Add Compact Tests (testc)

.github/workflows/unittest.yml

@@ -1,4 +1,4 @@
-name: Unit test and coverage
+name: Unit test
 
 on:
   push:
@@ -23,9 +23,10 @@ jobs:
     runs-on: ${{ matrix.os }}
     env:
       PYICU_WIN_VER: 2.14
-      INSTALL_PYICU_WIN: false
       INSTALL_TORCH: false
       INSTALL_FULL_DEPS: false
+      PYTHON_VERSION_LATEST: "3.13"
+      PYTHON_VERSION_LATEST_2: "3.12"
 
     steps:
     - name: Checkout
@@ -52,7 +53,7 @@ jobs:
         echo "ICU_VER=${ICU_VER}"
         echo "ICU_VER=${ICU_VER}" >> "${GITHUB_ENV}"
     - name: Install PyICU (Windows)
-      if: startsWith(matrix.os, 'windows-') && env.INSTALL_PYICU_WIN == 'true'
+      if: startsWith(matrix.os, 'windows-') && (matrix.python-version == '3.12' || matrix.python-version == '3.13')
       shell: powershell
       run: |
         $PYTHON_WIN_VER = "${{ matrix.python-version }}"
@@ -66,29 +67,42 @@ jobs:
       # If torch for the platform is not available in PyPI, use this command:
       # pip install "<torch_wheel_url>"
       # Get wheel URL from http://download.pytorch.org/whl/torch/
-    - name: Install dependencies
+    - name: Install dependencies from docker_requirements.txt
       if: env.INSTALL_FULL_DEPS == 'true'
       env:
         SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL: True
       run: pip install -r docker_requirements.txt
-    - name: Install PyThaiNLP
-      env:
-        PYTHONIOENCODING: utf-8
+    - name: Install PyThaiNLP + dependencies (minimum)
+      if: matrix.python-version != env.PYTHON_VERSION_LATEST && matrix.python-version != env.PYTHON_VERSION_LATEST_2
       run: pip install .
+    - name: Install PyThaiNLP + dependencies (compact)
+      if: matrix.python-version == env.PYTHON_VERSION_LATEST || matrix.python-version == env.PYTHON_VERSION_LATEST_2
+      run: pip install ".[compact]"
       # If you want to install a safe small set of optional dependencies, use:
-      # pip install .[compact]
-      # "compact" includes numpy, pyicu, and python-crfsuite.
+      # pip install ".[compact]"
       # We can gradually run more test cases by installing more optional
       # dependencies. But we should also consider to reduce the number
       # of dependencies to avoid the conflict between dependencies.
       # See: https://github.com/PyThaiNLP/pythainlp/issues/935
-    - name: Unit test and code coverage
-      run: coverage run -m unittest tests
-      # Use 'unittest tests' instead of 'unittest discover' to avoid loading
-      # tests with external imports.
+    - name: Unit test (core)
+      if: matrix.python-version != env.PYTHON_VERSION_LATEST && matrix.python-version != env.PYTHON_VERSION_LATEST_2
+      env:
+        PYTHONIOENCODING: utf-8
+      run: coverage run -m unittest tests.core
+    - name: Unit test (core + compact)
+      if: matrix.python-version == env.PYTHON_VERSION_LATEST || matrix.python-version == env.PYTHON_VERSION_LATEST_2
+      env:
+        PYTHONIOENCODING: utf-8
+      run: coverage run -m unittest tests.core tests.compact
+      # Only test "compact" set with the latest two stable Python versions.
+      # Use 'unittest <test_module>' instead of 'unittest discover' to avoid
+      # loading tests with dependencies more than expected.
       # Test cases loaded is defined in __init__.py in the tests directory.
+      # See also tests/README.md
     - name: Coverage report
+      if: matrix.python-version == env.PYTHON_VERSION_LATEST
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         COVERALLS_SERVICE_NAME: github
       run: coveralls
+      # Only submit a report from the latest Python version

pythainlp/util/strftime.py

@@ -4,6 +4,7 @@
 """
 Thai date/time formatting.
 """
+
 import warnings
 from datetime import datetime
 from string import digits
@@ -44,7 +45,7 @@ def _std_strftime(dt_obj: datetime, fmt_char: str) -> str:
         # in that case just use the fmt_char
         warnings.warn(
             (
-                f"String format directive unknown/not support: %{fmt_char}"
+                f"String format directive unknown/not support: %{fmt_char}\n"
                 f"The system raises this ValueError: {err}"
             ),
             UserWarning,
@@ -145,7 +146,8 @@ def _thai_strftime(dt_obj: datetime, fmt_char: str) -> str:
         )
     else:
         # No known localization available, use Python's default
-        str_ = _std_strftime(dt_obj, fmt_char)
+        # With a good _NEED_L10N and _EXTENSIONS, this should not happen
+        str_ = _std_strftime(dt_obj, fmt_char)  # pragma: no cover
 
     return str_
 

setup.py

@@ -101,7 +101,12 @@
     "wtp": ["transformers>=4.6.0", "wtpsplit>=1.0.1"],
     "wunsen": ["wunsen>=0.0.1"],
     # Compact dependencies, this one matches requirements.txt
-    "compact": ["PyYAML>=5.4.1", "numpy>=1.22", "pyicu>=2.3", "python-crfsuite>=0.9.7"],
+    "compact": [
+        "PyYAML>=5.4.1",
+        "numpy>=1.22",
+        "pyicu>=2.3",
+        "python-crfsuite>=0.9.7",
+    ],
     # Full dependencies
     "full": [
         "PyYAML>=5.4.1",

tests/README.md

@@ -1,19 +1,30 @@
 # Test cases
 
-Tests are categorized into two groups: fundamental and extra.
+Tests are categorized into three groups: core, compact, and extra.
 
-## Fundamental Tests (test_*.py)
+## Core Tests (test_*.py)
 
 - Focus on core functionalities.
-- Do not rely on additional dependencies beyond those listed in the
-  `requirements` section of `setup.py`.
+- Do not rely on external dependencies beyond the standard library,
+  except for `requests` which is used for corpus downloading.
+- Test with all officially supported Python versions
+  (currently 3.9, 3.10, 3.11, 3.12, and 3.13).
+
+## Compact Tests (testc_*.py)
+
+- Test a limited set of additional functionalities that rely on optional
+  dependencies specified in `requirements.txt`.
+- These dependencies are `PyYAML`, `numpy`, `pyicu`, `python-crfsuite`, and
+  `requests`.
+- Test with the latest two stable Python versions.
 
 ## Extra Tests (testx_*.py)
 
 - Explore functionalities that rely on optional dependencies specified in the
   `extras` section of `setup.py`.
-- These dependencies might include libraries like `nltk`, `pycrfsuite`, or
-  `torch`.
+- These dependencies might include libraries like `gensim`, `nltk`, or `torch`.
+- Due to dependency complexities, these functionalities are not part of the
+  automated test suite and will not be tested in the CI/CD pipeline.
 
 ## Default Test Suite
 

tests/__init__.py

@@ -11,18 +11,8 @@
 
 # Names of module to be tested
 test_packages: list[str] = [
-    "tests.test_ancient",
-    "tests.test_cli",
-    "tests.test_corpus",
-    "tests.test_khavee",
-    "tests.test_morpheme",
-    "tests.test_soundex",
-    "tests.test_spell",
-    "tests.test_tag",
-    "tests.test_tokenize",
-    "tests.test_tools",
-    "tests.test_transliterate",
-    "tests.test_util",
+    "tests.core",
+    "tests.compact",
 ]
 
 

tests/compact/__init__.py

@@ -0,0 +1,36 @@
+# -*- coding: utf-8 -*-
+# SPDX-FileCopyrightText: 2016-2024 PyThaiNLP Project
+# SPDX-License-Identifier: Apache-2.0
+"""
+Unit test. Compact version.
+
+Test functions that require "compact" dependencies (see setup.py).
+"""
+
+from unittest import TestLoader, TestSuite
+
+# Names of module to be tested
+test_packages: list[str] = [
+    "tests.compact.testc_tag",
+    "tests.compact.testc_tokenize",
+    "tests.compact.testc_util",
+]
+
+
+def load_tests(
+    loader: TestLoader, standard_tests: TestSuite, pattern: str
+) -> TestSuite:
+    """Load test protocol
+    See: https://docs.python.org/3/library/unittest.html#id1
+    """
+    suite = TestSuite()
+    for test_package in test_packages:
+        tests = loader.loadTestsFromName(test_package)
+        suite.addTests(tests)
+    return suite
+
+
+if __name__ == "__main__":
+    import unittest
+
+    unittest.main()

tests/compact/testc_tag.py

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+# SPDX-FileCopyrightText: 2016-2024 PyThaiNLP Project
+# SPDX-License-Identifier: Apache-2.0
+
+import unittest
+
+from pythainlp.tag import chunk_parse, pos_tag
+
+
+class TagTestCase(unittest.TestCase):
+    def test_chunk_parse(self):
+        tokens = ["ผม", "รัก", "คุณ"]
+
+        w_p = pos_tag(tokens, engine="perceptron", corpus="orchid")
+        self.assertIsNotNone(chunk_parse(w_p))

tests/compact/testc_tokenize.py

@@ -0,0 +1,85 @@
+# -*- coding: utf-8 -*-
+# SPDX-FileCopyrightText: 2016-2024 PyThaiNLP Project
+# SPDX-License-Identifier: Apache-2.0
+
+import unittest
+
+from pythainlp.tokenize import (
+    pyicu,
+    sent_tokenize,
+    subword_tokenize,
+    word_tokenize,
+)
+
+from ..core.test_tokenize import (
+    SENT_1,
+    SENT_1_TOKS,
+    SENT_2,
+    SENT_2_TOKS,
+    SENT_3,
+    SENT_3_TOKS,
+    SENT_4,
+    TEXT_1,
+)
+
+
+# Tests for functions that need "compact" dependencies
+class TokenizeTestCaseCompact(unittest.TestCase):
+    def test_icu(self):
+        self.assertEqual(pyicu.segment(None), [])
+        self.assertEqual(pyicu.segment(""), [])
+        self.assertEqual(
+            word_tokenize("ฉันรักภาษาไทยเพราะฉันเป็นคนไทย", engine="icu"),
+            ["ฉัน", "รัก", "ภาษา", "ไทย", "เพราะ", "ฉัน", "เป็น", "คน", "ไทย"],
+        )
+
+    def test_sent_tokenize(self):
+        # Use default engine (crfcut)
+        self.assertEqual(sent_tokenize(None), [])
+        self.assertEqual(sent_tokenize(""), [])
+        self.assertEqual(
+            sent_tokenize(SENT_1),
+            SENT_1_TOKS,
+        )
+        self.assertEqual(
+            sent_tokenize(SENT_2),
+            SENT_2_TOKS,
+        )
+        self.assertEqual(
+            sent_tokenize(SENT_3),
+            SENT_3_TOKS,
+        )
+
+        self.assertEqual(
+            sent_tokenize(SENT_1, engine="crfcut"),
+            SENT_1_TOKS,
+        )
+        self.assertEqual(
+            sent_tokenize(SENT_2, engine="crfcut"),
+            SENT_2_TOKS,
+        )
+        self.assertEqual(
+            sent_tokenize(SENT_3, engine="crfcut"),
+            SENT_3_TOKS,
+        )
+        self.assertEqual(
+            sent_tokenize(SENT_4, engine="crfcut"),
+            [["ผม", "กิน", "ข้าว", " ", "\n", "เธอ", "เล่น", "เกม"]],
+        )
+
+    def test_subword_tokenize(self):
+        self.assertEqual(subword_tokenize(None, engine="han_solo"), [])
+        self.assertEqual(
+            subword_tokenize("แมวกินปลา", engine="han_solo"),
+            ["แมว", "กิน", "ปลา"],
+        )
+        self.assertIn(
+            "ดาว", subword_tokenize("สวัสดีดาวอังคาร", engine="han_solo")
+        )
+
+        self.assertNotIn(
+            "า", subword_tokenize("สวัสดีดาวอังคาร", engine="han_solo")
+        )
+
+    def test_word_tokenize_icu(self):
+        self.assertIsNotNone(word_tokenize(TEXT_1, engine="icu"))

tests/testx_util.py → tests/compact/testc_util.py

@@ -8,25 +8,14 @@
 
 import unittest
 
-from pythainlp.util import rhyme, thai_word_tone_detector
 from pythainlp.util.spell_words import spell_word
 
 
 class UtilTestCaseX(unittest.TestCase):
-    def test_rhyme(self):
-        self.assertIsInstance(rhyme("แมว"), list)
-        self.assertTrue(len(rhyme("แมว")) > 2)
-
     def test_spell_word(self):
         self.assertEqual(spell_word("เสือ"), ["สอ", "เอือ", "เสือ"])
         self.assertEqual(spell_word("เสื้อ"), ["สอ", "เอือ", "ไม้โท", "เสื้อ"])
         self.assertEqual(spell_word("คน"), ["คอ", "นอ", "คน"])
         self.assertEqual(
             spell_word("คนดี"), ["คอ", "นอ", "คน", "ดอ", "อี", "ดี", "คนดี"]
         )
-
-    def test_thai_word_tone_detector(self):
-        self.assertIsNotNone(thai_word_tone_detector("คนดี"))
-        self.assertEqual(
-            thai_word_tone_detector("ราคา"), [("รา", "m"), ("คา", "m")]
-        )

tests/core/__init__.py

@@ -0,0 +1,45 @@
+# -*- coding: utf-8 -*-
+# SPDX-FileCopyrightText: 2016-2024 PyThaiNLP Project
+# SPDX-License-Identifier: Apache-2.0
+"""
+Unit test.
+
+Each file in tests/ is for each main package.
+"""
+
+from unittest import TestLoader, TestSuite
+
+# Names of module to be tested
+test_packages: list[str] = [
+    "tests.core.test_ancient",
+    "tests.core.test_cli",
+    "tests.core.test_corpus",
+    "tests.core.test_khavee",
+    "tests.core.test_morpheme",
+    "tests.core.test_soundex",
+    "tests.core.test_spell",
+    "tests.core.test_tag",
+    "tests.core.test_tokenize",
+    "tests.core.test_tools",
+    "tests.core.test_transliterate",
+    "tests.core.test_util",
+]
+
+
+def load_tests(
+    loader: TestLoader, standard_tests: TestSuite, pattern: str
+) -> TestSuite:
+    """Load test protocol
+    See: https://docs.python.org/3/library/unittest.html#id1
+    """
+    suite = TestSuite()
+    for test_package in test_packages:
+        tests = loader.loadTestsFromName(test_package)
+        suite.addTests(tests)
+    return suite
+
+
+if __name__ == "__main__":
+    import unittest
+
+    unittest.main()