fix test_path issues (#160)

* fix test_path issues

* changelog

Author: mathause

CHANGELOG.md

@@ -2,8 +2,9 @@
 
 ## v1.0.2 - unreleased
 
-- Restore behavior of `FileContainer.search(key=None)` - passing `None` to `search` is ignored again [#143](https://github.com/mpytools/filefisher/issues/143).
+- Restore behavior of `FileContainer.search(key=None)` - passing `None` to `search` is ignored again ([#143](https://github.com/mpytools/filefisher/issues/143)).
 - Added `FileContainer.search_single` to search for _exactly_ one path in a `FileContainer` ([#158](https://github.com/mpytools/filefisher/pull/158)).
+- Fixed issues of non-unique `test_path` ([#139](https://github.com/mpytools/filefisher/issues/139)).
 
 ## v1.0.1 - 16.01.205
 This version patches a bug from v1.0.0. that broke `FileContainer.concat()`.

filefisher/_filefinder.py

@@ -300,21 +300,28 @@ def __init__(
         self._full_pattern = self.full.pattern
 
         if test_paths is not None:
+
+            if isinstance(test_paths, str):
+                test_paths = [test_paths]
+
+            if len(test_paths) != len(set(test_paths)):
+                raise ValueError("`test_paths` are not unique")
+
             self._set_test_paths(test_paths)
 
     def _set_test_paths(self, test_paths):
 
-        if isinstance(test_paths, str):
-            test_paths = [test_paths]
-
         self._test_paths = test_paths
 
         # use fnmatch.filter to 'glob' pseudo-filenames
         def finder(pat):
 
             # make fnmatch work (almost) the same as glob
             if pat.endswith(os.path.sep):
-                paths_ = [os.path.dirname(s) + os.path.sep for s in test_paths]
+                # remove duplicate paths (i.e. if the filename made it non-unique)
+                paths_ = sorted(
+                    set(os.path.dirname(s) + os.path.sep for s in test_paths)
+                )
             else:
                 paths_ = test_paths
 

filefisher/tests/test_filefinder.py

@@ -134,6 +134,18 @@ def test_test_path_property():
     assert ff._test_paths == ["a", "b"]
 
 
+def test_test_path_assert_nonunique():
+
+    with pytest.raises(ValueError, match="`test_paths` are not unique"):
+        FileFinder("a", "", test_paths=["a", "a"])
+
+    with pytest.raises(ValueError, match="`test_paths` are not unique"):
+        FileFinder("a", "", test_paths=["a/", "a/"])
+
+    with pytest.raises(ValueError, match="`test_paths` are not unique"):
+        FileFinder("a", "", test_paths=["a/b", "a/b"])
+
+
 def test_file_pattern_no_sep():
 
     path_pattern = "path_pattern"
@@ -279,12 +291,15 @@ def test_find_path_none_found(tmp_path, test_paths):
 
 def test_find_paths_non_unique():
 
-    # test raises error for non-unique metadata - AFAIK not possible for real paths
+    # ensure find_paths works for duplicated folder names (made unique by the file name)
 
-    ff = FileFinder("{cat}", "", test_paths=["a/", "a/"])
+    ff = FileFinder("{cat}", "", test_paths=["a/a", "a/b"])
 
-    with pytest.raises(ValueError, match="Non-unique metadata detected"):
-        ff.find_paths()
+    result = ff.find_paths()
+
+    expected = {"path": {0: "a/*"}, "cat": {0: "a"}}
+    expected = pd.DataFrame.from_dict(expected).set_index("path")
+    pd.testing.assert_frame_equal(result.df, expected)
 
 
 def test_find_paths_simple(tmp_path, test_paths):
@@ -460,7 +475,9 @@ def test_find_files_non_unique():
 
     # test raises error for non-unique metadata - AFAIK not possible for real paths
 
-    ff = FileFinder("", "{cat}", test_paths=["/a", "/a"])
+    ff = FileFinder("", "{cat}")
+    # need to set via `_set_test_paths` to avoid duplicate check
+    ff._set_test_paths(test_paths=["/a", "/a"])
 
     with pytest.raises(ValueError, match="Non-unique metadata detected"):
         ff.find_files()