speed up testing

maps-as-data · Sep 6, 2024 · a9f54d9 · a9f54d9
1 parent 52b6c97
commit a9f54d9
Show file tree

Hide file tree

Showing 4 changed files with 66 additions and 63 deletions.
diff --git a/.github/workflows/mr_ci_text_spotting.yml b/.github/workflows/mr_ci_text_spotting.yml
@@ -57,7 +57,7 @@ jobs:
       - name: Install DPText-DETR
         run: |
           git clone https://github.com/maps-as-data/DPText-DETR.git
-          python -m pip install 'git+https://github.com/maps-as-data/DPText-DETR.git' --force-reinstall # Install DPText-DETR
+          python -m pip install 'git+https://github.com/maps-as-data/DPText-DETR.git' # Install DPText-DETR
           python -m pip install numpy==1.26.4
           wget https://huggingface.co/rwood-97/DPText_DETR_ArT_R_50_poly/resolve/main/art_final.pth
 
@@ -69,7 +69,7 @@ jobs:
       - name: Install DeepSolo
         run: |
           git clone https://github.com/maps-as-data/DeepSolo.git
-          python -m pip install 'git+https://github.com/maps-as-data/DeepSolo.git' --force-reinstall # Install DeepSolo
+          python -m pip install 'git+https://github.com/maps-as-data/DeepSolo.git' # Install DeepSolo
           python -m pip install numpy==1.26.4
           wget https://huggingface.co/rwood-97/DeepSolo_ic15_res50/resolve/main/ic15_res50_finetune_synth-tt-mlt-13-15-textocr.pth
 
@@ -80,7 +80,7 @@ jobs:
       - name: Install MapTextPipeline
         run: |
           git clone https://github.com/maps-as-data/MapTextPipeline.git
-          python -m pip install 'git+https://github.com/maps-as-data/MapTextPipeline.git' --force-reinstall # Install MapTextPipeline
+          python -m pip install 'git+https://github.com/maps-as-data/MapTextPipeline.git' # Install MapTextPipeline
           python -m pip install "numpy<2.0.0"
           wget https://huggingface.co/rwood-97/MapTextPipeline_rumsey/resolve/main/rumsey-finetune.pth
 

diff --git a/test_text_spotting/test_deepsolo_runner.py b/test_text_spotting/test_deepsolo_runner.py
@@ -23,12 +23,12 @@
 )
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def sample_dir():
     return pathlib.Path(__file__).resolve().parent.parent / "tests" / "sample_files"
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def init_dataframes(sample_dir, tmp_path):
     """Initializes MapImages object (with metadata from csv and patches) and creates parent and patch dataframes.
     Returns
@@ -44,7 +44,7 @@ def init_dataframes(sample_dir, tmp_path):
     return parent_df, patch_df
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def init_runner(init_dataframes):
     parent_df, patch_df = init_dataframes
     runner = DeepSoloRunner(
@@ -55,6 +55,13 @@ def init_runner(init_dataframes):
     return runner
 
 
+@pytest.fixture(scope="session")
+def runner_run_all(init_runner):
+    runner = init_runner
+    _ = runner.run_all()
+    return runner
+
+
 def test_deepsolo_init(init_dataframes):
     parent_df, patch_df = init_dataframes
     runner = DeepSoloRunner(
@@ -122,42 +129,37 @@ def test_deepsolo_run_all(init_runner):
     assert "patch-0-0-800-40-#mapreader_text.png#.png" in out.keys()
     assert isinstance(out["patch-0-0-800-40-#mapreader_text.png#.png"], list)
     # dataframe
-    runner.patch_predictions = {}
-    out = runner.run_all(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.patch_predictions, geo=False, parent=False)
     assert isinstance(out, pd.DataFrame)
     assert set(out.columns) == set(["image_id", "geometry", "text", "score"])
     assert "patch-0-0-800-40-#mapreader_text.png#.png" in out["image_id"].values
 
 
-def test_deepsolo_convert_to_parent(init_runner):
-    runner = init_runner
-    _ = runner.run_all()
+def test_deepsolo_convert_to_parent(runner_run_all):
+    runner = runner_run_all
     # dict
     out = runner.convert_to_parent_pixel_bounds()
     assert isinstance(out, dict)
     assert "mapreader_text.png" in out.keys()
     assert isinstance(out["mapreader_text.png"], list)
     # dataframe
-    runner.parent_predictions = {}
-    out = runner.convert_to_parent_pixel_bounds(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.parent_predictions, geo=False, parent=True)
     assert isinstance(out, pd.DataFrame)
     assert set(out.columns) == set(
         ["image_id", "patch_id", "geometry", "text", "score"]
     )
     assert "mapreader_text.png" in out["image_id"].values
 
 
-def test_deepsolo_convert_to_parent_coords(init_runner):
-    runner = init_runner
-    _ = runner.run_all()
+def test_deepsolo_convert_to_parent_coords(runner_run_all):
+    runner = runner_run_all
     # dict
     out = runner.convert_to_coords()
     assert isinstance(out, dict)
     assert "mapreader_text.png" in out.keys()
     assert isinstance(out["mapreader_text.png"], list)
     # dataframe
-    runner.parent_predictions = {}
-    out = runner.convert_to_coords(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.geo_predictions, geo=True, parent=True)
     assert isinstance(out, gpd.GeoDataFrame)
     assert set(out.columns) == set(
         ["image_id", "patch_id", "geometry", "crs", "text", "score"]
@@ -180,12 +182,12 @@ def test_deepsolo_deduplicate(sample_dir, tmp_path):
     _ = runner.run_all()
     out = runner.convert_to_parent_pixel_bounds(deduplicate=False)
     len_before = len(out["mapreader_text.png"])
-    runner.patch_predictions = {}
+    runner.parent_predictions = {}
     out_07 = runner.convert_to_parent_pixel_bounds(deduplicate=True)
     len_07 = len(out_07["mapreader_text.png"])
     print(len_before, len_07)
     assert len_before >= len_07
-    runner.patch_predictions = {}
+    runner.parent_predictions = {}
     out_05 = runner.convert_to_parent_pixel_bounds(deduplicate=True, min_ioa=0.5)
     len_05 = len(out_05["mapreader_text.png"])
     print(len_before, len_05)
@@ -203,9 +205,8 @@ def test_deepsolo_run_on_image(init_runner):
     assert isinstance(out["instances"], Instances)
 
 
-def test_deepsolo_save_to_geojson(init_runner, tmp_path):
-    runner = init_runner
-    _ = runner.run_all()
+def test_deepsolo_save_to_geojson(runner_run_all, tmp_path):
+    runner = runner_run_all
     _ = runner.convert_to_coords()
     runner.save_to_geojson(f"{tmp_path}/text.geojson")
     assert os.path.exists(f"{tmp_path}/text.geojson")

diff --git a/test_text_spotting/test_dptext_runner.py b/test_text_spotting/test_dptext_runner.py
@@ -23,12 +23,12 @@
 )
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def sample_dir():
     return pathlib.Path(__file__).resolve().parent.parent / "tests" / "sample_files"
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def init_dataframes(sample_dir, tmp_path):
     """Initializes MapImages object (with metadata from csv and patches) and creates parent and patch dataframes.
     Returns
@@ -44,7 +44,7 @@ def init_dataframes(sample_dir, tmp_path):
     return parent_df, patch_df
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def init_runner(init_dataframes):
     parent_df, patch_df = init_dataframes
     runner = DPTextDETRRunner(
@@ -55,6 +55,13 @@ def init_runner(init_dataframes):
     return runner
 
 
+@pytest.fixture(scope="session")
+def runner_run_all(init_runner):
+    runner = init_runner
+    _ = runner.run_all()
+    return runner
+
+
 def test_dptext_init(init_dataframes):
     parent_df, patch_df = init_dataframes
     runner = DPTextDETRRunner(
@@ -122,40 +129,35 @@ def test_dptext_run_all(init_runner):
     assert "patch-0-0-800-40-#mapreader_text.png#.png" in out.keys()
     assert isinstance(out["patch-0-0-800-40-#mapreader_text.png#.png"], list)
     # dataframe
-    runner.patch_predictions = {}
-    out = runner.run_all(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.patch_predictions, geo=False, parent=False)
     assert isinstance(out, pd.DataFrame)
     assert set(out.columns) == set(["image_id", "geometry", "score"])
     assert "patch-0-0-800-40-#mapreader_text.png#.png" in out["image_id"].values
 
 
-def test_dptext_convert_to_parent(init_runner):
-    runner = init_runner
-    _ = runner.run_all()
+def test_dptext_convert_to_parent(runner_run_all):
+    runner = runner_run_all
     # dict
     out = runner.convert_to_parent_pixel_bounds()
     assert isinstance(out, dict)
     assert "mapreader_text.png" in out.keys()
     assert isinstance(out["mapreader_text.png"], list)
     # dataframe
-    runner.parent_predictions = {}
-    out = runner.convert_to_parent_pixel_bounds(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.parent_predictions, geo=False, parent=True)
     assert isinstance(out, pd.DataFrame)
     assert set(out.columns) == set(["image_id", "patch_id", "geometry", "score"])
     assert "mapreader_text.png" in out["image_id"].values
 
 
-def test_dptext_convert_to_parent_coords(init_runner):
-    runner = init_runner
-    _ = runner.run_all()
+def test_dptext_convert_to_parent_coords(runner_run_all):
+    runner = runner_run_all
     # dict
     out = runner.convert_to_coords()
     assert isinstance(out, dict)
     assert "mapreader_text.png" in out.keys()
     assert isinstance(out["mapreader_text.png"], list)
     # dataframe
-    runner.parent_predictions = {}
-    out = runner.convert_to_coords(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.geo_predictions, geo=True, parent=True)
     assert isinstance(out, gpd.GeoDataFrame)
     assert set(out.columns) == set(["image_id", "patch_id", "geometry", "crs", "score"])
     assert "mapreader_text.png" in out["image_id"].values
@@ -176,12 +178,12 @@ def test_dptext_deduplicate(sample_dir, tmp_path):
     _ = runner.run_all()
     out = runner.convert_to_parent_pixel_bounds(deduplicate=False)
     len_before = len(out["mapreader_text.png"])
-    runner.patch_predictions = {}
+    runner.parent_predictions = {}
     out_07 = runner.convert_to_parent_pixel_bounds(deduplicate=True)
     len_07 = len(out_07["mapreader_text.png"])
     print(len_before, len_07)
     assert len_before >= len_07
-    runner.patch_predictions = {}
+    runner.parent_predictions = {}
     out_05 = runner.convert_to_parent_pixel_bounds(deduplicate=True, min_ioa=0.5)
     len_05 = len(out_05["mapreader_text.png"])
     print(len_before, len_05)
@@ -199,9 +201,8 @@ def test_dptext_run_on_image(init_runner):
     assert isinstance(out["instances"], Instances)
 
 
-def test_dptext_save_to_geojson(init_runner, tmp_path):
-    runner = init_runner
-    _ = runner.run_all()
+def test_dptext_save_to_geojson(runner_run_all, tmp_path):
+    runner = runner_run_all
     _ = runner.convert_to_coords()
     runner.save_to_geojson(f"{tmp_path}/text.geojson")
     assert os.path.exists(f"{tmp_path}/text.geojson")

diff --git a/test_text_spotting/test_maptext_runner.py b/test_text_spotting/test_maptext_runner.py
@@ -23,12 +23,12 @@
 )
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def sample_dir():
     return pathlib.Path(__file__).resolve().parent.parent / "tests" / "sample_files"
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def init_dataframes(sample_dir, tmp_path):
     """Initializes MapImages object (with metadata from csv and patches) and creates parent and patch dataframes.
     Returns
@@ -44,7 +44,7 @@ def init_dataframes(sample_dir, tmp_path):
     return parent_df, patch_df
 
 
-@pytest.fixture
+@pytest.fixture(scope="session")
 def init_runner(init_dataframes):
     parent_df, patch_df = init_dataframes
     runner = MapTextRunner(
@@ -55,6 +55,13 @@ def init_runner(init_dataframes):
     return runner
 
 
+@pytest.fixture(scope="session")
+def runner_run_all(init_runner):
+    runner = init_runner
+    _ = runner.run_all()
+    return runner
+
+
 def test_maptext_init(init_dataframes):
     parent_df, patch_df = init_dataframes
     runner = MapTextRunner(
@@ -122,42 +129,37 @@ def test_maptext_run_all(init_runner):
     assert "patch-0-0-800-40-#mapreader_text.png#.png" in out.keys()
     assert isinstance(out["patch-0-0-800-40-#mapreader_text.png#.png"], list)
     # dataframe
-    runner.patch_predictions = {}
-    out = runner.run_all(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.patch_predictions, geo=False, parent=False)
     assert isinstance(out, pd.DataFrame)
     assert set(out.columns) == set(["image_id", "geometry", "text", "score"])
     assert "patch-0-0-800-40-#mapreader_text.png#.png" in out["image_id"].values
 
 
-def test_maptext_convert_to_parent(init_runner):
-    runner = init_runner
-    _ = runner.run_all()
+def test_maptext_convert_to_parent(runner_run_all):
+    runner = runner_run_all
     # dict
     out = runner.convert_to_parent_pixel_bounds()
     assert isinstance(out, dict)
     assert "mapreader_text.png" in out.keys()
     assert isinstance(out["mapreader_text.png"], list)
     # dataframe
-    runner.parent_predictions = {}
-    out = runner.convert_to_parent_pixel_bounds(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.patch_predictions, geo=False, parent=True)
     assert isinstance(out, pd.DataFrame)
     assert set(out.columns) == set(
         ["image_id", "patch_id", "geometry", "text", "score"]
     )
     assert "mapreader_text.png" in out["image_id"].values
 
 
-def test_maptext_convert_to_parent_coords(init_runner):
-    runner = init_runner
-    _ = runner.run_all()
+def test_maptext_convert_to_parent_coords(runner_run_all):
+    runner = runner_run_all
     # dict
     out = runner.convert_to_coords()
     assert isinstance(out, dict)
     assert "mapreader_text.png" in out.keys()
     assert isinstance(out["mapreader_text.png"], list)
     # dataframe
-    runner.parent_predictions = {}
-    out = runner.convert_to_coords(return_dataframe=True)
+    out = runner._dict_to_dataframe(runner.parent_predictions, geo=True, parent=True)
     assert isinstance(out, gpd.GeoDataFrame)
     assert set(out.columns) == set(
         ["image_id", "patch_id", "geometry", "crs", "text", "score"]
@@ -180,12 +182,12 @@ def test_maptext_deduplicate(sample_dir, tmp_path):
     _ = runner.run_all()
     out = runner.convert_to_parent_pixel_bounds(deduplicate=False)
     len_before = len(out["mapreader_text.png"])
-    runner.patch_predictions = {}
+    runner.parent_predictions = {}
     out_07 = runner.convert_to_parent_pixel_bounds(deduplicate=True)
     len_07 = len(out_07["mapreader_text.png"])
     print(len_before, len_07)
     assert len_before >= len_07
-    runner.patch_predictions = {}
+    runner.parent_predictions = {}
     out_05 = runner.convert_to_parent_pixel_bounds(deduplicate=True, min_ioa=0.5)
     len_05 = len(out_05["mapreader_text.png"])
     print(len_before, len_05)
@@ -203,9 +205,8 @@ def test_maptext_run_on_image(init_runner):
     assert isinstance(out["instances"], Instances)
 
 
-def test_maptext_save_to_geojson(init_runner, tmp_path):
-    runner = init_runner
-    _ = runner.run_all()
+def test_maptext_save_to_geojson(runner_run_all, tmp_path):
+    runner = runner_run_all
     _ = runner.convert_to_coords()
     runner.save_to_geojson(f"{tmp_path}/text.geojson")
     assert os.path.exists(f"{tmp_path}/text.geojson")