Merge pull request #12979 from adrianeboyd/feature/cython-profile-312

Redesigned cython profiling and other minor updates for python 3.12

.github/workflows/tests.yml

@@ -58,7 +58,7 @@ jobs:
       fail-fast: true
       matrix:
         os: [ubuntu-latest, windows-latest, macos-latest]
-        python_version: ["3.11"]
+        python_version: ["3.11", "3.12.0-rc.2"]
         include:
           - os: windows-latest
             python_version: "3.7"
@@ -93,6 +93,7 @@ jobs:
       - name: Run mypy
         run: |
           python -m mypy spacy
+        if: matrix.python_version != '3.7'
 
       - name: Delete source directory and .egg-info
         run: |

requirements.txt

@@ -33,7 +33,7 @@ pytest-timeout>=1.3.0,<2.0.0
 mock>=2.0.0,<3.0.0
 flake8>=3.8.0,<6.0.0
 hypothesis>=3.27.0,<7.0.0
-mypy>=0.990,<1.1.0; platform_machine != "aarch64"
+mypy>=1.5.0,<1.6.0; platform_machine != "aarch64" and python_version >= "3.8"
 types-mock>=0.1.1
 types-setuptools>=57.0.0
 types-requests

setup.py

@@ -78,6 +78,7 @@ COMPILER_DIRECTIVES = {
     "language_level": -3,
     "embedsignature": True,
     "annotation_typing": False,
+    "profile": sys.version_info < (3, 12),
 }
 # Files to copy into the package that are otherwise not included
 COPY_FILES = {

spacy/attrs.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 from .errors import Errors
 
 IOB_STRINGS = ("", "I", "O", "B")

spacy/cli/apply.py

@@ -133,7 +133,9 @@ def apply(
     if len(text_files) > 0:
         streams.append(_stream_texts(text_files))
     datagen = cast(DocOrStrStream, chain(*streams))
-    for doc in tqdm.tqdm(nlp.pipe(datagen, batch_size=batch_size, n_process=n_process)):
+    for doc in tqdm.tqdm(
+        nlp.pipe(datagen, batch_size=batch_size, n_process=n_process), disable=None
+    ):
         docbin.add(doc)
     if output_file.suffix == "":
         output_file = output_file.with_suffix(".spacy")

spacy/cli/benchmark_speed.py

@@ -89,7 +89,7 @@ class Quartiles:
 def annotate(
     nlp: Language, docs: List[Doc], batch_size: Optional[int]
 ) -> numpy.ndarray:
-    docs = nlp.pipe(tqdm(docs, unit="doc"), batch_size=batch_size)
+    docs = nlp.pipe(tqdm(docs, unit="doc", disable=None), batch_size=batch_size)
     wps = []
     while True:
         with time_context() as elapsed:

spacy/cli/profile.py

@@ -71,7 +71,7 @@ def profile(model: str, inputs: Optional[Path] = None, n_texts: int = 10000) ->
 
 
 def parse_texts(nlp: Language, texts: Sequence[str]) -> None:
-    for doc in nlp.pipe(tqdm.tqdm(texts), batch_size=16):
+    for doc in nlp.pipe(tqdm.tqdm(texts, disable=None), batch_size=16):
         pass
 
 

spacy/kb/candidate.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True
+# cython: infer_types=True
 
 from typing import Iterable
 

spacy/kb/kb.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True
+# cython: infer_types=True
 
 from pathlib import Path
 from typing import Iterable, Tuple, Union

spacy/kb/kb_in_memory.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True
+# cython: infer_types=True
 from typing import Any, Callable, Dict, Iterable
 
 import srsly

spacy/lexeme.pyx

@@ -1,4 +1,5 @@
 # cython: embedsignature=True
+# cython: profile=False
 # Compiler crashes on memory view coercion without this. Should report bug.
 cimport numpy as np
 from libc.string cimport memset

spacy/matcher/dependencymatcher.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True
+# cython: infer_types=True
 import warnings
 from collections import defaultdict
 from itertools import product

spacy/matcher/levenshtein.pyx

@@ -1,4 +1,4 @@
-# cython: profile=True, binding=True, infer_types=True
+# cython: binding=True, infer_types=True
 from cpython.object cimport PyObject
 from libc.stdint cimport int64_t
 

spacy/matcher/matcher.pyx

@@ -1,4 +1,4 @@
-# cython: binding=True, infer_types=True, profile=True
+# cython: binding=True, infer_types=True
 from typing import Iterable, List
 
 from cymem.cymem cimport Pool

spacy/matcher/phrasematcher.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True
+# cython: infer_types=True
 from preshed.maps cimport map_clear, map_get, map_init, map_iter, map_set
 
 import warnings

spacy/ml/parser_model.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True, cdivision=True, boundscheck=False
+# cython: profile=False
 cimport numpy as np
 from libc.math cimport exp
 from libc.stdlib cimport calloc, free, realloc

spacy/morphology.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types
+# cython: profile=False
 import warnings
 
 import numpy

spacy/parts_of_speech.pyx

@@ -1,4 +1,4 @@
-
+# cython: profile=False
 IDS = {
     "": NO_TAG,
     "ADJ": ADJ,

spacy/pipeline/_edit_tree_internals/edit_trees.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True, binding=True
+# cython: profile=False
 from cython.operator cimport dereference as deref
 from libc.stdint cimport UINT32_MAX, uint32_t
 from libc.string cimport memset

spacy/pipeline/_parser_internals/_beam_utils.pyx

@@ -1,5 +1,4 @@
 # cython: infer_types=True
-# cython: profile=True
 import numpy
 
 from thinc.extra.search cimport Beam

spacy/pipeline/_parser_internals/_state.pyx

@@ -0,0 +1 @@
+# cython: profile=False

spacy/pipeline/_parser_internals/arc_eager.pyx

@@ -1,4 +1,4 @@
-# cython: profile=True, cdivision=True, infer_types=True
+# cython: cdivision=True, infer_types=True
 from cymem.cymem cimport Address, Pool
 from libc.stdint cimport int32_t
 from libcpp.vector cimport vector

spacy/pipeline/_parser_internals/ner.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 from cymem.cymem cimport Pool
 from libc.stdint cimport int32_t
 

spacy/pipeline/_parser_internals/nonproj.pyx

@@ -1,4 +1,4 @@
-# cython: profile=True, infer_types=True
+# cython: infer_types=True
 """Implements the projectivize/deprojectivize mechanism in Nivre & Nilsson 2005
 for doing pseudo-projective parsing implementation uses the HEAD decoration
 scheme.

spacy/pipeline/_parser_internals/stateclass.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True
+# cython: profile=False
 from libcpp.vector cimport vector
 
 from ...tokens.doc cimport Doc

spacy/pipeline/_parser_internals/transition_system.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True
+# cython: profile=False
 from __future__ import print_function
 
 from cymem.cymem cimport Pool

spacy/pipeline/dep_parser.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from collections import defaultdict
 from typing import Callable, Optional
 

spacy/pipeline/morphologizer.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from itertools import islice
 from typing import Callable, Dict, Optional, Union
 

spacy/pipeline/multitask.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from typing import Optional
 
 import numpy

spacy/pipeline/ner.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from collections import defaultdict
 from typing import Callable, Optional
 

spacy/pipeline/pipe.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 import warnings
 from typing import Callable, Dict, Iterable, Iterator, Tuple, Union
 

spacy/pipeline/sentencizer.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from typing import Callable, List, Optional
 
 import srsly

spacy/pipeline/senter.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from itertools import islice
 from typing import Callable, Optional
 

spacy/pipeline/tagger.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from itertools import islice
 from typing import Callable, Optional
 

spacy/pipeline/trainable_pipe.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from typing import Callable, Dict, Iterable, Iterator, Optional, Tuple
 
 import srsly

spacy/pipeline/transition_parser.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True, cdivision=True, boundscheck=False, binding=True
+# cython: profile=False
 from __future__ import print_function
 
 cimport numpy as np

spacy/strings.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True
+# cython: profile=False
 cimport cython
 from libc.stdint cimport uint32_t
 from libc.string cimport memcpy

spacy/symbols.pyx

@@ -1,4 +1,5 @@
 # cython: optimize.unpack_method_calls=False
+# cython: profile=False
 IDS = {
     "": NIL,
     "IS_ALPHA": IS_ALPHA,

spacy/tokenizer.pyx

@@ -1,4 +1,4 @@
-# cython: embedsignature=True, profile=True, binding=True
+# cython: embedsignature=True, binding=True
 cimport cython
 from cymem.cymem cimport Pool
 from cython.operator cimport dereference as deref

spacy/tokens/_retokenize.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, bounds_check=False, profile=True
+# cython: infer_types=True, bounds_check=False
 from cymem.cymem cimport Pool
 from libc.string cimport memset
 

spacy/tokens/doc.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, bounds_check=False, profile=True
+# cython: infer_types=True, bounds_check=False
 from typing import Set
 
 cimport cython

spacy/tokens/graph.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True, cdivision=True, boundscheck=False, binding=True
+# cython: profile=False
 from typing import Generator, List, Tuple
 
 cimport cython

spacy/tokens/morphanalysis.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 cimport numpy as np
 from libc.string cimport memset
 

spacy/tokens/span.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 cimport numpy as np
 
 import copy

spacy/tokens/span_group.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 import struct
 import weakref
 from copy import deepcopy

spacy/tokens/token.pyx

@@ -1,4 +1,5 @@
 # cython: infer_types=True
+# cython: profile=False
 # Compiler crashes on memory view coercion without this. Should report bug.
 cimport numpy as np
 

spacy/training/align.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 import re
 from itertools import chain
 from typing import List, Tuple

spacy/training/alignment_array.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 from typing import List
 
 import numpy

spacy/training/example.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 from collections.abc import Iterable as IterableInstance
 
 import numpy

spacy/training/gold_io.pyx

@@ -1,3 +1,4 @@
+# cython: profile=False
 import warnings
 
 import srsly

spacy/training/initialize.py

@@ -302,7 +302,7 @@ def read_vectors(
             shape = (truncate_vectors, shape[1])
     vectors_data = numpy.zeros(shape=shape, dtype="f")
     vectors_keys = []
-    for i, line in enumerate(tqdm.tqdm(f)):
+    for i, line in enumerate(tqdm.tqdm(f, disable=None)):
         line = line.rstrip()
         pieces = line.rsplit(" ", vectors_data.shape[1])
         word = pieces.pop(0)

spacy/typedefs.pyx

@@ -0,0 +1 @@
+# cython: profile=False

spacy/util.py

@@ -1068,6 +1068,9 @@ def make_tempdir() -> Generator[Path, None, None]:
         rmfunc(path)
 
     try:
+        if sys.version_info >= (3, 12):
+            shutil.rmtree(str(d), onexc=force_remove)
+        else:
             shutil.rmtree(str(d), onerror=force_remove)
     except PermissionError as e:
         warnings.warn(Warnings.W091.format(dir=d, msg=e))

spacy/vectors.pyx

@@ -1,4 +1,4 @@
-# cython: infer_types=True, profile=True, binding=True
+# cython: infer_types=True, binding=True
 from typing import Callable
 
 from cython.operator cimport dereference as deref

spacy/vocab.pyx

@@ -1,4 +1,3 @@
-# cython: profile=True
 import functools
 
 import numpy