Build wheels for python 3.13 and python 3.14 (#743)

benfred · web-flow · commit 0d227fe32eb5 · 2026-04-29T21:54:38.000-07:00
* build wheels for python 3.13 and python 3.14
* drop support for python 3.6 and python 3.7
* Update github actions versions
* Update pre-commit config
* Various minor fixes to work with newer versions of dependencies
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
@@ -16,10 +16,10 @@ jobs:
   lint:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v6
 
     - name: Set up Python
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v6
       with:
         python-version: 3.9
 
@@ -57,29 +57,13 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [macos-latest, ubuntu-latest, windows-latest]
-        python-version: [36, 37, 38, 39, 310, 311, 312]
-        cibw-arch: [auto, aarch64]
-        exclude:
-          - os: macos-latest
-            cibw-arch: aarch64
-          - os: windows-latest
-            cibw-arch: aarch64
+        os: [macos-latest, ubuntu-latest, windows-latest, ubuntu-24.04-arm]
 
     steps:
-      - uses: actions/checkout@v3
-
-      - name: Set up QEMU
-        if: runner.os == 'Linux'
-        uses: docker/setup-qemu-action@v3
-        with:
-          platforms: all
+      - uses: actions/checkout@v6
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.16.2
-        env:
-          CIBW_BUILD: cp${{matrix.python-version}}-*
-          CIBW_ARCHS_LINUX: ${{ matrix.cibw-arch }}
+        uses: pypa/cibuildwheel@v3.4.1
 
       - name: Build sdist
         run: |
@@ -93,42 +77,38 @@ jobs:
         if: runner.os == 'Linux'
 
       - name: Upload Binaries
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v7
         with:
-          name: wheels
-          path: wheelhouse
+          name: wheels-${{ matrix.os }}
+          path: ./wheelhouse/*.whl
 
   test-wheels:
     needs: [build-wheels]
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: [3.7, 3.8, 3.9, '3.10', '3.11', '3.12']
-        os: [macos-latest, ubuntu-latest, windows-latest]
+        python-version: [3.8, 3.9, '3.10', '3.11', '3.12', '3.13', '3.14']
+        os: [macos-latest, ubuntu-latest, windows-latest, ubuntu-24.04-arm]
 
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v6
       with:
         path: implicit_source
-    - uses: actions/download-artifact@v3
+    - uses: actions/download-artifact@v8
       with:
-        name: wheels
+        name: wheels-${{ matrix.os }}
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v6
       with:
         python-version: ${{ matrix.python-version }}
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
         pip install pytest
         pip install -r implicit_source/requirements.txt
-    - name: Install h5py
-      run: pip install h5py
-
     - name: Install ANN Libraries
       run: pip install annoy nmslib
-      if: ${{ matrix.python-version != '3.12' &&  matrix.python-version != '3.11' && runner.os == 'Linux' }}
-
+      if: ${{ matrix.python-version != '3.14' && runner.os == 'Linux' }}
     - name: Install wheel
       run: |
         pip install --force-reinstall --no-deps --no-index --find-links . implicit
@@ -142,15 +122,16 @@ jobs:
     if: "startsWith(github.ref, 'refs/tags/')"
     needs: [test-wheels]
     steps:
-      - uses: actions/download-artifact@v3
+      - uses: actions/download-artifact@v8
         with:
-          name: wheels
+          pattern: wheels-*
+          merge-multiple: true
       - name: Create GitHub Release
         uses: fnkr/github-action-ghr@v1.3
         env:
           GHR_PATH: .
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      - uses: actions/setup-python@v2
+      - uses: actions/setup-python@v6
         with:
           python-version: 3.9
       - name: Push to PyPi
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -1,19 +1,19 @@
 repos:
       - repo: https://github.com/timothycrosley/isort
-        rev: 5.12.0
+        rev: 8.0.1
         hooks:
         - id: isort
           additional_dependencies: [toml]
       - repo: https://github.com/python/black
-        rev: 23.3.0
+        rev: 26.1.0
         hooks:
         - id: black
       - repo: https://github.com/pycqa/flake8
-        rev: 6.0.0
+        rev: 7.3.0
         hooks:
         - id: flake8
       - repo: https://github.com/pycqa/pylint
-        rev: v2.17.4
+        rev: v4.0.5
         hooks:
         - id: pylint
       - repo: https://github.com/codespell-project/codespell
diff --git a/.pylintrc b/.pylintrc
@@ -45,3 +45,6 @@ disable=fixme,
 min-similarity-lines=64
 ignore-docstrings=yes
 ignore-imports=yes
+
+[DESIGN]
+max-positional-arguments=12
diff --git a/benchmarks/benchmark_als.py b/benchmarks/benchmark_als.py
@@ -1,4 +1,5 @@
-""" test script to verify the CG method works, and time it versus cholesky """
+"""test script to verify the CG method works, and time it versus cholesky"""
+
 import argparse
 import json
 import logging
diff --git a/examples/lastfm.py b/examples/lastfm.py
@@ -1,13 +1,13 @@
-""" An example of using this library to calculate related artists
+"""An example of using this library to calculate related artists
 from the last.fm dataset. More details can be found
 at http://www.benfrederickson.com/matrix-factorization/
 
 This code will automatically download a HDF5 version of the dataset from
 GitHub when it is first run. The original dataset can also be found at
 http://ocelma.net/MusicRecommendationDataset/lastfm-360K.html
 """
+
 import argparse
-import codecs
 import logging
 import time
 
@@ -102,7 +102,7 @@ def calculate_similar_artists(output_filename, model_name="als"):
     # write out as a TSV of artistid, otherartistid, score
     logging.debug("writing similar items")
     with tqdm.tqdm(total=len(to_generate)) as progress:
-        with codecs.open(output_filename, "w", "utf8") as o:
+        with open(output_filename, "w", encoding="utf8") as o:
             batch_size = 1000
             for startidx in range(0, len(to_generate), batch_size):
                 batch = to_generate[startidx : startidx + batch_size]
@@ -146,7 +146,7 @@ def calculate_recommendations(output_filename, model_name="als"):
     # generate recommendations for each user and write out to a file
     start = time.time()
     with tqdm.tqdm(total=len(users)) as progress:
-        with codecs.open(output_filename, "w", "utf8") as o:
+        with open(output_filename, "w", encoding="utf8") as o:
             batch_size = 1000
             to_generate = np.arange(len(users))
             for startidx in range(0, len(to_generate), batch_size):
diff --git a/examples/movielens.py b/examples/movielens.py
@@ -13,7 +13,6 @@
 from __future__ import print_function
 
 import argparse
-import codecs
 import logging
 import time
 
@@ -87,7 +86,7 @@ def calculate_similar_movies(output_filename, model_name="als", min_rating=4.0,
 
     log.debug("calculating similar movies")
     with tqdm.tqdm(total=len(to_generate)) as progress:
-        with codecs.open(output_filename, "w", "utf8") as o:
+        with open(output_filename, "w", encoding="utf8") as o:
             batch_size = 1000
             for startidx in range(0, len(to_generate), batch_size):
                 batch = to_generate[startidx : startidx + batch_size]
diff --git a/examples/tutorial_lastfm.ipynb b/examples/tutorial_lastfm.ipynb
@@ -293,7 +293,7 @@
     "# Use pandas to display the output in a table, pandas isn't a dependency of implicit otherwise\n",
     "import numpy as np\n",
     "import pandas as pd\n",
-    "pd.DataFrame({\"artist\": artists[ids], \"score\": scores, \"already_liked\": np.in1d(ids, user_plays[userid].indices)})"
+    "pd.DataFrame({\"artist\": artists[ids], \"score\": scores, \"already_liked\": np.isin(ids, user_plays[userid].indices)})"
    ]
   },
   {
diff --git a/implicit/_nearest_neighbours.pyx b/implicit/_nearest_neighbours.pyx
@@ -11,6 +11,7 @@ from cython cimport floating, integral
 from cython.operator import dereference
 from cython.parallel import parallel, prange
 
+from libc.stdint cimport int32_t
 from libcpp cimport bool
 from libcpp.algorithm cimport sort_heap
 from libcpp.utility cimport pair
@@ -131,8 +132,8 @@ def all_pairs_knn(users, unsigned int K=100, int num_threads=0, show_progress=Tr
 
     # holds triples of output
     cdef double[:] values = np.zeros(item_count * K)
-    cdef long[:] rows = np.zeros(item_count * K, dtype=int)
-    cdef long[:] cols = np.zeros(item_count * K, dtype=int)
+    cdef int32_t[:] rows = np.zeros(item_count * K, dtype="int32")
+    cdef int32_t[:] cols = np.zeros(item_count * K, dtype="int32")
 
     progress = tqdm(total=item_count, disable=not show_progress)
     with nogil, parallel(num_threads=num_threads):
diff --git a/implicit/ann/annoy.py b/implicit/ann/annoy.py
@@ -12,7 +12,6 @@
 
 
 class AnnoyModel(RecommenderBase):
-
     """Speeds up inference calls to MatrixFactorization models by using an
     `Annoy <https://github.com/spotify/annoy>`_ index to calculate similar items and
     recommend items.
diff --git a/implicit/ann/nmslib.py b/implicit/ann/nmslib.py
@@ -12,7 +12,6 @@
 
 
 class NMSLibModel(RecommenderBase):
-
     """Speeds up inference calls to MatrixFactorization models by using
     `NMSLib <https://github.com/nmslib/nmslib>`_ to create approximate nearest neighbours
     indices of the latent factors.
diff --git a/implicit/approximate_als.py b/implicit/approximate_als.py
@@ -1,8 +1,9 @@
-""" Models that use various Approximate Nearest Neighbours libraries in order to quickly
+"""Models that use various Approximate Nearest Neighbours libraries in order to quickly
 generate recommendations and lists of similar items.
 
 See http://www.benfrederickson.com/approximate-nearest-neighbours-for-recommender-systems/
 """
+
 import implicit.gpu
 
 
diff --git a/implicit/cpu/als.py b/implicit/cpu/als.py
@@ -1,4 +1,5 @@
-""" Implicit Alternating Least Squares """
+"""Implicit Alternating Least Squares"""
+
 import functools
 import heapq
 import logging
diff --git a/implicit/cpu/matrix_factorization_base.py b/implicit/cpu/matrix_factorization_base.py
@@ -1,4 +1,5 @@
-""" Base class for recommendation algorithms in this package """
+"""Base class for recommendation algorithms in this package"""
+
 import warnings
 
 import numpy as np
diff --git a/implicit/evaluation.pyx b/implicit/evaluation.pyx
@@ -431,7 +431,7 @@ def ranking_metrics_at_k(model, train_user_items, test_user_items, int K=10,
 
     while start_idx < len(to_generate):
         batch = to_generate[start_idx: start_idx + batch_size]
-        ids, _ = model.recommend(batch, train_user_items[batch], N=K)
+        ids, _ = model.recommend(batch, train_user_items[np.asarray(batch)], N=K)
         start_idx += batch_size
 
         with nogil:
diff --git a/implicit/nearest_neighbours.py b/implicit/nearest_neighbours.py
@@ -89,15 +89,15 @@ def recommend(
         )
 
         if filter_items is not None:
-            mask = np.in1d(ids, filter_items, invert=True)
+            mask = np.isin(ids, filter_items, invert=True)
             ids, scores = ids[mask][:N], scores[mask][:N]
 
         elif items is not None:
-            mask = np.in1d(ids, items)
+            mask = np.isin(ids, items)
             ids, scores = ids[mask], scores[mask]
 
             # returned items should be equal to input selected items
-            missing = items[np.in1d(items, ids, invert=True)]
+            missing = items[np.isin(items, ids, invert=True)]
             if missing.size:
                 ids = np.append(ids, missing)
                 scores = np.append(scores, np.full(missing.size, -np.finfo(scores.dtype).max))
@@ -128,15 +128,15 @@ def similar_items(
         scores = self.similarity[itemid].data
 
         if filter_items is not None:
-            mask = np.in1d(ids, filter_items, invert=True)
+            mask = np.isin(ids, filter_items, invert=True)
             ids, scores = ids[mask], scores[mask]
 
         elif items is not None:
-            mask = np.in1d(ids, items)
+            mask = np.isin(ids, items)
             ids, scores = ids[mask], scores[mask]
 
             # returned items should be equal to input selected items
-            missing = items[np.in1d(items, ids, invert=True)]
+            missing = items[np.isin(items, ids, invert=True)]
             if missing.size:
                 ids = np.append(ids, missing)
                 scores = np.append(scores, np.full(missing.size, -np.finfo(scores.dtype).max))
diff --git a/implicit/recommender_base.py b/implicit/recommender_base.py
@@ -1,4 +1,5 @@
-""" Base class for recommendation algorithms in this package """
+"""Base class for recommendation algorithms in this package"""
+
 import warnings
 from abc import ABCMeta, abstractmethod
 
diff --git a/implicit/utils.py b/implicit/utils.py
@@ -138,14 +138,14 @@ def _batch_call(func, ids, *args, N=10, **kwargs):
 
 def _filter_items_from_results(queryid, ids, scores, filter_items, N):
     if np.isscalar(queryid):
-        mask = np.in1d(ids, filter_items, invert=True)
+        mask = np.isin(ids, filter_items, invert=True)
         ids, scores = ids[mask][:N], scores[mask][:N]
     else:
         rows = len(queryid)
         filtered_scores = np.zeros((rows, N), dtype=scores.dtype)
         filtered_ids = np.zeros((rows, N), dtype=ids.dtype)
         for row in range(rows):
-            mask = np.in1d(ids[row], filter_items, invert=True)
+            mask = np.isin(ids[row], filter_items, invert=True)
             filtered_ids[row] = ids[row][mask][:N]
             filtered_scores[row] = scores[row][mask][:N]
         ids, scores = filtered_ids, filtered_scores
diff --git a/pyproject.toml b/pyproject.toml
@@ -18,7 +18,7 @@ build-backend = "setuptools.build_meta"
 # skip testing in the cibuildwheel phase, will install the wheels later
 # and verify
 test-command = ""
-skip = ["pp*", "*musl*",  "*-manylinux_i686", "*win32"]
+skip = ["*t-*", "*musl*",  "*-manylinux_i686", "*win32"]
 
 [[tool.cibuildwheel.overrides]]
 select = "*-manylinux_x86_64*"
diff --git a/tests/recommender_base_test.py b/tests/recommender_base_test.py
@@ -1,4 +1,5 @@
-""" Common test functions for all recommendation models """
+"""Common test functions for all recommendation models"""
+
 import os
 import pickle
 import random

Original file line number	Diff line number	Diff line change
`@@ -293,7 +293,7 @@`
`293`	`293`	`"# Use pandas to display the output in a table, pandas isn't a dependency of implicit otherwise\n",`
`294`	`294`	`"import numpy as np\n",`
`295`	`295`	`"import pandas as pd\n",`
`296`		`- "pd.DataFrame({\"artist\": artists[ids], \"score\": scores, \"already_liked\": np.in1d(ids, user_plays[userid].indices)})"`
	`296`	`+ "pd.DataFrame({\"artist\": artists[ids], \"score\": scores, \"already_liked\": np.isin(ids, user_plays[userid].indices)})"`
`297`	`297`	`]`
`298`	`298`	`},`
`299`	`299`	`{`