TinySemVer is a minimalistic Semantic Versioning package for projects following Conventional Commits in a single short Python file, capable of generating changelogs and release notes with the use of OpenAI-compatible LLM APIs, or without them.
In plain English, if your commit messages look like feat: add new feature
or fix: bugfix
, this package will automate releasing new "GIT tags" based on the commit messages.
Here is how to integrate it into your project CI:
$ pip install tinysemver
$ tinysemver --dry-run --verbose
> Current version: 1.2.2
> Next version: 1.3.0
The --dry-run
flag will only print the next version without changing any files.
Great for pre-release CI pipelines.
If you need more control over the default specification, here are more options you can run against the files in this repository:
# This won't push
$ tinysemver --verbose \
--major-verbs 'breaking,break,major' \
--minor-verbs 'feature,minor,add,new' \
--patch-verbs 'fix,patch,bug,improve,docs,make' \
--changelog-file 'CHANGELOG.md' \
--version-file 'VERSION' \
--update-version-in 'pyproject.toml' '^version = "(\d+\.\d+\.\d+)"' \
--github-repository 'ashvardanian/tinysemver'
# Revert to the previous commit
$ git reset --soft HEAD~1
It's recommended to use strict version matching with \d+\.\d+\.\d+
instead of a generic wildcard like .*
, but both would work.
Here is an example of passing even more parameters for a project like stringzilla
:
$ tinysemver --verbose \
--major-verbs 'breaking,break,major' \
--minor-verbs 'feature,minor,add,new' \
--patch-verbs 'fix,patch,bug,improve,docs,make' \
--changelog-file 'CHANGELOG.md' \
--version-file 'VERSION' \
--update-version-in 'pyproject.toml' '^version = "(\d+\.\d+\.\d+)"' \
--update-version-in 'package.json' '"version": "(.*)"' \
--update-version-in 'CITATION.cff' '^version: (.*)' \
--update-major-version-in 'include/stringzilla/stringzilla.h' '^#define STRINGZILLA_VERSION_MAJOR (.*)' \
--update-minor-version-in 'include/stringzilla/stringzilla.h' '^#define STRINGZILLA_VERSION_MINOR (.*)' \
--update-patch-version-in 'include/stringzilla/stringzilla.h' '^#define STRINGZILLA_VERSION_PATCH (.*)' \
--github-repository 'ashvardanian/stringzilla' \
--push
> Current version: 1.2.2
> ? Commits since last tag: 3 # Only in verbose mode
> # 5579972: Improve: Log file patches # Only in verbose mode
> # de645ea: Improve: Grouping CHANGELOG # Only in verbose mode
> Next version: 1.3.0
> Will update file: VERSION:0
> - 1.2.2 # Only in verbose mode
> + 1.3.0 # Only in verbose mode
> Will update file: package.json:5
> - "version": "1.2.2" # Only in verbose mode
> + "version": "1.3.0" # Only in verbose mode
> Will update file: pyproject.toml:7
> - version = "1.2.2" # Only in verbose mode
> + version = "1.3.0" # Only in verbose mode
> Will update file: CITATION.cff:7
> - version: 1.2.2 # Only in verbose mode
> + version: 1.3.0 # Only in verbose mode
> Appending to changelog file: CHANGELOG.md
> = skipping 250 lines # Only in verbose mode
> + adding 30 lines # Only in verbose mode
Alternatively, you can just ask for --help
:
$ tinysemver --help
TinySemVer can leverage a language model to validate the commits and generate clean and infromative release notes.
$ tinysemver --verbose \
--github-repository 'ashvardanian/tinysemver' \
--openai-base-url 'https://api.groq.com/openai/v1' \
--openai-api-key 'GET_YOURSELF_A_KEY' \
--openai-model 'llama-3.2-11b-text-preview' \
--dry-run
TinySemVer can be easily integrated into your GitHub Actions CI pipeline.
Assuming the differences between YAML and shell notation, some arguments are passed in a different form, like --update-version-in
.
name: Release
on:
push:
branches: [ main ]
jobs:
semver:
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
with:
persist-credentials: false # Only if main branch if protected
- name: Run TinySemVer
uses: ashvardanian/tinysemver@v2.0.1
with:
major-verbs: 'breaking,break,major'
minor-verbs: 'feature,minor,add,new'
patch-verbs: 'fix,patch,bug,improve,docs,make'
changelog-file: 'CHANGELOG.md'
version-file: 'VERSION'
update-version-in: 'pyproject.toml:version = "(.*)"' # Use colon instead of space
git-user-name: 'GitHub Actions'
git-user-email: 'actions@github.com'
github-token: ${{ secrets.GITHUB_TOKEN }}
verbose: 'true'
push: 'true'
create-release: 'true'
dry-run: 'false'
publish:
needs: semver # Depends on the previous job
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
with:
ref: main # Take the most recent updated version
Every team has a different workflow, but a common pattern is to have one release.yml
for the main
branch and another prerelease.yml
for the main-dev
branch used as a staging area.
The latter would run with dry-run: 'true'
and push: 'false'
to prevent pushing changes to the main repository.
The create-release
flag is optional and can be set to false
if you don't want to create a new release on GitHub.
If you need to update the version in multiple files, pass a multiline string with the |
operator:
update-version-in: |
pyproject.toml:version = "(.*)"
package.json:"version": "(.*)"
CITATION.cff:version: "(.*)"
For examples, consider checking StringZilla, USearch, and other libraries using TinySemVer.
If your default branch is protected with a "pull request before merging" rule:
- A repository-scoped Personal Access Token (PAT) is required to push to the branch.
- Set
persist-credentials: false
in theactions/checkout
step.
Also keep in mind:
- The default
GITHUB_TOKEN
cannot be used with protected branches. - Using a PAT instead of
GITHUB_TOKEN
poses security risks:- Workflows from any branch can access secret variables.
- This could allow non-protected branches to use elevated permissions.
- Mitigation:
- Use a fine-grained PAT with minimal necessary permissions.
- Prefer the
pull_request
workflow trigger, which limits permissions. - Be cautious: users with write access could still potentially exploit workflows to expose the PAT.
Tip
Always follow the principle of least privilege when setting up tokens and permissions.
For more information on CI configurations and pushing changes in GitHub Actions, see the semantic-release GitHub Actions guide.
In the past I was using semantic-release for my 10+ projects. At some point, a breaking change in the dependencies broke all my projects CI pipelines for a month, affecting dozens of tech companies using those libraries. I felt miserable trying to trace the issue and reluctant to go through 363K lines of low-quality JavaScript code to find the bug. Yes, it's 363K lines of code:
$ .../node_modules$ cloc .
10751 text files.
7809 unique files.
3498 files ignored.
github.com/AlDanial/cloc v 1.90 T=2.96 s (2450.6 files/s, 300331.1 lines/s)
--------------------------------------------------------------------------------
Language files blank comment code
--------------------------------------------------------------------------------
JavaScript 4902 48080 81205 363424
TypeScript 732 7008 73034 79367
... ... ... ... ...
--------------------------------------------------------------------------------
SUM: 7256 90782 164390 634071
--------------------------------------------------------------------------------
Here is the cloc
output for tinysemver
:
$ tinysemver$ cloc .
17 text files.
13 unique files.
6 files ignored.
github.com/AlDanial/cloc v 1.96 T=0.01 s (660.7 files/s, 44267.6 lines/s)
-------------------------------------------------------------------------------
Language files blank comment code
-------------------------------------------------------------------------------
Python 1 79 93 493
... ... ... ... ...
-------------------------------------------------------------------------------
SUM: 13 227 107 1124
-------------------------------------------------------------------------------
- Optional commit scopes, like
feat(scope): add new feature
. Doesn't make sense for most projects. - Pre-release versions, like
1.2.3-alpha.1
. Not needed for most projects. - GenAI.
For reference, according to SemVer 2.0, all following versions are valid:
1.1.2-prerelease+meta
,1.1.2+meta
,1.1.2+meta-valid
,1.0.0-alpha
,1.0.0-beta
,1.0.0-alpha.beta.1
,1.0.0-alpha.1
,1.0.0-alpha0.valid
,1.0.0-alpha.0valid
,1.0.0-alpha-a.b-c-somethinglong+build.1-aef.1-its-okay
,1.0.0-rc.1+build.1
,2.0.0-rc.1+build.123
,1.2.3-beta
,10.2.3-DEV-SNAPSHOT
,1.2.3-SNAPSHOT-123
,2.0.0+build.1848
,2.0.1-alpha.1227
,1.0.0-alpha+beta
,1.2.3----RC-SNAPSHOT.12.9.1--.12+788
,1.2.3----R-S.12.9.1--.12+meta
,1.2.3----RC-SNAPSHOT.12.9.1--.12
,1.0.0+0.build.1-rc.10000aaa-kk-0.1
,1.0.0-0A.is.legal
.
Assembling RegEx queries can be hard. Luckily, there aren't too many files to update in most projects. Below is an example of a pipeline for the USearch project, that has bindings to 10 programming languages. Feel free to add other sources and examples.
$ mkdir -p example
$ wget https://github.com/unum-cloud/usearch/raw/main/VERSION -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/CHANGELOG.md -P example/ # Missing
$ wget https://github.com/unum-cloud/usearch/raw/main/CITATION.cff -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/CMakeLists.txt -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/Cargo.toml -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/package.json -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/conanfile.py -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/README.md -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/wasmer.toml -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/csharp/nuget/nuget-package.props -P example/
$ wget https://github.com/unum-cloud/usearch/raw/main/include/usearch/index.hpp -P example/
# You can match the semantic version part with a generic wildcard like: .*
# But it's recommended to stick to a stricter format: \d+\.\d+\.\d+
$ tinysemver --dry-run --verbose \
--major-verbs 'breaking,break,major' \
--minor-verbs 'feature,minor,add,new' \
--patch-verbs 'fix,patch,bug,improve,docs,make' \
--version-file 'example/VERSION' \
--changelog-file 'example/CHANGELOG.md' \
--update-version-in 'example/CITATION.cff' '^version: (\d+\.\d+\.\d+)' \
--update-version-in 'example/CMakeLists.txt' '\sVERSION (\d+\.\d+\.\d+)' \
--update-version-in 'example/Cargo.toml' '^version = "(\d+\.\d+\.\d+)"' \
--update-version-in 'example/package.json' '"version": "(\d+\.\d+\.\d+)"' \
--update-version-in 'example/conanfile.py' '\sversion = "(\d+\.\d+\.\d+)"' \
--update-version-in 'example/README.md' '^version = \{(\d+\.\d+\.\d+)\}' \
--update-version-in 'example/wasmer.toml' '^version = "(\d+\.\d+\.\d+)"' \
--update-version-in 'example/nuget-package.props' '(\d+\.\d+\.\d+)\<\/Version\>' \
--update-major-version-in 'example/index.hpp' '^#define USEARCH_VERSION_MAJOR (\d+)' \
--update-minor-version-in 'example/index.hpp' '^#define USEARCH_VERSION_MINOR (\d+)' \
--update-patch-version-in 'example/index.hpp' '^#define USEARCH_VERSION_PATCH (\d+)' \
--path .
Feel free to open an issue or a pull request.
If you need to bump the version of tinysemver
itself:
tinysemver --verbose \
--version-file 'VERSION' \
--changelog-file 'CHANGELOG.md' \
--update-version-in 'pyproject.toml' 'version = "(.*)"' \
--github-repository 'ashvardanian/tinysemver' --push