Merge pull request #116 from lguerard/listdir_matching_recursive

feat(pathtools): add recursive option to listdir_matching

Author: ehrenfeu

src/imcflibs/pathtools.py

@@ -180,7 +180,9 @@ def jython_fiji_exists(path):
         return False
 
 
-def listdir_matching(path, suffix, fullpath=False, sort=False, regex=False):
+def listdir_matching(
+    path, suffix, fullpath=False, sort=False, regex=False, recursive=False
+):
     """Get a list of files in a directory matching a given suffix.
 
     Parameters
@@ -199,25 +201,65 @@ def listdir_matching(path, suffix, fullpath=False, sort=False, regex=False):
     regex : bool, optional
         If set to True, uses the suffix-string as regular expression to match
         filenames. By default False.
+    recursive : bool, optional
+        If set to True, the directory tree will be traversed recursively and
+        files in subfolders will be included. When `fullpath` is False and
+        `recursive` is True the returned file names are relative to `path`
+        (e.g. `subdir/file.ext`). Default is False.
 
     Returns
     -------
     list
         All file names in the directory matching the suffix (without path!).
     """
     matching_files = list()
-    for candidate in os.listdir(path):
-        if not regex and candidate.lower().endswith(suffix.lower()):
-            # log.debug("Found file %s", candidate)
-            if fullpath:
-                matching_files.append(os.path.join(path, candidate))
-            else:
-                matching_files.append(candidate)
-        if regex and re.match(suffix.lower(), candidate.lower()):
-            if fullpath:
-                matching_files.append(os.path.join(path, candidate))
+
+    # Prepare regex if requested (case-insensitive)
+    regex_compiled = None
+    if regex:
+        try:
+            regex_compiled = re.compile(suffix, re.IGNORECASE)
+        except re.error:
+            # If provided regex is invalid, fall back to no matches
+            return matching_files
+
+    if recursive:
+        # Walk directory tree and test each filename
+        for dirpath, _, filenames in os.walk(path):
+            for candidate in filenames:
+                if not regex_compiled:
+                    if candidate.lower().endswith(suffix.lower()):
+                        if fullpath:
+                            matching_files.append(os.path.join(dirpath, candidate))
+                        else:
+                            rel = os.path.relpath(
+                                os.path.join(dirpath, candidate), path
+                            )
+                            matching_files.append(rel)
+                else:
+                    if regex_compiled.match(candidate):
+                        if fullpath:
+                            matching_files.append(os.path.join(dirpath, candidate))
+                        else:
+                            rel = os.path.relpath(
+                                os.path.join(dirpath, candidate), path
+                            )
+                            matching_files.append(rel)
+    else:
+        # Non-recursive: only list entries in the given directory
+        for candidate in os.listdir(path):
+            if not regex_compiled:
+                if candidate.lower().endswith(suffix.lower()):
+                    if fullpath:
+                        matching_files.append(os.path.join(path, candidate))
+                    else:
+                        matching_files.append(candidate)
             else:
-                matching_files.append(candidate)
+                if regex_compiled.match(candidate):
+                    if fullpath:
+                        matching_files.append(os.path.join(path, candidate))
+                    else:
+                        matching_files.append(candidate)
 
     if sort:
         matching_files = strtools.sort_alphanumerically(matching_files)