feat: agent-side global exclusions configuration matching Karen's Replicator Other Settings

windows-agent/agent/scanner.py

@@ -48,6 +48,18 @@ def is_file_stable(filepath: Path, min_stable_seconds: int = 60, sample_interval
 import platform
 import stat
 
+DEFAULT_GLOBAL_EXCLUSIONS = [
+    "pagefile.sys",
+    "hiberfil.sys",
+    "thumbs.db",
+    "desktop.ini",
+    "recycle.bin",
+    "recycler",
+    "$recycle.bin",
+    "temp",
+    "*.tmp"
+]
+
 class DirectoryScanner:
     """Scans Windows source paths for files matching specific backup patterns."""
 
@@ -59,7 +71,8 @@ class DirectoryScanner:
         skip_hidden: bool = False,
         skip_system: bool = False,
         skip_readonly: bool = False,
-        min_stable_seconds: int = 60
+        min_stable_seconds: int = 60,
+        enable_global_exclusions: bool = True
     ):
         self.source_path = Path(source_path).resolve()
         self.patterns = [p.strip() for p in file_patterns.split(",") if p.strip()]
@@ -68,6 +81,7 @@ class DirectoryScanner:
         self.skip_system = skip_system
         self.skip_readonly = skip_readonly
         self.min_stable_seconds = min_stable_seconds
+        self.enable_global_exclusions = enable_global_exclusions
 
     def scan(self) -> List[Path]:
         """Returns list of all matching files that are stable and ready for backup."""
@@ -78,16 +92,19 @@ class DirectoryScanner:
 
         if self.source_path.is_file():
             if self._matches_patterns(self.source_path.name) and not self._matches_exclude(self.source_path.name):
-                if not self._should_skip_by_attributes(self.source_path):
-                    matched_files.append(self.source_path)
+                if not (self.enable_global_exclusions and self._is_globally_excluded(self.source_path.name)):
+                    if not self._should_skip_by_attributes(self.source_path):
+                        matched_files.append(self.source_path)
             return matched_files
 
         for root, dirs, files in os.walk(self.source_path):
             # Prune excluded directories from search in-place
-            dirs[:] = [d for d in dirs if not self._matches_exclude(d)]
+            dirs[:] = [d for d in dirs if not self._matches_exclude(d) and not (self.enable_global_exclusions and self._is_globally_excluded(d))]
             
             for file in files:
                 if self._matches_patterns(file) and not self._matches_exclude(file):
+                    if self.enable_global_exclusions and self._is_globally_excluded(file):
+                        continue
                     full_path = Path(root) / file
                     if not self._should_skip_by_attributes(full_path):
                         matched_files.append(full_path)
@@ -108,6 +125,12 @@ class DirectoryScanner:
                 return True
         return False
 
+    def _is_globally_excluded(self, name: str) -> bool:
+        for pattern in DEFAULT_GLOBAL_EXCLUSIONS:
+            if fnmatch.fnmatch(name.lower(), pattern.lower()):
+                return True
+        return False
+
     def _should_skip_by_attributes(self, filepath: Path) -> bool:
         if platform.system() != "Windows":
             return False