Merge pull request #4 from Wonshtrum/name_with_spaces

Added git support for files with names containing spaces

codemetrics/git.py

@@ -42,8 +42,14 @@ def __init__(self, git_client=default_client, cwd: pl.Path = None, _pdb=False):
         super().__init__(cwd=cwd)
         self._pdb = _pdb
         self.git_client = git_client
+        self.log_re = re.compile(
+            r"([-\d]+)\s+([-\d]+)\s+(.*)"
+        )
         self.log_moved_re = re.compile(
-            r"([-\d]+)\s+([-\d]+)\s+(\S*)\{(\S*) => (\S*)\}(\S*)"
+            r"([-\d]+)\s+([-\d]+)\s+(.*) => (.*)"
+        )
+        self.log_curly_re = re.compile(
+            r"([-\d]+)\s+([-\d]+)\s+(.*)\{(.*) => (.*)\}(.*)"
         )
 
     def parse_path_elem(self, path_elem: str):
@@ -60,22 +66,19 @@ def parse_path_elem(self, path_elem: str):
 
         """
         copy_from_path: typing.Optional[str] = None
-        if "{" not in path_elem:
-            if "=>" in path_elem:
-                added, removed, copy_from_path, _, rel_path = path_elem.split()
-            else:
-                added, removed, rel_path = path_elem.split()
+        match_log = self.log_re.match(path_elem)
+        match_log_moved = self.log_moved_re.match(path_elem)
+        match_log_curly = self.log_curly_re.match(path_elem)
+        if match_log_curly:
+            added, removed, base, from_mid, rel_mid, end = match_log_curly.groups()
+            rel_path = (base + rel_mid + end).replace("//", "/")
+            copy_from_path = (base + from_mid + end).replace("//", "/")
+        elif match_log_moved:
+            added, removed, copy_from_path, rel_path = match_log_moved.groups()
+        elif match_log:
+            added, removed, rel_path = match_log.groups()
         else:
-            match = self.log_moved_re.match(path_elem)
-            if not match:
-                raise ValueError(f"{path_elem} not understood")
-            added = match.group(1)
-            removed = match.group(2)
-            rel_path = match.group(3) + match.group(5) + match.group(6)
-            rel_path = rel_path.replace("//", "/")
-            copy_from_path = (match.group(3) + match.group(4) + match.group(6)).replace(
-                "//", "/"
-            )
+            raise ValueError(f"{path_elem} not understood")
         added_as_int = int(added) if added != "-" else np.nan
         removed_as_int = int(removed) if removed != "-" else np.nan
         return added_as_int, removed_as_int, rel_path, copy_from_path

tests/test_git.py

@@ -33,6 +33,33 @@ def test_parse_path_elem(self):
         self.assertEqual("dir/test.py", relpath)
         self.assertIsNone(copyfrompath)
 
+    def test_parse_path_elem_with_spaces(self):
+        """Parsing of path element."""
+        pe = "21        2   	dir 1/test 1.py"
+        added, removed, relpath, copyfrompath = self.git.parse_path_elem(pe)
+        self.assertEqual(21, added)
+        self.assertEqual(2, removed)
+        self.assertEqual("dir 1/test 1.py", relpath)
+        self.assertIsNone(copyfrompath)
+
+    def test_parse_renamed_path_no_curly(self):
+        """Parsing of path element."""
+        pe = "1       1       a.py => b.py"
+        added, removed, relpath, copyfrompath = self.git.parse_path_elem(pe)
+        self.assertEqual(1, added)
+        self.assertEqual(1, removed)
+        self.assertEqual("b.py", relpath)
+        self.assertEqual("a.py", copyfrompath)
+
+    def test_parse_renamed_path_no_curly_with_spaces(self):
+        """Parsing of path element."""
+        pe = "1       1       a 1.py => b 1.py"
+        added, removed, relpath, copyfrompath = self.git.parse_path_elem(pe)
+        self.assertEqual(1, added)
+        self.assertEqual(1, removed)
+        self.assertEqual("b 1.py", relpath)
+        self.assertEqual("a 1.py", copyfrompath)
+
     def test_parse_renamed_path(self):
         """Parsing of path element."""
         pe = "1       1       dir/{b/a.py => a/b.py}"
@@ -42,6 +69,15 @@ def test_parse_renamed_path(self):
         self.assertEqual("dir/a/b.py", relpath)
         self.assertEqual("dir/b/a.py", copyfrompath)
 
+    def test_parse_renamed_path_with_spaces(self):
+        """Parsing of path element."""
+        pe = "1       1       dir 1/{b 1/a a.py => a 1/b b.py}"
+        added, removed, relpath, copyfrompath = self.git.parse_path_elem(pe)
+        self.assertEqual(1, added)
+        self.assertEqual(1, removed)
+        self.assertEqual("dir 1/a 1/b b.py", relpath)
+        self.assertEqual("dir 1/b 1/a a.py", copyfrompath)
+
     def test_parse_renamed_path_empty_right(self):
         """Parsing of path element."""
         pe = "21        2   	dir/{category => }/test.py"
@@ -51,6 +87,15 @@ def test_parse_renamed_path_empty_right(self):
         self.assertEqual("dir/test.py", relpath)
         self.assertEqual("dir/category/test.py", copyfrompath)
 
+    def test_parse_renamed_path_empty_right_with_spaces(self):
+        """Parsing of path element."""
+        pe = "21        2   	dir 1/{category 1 => }/test 1.py"
+        added, removed, relpath, copyfrompath = self.git.parse_path_elem(pe)
+        self.assertEqual(21, added)
+        self.assertEqual(2, removed)
+        self.assertEqual("dir 1/test 1.py", relpath)
+        self.assertEqual("dir 1/category 1/test 1.py", copyfrompath)
+
     def test_parse_renamed_path_empty_left(self):
         """Parsing of path element."""
         pe = "-       -       dir/{ => subdir}/file.py"
@@ -60,6 +105,15 @@ def test_parse_renamed_path_empty_left(self):
         self.assertEqual("dir/subdir/file.py", relpath)
         self.assertEqual("dir/file.py", copyfrompath)
 
+    def test_parse_renamed_path_empty_left_with_spaces(self):
+        """Parsing of path element."""
+        pe = "-       -       dir 1/{ => subdir 1}/file 1.py"
+        added, removed, relpath, copyfrompath = self.git.parse_path_elem(pe)
+        self.assertTrue(np.isnan(added))
+        self.assertTrue(np.isnan(removed))
+        self.assertEqual("dir 1/subdir 1/file 1.py", relpath)
+        self.assertEqual("dir 1/file 1.py", copyfrompath)
+
 
 def get_log():
     retval = textwrap.dedent(