Duplicate Changelog Entries for Merged Branches

[lovetheguitar]

Description

Somehow I end up with duplicate changelog entries after merging a MR.
E.g.

## v0.1.2 (2022-05-02)

### Fix

- **release**: do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH
- **release**: do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH
- **release**: try to not escape $CI_DEFAULT_BRANCH default value
- **release**: do not run run bump_version/dry_run in tag pipeline

Steps to reproduce

Use gitlab and add some commits to a new branch.
Open a MR and merge it.
Run commitizen on the main branch and get a changelog with duplicated entries.

Current behavior

Duplicate Changelog Entries:

## v0.1.2 (2022-05-02)

### Fix

- **release**: do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH
- **release**: do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH
- **release**: try to not escape $CI_DEFAULT_BRANCH default value
- **release**: do not run run bump_version/dry_run in tag pipeline

Desired behavior

No duplicates from merging a MR:

## v0.1.2 (2022-05-02)

### Fix

- **release**: do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH
- **release**: try to not escape $CI_DEFAULT_BRANCH default value
- **release**: do not run run bump_version/dry_run in tag pipeline

Screenshots

Problem appears to be that commitizen seems to parse the commit body also?!

Git log shows:

Date:   Mon May 2 15:28:31 2022 +0200

    Merge branch 'fix/improve_release_jobs' into 'master'

    fix(release): do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH

    Closes #4

    See merge request group/ci_templates!10

Date:   Mon May 2 15:12:10 2022 +0200

    fix(release): do version bump on $RELEASE_BRANCH (if defined) else on $CI_DEFAULT_BRANCH

    else dry run version bump
...

Environment

Commitizen Version: 2.24.0
Python Version: 3.8.10 (tags/v3.8.10:3d8993a, May 3 2021, 11:48:03) [MSC v.1928 64 bit (AMD64)]
Operating System: Windows

Thanks for the nice tool. 😃

[tanguy-a]

Hey, got the exact same issue here:
Commitizen Version: 2.24.0
Python Version: 3.7.12 && 3.9.1

I tried with pip3 install -U Commitizen==2.20.0 and it did not duplicate.
The versions after this one seem to have the duplication issue. So 2.20.1 duplicates.

[adam-grant-hendry]

@lovetheguitar Thanks for submitting this! I think I'll downgrade until a fix is submitted.

@tanguy-a Thanks for the tip!

[Lee-W]

Hi @lovetheguitar could you please provide concrete steps for us to reproduce this issue on our side? Thanks !

[woile]

I think this is related to the fact that we process the body, even though the subject of a commit message may not match conventional commits.

commitizen/commitizen/changelog.py

Lines 116 to 127 in afa0d93

	# Process subject from commit message
	message = map_pat.match(commit.message)
	if message:
	parsed_message: Dict = message.groupdict()
	# change_type becomes optional by providing None
	change_type = parsed_message.pop("change_type", None)
	if change_type_map:
	change_type = change_type_map.get(change_type, change_type)
	if changelog_message_builder_hook:
	parsed_message = changelog_message_builder_hook(parsed_message, commit)
	changes[change_type].append(parsed_message)

commitizen/commitizen/changelog.py

Lines 129 to 140 in afa0d93

	# Process body from commit message
	body_parts = commit.body.split("\n\n")
	for body_part in body_parts:
	message_body = body_map_pat.match(body_part)
	if not message_body:
	continue
	parsed_message_body: Dict = message_body.groupdict()
	change_type = parsed_message_body.pop("change_type", None)
	if change_type_map:
	change_type = change_type_map.get(change_type, change_type)
	changes[change_type].append(parsed_message_body)

I think if the subject doesn't match, we should skip directly, no?

I see this happening also now because github is adding the commit message in the body.