Merge branch 'scribe-org:main' into Hindi

src/scribe_data/cli/get.py

@@ -92,7 +92,7 @@ def get_data(
 
     if all:
         print("Updating all languages and data types ...")
-        query_data(None, None, overwrite)
+        query_data(None, None, None, overwrite)
         subprocess_result = True
 
     # MARK: Emojis

src/scribe_data/wikidata/query_data.py

@@ -22,6 +22,8 @@
 
 import json
 import os
+import subprocess
+import sys
 from pathlib import Path
 from urllib.error import HTTPError
 
@@ -33,6 +35,52 @@
 from scribe_data.wikidata.wikidata_utils import sparql
 
 
+def execute_formatting_script(formatting_file_path, output_dir):
+    """
+    Executes a formatting script given a filepath and output directory for the process.
+
+    Parameters
+    ----------
+        formatting_file_path : str
+            The formatting file to run.
+
+        output_dir : str
+            The output directory path for results.
+
+    Returns
+    -------
+        The results of the formatting script saved in the given output directory.
+    """
+    # Determine the root directory of the project.
+    project_root = Path(__file__).parent.parent
+
+    if sys.platform.startswith("win"):
+        python_executable = sys.executable
+        pythonpath = str(project_root)
+
+        # Create environment with updated PYTHONPATH.
+        env = os.environ.copy()
+        if "PYTHONPATH" in env:
+            env["PYTHONPATH"] = f"{pythonpath};{env['PYTHONPATH']}"
+
+        else:
+            env["PYTHONPATH"] = pythonpath
+
+        # Use subprocess.run instead of os.system.
+        subprocess.run(
+            [python_executable, str(formatting_file_path), "--file-path", output_dir],
+            env=env,
+            check=True,
+        )
+
+    else:
+        # Unix-like systems (Linux, macOS).
+        subprocess.run(
+            ["python3", str(formatting_file_path), "--file-path", output_dir],
+            check=True,
+        )
+
+
 def query_data(
     languages: str = None,
     data_type: str = None,
@@ -64,7 +112,7 @@ def query_data(
     PATH_TO_LANGUAGE_EXTRACTION_FILES = (
         SCRIBE_DATA_SRC_PATH / "language_data_extraction"
     )
-
+    languages = [lang.capitalize() for lang in languages]
     current_languages = list(language_metadata["languages"])
     current_data_type = ["nouns", "verbs", "prepositions"]
 
@@ -262,7 +310,9 @@ def query_data(
                 / target_type
                 / f"format_{target_type}.py"
             )
-            os.system(f"python3 {formatting_file_path} --file-path {output_dir}")
+            execute_formatting_script(
+                formatting_file_path=formatting_file_path, output_dir=output_dir
+            )
 
 
 if __name__ == "__main__":

tests/cli/test_get.py

@@ -26,27 +26,102 @@
 from scribe_data.cli.get import get_data
 
 
-class TestCLIGetCommand(unittest.TestCase):
-    @unittest.skip("Mocking doesn't work as expected.")
-    def test_get_command(self):
-        with patch("scribe_data.cli.get.get_data") as mock_get_data:
-            # Call the function you're testing
-            get_data(
-                language="English",
-                data_type="nouns",
-                output_dir="tests_output",
-                output_type="json",
-            )
-
-            get_data(all=True)
-
-            # Validate the calls.
-            assert mock_get_data.call_count == 2
-
-            args, kwargs = mock_get_data.mock_calls[0]
-            self.assertEqual(args, ("English", "nouns", "tests_output"))
-            self.assertFalse(kwargs.get("all"))
-
-            args, kwargs = mock_get_data.mock_calls[-1]  # Get the last call
-            self.assertIsNone(args)
-            self.assertTrue(kwargs["all"])
+class TestGetData(unittest.TestCase):
+    # MARK: Subprocess Patching
+
+    @patch("subprocess.run")
+    def test_get_emoji_keywords(self, mock_subprocess_run):
+        get_data(language="English", data_type="emoji-keywords")
+        self.assertTrue(mock_subprocess_run.called)
+
+    # MARK: Invalid Arguments
+
+    def test_invalid_arguments(self):
+        with self.assertRaises(ValueError):
+            get_data()
+
+    # MARK: All Data
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_all_data(self, mock_query_data):
+        get_data(all=True)
+        mock_query_data.assert_called_once_with(None, None, None, False)
+
+    # MARK: Language and Data Type
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_specific_language_and_data_type(self, mock_query_data):
+        get_data(language="german", data_type="nouns", output_dir="./test_output")
+        mock_query_data.assert_called_once_with(
+            languages=["german"],
+            data_type=["nouns"],
+            output_dir="./test_output",
+            overwrite=False,
+        )
+
+    # MARK: Capitalized Language
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_data_with_capitalized_language(self, mock_query_data):
+        get_data(language="German", data_type="nouns")
+        mock_query_data.assert_called_once_with(
+            languages=["German"],
+            data_type=["nouns"],
+            output_dir="scribe_data_json_export",
+            overwrite=False,
+        )
+
+    # MARK: Lowercase Language
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_data_with_lowercase_language(self, mock_query_data):
+        get_data(language="german", data_type="nouns")
+        mock_query_data.assert_called_once_with(
+            languages=["german"],
+            data_type=["nouns"],
+            output_dir="scribe_data_json_export",
+            overwrite=False,
+        )
+
+    # MARK: Output Directory
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_data_with_different_output_directory(self, mock_query_data):
+        get_data(
+            language="german", data_type="nouns", output_dir="./custom_output_test"
+        )
+        mock_query_data.assert_called_once_with(
+            languages=["german"],
+            data_type=["nouns"],
+            output_dir="./custom_output_test",
+            overwrite=False,
+        )
+
+    # MARK: Overwrite is True
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_data_with_overwrite_true(self, mock_query_data):
+        get_data(language="English", data_type="verbs", overwrite=True)
+        mock_query_data.assert_called_once_with(
+            languages=["English"],
+            data_type=["verbs"],
+            output_dir="scribe_data_json_export",
+            overwrite=True,
+        )
+
+    # MARK: Overwrite is False
+
+    @patch("scribe_data.cli.get.query_data")
+    def test_get_data_with_overwrite_false(self, mock_query_data):
+        get_data(
+            language="English",
+            data_type="verbs",
+            overwrite=False,
+            output_dir="./custom_output_test",
+        )
+        mock_query_data.assert_called_once_with(
+            languages=["English"],
+            data_type=["verbs"],
+            output_dir="./custom_output_test",
+            overwrite=False,
+        )