From 16310ed2c7b60147f790433da5e4d17f5724a7b8 Mon Sep 17 00:00:00 2001
From: Raphael Roberts <raphael.roberts48@gmail.com>
Date: Fri, 7 May 2021 23:48:04 -0500
Subject: [PATCH] Add reflink support for unlinkable files and a set of all
 property files processed that can be safely skipped

---
 requirements.txt           |  1 +
 tibi_hardlinks/__init__.py | 25 +++++++++++++++----------
 tibi_hardlinks/backups.py  | 21 +++++++++++++++++----
 3 files changed, 33 insertions(+), 14 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 81dd2f5..a01f88b 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -2,3 +2,4 @@ pytz
 python-dateutil
 PyYAML
 appdirs
+reflink
diff --git a/tibi_hardlinks/__init__.py b/tibi_hardlinks/__init__.py
index 457529b..f2f5ed4 100644
--- a/tibi_hardlinks/__init__.py
+++ b/tibi_hardlinks/__init__.py
@@ -16,19 +16,24 @@ def find_all_property_files():
 
 def hardlink_all_files(files, dry=False):
     for prop_file in files:
-        backup = Backup(prop_file)
-        dest_dir = create_backup_directory_name(
-            backup.backup_info, CONFIG_DATA["customization"]["time_format"]
-        )
-        dest_dir = Path(CONFIG_DATA["output"]["output_root"]) / dest_dir
-        if dry:
-            print(f"mkdir: {dest_dir}")
-        else:
-            dest_dir.mkdir(parents=True, exist_ok=True)
-        backup.hardlink_to(dest_dir, dry)
+        try:
             already_processed = cache_manager.tibi_cache["already_processed"]
+        except KeyError:
+            already_processed = set()
             cache_manager.tibi_cache["already_processed"] = already_processed
             cache_manager.tibi_cache.sync()
+        if prop_file not in already_processed:
+            backup = Backup(prop_file)
+            dest_dir = create_backup_directory_name(
+                backup.backup_info, CONFIG_DATA["customization"]["time_format"]
+            )
+            dest_dir = Path(CONFIG_DATA["output"]["output_root"]) / dest_dir
+            if not dest_dir.exists():
+                if dry:
+                    print(f"mkdir: {dest_dir}")
+                else:
+                    dest_dir.mkdir(parents=True)
+            backup.hardlink_to(dest_dir, dry)
 
 
 def main():
diff --git a/tibi_hardlinks/backups.py b/tibi_hardlinks/backups.py
index 5bb20fc..c17b2ff 100644
--- a/tibi_hardlinks/backups.py
+++ b/tibi_hardlinks/backups.py
@@ -1,5 +1,8 @@
 from pathlib import Path
 import datetime
+import re
+
+from reflink import reflink
 
 from tibi_hardlinks.backup_parser import read_backup_properties
 from tibi_hardlinks.exceptions import ConfigurationException
@@ -99,8 +102,18 @@ class Backup:
         for filepath in self.related_files.values():
             if filepath:
                 filename = filepath.name
-                if dry:
-                    print(f'link: "{filepath}" => "{dest_dir / filename}"')
-                else:
-                    filepath.link_to(dest_dir / filename)
+                dest_path = dest_dir / filename
+                can_skip = True
+                if not dest_path.exists():
+                    can_skip = False
+                    if dry:
+                        print(f'link: "{filepath}" => "{dest_path}"')
+                    else:
+                        try:
+                            filepath.link_to(dest_path)
+                        except OSError:
+                            reflink(str(filepath), str(dest_path))
+                        except Exception:
+                            raise
+        if can_skip and not dry:
             cache_manager.tibi_cache["already_processed"].add(self.property_file)