Merge pull request #147 from initze/feature/tests

Add testing infrastructure

README.md

@@ -89,30 +89,30 @@ Run in dev:
 
 ```sh
 $ rye run thaw-slump-segmentation hello tobi
-Hello tobi
+Hello, tobi!
 ```
 
 or run as python module:
 
 ```sh
 $ rye run python -m thaw_slump_segmentation hello tobi
-Hello tobi
+Hello, tobi!
 ```
 
 With activated env, e.g. after installation, just remove the `rye run`:
 
 ```sh
 $ source .venv/bin/activate
 $ thaw-slump-segmentation hello tobi
-Hello tobi
+Hello, tobi!
 ```
 
 or
 
 ```sh
 $ source .venv/bin/activate
 $ python -m thaw_slump_segmentation hello tobi
-Hello tobi
+Hello, tobi!
 ```
 
 ## Data Processing
@@ -231,3 +231,59 @@ visualization_tiles:
   20190727_160426_104e: [5, 52, 75, 87, 113, 139, 239, 270, 277, 291, 305]
 
 ```
+
+### Tests
+
+We use [Pytest](https://pytest.org/) as testing library for unit tests.
+
+Run tests (via rye, for other environments activate the environment
+with `.venv/bin/activate` or similar and then simply remove the `rye run` before the command):
+
+```sh
+rye run pytest
+```
+
+Run all tests from a specific file:
+
+```sh
+rye run pytest tests/test_hello.py
+```
+
+Run a specific test from a specific file:
+
+```sh
+rye run pytest 
+```
+
+#### Test Parameters
+
+Several arguments can be passed to pytest to configure the environment for the tests, but not all test parameters are needed by all tests. Tests needing a parameter are skipped if the parameter is missing. In case the reason for skipping a test is of interest, [the argument `-rs`](https://docs.pytest.org/en/6.2.x/usage.html#detailed-summary-report) can be passed to pytest.
+
+##### `--data_dir`
+
+Most tests need input data which will be imported from a data directory as root. The tests expect a specific folder structure below the root. For now this is:
+* a folder `raw_data_dir` where the folders `scenes` and `tiles` exist for the planet source data
+* a folder `auxiliary` with the subfolder `ArcticDEM` where the virtual raster files reside, which are pointing to the slope and relative elevation data.
+
+No files are changed within those directories. The data is copied to a temporary directory and the tests are run there.
+
+##### `--gdal_bin`, `--gdal_path`
+
+Paths to the folders where the gdal executables (like `gdaltransform`, `--gdal_bin`) and gdal scripts like `gdal_retile.py` (`--gdal_path`) reside.
+
+##### `--proj_data_env`
+
+The proj library is needed for some processes by the gdal binaries and scripts. In case it cannot find or open its data folder, the PROJ_DATA environment variable needs to be set. Passing the parameter `--proj_data_env` to pytest will set this environment variable during the tests. 
+
+### Todos testing
+
+* [ ] Add GitHub Workflow for [Pytest](https://docs.github.com/en/actions/automating-builds-and-tests/building-and-testing-python#testing-your-code) and [Ruff](https://github.com/chartboost/ruff-action)
+* [ ] Add tests for data-related scripts (single and multiple files (for multiprocessing) each):
+  * [ ] Setup Raw Data
+  * [x] Prepare data
+  * [ ] Inference
+  * [ ] (S2 related scripts)
+* [ ] Utilities
+  * [ ] GDAL init
+  * [ ] EE auth + init
+  * [ ] TBD

pyproject.toml

@@ -44,8 +44,6 @@ dependencies = [
     "opencv-python>=4.9.0.80",
     "swifter>=1.4.0",
     "mkdocs-awesome-pages-plugin>=2.9.2",
-    "ruff>=0.4.8",
-    "ipykernel>=6.29.4",
     "rich>=13.7.1",
     "torchsummary>=1.5.1",
     "typer>=0.12.3",
@@ -72,10 +70,13 @@ build-backend = "hatchling.build"
 [tool.rye]
 managed = true
 dev-dependencies = [
+    "ruff>=0.4.2",
+    "ipykernel>=6.29.4",
     "ipython>=8.23.0",
     "mkdocs-material>=9.5.19",
     "mkdocstrings[python]>=0.24.3",
     "mkdocs-gen-files>=0.5.0",
+    "pytest>=8.2.2",
 ]
 
 [tool.hatch.metadata]
@@ -94,3 +95,6 @@ line-length = 120
 [tool.ruff.format]
 quote-style = "single"
 docstring-code-format = true
+
+[tool.pytest.ini_options]
+addopts = ["--import-mode=importlib"]

src/thaw_slump_segmentation/main.py

@@ -25,7 +25,8 @@
 
 @cli.command()
 def hello(name: str):
-    typer.echo(f'Hello {name}')
+    typer.echo(f'Hello, {name}!')
+    return f'Hello, {name}!'
 
 
 cli.command()(train)

tests/conftest.py

@@ -0,0 +1,124 @@
+import pytest
+from pathlib import Path
+import os, warnings, subprocess
+
+def pytest_addoption(parser):
+    # adds a test parameter --data-dir to pass a directory where predefined testfiles reside
+    # the testfiles will be configured as a fixture
+    # tests which request this directory will be skipped if this parameter isn't passed.
+    parser.addoption(
+        "--data_dir",
+        action="store",
+        default=None,
+        help="Directory containing the predefined data files"
+    )
+
+    # GDAL config :
+    parser.addoption(
+        "--gdal_bin",
+        action="store",
+        default=None,
+        help='Path to gdal binaries'
+    )
+    parser.addoption(
+        "--gdal_path",
+        action="store",
+        default=None,
+        help='Path to gdal python scripts (e.g. gdal_retile.py), might be the same as --gdal_bin'
+    )
+    parser.addoption(
+        "--proj_data_env",
+        action="store",
+        default=None,
+        help='path to data of the proj library, will be set as PROJ_DATA environment variable'
+    )
+
+
+# the fixture for the data dir:
+@pytest.fixture
+def data_dir(request):
+    p = request.config.getoption("--data_dir")
+
+    if p is None:
+        pytest.skip( f"parameter --data_dir needed for this test" )
+
+    data_path_root = Path(p)
+
+    if not data_path_root.exists():
+        pytest.skip( f" {data_path_root} does not exist " )
+
+    return data_path_root
+
+
+#
+# ========== GDAL related fixtures =========================================
+#
+# take note if we ought to test if GDAL is available:
+#  0  -> GDAL not tested
+#  1  -> GDAL tested and works
+#  -1 -> GDAL tested and does NOT work
+GDAL_TEST = 0
+
+# the proj_data fixture will be imported by the gdal_{bin|path} fixtures
+@pytest.fixture()
+def proj_data(request):
+    proj_data = request.config.getoption("--proj_data_env")
+    if proj_data is not None:
+        os.environ["PROJ_DATA"] = proj_data   
+
+@pytest.fixture
+def gdal_bin(request, proj_data):
+    global GDAL_TEST
+    gdal_bin = request.config.getoption("--gdal_bin")
+    if gdal_bin is None:
+        pytest.skip( f"parameter --gdal_bin needed for this test" )
+
+    if GDAL_TEST == 0:
+        # not tested yet
+        # check if gdal via subprocess (as we do it in the scripts) works:
+        gdal_bin_path = Path(gdal_bin)
+
+        for bin_tool in [ 'gdal_rasterize', 'gdal_translate', 'gdalwarp', 'gdaltransform' ]:
+            bin_tool_path = gdal_bin_path / bin_tool
+
+            if not bin_tool_path.exists():
+                pytest.skip( f"--gdal_bin is required to point to the folder of the gdal binaries (e.g. gdalwarp) " )
+
+
+        gdaltransform = gdal_bin_path / "gdaltransform"
+        shell_command = f'echo "12 34" | {gdaltransform} -s_srs EPSG:4326 -t_srs EPSG:3857'
+        proc = subprocess.run(shell_command, shell=True, capture_output=True, text=True)
+
+        if proc.returncode > 0:
+            GDAL_TEST = -1
+            warning_message = (f"testing GDAL with {shell_command} failed: \n"
+            f"{proc.stderr}\n"
+            "You might want to pass --proj_data_env to pytest pointing to the proj database folder"
+            )
+            warnings.warn(warning_message)
+        else:
+            GDAL_TEST = 1
+
+    if GDAL_TEST == -1:
+        pytest.skip( f"a working gdal is required for this test"  )
+
+    return gdal_bin
+
+@pytest.fixture
+def gdal_path(request, proj_data):
+    gdal_path = request.config.getoption("--gdal_path")
+    if gdal_path is None:
+        pytest.skip( f"parameter --gdal_path needed for this test" )
+
+
+    pygdal_bin_path = Path(gdal_path)
+
+    for py_tool in [ 'gdal_merge.py', 'gdal_retile.py', 'gdal_polygonize.py' ]:
+        pygdal_tool_path = pygdal_bin_path / py_tool
+
+        if not pygdal_tool_path.exists():
+            pytest.skip( f"--gdal_path is required to point to the folder of the gdal python scripts (e.g. gdal_merge.py) " )
+
+    return gdal_path
+
+

tests/test_hello.py

@@ -0,0 +1,5 @@
+from thaw_slump_segmentation.main import hello
+
+
+def test_hello():
+    assert hello('Tobias') == 'Hello, Tobias!'

tests/test_preprocessDirectory.py

@@ -0,0 +1,56 @@
+import pytest
+from pathlib import Path
+import tempfile, shutil
+
+from thaw_slump_segmentation.scripts.setup_raw_data import preprocess_directory
+
+
+def testCompleteProcessing(data_dir:Path, gdal_bin, gdal_path):
+
+    # check if data_dir this is just a basic image folder formatted from planet download
+
+    image_dir = data_dir / "raw_data_dir" / "scenes" / "20230807_191420_44_241d"
+    aux_dir = data_dir / "auxiliary"
+    if not image_dir.exists():
+        pytest.skip(f"could not find predefined image dir {image_dir}")
+
+    # create a working directory and copy images there
+    temp_path = Path(tempfile.mkdtemp())
+
+    source_dir = temp_path / "input"
+    source_dir.mkdir()
+    [shutil.copy(f, source_dir) for f in image_dir.glob("*")]
+
+    # target directory of the preprocess_directory processing
+    target_dir = temp_path / "output"
+    target_dir.mkdir()
+
+    backup_dir = temp_path / "backup"
+    backup_dir.mkdir()
+
+    preprocess_result = preprocess_directory(
+        image_dir=source_dir, 
+        data_dir=target_dir, 
+        aux_dir=aux_dir,
+        backup_dir=backup_dir,
+        log_path=temp_path / "preprocess.log",
+        gdal_bin=gdal_bin, gdal_path=gdal_path, label_required=False
+        )
+
+    # {'rename': 2, 'label': 2, 'ndvi': 1, 'tcvis': 1, 'rel_dem': 1, 'slope': 1, 'mask': 1, 'move': 1}
+    # assert that every result entry except "rename" (was for chopping of _clip) and "label" is 1
+    for k,v in preprocess_result.items():
+        if k not in ["rename", "label"]:
+            assert v == 1
+
+    #assert that the files ndvi.tif, relative_elevation.tif, slope.tif and tcvis.tif exist in backup/input
+    # and their size is non-zero
+
+    for f in ["ndvi.tif", "relative_elevation.tif", "slope.tif", "tcvis.tif"]:
+        testfile = backup_dir / "input" / f
+        assert testfile.exists()
+        assert testfile.stat().st_size > 0
+
+
+    shutil.rmtree(temp_path)
+