Merge remote-tracking branch 'refs/remotes/origin/main'

Reviewed-on: #6

.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.

README.md

@@ -58,7 +58,7 @@ The GPG fingerprint is `0032C71FA6A11EF9567D4434C5C06BD4603C28B1`.
 #### 2) Add the APT source
 
 ```bash
-echo "deb [arch=amd64 signed-by=/usr/share/keyrings/guardutils.gpg] https://repo.sysmd.uk debian main" | sudo tee /etc/apt/sources.list.d/guardutils.list
+echo "deb [arch=amd64 signed-by=/usr/share/keyrings/guardutils.gpg] https://repo.sysmd.uk/guardutils/debian stable main" | sudo tee /etc/apt/sources.list.d/guardutils.list
 ```
 
 #### 3) Update and install
@@ -82,13 +82,11 @@ sudo rpm --import https://repo.sysmd.uk/guardutils/guardutils.gpg
 sudo tee /etc/yum.repos.d/guardutils.repo > /dev/null << 'EOF'
 [guardutils]
 name=GuardUtils Repository
-baseurl = https://repo.sysmd.uk/rpm/$basearch
-
+baseurl=https://repo.sysmd.uk/guardutils/rpm/$basearch
 enabled=1
 gpgcheck=1
-gpgkey = https://repo.sysmd.uk/guardutils/guardutils.gpg
-
 repo_gpgcheck=1
+gpgkey=https://repo.sysmd.uk/guardutils/guardutils.gpg
 EOF
 ```
 
@@ -106,7 +104,7 @@ pip install filedust
 
 ### From this repository
 ```
-git clone https://github.com/guardutils/filedust.git
+git clone https://git.sysmd.uk/guardutils/filedust.git
 cd filedust/
 poetry install
 ```

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "filedust"
-version = "0.3.1"
+version = "0.4.0"
 description = "Opinionated junk cleaner for dev machines (caches, build artifacts, editor backups)."
 authors = ["Marco D'Aleo <marco@marcodaleo.com>"]
 license = "GPL-3.0-or-later"

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):