Merge pull request #1576 from itsluketwist/fix-trailers

Fix up the commit trailers functionality

Author: Byron

git/objects/commit.py

@@ -26,6 +26,7 @@
 import os
 from io import BytesIO
 import logging
+from collections import defaultdict
 
 
 # typing ------------------------------------------------------------------
@@ -335,8 +336,72 @@ def stats(self) -> Stats:
         return Stats._list_from_string(self.repo, text)
 
     @property
-    def trailers(self) -> Dict:
-        """Get the trailers of the message as dictionary
+    def trailers(self) -> Dict[str, str]:
+        """Get the trailers of the message as a dictionary
+
+        :note: This property is deprecated, please use either ``Commit.trailers_list`` or ``Commit.trailers_dict``.
+
+        :return:
+            Dictionary containing whitespace stripped trailer information.
+            Only contains the latest instance of each trailer key.
+        """
+        return {
+            k: v[0] for k, v in self.trailers_dict.items()
+        }
+
+    @property
+    def trailers_list(self) -> List[Tuple[str, str]]:
+        """Get the trailers of the message as a list
+
+        Git messages can contain trailer information that are similar to RFC 822
+        e-mail headers (see: https://git-scm.com/docs/git-interpret-trailers).
+
+        This functions calls ``git interpret-trailers --parse`` onto the message
+        to extract the trailer information, returns the raw trailer data as a list.
+
+        Valid message with trailer::
+
+            Subject line
+
+            some body information
+
+            another information
+
+            key1: value1.1
+            key1: value1.2
+            key2 :    value 2 with inner spaces
+
+
+        Returned list will look like this::
+
+            [
+                ("key1", "value1.1"),
+                ("key1", "value1.2"),
+                ("key2", "value 2 with inner spaces"),
+            ]
+
+
+        :return:
+            List containing key-value tuples of whitespace stripped trailer information.
+        """
+        cmd = ["git", "interpret-trailers", "--parse"]
+        proc: Git.AutoInterrupt = self.repo.git.execute(cmd, as_process=True, istream=PIPE)  # type: ignore
+        trailer: str = proc.communicate(str(self.message).encode())[0].decode("utf8")
+        trailer = trailer.strip()
+
+        if not trailer:
+            return []
+
+        trailer_list = []
+        for t in trailer.split("\n"):
+            key, val = t.split(":", 1)
+            trailer_list.append((key.strip(), val.strip()))
+
+        return trailer_list
+
+    @property
+    def trailers_dict(self) -> Dict[str, List[str]]:
+        """Get the trailers of the message as a dictionary
 
         Git messages can contain trailer information that are similar to RFC 822
         e-mail headers (see: https://git-scm.com/docs/git-interpret-trailers).
@@ -345,42 +410,35 @@ def trailers(self) -> Dict:
         to extract the trailer information. The key value pairs are stripped of
         leading and trailing whitespaces before they get saved into a dictionary.
 
-        Valid message with trailer:
-
-        .. code-block::
+        Valid message with trailer::
 
             Subject line
 
             some body information
 
             another information
 
-            key1: value1
+            key1: value1.1
+            key1: value1.2
             key2 :    value 2 with inner spaces
 
-        dictionary will look like this:
 
-        .. code-block::
+        Returned dictionary will look like this::
 
             {
-                "key1": "value1",
-                "key2": "value 2 with inner spaces"
+                "key1": ["value1.1", "value1.2"],
+                "key2": ["value 2 with inner spaces"],
             }
 
-        :return: Dictionary containing whitespace stripped trailer information
 
+        :return:
+            Dictionary containing whitespace stripped trailer information.
+            Mapping trailer keys to a list of their corresponding values.
         """
-        d = {}
-        cmd = ["git", "interpret-trailers", "--parse"]
-        proc: Git.AutoInterrupt = self.repo.git.execute(cmd, as_process=True, istream=PIPE)  # type: ignore
-        trailer: str = proc.communicate(str(self.message).encode())[0].decode()
-        if trailer.endswith("\n"):
-            trailer = trailer[0:-1]
-        if trailer != "":
-            for line in trailer.split("\n"):
-                key, value = line.split(":", 1)
-                d[key.strip()] = value.strip()
-        return d
+        d = defaultdict(list)
+        for key, val in self.trailers_list:
+            d[key].append(val)
+        return dict(d)
 
     @classmethod
     def _iter_from_process_or_stream(cls, repo: "Repo", proc_or_stream: Union[Popen, IO]) -> Iterator["Commit"]:

test/test_commit.py

@@ -494,52 +494,57 @@ def test_datetimes(self):
 
     def test_trailers(self):
         KEY_1 = "Hello"
-        VALUE_1 = "World"
+        VALUE_1_1 = "World"
+        VALUE_1_2 = "Another-World"
         KEY_2 = "Key"
         VALUE_2 = "Value with inner spaces"
 
-        # Check if KEY 1 & 2 with Value 1 & 2 is extracted from multiple msg variations
-        msgs = []
-        msgs.append(f"Subject\n\n{KEY_1}: {VALUE_1}\n{KEY_2}: {VALUE_2}\n")
-        msgs.append(f"Subject\n  \nSome body of a function\n \n{KEY_1}: {VALUE_1}\n{KEY_2}: {VALUE_2}\n")
-        msgs.append(
-            f"Subject\n  \nSome body of a function\n\nnon-key: non-value\n\n{KEY_1}: {VALUE_1}\n{KEY_2}: {VALUE_2}\n"
-        )
-        msgs.append(
-            f"Subject\n  \nSome multiline\n body of a function\n\nnon-key: non-value\n\n{KEY_1}: {VALUE_1}\n{KEY_2} :      {VALUE_2}\n"
-        )
-
+        # Check the following trailer example is extracted from multiple msg variations
+        TRAILER = f"{KEY_1}: {VALUE_1_1}\n{KEY_2}: {VALUE_2}\n{KEY_1}: {VALUE_1_2}"
+        msgs = [
+            f"Subject\n\n{TRAILER}\n",
+            f"Subject\n  \nSome body of a function\n \n{TRAILER}\n",
+            f"Subject\n  \nSome body of a function\n\nnon-key: non-value\n\n{TRAILER}\n",
+            (
+                # check when trailer has inconsistent whitespace
+                f"Subject\n  \nSome multiline\n body of a function\n\nnon-key: non-value\n\n"
+                f"{KEY_1}:{VALUE_1_1}\n{KEY_2} :      {VALUE_2}\n{KEY_1}:    {VALUE_1_2}\n"
+            ),
+        ]
         for msg in msgs:
-            commit = self.rorepo.commit("master")
-            commit = copy.copy(commit)
+            commit = copy.copy(self.rorepo.commit("master"))
             commit.message = msg
-            assert KEY_1 in commit.trailers.keys()
-            assert KEY_2 in commit.trailers.keys()
-            assert commit.trailers[KEY_1] == VALUE_1
-            assert commit.trailers[KEY_2] == VALUE_2
-
-        # Check that trailer stays empty for multiple msg combinations
-        msgs = []
-        msgs.append(f"Subject\n")
-        msgs.append(f"Subject\n\nBody with some\nText\n")
-        msgs.append(f"Subject\n\nBody with\nText\n\nContinuation but\n doesn't contain colon\n")
-        msgs.append(f"Subject\n\nBody with\nText\n\nContinuation but\n only contains one :\n")
-        msgs.append(f"Subject\n\nBody with\nText\n\nKey: Value\nLine without colon\n")
-        msgs.append(f"Subject\n\nBody with\nText\n\nLine without colon\nKey: Value\n")
+            assert commit.trailers_list == [
+                (KEY_1, VALUE_1_1),
+                (KEY_2, VALUE_2),
+                (KEY_1, VALUE_1_2),
+            ]
+            assert commit.trailers_dict == {
+                KEY_1: [VALUE_1_1, VALUE_1_2],
+                KEY_2: [VALUE_2],
+            }
+
+        # check that trailer stays empty for multiple msg combinations
+        msgs = [
+            f"Subject\n",
+            f"Subject\n\nBody with some\nText\n",
+            f"Subject\n\nBody with\nText\n\nContinuation but\n doesn't contain colon\n",
+            f"Subject\n\nBody with\nText\n\nContinuation but\n only contains one :\n",
+            f"Subject\n\nBody with\nText\n\nKey: Value\nLine without colon\n",
+            f"Subject\n\nBody with\nText\n\nLine without colon\nKey: Value\n",
+        ]
 
         for msg in msgs:
-            commit = self.rorepo.commit("master")
-            commit = copy.copy(commit)
+            commit = copy.copy(self.rorepo.commit("master"))
             commit.message = msg
-            assert len(commit.trailers.keys()) == 0
+            assert commit.trailers_list == []
+            assert commit.trailers_dict == {}
 
         # check that only the last key value paragraph is evaluated
-        commit = self.rorepo.commit("master")
-        commit = copy.copy(commit)
-        commit.message = f"Subject\n\nMultiline\nBody\n\n{KEY_1}: {VALUE_1}\n\n{KEY_2}: {VALUE_2}\n"
-        assert KEY_1 not in commit.trailers.keys()
-        assert KEY_2 in commit.trailers.keys()
-        assert commit.trailers[KEY_2] == VALUE_2
+        commit = copy.copy(self.rorepo.commit("master"))
+        commit.message = f"Subject\n\nMultiline\nBody\n\n{KEY_1}: {VALUE_1_1}\n\n{KEY_2}: {VALUE_2}\n"
+        assert commit.trailers_list == [(KEY_2, VALUE_2)]
+        assert commit.trailers_dict == {KEY_2: [VALUE_2]}
 
     def test_commit_co_authors(self):
         commit = copy.copy(self.rorepo.commit("4251bd5"))