Chore: Use pytest fixtures effectively

CHANGELOG.md

@@ -20,6 +20,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Bump pypa/gh-action-pypi-publish from 1.9.0 to 1.10.2 (by [@dependabot](https://github.com/apps/dependabot) in [#22](https://github.com/stumpylog/tika-client/pull/22))
 - Update `pre-commit` to 4.0.1 ([#23](https://github.com/stumpylog/tika-client/pull/23))
+- Use pytest fixtures effectively ([#24](https://github.com/stumpylog/tika-client/pull/24))
 
 ## [0.6.0] - 2024-07-18
 

tests/conftest.py

@@ -1,24 +1,85 @@
 import logging
 import os
 from pathlib import Path
-from typing import Final
+from typing import Generator
 
 import pytest
 
 from tika_client.client import TikaClient
 
-TIKA_URL: Final[str] = os.getenv("TIKA_URL", "http://localhost:9998")
 
-SAMPLE_DIR: Final[Path] = Path(__file__).parent.resolve() / "samples"
+@pytest.fixture(scope="session")
+def tika_host() -> str:
+    return os.getenv("TIKA_URL", "http://localhost:9998")
+
+
+@pytest.fixture(scope="session")
+def samples_dir() -> Path:
+    return Path(__file__).parent.resolve() / "samples"
+
+
+@pytest.fixture(scope="session")
+def sample_libre_office_writer_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample-libre-office.odt"
+
+
+@pytest.fixture(scope="session")
+def sample_google_docs_to_libre_office_writer_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample.odt"
+
+
+@pytest.fixture(scope="session")
+def sample_google_docs_to_docx_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample.docx"
+
+
+@pytest.fixture(scope="session")
+def sample_docx_file(samples_dir: Path) -> Path:
+    return samples_dir / "microsoft-sample.docx"
+
+
+@pytest.fixture(scope="session")
+def sample_doc_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample.doc"
+
+
+@pytest.fixture(scope="session")
+def sample_html_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample.html"
+
+
+@pytest.fixture(scope="session")
+def sample_office_doc_with_images_file(samples_dir: Path) -> Path:
+    return samples_dir / "test-document-images.odt"
+
+
+@pytest.fixture(scope="session")
+def sample_jpeg_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample.jpg"
+
+
+@pytest.fixture(scope="session")
+def sample_png_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample.png"
+
+
+@pytest.fixture(scope="session")
+def sample_ods_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample-spreadsheet.ods"
+
+
+@pytest.fixture(scope="session")
+def sample_xlsx_file(samples_dir: Path) -> Path:
+    return samples_dir / "sample-spreadsheet.xlsx"
 
 
 @pytest.fixture
-def tika_client() -> TikaClient:
-    with TikaClient(tika_url=TIKA_URL, log_level=logging.INFO) as client:
+def tika_client(tika_host: str) -> Generator[TikaClient, None, None]:
+    with TikaClient(tika_url=tika_host, log_level=logging.INFO) as client:
         yield client
 
 
 @pytest.fixture
-def tika_client_compressed() -> TikaClient:
-    with TikaClient(tika_url=TIKA_URL, log_level=logging.INFO, compress=True) as client:
+def tika_client_compressed(tika_host: str) -> Generator[TikaClient, None, None]:
+    with TikaClient(tika_url=tika_host, log_level=logging.INFO, compress=True) as client:
         yield client

tests/test_datetime_formats.py

@@ -1,29 +1,36 @@
 from datetime import datetime
 from datetime import timedelta
 from datetime import timezone
+from pathlib import Path
 
 import magic
 import pytest
 from pytest_httpx import HTTPXMock
 
-from tests.conftest import SAMPLE_DIR
 from tika_client.client import TikaClient
 from tika_client.data_models import DublinCoreKey
 from tika_client.data_models import TikaKey
 
 
 class TestDateTimeFormat:
-    def test_parse_offset_date_format_utc(self, tika_client: TikaClient, httpx_mock: HTTPXMock):
+    def test_parse_offset_date_format_utc(
+        self,
+        tika_client: TikaClient,
+        sample_libre_office_writer_file: Path,
+        httpx_mock: HTTPXMock,
+    ):
         """
         Test the datetime parsing properly handles a time with a UTC timezone in the +xx:yy format
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
 
         httpx_mock.add_response(
             json={TikaKey.ContentType: "test", TikaKey.Parsers: [], DublinCoreKey.Created: "2023-05-17T16:30:44+00:00"},
         )
 
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.created == datetime(
             year=2023,
@@ -35,17 +42,24 @@ def test_parse_offset_date_format_utc(self, tika_client: TikaClient, httpx_mock:
             tzinfo=timezone.utc,
         )
 
-    def test_parse_offset_date_format_zulu(self, tika_client: TikaClient, httpx_mock: HTTPXMock):
+    def test_parse_offset_date_format_zulu(
+        self,
+        tika_client: TikaClient,
+        sample_libre_office_writer_file: Path,
+        httpx_mock: HTTPXMock,
+    ):
         """
         Test the datetime parsing properly handles a time with a UTC timezone in the Z format
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
 
         httpx_mock.add_response(
             json={TikaKey.ContentType: "test", TikaKey.Parsers: [], DublinCoreKey.Created: "2023-01-17T16:35:44Z"},
         )
 
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.created == datetime(
             year=2023,
@@ -57,34 +71,48 @@ def test_parse_offset_date_format_zulu(self, tika_client: TikaClient, httpx_mock
             tzinfo=timezone.utc,
         )
 
-    def test_parse_offset_date_format_positive(self, tika_client: TikaClient, httpx_mock: HTTPXMock):
+    def test_parse_offset_date_format_positive(
+        self,
+        tika_client: TikaClient,
+        sample_libre_office_writer_file: Path,
+        httpx_mock: HTTPXMock,
+    ):
         """
         Test the datetime parsing properly handles a time with a timezone in the +xx:yy format offset from UTC
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
 
         httpx_mock.add_response(
             json={TikaKey.ContentType: "test", TikaKey.Parsers: [], DublinCoreKey.Created: "2023-06-17T16:30:44+08:00"},
         )
 
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.created == pytest.approx(
             datetime(year=2023, month=6, day=17, hour=16, minute=30, second=44, tzinfo=timezone(timedelta(hours=8))),
             rel=timedelta(seconds=1),
         )
 
-    def test_parse_offset_date_format_negative(self, tika_client: TikaClient, httpx_mock: HTTPXMock):
+    def test_parse_offset_date_format_negative(
+        self,
+        tika_client: TikaClient,
+        sample_libre_office_writer_file: Path,
+        httpx_mock: HTTPXMock,
+    ):
         """
         Test the datetime parsing properly handles a time with a timezone in the -xx:yy format offset from UTC
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
 
         httpx_mock.add_response(
             json={TikaKey.ContentType: "test", TikaKey.Parsers: [], DublinCoreKey.Created: "2023-06-17T16:30:44-08:00"},
         )
 
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.created == pytest.approx(
             datetime(
@@ -99,32 +127,46 @@ def test_parse_offset_date_format_negative(self, tika_client: TikaClient, httpx_
             rel=timedelta(seconds=1),
         )
 
-    def test_parse_offset_date_format_python_isoformat(self, tika_client: TikaClient, httpx_mock: HTTPXMock):
+    def test_parse_offset_date_format_python_isoformat(
+        self,
+        tika_client: TikaClient,
+        sample_libre_office_writer_file: Path,
+        httpx_mock: HTTPXMock,
+    ):
         """
         Test the datetime parsing properly handles a time with a timezone in the ISO 8061 format (as done by Python)
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
 
         expected = datetime.now(tz=timezone.utc)
 
         httpx_mock.add_response(
             json={TikaKey.ContentType: "test", TikaKey.Parsers: [], DublinCoreKey.Created: expected.isoformat()},
         )
 
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.created == pytest.approx(expected, rel=timedelta(seconds=1))
 
-    def test_parse_offset_date_no_match(self, tika_client: TikaClient, httpx_mock: HTTPXMock):
+    def test_parse_offset_date_no_match(
+        self,
+        tika_client: TikaClient,
+        sample_libre_office_writer_file: Path,
+        httpx_mock: HTTPXMock,
+    ):
         """
         Test the datetime parsing properly handles a time string which doesn't match the correct formats
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
 
         httpx_mock.add_response(
             json={TikaKey.ContentType: "test", TikaKey.Parsers: [], DublinCoreKey.Created: "202-06-17T16:30:44-0"},
         )
 
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.created is None

tests/test_file_formats.py

@@ -1,20 +1,23 @@
 from datetime import datetime
+from pathlib import Path
 
 import magic
 
-from tests.conftest import SAMPLE_DIR
 from tika_client.client import TikaClient
 
 
 class TestLibreOfficeFormats:
-    def test_parse_libre_office_writer_document(self, tika_client: TikaClient):
+    def test_parse_libre_office_writer_document(self, tika_client: TikaClient, sample_libre_office_writer_file: Path):
         """
         Test handling of a ODT document produced by LibreOffice
         """
-        test_file = SAMPLE_DIR / "sample-libre-office.odt"
-        resp = tika_client.tika.as_html.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.tika.as_html.from_file(
+            sample_libre_office_writer_file,
+            magic.from_file(str(sample_libre_office_writer_file), mime=True),
+        )
 
         assert resp.type == "application/vnd.oasis.opendocument.text"
+        assert resp.content is not None
         assert (
             "<body><p>This is a document created by LibreOffice Writer 7.5.12, on July 19th, 2023</p>\n</body>"
             in resp.content

tests/test_image_files.py

@@ -1,24 +1,23 @@
+from pathlib import Path
+
 import magic
 
-from tests.conftest import SAMPLE_DIR
 from tika_client.client import TikaClient
 
 
 class TestParseImageMetadata:
-    def test_image_jpeg(self, tika_client: TikaClient):
+    def test_image_jpeg(self, tika_client: TikaClient, sample_jpeg_file: Path):
         """
         Test the handling of a JPEG file metadata retrieval
         """
-        test_file = SAMPLE_DIR / "sample.jpg"
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(sample_jpeg_file, magic.from_file(str(sample_jpeg_file), mime=True))
 
         assert resp.type == "image/jpeg"
 
-    def test_image_png(self, tika_client: TikaClient):
+    def test_image_png(self, tika_client: TikaClient, sample_png_file: Path):
         """
         Test the handling of a PNG file metadata retrieval
         """
-        test_file = SAMPLE_DIR / "sample.png"
-        resp = tika_client.metadata.from_file(test_file, magic.from_file(str(test_file), mime=True))
+        resp = tika_client.metadata.from_file(sample_png_file, magic.from_file(str(sample_png_file), mime=True))
 
         assert resp.type == "image/png"