From 5e9a55dcc6b24c9d82de1aa12d5e5298fff3b977 Mon Sep 17 00:00:00 2001
From: Marco D'Aleo <marco@marcodaleo.com>
Date: Mon, 15 Dec 2025 15:39:14 +0000
Subject: [PATCH] Add globstar filtering

---
 .filedust.conf.example |  7 ++++--
 src/filedust/junk.py   | 53 ++++++++++++++++++++++++++++++++++++------
 2 files changed, 51 insertions(+), 9 deletions(-)

diff --git a/.filedust.conf.example b/.filedust.conf.example
index 67bd655..f260ef9 100644
--- a/.filedust.conf.example
+++ b/.filedust.conf.example
@@ -2,9 +2,12 @@
 # Place at: ~/.filedust.conf
 #
 # Use this file to customize cleanup behavior.
-# Only keys matter (no values). Paths are relative to $HOME.
 #
-# Patterns (globs) are allowed.
+# Patterns are matched against paths relative to $HOME
+# Supports:
+#   *   = one path segment
+#   **  = zero or more path segments (recursive)
+# Matching is case-sensitive
 
 [exclude]
 # Add directories or patterns you want filedust to ignore.
diff --git a/src/filedust/junk.py b/src/filedust/junk.py
index 19f572a..3ec8383 100644
--- a/src/filedust/junk.py
+++ b/src/filedust/junk.py
@@ -20,6 +20,7 @@ def load_user_rules() -> UserRules:
 
     if cfg_path.exists():
         parser = configparser.ConfigParser(allow_no_value=True)
+        parser.optionxform = str
         parser.read(cfg_path)
 
         if parser.has_section("include"):
@@ -32,8 +33,46 @@ def load_user_rules() -> UserRules:
 
 
 def matches_any(patterns: list[str], relpath: Path) -> bool:
-    posix = relpath.as_posix()
-    return any(fnmatch(posix, p) for p in patterns)
+    """
+    True globstar matcher.
+
+    Rules:
+    - *  matches exactly one path segment
+    - ** matches zero or more segments
+    - Patterns are relative to $HOME
+    """
+
+    path_parts = relpath.parts
+
+    for pat in patterns:
+        pat = pat.strip("/")
+
+        pat_parts = tuple(pat.split("/"))
+
+        if _match_parts(pat_parts, path_parts):
+            return True
+
+    return False
+
+
+def _match_parts(pat: tuple[str, ...], path: tuple[str, ...]) -> bool:
+    """Recursive glob matcher with ** support."""
+    if not pat:
+        return not path
+
+    if pat[0] == "**":
+        # ** matches zero or more segments
+        return _match_parts(pat[1:], path) or (
+            bool(path) and _match_parts(pat, path[1:])
+        )
+
+    if not path:
+        return False
+
+    if fnmatch(path[0], pat[0]):
+        return _match_parts(pat[1:], path[1:])
+
+    return False
 
 
 @dataclass
@@ -148,11 +187,11 @@ def iter_junk(root: Path, rules: UserRules | None = None) -> Iterable[Finding]:
     for dirpath, dirnames, filenames in os.walk(root, followlinks=False):
         dirpath_p = Path(dirpath)
 
-        # Fast relative path computation
-        if dirpath == root_str:
-            rel_dir = Path(".")
-        else:
-            rel_dir = Path(dirpath[len(root_str) :].lstrip("/"))
+        try:
+            rel_dir = dirpath_p.resolve().relative_to(HOME)
+        except ValueError:
+            # Should never happen due to earlier checks
+            continue
 
         # USER EXCLUDE → skip entire subtree
         if matches_any(rules.exclude, rel_dir):
-- 
2.49.1