Prepare ThinkDSP for PyPI packaging by hoppa1231 · Pull Request #134 · AllenDowney/ThinkDSP

hoppa1231 · 2026-02-11T17:32:14Z

No description provided.

- Implement tests for spectrum generation with both even and odd lengths. - Add tests for ComplexSinusoid, Chirp, ExpoChirp, and wave addition. - Include tests for DCT and impulse signal generation. - Ensure numerical accuracy with assertions for expected results.

…tion

hoppa1231 · 2026-02-11T19:36:59Z

Summary

Prepare ThinkDSP for PyPI packaging with a src/ layout.
Update metadata, dependencies, and classifiers to be PyPI‑valid.
Add packaging files and a GitHub Actions publish workflow.

Motivation

I wanted to make ThinkDSP installable as a standard Python package from PyPI.
While doing this, I also fixed invalid Trove classifiers and aligned metadata
with current packaging tooling. This allows automated builds and publishing.

Changes

Added src/thinkdsp/__init__.py (copied from code/thinkdsp.py) for a
conventional src/ package layout.
Added tests/test_thinkdsp.py (copied from code/thinkdsp_test.py).
Updated pyproject.toml:
- project metadata (readme, license, packages, include, exclude)
- valid PyPI classifiers
- optional extras for notebooks
- updated dependencies to allow NumPy 2.x
- switched build backend to poetry-core
Added LICENSE (MIT).
Updated .gitignore to ignore common Python/venv/build/test artifacts.
Updated README.md with an attribution block.
Added .github/workflows/publish.yml for PyPI publishing via OIDC.

Testing

Not run (packaging/metadata changes only).

Notes / Potential Concerns

Package name on PyPI had to be unique (I used thinkdsp-kim for publishing),
but the import name remains thinkdsp.
Dependencies now require Python >=3.9 to support NumPy 2.x.
If you prefer to avoid a src/ layout or PyPI publishing workflow, I can
split this into smaller PRs or drop those parts.

AllenDowney · 2026-02-13T14:28:24Z

This looks really good, thank you!

I have held off on making thinkdsp an installable package because (1) it's not necessary for readers, and (2) many readers have problems installing packages. So I don't want to encourage readers to go down this road.

But it is nice to give people the option, so I am inclined to merge this PR and put the package on PyPI. In that case we can use the name thinkdsp without the suffix, and drop the language about the distribution being unofficial (also still crediting you as maintainer).

What do you think?

Copilot

Pull request overview

This pull request prepares the ThinkDSP codebase for PyPI packaging by restructuring the project into a proper Python package with a src/ layout, adding packaging configuration, tests, CI/CD workflows, and license files.

Changes:

Reorganized code into a src/thinkdsp/ package structure with proper Python packaging setup
Added pyproject.toml with Poetry configuration for PyPI publishing as "thinkdsp-kim"
Created GitHub Actions workflow for automated PyPI publishing and added comprehensive .gitignore
Updated code to use modern NumPy methods (tobytes/frombuffer) and improved IPython Audio import handling
Added MIT License file, attribution section in README, and pytest-based test suite

Reviewed changes

Copilot reviewed 7 out of 8 changed files in this pull request and generated 11 comments.

Show a summary per file

File	Description
pyproject.toml	Configures Poetry packaging with metadata, dependencies, and build system for PyPI distribution
src/thinkdsp/init.py	Main package module with DSP utilities, updated with modern NumPy methods
tests/test_thinkdsp.py	Unit tests for core DSP functionality (Wave, Spectrum, Signal classes)
code/thinkdsp.py	Updated original code file with deprecation fixes (tobytes, frombuffer, Audio import)
README.md	Added attribution section explaining this is a packaging fork maintained by hoppa1231
LICENSE	Added MIT License file with copyright information for Allen B. Downey
.gitignore	Expanded to include Python, Poetry, and packaging-related ignore patterns
.github/workflows/publish.yml	GitHub Actions workflow for building and publishing to PyPI on version tags

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-13T14:36:13Z

pyproject.toml

-pandas = "^1.1.4"
-scipy = "^1.5.4"
+python = ">=3.9,<4.0"
+numpy = ">=1.22.4,<3.0"


The numpy version constraint ">=1.22.4,<3.0" is reasonable, but NumPy 1.22.4 was released in 2022 and requires Python 3.8+. Since the minimum Python version is set to 3.9, consider aligning the numpy minimum version with what's typically available for Python 3.9 environments. NumPy 1.22 dropped support for deprecated functions that might be used in older code. Verify that all numpy usage in the codebase is compatible with numpy 1.22+, particularly the np.fromstring usage (deprecated in favor of np.frombuffer).

Suggested change

numpy = ">=1.22.4,<3.0"

numpy = ">=1.20,<3.0"

Copilot · 2026-02-13T14:36:14Z

pyproject.toml

+scipy = ">=1.13.0,<2.0"
+matplotlib = ">=3.8.0,<4.0"
+jupyter = { version = "^1.0.0", optional = true }
+pandas = { version = "^1.1.4", optional = true }


The pandas version constraint uses "^1.1.4" which means ">=1.1.4,<2.0.0". However, pandas 1.1.4 was released in 2020 and is quite old. This package may not be compatible with pandas 2.x which was released in April 2023. Consider testing compatibility with pandas 2.x and updating the constraint to ">=1.1.4,<3.0" if compatible, or explicitly documenting the pandas 2.x incompatibility.

Suggested change

pandas = { version = "^1.1.4", optional = true }

pandas = { version = ">=1.1.4,<3.0", optional = true }

Copilot · 2026-02-13T14:36:14Z

tests/test_thinkdsp.py

+"""This file contains code for use with "Think Stats",
+by Allen B. Downey, available from greenteapress.com
+
+Copyright 2014 Allen B. Downey
+License: GNU GPLv3 http://www.gnu.org/licenses/gpl.html
+"""


The test file header incorrectly references "Think Stats" instead of "Think DSP" and mentions GNU GPLv3 license while the LICENSE file and pyproject.toml specify MIT License. This is inconsistent with the actual source code which uses MIT License and references "Think DSP".

Copilot · 2026-02-13T14:36:14Z

pyproject.toml

-scipy = "^1.5.4"
+python = ">=3.9,<4.0"
+numpy = ">=1.22.4,<3.0"
+scipy = ">=1.13.0,<2.0"


The scipy version constraint ">=1.13.0,<2.0" is very restrictive and may cause compatibility issues. SciPy 1.13.0 was released in May 2024, which means users with older Python environments (even Python 3.9-3.11 that are supported) may not be able to install this package if they have older scipy versions. Consider using a more permissive constraint like ">=1.5.4,<2.0" to match the minimum Python 3.9 requirement, or document why scipy 1.13.0 is specifically required.

Suggested change

scipy = ">=1.13.0,<2.0"

scipy = ">=1.5.4,<2.0"

Copilot · 2026-02-13T14:36:14Z

src/thinkdsp/__init__.py

+        raise ValueError("sampwidth %d unknown" % sampwidth)
+
+    if sampwidth == 3:
+        xs = np.fromstring(z_str, dtype=np.int8).astype(np.int32)


The code uses np.fromstring which was deprecated in NumPy 1.14 and removed in NumPy 1.24. Since the pyproject.toml specifies numpy>=1.22.4, this will break with NumPy 1.24+. The code should be updated to use np.frombuffer instead, similar to line 113.

Suggested change

xs = np.fromstring(z_str, dtype=np.int8).astype(np.int32)

xs = np.frombuffer(z_str, dtype=np.int8).astype(np.int32)

Copilot · 2026-02-13T14:36:15Z

README.md

+## Attribution
+
+This repository packages the `thinkdsp` code from *Think DSP* by Allen B. Downey.
+Original code and copyright remain with the author under the MIT License.
+This packaging fork is maintained by `hoppa1231` and is not affiliated with the
+original author or publisher.
+


The README does not include installation instructions for the PyPI package (thinkdsp-kim). Consider adding a section explaining how to install the package using pip, such as "pip install thinkdsp-kim" or "pip install thinkdsp-kim[notebooks]" for the optional dependencies. This would help users understand how to use the packaged version.

Copilot · 2026-02-13T14:36:15Z

tests/test_thinkdsp.py

+from __future__ import print_function, division
+


The test file includes "from future import print_function, division" which is only needed for Python 2 compatibility. Since pyproject.toml specifies python = ">=3.9,<4.0", this import is unnecessary and should be removed as it's redundant in Python 3.9+.

Suggested change

from __future__ import print_function, division

Copilot · 2026-02-13T14:36:15Z

src/thinkdsp/__init__.py

+        return ys
+
+
+class PinkNoise(Noise):


This class does not call Noise.init during initialization. (PinkNoise.init may be missing a call to a base class init)

Copilot · 2026-02-13T14:36:16Z

src/thinkdsp/__init__.py

+    print(cos_cov, sin_cov, mag((cos_cov, sin_cov)))
+    return
+
+    wfile = WavFileWriter()


This statement is unreachable.

Copilot · 2026-02-13T14:36:16Z

src/thinkdsp/__init__.py

+
+    __radd__ = __add__
+
+    def __or__(self, other):


This method raises ValueError - should raise an ArithmeticError or return NotImplemented instead.

hoppa1231 · 2026-02-13T20:24:36Z

Thanks, AllenDowney — I agree with your reasoning.

I’m happy to proceed with an official PyPI release. Publishing as thinkdsp (no suffix) makes sense, and we can keep the book workflow unchanged while offering pip install as an optional convenience for those who want it.

If you’re good with that, I’ll update this PR to:

rename the distribution from thinkdsp-kim → thinkdsp (and adjust the publish workflow accordingly),
update the README attribution to remove the “unofficial/not affiliated” wording while keeping clear credit + maintainer info,
keep install instructions minimal and explicitly optional (so we don’t push readers into packaging issues).

I’ll also address the Copilot review items in the same update:

clean up pyproject.toml (remove duplicate constraints; loosen pins for numpy/scipy/matplotlib to be less restrictive; keep notebook deps optional, and broaden pandas if compatible),
fix the test header to “Think DSP” + MIT and remove the Python 2 __future__ import,
add the missing base __init__ call for PinkNoise (or remove the override if not needed),
remove the unreachable debug snippet,
and adjust __or__ to return NotImplemented / raise TypeError instead of ValueError.

Let me know if you’d prefer dropping the PyPI publish workflow from this PR and handling publishing separately.

AllenDowney · 2026-02-18T14:56:47Z

@hoppa1231 I'm working on getting the thinkdsp name back pypi/support#9359

If that gets resolved, we can pick it up from there. So hold off for now, thanks!

hoppa1231 added 5 commits February 11, 2026 20:28

Fix deprecated numpy methods in WavFileWriter and read_wave functions

de5a0d1

Update package name to thinkdsp-kim in pyproject.toml and publish.yml

775e700

Update classifiers in pyproject.toml for accuracy in topic categoriza…

0aef5ad

…tion

Bump version to 0.1.1 and update dependency versions in pyproject.toml

9ce669c

hoppa1231 changed the title ~~Fix deprecated numpy methods in WavFileWriter and read_wave functions~~ Prepare ThinkDSP for PyPI packaging Feb 11, 2026

AllenDowney requested a review from Copilot February 13, 2026 14:29

AllenDowney self-assigned this Feb 13, 2026

Copilot started reviewing on behalf of AllenDowney February 13, 2026 14:29 View session

Copilot AI reviewed Feb 13, 2026

View reviewed changes

AllenDowney added a commit that referenced this pull request Feb 13, 2026

Merge PR #134: Prepare ThinkDSP for PyPI packaging

ae327bb

Update package name to thinkdsp and PyPI URL in configuration files

5c52f19

	pandas = { version = "^1.1.4", optional = true }
	pandas = { version = ">=1.1.4,<3.0", optional = true }

	xs = np.fromstring(z_str, dtype=np.int8).astype(np.int32)
	xs = np.frombuffer(z_str, dtype=np.int8).astype(np.int32)

Conversation

hoppa1231 commented Feb 11, 2026

Uh oh!

hoppa1231 commented Feb 11, 2026

Summary

Motivation

Changes

Testing

Notes / Potential Concerns

Uh oh!

AllenDowney commented Feb 13, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 13, 2026

Choose a reason for hiding this comment

Uh oh!

hoppa1231 commented Feb 13, 2026

Uh oh!

AllenDowney commented Feb 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Comments