feat: configurable commit validation

Allow clients overriding the BaseCommitizen object to have more custom
commit message validation beyond just matching a regex schema. This lets
clients have custom error messages and more intelligent pattern
matching.

Author: protonjohn

commitizen/commands/check.py

@@ -1,15 +1,10 @@
 import os
-import re
 import sys
 from typing import Any, Dict, Optional
 
 from commitizen import factory, git, out
 from commitizen.config import BaseConfig
-from commitizen.exceptions import (
-    InvalidCommandArgumentError,
-    InvalidCommitMessageError,
-    NoCommitsFoundError,
-)
+from commitizen.exceptions import InvalidCommandArgumentError, NoCommitsFoundError
 
 
 class Check:
@@ -54,31 +49,13 @@ def __call__(self):
         """Validate if commit messages follows the conventional pattern.
 
         Raises:
-            InvalidCommitMessageError: if the commit provided not follows the conventional pattern
+            InvalidCommitMessageError: if the provided commit does not follow the conventional pattern
         """
         commits = self._get_commits()
         if not commits:
             raise NoCommitsFoundError(f"No commit found with range: '{self.rev_range}'")
 
-        pattern = self.cz.schema_pattern()
-        ill_formated_commits = [
-            commit
-            for commit in commits
-            if not self.validate_commit_message(commit.message, pattern)
-        ]
-        displayed_msgs_content = "\n".join(
-            [
-                f'commit "{commit.rev}": "{commit.message}"'
-                for commit in ill_formated_commits
-            ]
-        )
-        if displayed_msgs_content:
-            raise InvalidCommitMessageError(
-                "commit validation: failed!\n"
-                "please enter a commit message in the commitizen format.\n"
-                f"{displayed_msgs_content}\n"
-                f"pattern: {pattern}"
-            )
+        self.cz.validate_commits(commits, self.allow_abort)
         out.success("Commit validation: successful!")
 
     def _get_commits(self):
@@ -128,16 +105,3 @@ def _filter_comments(msg: str) -> str:
             if not line.startswith("#"):
                 lines.append(line)
         return "\n".join(lines)
-
-    def validate_commit_message(self, commit_msg: str, pattern: str) -> bool:
-        if not commit_msg:
-            return self.allow_abort
-        if (
-            commit_msg.startswith("Merge")
-            or commit_msg.startswith("Revert")
-            or commit_msg.startswith("Pull request")
-            or commit_msg.startswith("fixup!")
-            or commit_msg.startswith("squash!")
-        ):
-            return True
-        return bool(re.match(pattern, commit_msg))

commitizen/cz/__init__.py

@@ -5,7 +5,7 @@
 
 from commitizen.cz.base import BaseCommitizen
 from commitizen.cz.conventional_commits import ConventionalCommitsCz
-from commitizen.cz.customize import CustomizeCommitsCz
+from commitizen.cz.customize import CustomizeCommitsCz, CustomizeCommitValidationCz
 from commitizen.cz.jira import JiraSmartCz
 
 
@@ -34,6 +34,7 @@ def discover_plugins(path: Iterable[str] = None) -> Dict[str, Type[BaseCommitize
     "cz_conventional_commits": ConventionalCommitsCz,
     "cz_jira": JiraSmartCz,
     "cz_customize": CustomizeCommitsCz,
+    "cz_custom_validation": CustomizeCommitValidationCz,
 }
 
 registry.update(discover_plugins())

commitizen/cz/base.py

@@ -1,3 +1,4 @@
+import re
 from abc import ABCMeta, abstractmethod
 from typing import Callable, Dict, List, Optional, Tuple
 
@@ -6,6 +7,7 @@
 from commitizen import git
 from commitizen.config.base_config import BaseConfig
 from commitizen.defaults import Questions
+from commitizen.exceptions import InvalidCommitMessageError
 
 
 class BaseCommitizen(metaclass=ABCMeta):
@@ -74,6 +76,47 @@ def schema_pattern(self) -> Optional[str]:
         """Regex matching the schema used for message validation."""
         raise NotImplementedError("Not Implemented yet")
 
+    def validate_commit_message(
+        self, commit: git.GitCommit, pattern: str, allow_abort: bool
+    ) -> bool:
+        if not commit.message:
+            return allow_abort
+        if (
+            commit.message.startswith("Merge")
+            or commit.message.startswith("Revert")
+            or commit.message.startswith("Pull request")
+            or commit.message.startswith("fixup!")
+            or commit.message.startswith("squash!")
+        ):
+            return True
+        return bool(re.match(pattern, commit.message))
+
+    def validate_commits(self, commits: List[git.GitCommit], allow_abort: bool):
+        """
+        Validate a commit. Invokes schema_pattern by default.
+        Raises:
+            InvalidCommitMessageError: if the provided commit does not follow the conventional pattern
+        """
+
+        pattern = self.schema_pattern()
+        assert pattern is not None
+
+        displayed_msgs_content = "\n".join(
+            [
+                f'commit "{commit.rev}": "{commit.message}"'
+                for commit in commits
+                if not self.validate_commit_message(commit, pattern, allow_abort)
+            ]
+        )
+
+        if displayed_msgs_content:
+            raise InvalidCommitMessageError(
+                "commit validation: failed!\n"
+                "please enter a commit message in the commitizen format.\n"
+                f"{displayed_msgs_content}\n"
+                f"pattern: {pattern}"
+            )
+
     def info(self) -> Optional[str]:
         """Information about the standardized commit message."""
         raise NotImplementedError("Not Implemented yet")

commitizen/cz/customize/__init__.py

@@ -1 +1 @@
-from .customize import CustomizeCommitsCz  # noqa
+from .customize import CustomizeCommitsCz, CustomizeCommitValidationCz  # noqa

commitizen/cz/customize/customize.py

@@ -3,15 +3,19 @@
 except ImportError:
     from string import Template  # type: ignore
 
+import re
 from typing import Optional
 
 from commitizen import defaults
 from commitizen.config import BaseConfig
 from commitizen.cz.base import BaseCommitizen
 from commitizen.defaults import Questions
-from commitizen.exceptions import MissingCzCustomizeConfigError
+from commitizen.exceptions import (
+    InvalidCommitMessageError,
+    MissingCzCustomizeConfigError,
+)
 
-__all__ = ["CustomizeCommitsCz"]
+__all__ = ["CustomizeCommitsCz", "CustomizeCommitValidationCz"]
 
 
 class CustomizeCommitsCz(BaseCommitizen):
@@ -79,3 +83,53 @@ def info(self) -> Optional[str]:
         elif info:
             return info
         return None
+
+
+class CustomizeCommitValidationCz(CustomizeCommitsCz):
+    def validate_commit(self, pattern, commit, allow_abort):
+        """
+        Validates that a commit message doesn't contain certain tokens.
+        """
+        if not commit.message:
+            return allow_abort
+
+        invalid_tokens = ["Merge", "Revert", "Pull request", "fixup!", "squash!"]
+        for token in invalid_tokens:
+            if commit.message.startswith(token):
+                raise InvalidCommitMessageError(
+                    f"Commits may not start with the token {token}."
+                )
+
+        return re.match(pattern, commit.message)
+
+    def validate_commits(self, commits, allow_abort):
+        """
+        Validates a list of commits against the configured commit validation schema.
+        See the schema() and example() functions for examples.
+        """
+
+        pattern = self.schema_pattern()
+
+        displayed_msgs_content = []
+        for commit in commits:
+            message = ""
+            valid = False
+            try:
+                valid = self.validate_commit(pattern, commit, allow_abort)
+            except InvalidCommitMessageError as e:
+                message = e.message
+
+            if not valid:
+                displayed_msgs_content.append(
+                    f"commit {commit.rev}\n"
+                    f"Author: {commit.author} <{commit.author_email}>\n\n"
+                    f"{commit.message}\n\n"
+                    f"commit validation: failed! {message}\n"
+                )
+
+        if displayed_msgs_content:
+            displayed_msgs = "\n".join(displayed_msgs_content)
+            raise InvalidCommitMessageError(
+                f"{displayed_msgs}\n"
+                "Please enter a commit message in the commitizen format.\n"
+            )

docs/customization.md

@@ -317,6 +317,46 @@ cz -n cz_strange bump
 
 [convcomms]: https://github.com/commitizen-tools/commitizen/blob/master/commitizen/cz/conventional_commits/conventional_commits.py
 
+### Custom Commit Validation
+
+The default implementation of `validate_commits` will call `validate_commit` for each commit passed as input to the `check` command, and evaluate
+it against the `schema_pattern` regular expression. You can override those 3 functions on `BaseCommitizen` for different levels of flexibility.
+
+```python
+import re
+from commitizen.cz.base import BaseCommitizen
+from commitizen.git import GitCommit
+
+class StrangeCommitizen(BaseCommitizen):
+    def schema_pattern(self) -> Optional[str]:
+        # collect the different prefix options into a regex
+        choices = ["feat", "fix", "chore", "bump"]
+        prefixes_regex = r"|".join(choices)
+
+        return (
+            f"({prefixes_regex})" # different prefixes
+            "(\(\S+\))?!?:(\s.*)" # scope & subject body
+        )
+
+    def validate_commit(self, pattern: str, commit: GitCommit, allow_abort: bool) -> bool:
+        return allow_abort if not re.match(pattern, commit.message) else True
+    
+    def validate_commits(self, commits: List[GitCommit], allow_abort: bool):
+        pattern = self.schema_pattern()
+        invalid_commits = "\n".join([
+            (
+                f'"{commit.rev}": "{commit.message}"\n'
+                "commit validation failed!"
+            )
+            for commit in commits
+            if not self.validate_commit(pattern, commit, allow_abort)
+        ])
+        
+        if invalid_commits:
+            raise InvalidCommitMessageError(invalid_commits)
+
+```
+
 ### Custom changelog generator
 
 The changelog generator should just work in a very basic manner without touching anything.

tests/commands/test_check_command.py

@@ -128,6 +128,46 @@ def test_check_conventional_commit_succeeds(mocker: MockFixture, capsys):
     assert "Commit validation: successful!" in out
 
 
+def test_check_custom_validation_succeeds(mocker, capsys, config_customize):
+    testargs = [
+        "cz",
+        "-n",
+        "cz_custom_validation",
+        "check",
+        "--commit-msg-file",
+        "some_file",
+    ]
+    mocker.patch.object(sys, "argv", testargs)
+    mocker.patch(
+        "commitizen.commands.check.open",
+        mocker.mock_open(read_data="fix(scope): some commit message"),
+    )
+    mocker.patch("commitizen.config.read_cfg", return_value=config_customize)
+    cli.main()
+    out, _ = capsys.readouterr()
+    assert "Commit validation: successful!" in out
+
+
+def test_check_custom_validation_fails(mocker, config_customize):
+    testargs = [
+        "cz",
+        "-n",
+        "cz_custom_validation",
+        "check",
+        "--commit-msg-file",
+        "some_file",
+    ]
+    mocker.patch.object(sys, "argv", testargs)
+    mocker.patch(
+        "commitizen.commands.check.open",
+        mocker.mock_open(read_data="fixup! fix(scope): some commit message"),
+    )
+    mocker.patch("commitizen.config.read_cfg", return_value=config_customize)
+    with pytest.raises(InvalidCommitMessageError) as excinfo:
+        cli.main()
+    assert "commit validation: failed!" in str(excinfo.value)
+
+
 @pytest.mark.parametrize(
     "commit_msg",
     (
@@ -332,7 +372,7 @@ def test_check_command_with_pipe_message_and_failed(mocker: MockFixture):
     assert "commit validation: failed!" in str(excinfo.value)
 
 
-def test_check_command_with_comment_in_messege_file(mocker: MockFixture, capsys):
+def test_check_command_with_comment_in_message_file(mocker: MockFixture, capsys):
     testargs = ["cz", "check", "--commit-msg-file", "some_file"]
     mocker.patch.object(sys, "argv", testargs)
     mocker.patch(