string_buffer(length)
+        res = _lgetxattr(path, name, buf, ctypes.c_size_t(length))
+        if res >= 0:
+            return buf.raw[0:res]
+        else:
+            err = ctypes.get_errno()
+            if err == errno.ERANGE:
+                length *= 2
+            elif err == errno.ENODATA:
+                return None
+            else:
+                raise IOError(err, os.strerror(err), path)
+
+def xattr_basepath(fullpath):
+    basepath = lgetxattr(fullpath, 'user.mergerfs.basepath')
+    if basepath is not None:
+        return basepath.decode(errors='backslashreplace')
+    return None
+
+def xattr_relpath(fullpath):
+    relpath = lgetxattr(fullpath, 'user.mergerfs.relpath')
+    if relpath is not None:
+        return relpath.decode(errors='backslashreplace')
+    return None
+
+def mergerfs_srcmounts(ctrlfile):
+    srcmounts_raw = lgetxattr(ctrlfile, 'user.mergerfs.srcmounts')
+    if srcmounts_raw is not None:
+        srcmounts = srcmounts_raw.decode(errors='backslashreplace').split(':')
+        expanded_srcmounts = []
+        for srcmount in srcmounts:
+            expanded_srcmounts.extend(glob.glob(srcmount))
+        return expanded_srcmounts
+    return []
+
+def mergerfs_control_file(path):
+    while path != os.path.sep:
+        ctrlfile = os.path.join(path, '.mergerfs')
+        if os.path.exists(ctrlfile):
+            return ctrlfile
+        path = os.path.abspath(os.path.join(path, os.pardir))
+    return None
+
+def parse_arguments():
+    parser = argparse.ArgumentParser(description='Find file location in mergerfs and handle duplicates with deletion option.')
+    parser.add_argument('file_path', type=str, help='File path to check')
+    parser.add_argument('-d', '--delete', action='store_true', help='Generate delete commands for duplicates')
+    parser.add_argument('-v', '--verbose', action='store_true', help='Verbose mode to show searching process')
+    return parser.parse_args()
+
+if __name__ == "__main__":
+    args = parse_arguments()
+
+    file_path = args.file_path
+    relpath = xattr_relpath(file_path)
+    ctrlfile = mergerfs_control_file(file_path)
+    if ctrlfile is None:
+        print("# Error: Could not find the .mergerfs control file.")
+        sys.exit(1)
+
+    srcmounts = mergerfs_srcmounts(ctrlfile)
+    basepath = xattr_basepath(file_path)
+    found_files = []
+
+    if basepath and relpath:
+        first_found = True
+        for srcmount in srcmounts:
+            potential_path = os.path.join(srcmount, relpath.lstrip('/'))
+            if args.verbose:
+                print(f"{'# ' if args.delete else ''}Searching {srcmount}")
+            if os.path.exists(potential_path):
+                found_files.append(potential_path)
+                escaped_path = potential_path.replace('"', '\\"')
+                if first_found:
+                    print(f"{'# ' if args.delete else ''}Full physical path of the file: {escaped_path}")
+                    first_found = False
+                else:
+                    if args.delete:
+                        print(f'rm -vf "{escaped_path}"')
+                    elif not args.delete:
+                        print(f"{'# ' if args.delete else ''}***Duplicate: {escaped_path}")
+    else:
+        if not args.delete:
+            print("Could not find the full physical path for the file or the file is not part of a mergerfs pool.")
+