PyObjC NaturalLanguage Framework

Warnings

• breaking PyObjC 12.0 dropped support for Python 3.9, and PyObjC 11.0 previously dropped support for Python 3.8. Users upgrading must ensure their environment uses Python 3.10 or newer.
  Fix: Upgrade Python to 3.10 or later before upgrading PyObjC to versions 11.0 or higher.
• gotcha This library wraps macOS-specific frameworks and will only function on macOS. It is not compatible with other operating systems.
  Fix: Ensure the application is run on a macOS environment. Consider OS-agnostic alternatives if cross-platform compatibility is required.
• breaking PyObjC 10.3 introduced breaking changes to how `__init__` is handled in Python classes with custom `__new__` implementations. While 10.3.1 partially reverted this to allow `__init__` when a *user* implements `__new__`, classes relying on PyObjC's provided `__new__` still cannot use `__init__`.
  Fix: Review classes with custom `__new__` methods and their interaction with `__init__`. Adapt to PyObjC's specific rules or consider alternative initialization patterns.
• breaking PyObjC 11.1 updated its Automatic Reference Counting (ARC) model for Objective-C initializer methods (those prefixed with 'init'). These methods now correctly model the behavior of 'stealing a reference to self and returning a new reference', which may affect memory management and require review of manual `alloc()`/`init_()` patterns.
  Fix: Test existing code thoroughly, especially where `alloc()` and `init_()` are used directly, to ensure correct reference counting behavior. Consult PyObjC's ARC documentation.
• gotcha Apple's `NLTokenizer` documentation specifies that instances should be used on one thread or dispatch queue at a time. Failing to adhere to this can lead to unpredictable behavior or crashes in multi-threaded Python applications using this wrapper.
  Fix: Ensure each `NLTokenizer` instance is confined to a single thread or dispatch queue, or create separate instances for each concurrent task.

Install

• pip install pyobjc-framework-naturallanguage Install stable version

Imports

• NaturalLanguage

  import NaturalLanguage

• NLTokenizer

  from NaturalLanguage import NLTokenizer

• NLLanguageRecognizer

  from NaturalLanguage import NLLanguageRecognizer

• NLTokenUnit

  from NaturalLanguage import NLTokenUnit

Quickstart

This quickstart demonstrates basic text tokenization and language recognition using the NaturalLanguage framework. It first tokenizes a sentence into individual words and then identifies the dominant language of the text.

import NaturalLanguage
from Foundation import NSMakeRange

text_to_analyze = "The quick brown fox jumps over the lazy dog. This is a second sentence."

# --- Tokenization Example ---
# Create an NLTokenizer instance for word units
tokenizer = NaturalLanguage.NLTokenizer.alloc().initWithUnit_(NaturalLanguage.NLTokenUnitWord)

# Set the string to be tokenized
tokenizer.setString_(text_to_analyze)

tokens = []
# Enumerate tokens using a Python callable as the Objective-C block
def token_block_handler(token_range, flags):
    start = token_range.location
    length = token_range.length
    token_text = text_to_analyze[start : start + length]
    tokens.append(token_text)
    return True  # Return True to continue enumeration

tokenizer.enumerateTokensInRange_usingBlock_(
    NSMakeRange(0, len(text_to_analyze)),
    token_block_handler
)

print(f"Original text: '{text_to_analyze}'")
print(f"Tokens (words): {tokens}")

# --- Language Recognition Example ---
lang_recognizer = NaturalLanguage.NLLanguageRecognizer.alloc().init()
lang_recognizer.processString_(text_to_analyze)
dominant_language = lang_recognizer.dominantLanguage()
print(f"Dominant language: {dominant_language}")