feat(plaintext): add typer cli

alto2txt2fixture/cli.py

@@ -1,11 +1,39 @@
 import os
+from pathlib import Path
 
+import typer
 from rich.table import Table
+from typing_extensions import Annotated
 
+from .plaintext import (
+    DEFAULT_EXTRACTED_SUBDIR,
+    DEFAULT_PLAINTEXT_FIXTURE_OUTPUT,
+    PlainTextFixture,
+)
 from .settings import DATA_PROVIDER_INDEX, SETUP_TITLE, settings
 from .types import dotdict
 from .utils import check_newspaper_collection_configuration, console, gen_fixture_tables
 
+cli = typer.Typer(pretty_exceptions_show_locals=False)
+
+
+@cli.command()
+def plaintext(
+    path: Annotated[Path, typer.Argument()],
+    save_path: Annotated[Path, typer.Option()] = Path(DEFAULT_PLAINTEXT_FIXTURE_OUTPUT),
+    data_provider_code: Annotated[str, typer.Option()] = "",
+    extract_path: Annotated[Path, typer.Argument()] = Path(DEFAULT_EXTRACTED_SUBDIR),
+) -> None:
+    """Create a PlainTextFixture and save to `save_path`."""
+    plaintext_fixture = PlainTextFixture(
+        path=path,
+        data_provider_code=data_provider_code,
+        extract_subdir=extract_path,
+        export_directory=save_path,
+    )
+    plaintext_fixture.extract_compressed()
+    plaintext_fixture.export_to_json_fixtures()
+
 
 def show_setup(clear: bool = True, title: str = SETUP_TITLE, **kwargs) -> None:
     """Generate a `rich.table.Table` for printing configuration to console."""

alto2txt2fixture/plaintext.py

@@ -28,10 +28,10 @@
 
 FULLTEXT_DJANGO_MODEL: Final[str] = "fulltext.fulltext"
 
-HOME_DIR: PathLike = Path.home()
-DOWNLOAD_DIR: PathLike = HOME_DIR / "metadata-db/"
-ARCHIVE_SUBDIR: PathLike = Path("archives")
-EXTRACTED_SUBDIR: PathLike = Path("extracted")
+# HOME_DIR: PathLike = Path.home()
+# DOWNLOAD_DIR: PathLike = HOME_DIR / "metadata-db/"
+# ARCHIVE_SUBDIR: PathLike = Path("archives")
+DEFAULT_EXTRACTED_SUBDIR: Final[PathLike] = Path("extracted")
 FULLTEXT_METHOD: str = "download"
 FULLTEXT_CONTAINER_SUFFIX: str = "-alto2txt"
 FULLTEXT_CONTAINER_PATH: PathLike = Path("plaintext/")
@@ -183,12 +183,12 @@ class PlainTextFixture:
     # mount_path: PathLike | None = Path(settings.MOUNTPOINT)
     data_provider: DataProviderFixtureDict | None = None
     model_str: str = FULLTEXT_DJANGO_MODEL
-    archive_subdir: PathLike = ARCHIVE_SUBDIR
-    extract_subdir: PathLike = EXTRACTED_SUBDIR
+    # archive_subdir: PathLike = ARCHIVE_SUBDIR
+    extract_subdir: PathLike = DEFAULT_EXTRACTED_SUBDIR
     plaintext_extension: str = TXT_FIXTURE_FILE_EXTENSION
     plaintext_glob_regex: str = TXT_FIXTURE_FILE_GLOB_REGEX
     # decompress_subdir: PathLike = FULLTEXT_DECOMPRESSED_PATH
-    download_dir: PathLike = DOWNLOAD_DIR
+    # download_dir: PathLike = DOWNLOAD_DIR
     fulltext_container_suffix: str = FULLTEXT_CONTAINER_SUFFIX
     data_provider_code_dict: dict[str, DataProviderFixtureDict] = field(
         default_factory=lambda: NEWSPAPER_DATA_PROVIDER_CODE_DICT

pyproject.toml

@@ -14,6 +14,7 @@ pytz = "^2022.7.1"
 rich = "^12.6.0"
 types-pytz = "^2023.3.0.0"
 python-slugify = "^8.0.1"
+typer = {extras = ["all"], version = "^0.9.0"}
 
 [tool.poetry.group.dev.dependencies]
 pytest-sugar = "^0.9.7"
@@ -45,7 +46,8 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry.scripts]
 a2t2f-news = "alto2txt2fixture.__main__:run"
-a2tsf-adj = "alto2txt2fixture.create_adjacent_tables:run"
+a2t2f-adj = "alto2txt2fixture.create_adjacent_tables:run"
+a2t2f-plaintext = "alto2txt2fixture.cli:cli"
 
 [tool.pytest.ini_options]
 xfail_strict = true

tests/test_cli.py

@@ -0,0 +1,41 @@
+import json
+
+from typer.testing import CliRunner
+
+from alto2txt2fixture.cli import cli
+from alto2txt2fixture.types import FixtureDict
+
+runner = CliRunner()
+
+
+def test_plaintext_cli(tmpdir):
+    """Test running `plaintext` file export via `cli`."""
+    result = runner.invoke(
+        cli,
+        [
+            "tests/bl_lwm/",
+            "--save-path",
+            tmpdir / "test-cli-plaintext-fixture",
+            "--data-provider-code",
+            "bl_lwm",
+        ],
+    )
+    assert result.exit_code == 0
+    assert "Extract path: tests/bl_lwm/extracted" in result.stdout
+    exported_json: list[FixtureDict] = json.load(
+        tmpdir / "test-cli-plaintext-fixture" / "plaintext_fixture-1.json"
+    )
+    assert exported_json[0]["model"] == "fulltext.fulltext"
+    assert "NEW TREDEGAR & BARGOED" in exported_json[0]["fields"]["text"]
+    assert (
+        exported_json[0]["fields"]["path"]
+        == "tests/bl_lwm/extracted/0003548/1904/0630/0003548_19040630_art0002.txt"
+    )
+    assert (
+        exported_json[0]["fields"]["compressed_path"]
+        == "tests/bl_lwm/0003548-test_plaintext.zip"
+    )
+    assert (
+        exported_json[0]["fields"]["updated_at"]
+        == exported_json[0]["fields"]["updated_at"]
+    )