full fua testing for simplify_network() (#41)

* curate full FUA testing data * add full fua simplify_network() tests * bump 4/5 tolerance to 0.2 – test_simplify_network_full_fua() * try uploading osberved artifacts * try uploading osberved artifacts [2] * try uploading osberved artifacts [3] * try uploading osberved artifacts [4] * try uploading osberved artifacts [5] * try uploading osberved artifacts [6] * try uploading osberved artifacts [7] * try uploading osberved artifacts [8] * try uploading osberved artifacts [9] * try uploading osberved artifacts [10] * upload osberved artifacts * try uploading osberved artifacts [11] * try uploading osberved artifacts [12] * known simplified from ci_artifacts-ubuntu-latest-py312_sgeop-latest * run equality tests only on Ubuntu + latest/dev CI envs * codecov --------- Co-authored-by: Martin Fleischmann <[email protected]>
uscuni · Oct 16, 2024 · 5c033a4 · 5c033a4
1 parent 92579a9
commit 5c033a4
Show file tree

Hide file tree

Showing 17 changed files with 202 additions and 12 deletions.
diff --git a/.github/workflows/testing.yml b/.github/workflows/testing.yml
@@ -20,21 +20,22 @@
      name: ${{ matrix.os }}, ${{ matrix.environment-file }}
      runs-on: ${{ matrix.os }}
      timeout-minutes: 30
+     continue-on-error: true
      strategy:
        matrix:
          os: [ubuntu-latest]
          environment-file: [
-            ci/py311_sgeop-oldest.yaml,
-            ci/py311_sgeop-latest.yaml,
-            ci/py312_sgeop-latest.yaml,
-            ci/py312_sgeop-dev.yaml,
+            py311_sgeop-oldest,
+            py311_sgeop-latest,
+            py312_sgeop-latest,
+            py312_sgeop-dev,
          ]
          include:
-           - environment-file: ci/py312_sgeop-latest.yaml
+           - environment-file: py312_sgeop-latest
              os: macos-13 # Intel
-           - environment-file: ci/py312_sgeop-latest.yaml
+           - environment-file: py312_sgeop-latest
              os: macos-14 # Apple Silicon
-           - environment-file: ci/py312_sgeop-latest.yaml
+           - environment-file: py312_sgeop-latest
              os: windows-latest
        fail-fast: false
 
@@ -51,7 +52,7 @@
        - name: setup micromamba
          uses: mamba-org/setup-micromamba@v2
          with:
-           environment-file: ${{ matrix.environment-file }}
+           environment-file: ci/${{ matrix.environment-file }}.yaml
            micromamba-version: "latest"
 
        - name: install package
@@ -71,7 +72,24 @@
            --cov core \
            --cov-append \
            --cov-report term-missing \
-           --cov-report xml .
+           --cov-report xml . \
+           --env_type ${{ matrix.environment-file }}
+       
+       - name: zip artifacts - Ubuntu & macOS
+         run: zip ci_artifacts.zip ci_artifacts -r
+         if: matrix.os != 'windows-latest' && (success() || failure())
+
+       - name: zip artifacts - Windows
+         shell: powershell
+         run: Compress-Archive -Path ci_artifacts -Destination ci_artifacts.zip
+         if: matrix.os == 'windows-latest' && (success() || failure())
+
+       - name: archive observed simplified networks
+         uses: actions/upload-artifact@v4
+         with:
+           name: ci_artifacts-${{ matrix.os }}-${{ matrix.environment-file }}
+           path: ci_artifacts.zip
+         if: success() || failure()
 
        - name: codecov
          uses: codecov/codecov-action@v4
diff --git a/data/README.md b/data/README.md
@@ -0,0 +1,25 @@
+# FUA testing data
+
+## Contents
+
+* This `README.md`
+* `generate_simplified.py` – see [sgeop#7](https://github.com/uscuni/sgeop/issues/7)
+* Data
+   * There is a directory for each FUA listed below that contains 2 files:
+      * `original.parquet`: The original input street network derived from [OSM](https://www.openstreetmap.org/about) via [OSMNX](https://osmnx.readthedocs.io/en/stable/).
+      * `simplified.parquet`: The simplified street network following our algorithm with *default parameters*.
+
+## FUA Information
+
+| FUA  | City                                   | Shortand              |
+| ---  | ---                                    | ---                   |
+| 1133 | Aleppo, Syria, Middle East / Asia      | `aleppo_1133`         |
+| 869  | Auckland, New Zealand, Oceania / Asia  | `auckland_869`        |
+| 809  | Douala, Cameroon, Africa               | `douala_809`          |
+| 1656 | Liège, Belgium, Europe                 | `liege_1656`          |
+| 4617 | Bucaramanga, Colombia, S. America      | `bucaramanga_4617`    |
+| 4881 | Salt Lake City, Utah, USA, N. America  | `slc_4881`            |
+
+---------------------------------------
+
+Copyright (c) 2024-, sgeop Developers
diff --git a/data/aleppo_1133/original.parquet b/data/aleppo_1133/original.parquet
diff --git a/data/aleppo_1133/simplified.parquet b/data/aleppo_1133/simplified.parquet
diff --git a/data/auckland_869/original.parquet b/data/auckland_869/original.parquet
diff --git a/data/auckland_869/simplified.parquet b/data/auckland_869/simplified.parquet
diff --git a/data/bucaramanga_4617/original.parquet b/data/bucaramanga_4617/original.parquet
diff --git a/data/bucaramanga_4617/simplified.parquet b/data/bucaramanga_4617/simplified.parquet
diff --git a/data/douala_809/original.parquet b/data/douala_809/original.parquet
diff --git a/data/douala_809/simplified.parquet b/data/douala_809/simplified.parquet
diff --git a/data/generate_simplified.py b/data/generate_simplified.py
@@ -0,0 +1,83 @@
+import logging
+import pathlib
+import time
+
+import geopandas
+
+import sgeop
+
+start_time = time.time()
+
+logging.basicConfig(
+    filename="simplified_generation.log",
+    filemode="a",
+    format="%(asctime)s,%(msecs)d %(name)s %(levelname)s %(message)s",
+    datefmt="%H:%M:%S",
+    level=logging.NOTSET,
+)
+logger = logging.getLogger(__name__)
+
+logging.info("")
+logging.info("")
+logging.info(" |‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾‾|")
+logging.info(" | Generating Simplified Street Networks |")
+logging.info(" |_______________________________________|")
+logging.info("")
+logging.info("")
+logging.info("")
+
+fua_city = {
+    1133: "aleppo",
+    869: "auckland",
+    4617: "bucaramanga",
+    809: "douala",
+    1656: "liege",
+    4881: "slc",
+}
+
+# dict of cityname: fua ID
+city_fua = {c: f for f, c in fua_city.items()}
+
+for city, fua in city_fua.items():
+
+    t1 = time.time()
+    aoi = f"{city}_{fua}"
+
+    logging.info("")
+    logging.info("")
+    logging.info("")
+    logging.info("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >>>>")
+    logging.info("")
+    logging.info("")
+    logging.info(f"  ** {aoi} **")
+    logging.info("")
+    logging.info("")
+
+    # input data
+    original = geopandas.read_parquet(pathlib.Path(aoi, "original.parquet"))
+
+    # output data
+    simplified = sgeop.simplify_network(original)
+    simplified.to_parquet(pathlib.Path(aoi, "simplified.parquet"))
+
+    t2 = round((time.time() - t1) / 60.0, 2)
+
+    logging.info("")
+    logging.info("")
+    logging.info(f"\t{aoi} runtime: {t2} minutes")
+    logging.info("")
+    logging.info("")
+    logging.info("")
+    logging.info("<<<< ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
+    logging.info("")
+
+endtime_time = round((time.time() - start_time) / 60.0, 2)
+
+logging.info("")
+logging.info("")
+logging.info(f"Total runtime: {endtime_time} minutes")
+logging.info(
+    "========================================================================="
+)
+logging.info("")
+logging.info("")
diff --git a/data/liege_1656/original.parquet b/data/liege_1656/original.parquet
diff --git a/data/liege_1656/simplified.parquet b/data/liege_1656/simplified.parquet
diff --git a/data/slc_4881/original.parquet b/data/slc_4881/original.parquet
diff --git a/data/slc_4881/simplified.parquet b/data/slc_4881/simplified.parquet
diff --git a/sgeop/tests/conftest.py b/sgeop/tests/conftest.py
@@ -1,3 +1,5 @@
+import platform
+
 import geopandas.testing
 import numpy
 import pandas
@@ -59,8 +61,27 @@ def geom_test(
     ).all()
 
 
+def pytest_addoption(parser):
+    """Add custom command line arguments."""
+
+    # flag for determining CI environment
+    parser.addoption(
+        "--env_type",
+        action="store",
+        default="latest",
+        help="Testing environment type label",
+        type=str,
+    )
+
+
 def pytest_configure(config):  # noqa: ARG001
     """PyTest session attributes, methods, etc."""
 
+    valid_env_types = ["oldest", "latest", "dev"]
+    pytest.env_type = config.getoption("env_type").split("-")[-1]
+    assert pytest.env_type in valid_env_types
+
+    pytest.ubuntu = "ubuntu" in platform.version().lower()
+
     pytest.polygonize = polygonize
     pytest.geom_test = geom_test
diff --git a/sgeop/tests/test_simplify.py b/sgeop/tests/test_simplify.py
@@ -7,15 +7,58 @@
 import sgeop
 
 test_data = pathlib.Path("sgeop", "tests", "data")
+full_fua_data = pathlib.Path("data")
 
-ac = "apalachicola"
+ci_artifacts = pathlib.Path("ci_artifacts")
 
 
-def test_simplify():
+def test_simplify_network_small():
+    ac = "apalachicola"
     known = geopandas.read_parquet(test_data / f"{ac}_simplified.parquet")
+    known_length = 69477.0
+
     observed = sgeop.simplify_network(
         geopandas.read_parquet(test_data / f"{ac}_original.parquet")
     )
+    observed_length = observed.geometry.length.sum()
 
-    pytest.geom_test(known, observed)
+    # storing GH artifacts
+    artifact_dir = ci_artifacts / ac
+    artifact_dir.mkdir(parents=True, exist_ok=True)
+    observed.to_parquet(artifact_dir / "simplified.parquet")
+
+    assert pytest.approx(observed_length, rel=0.00001) == known_length
+
+    assert observed.shape == known.shape
     assert_series_equal(known._status, observed._status)
+    pytest.geom_test(known, observed)
+
+
+@pytest.mark.parametrize(
+    "aoi,tol,known_length",
+    [
+        ("aleppo_1133", 2e-1, 4_361_625),
+        ("auckland_869", 2e-1, 1_268_048),
+        ("bucaramanga_4617", 2e-1, 1_681_011),
+        ("douala_809", 1e-1, 2_961_364),
+        ("liege_1656", 2e-1, 2_350_782),
+        ("slc_4881", 2e-1, 1_762_456),
+    ],
+)
+def test_simplify_network_full_fua(aoi, tol, known_length):
+    known = geopandas.read_parquet(full_fua_data / aoi / "simplified.parquet")
+    observed = sgeop.simplify_network(
+        geopandas.read_parquet(full_fua_data / aoi / "original.parquet")
+    )
+    observed_length = observed.geometry.length.sum()
+
+    # storing GH artifacts
+    artifact_dir = ci_artifacts / aoi
+    artifact_dir.mkdir(parents=True, exist_ok=True)
+    observed.to_parquet(artifact_dir / "simplified.parquet")
+
+    assert pytest.approx(observed_length, rel=0.0001) == known_length
+
+    if pytest.ubuntu and pytest.env_type != "oldest":
+        assert_series_equal(known._status, observed._status)
+        pytest.geom_test(known, observed, tolerance=tol)