Support glob pattern in seismic responses

Changes to other tests are mostly caused by additional check for
directory existence that doesn't play nicely with mocked_files.

Author: achaikou

src/ert/config/seismic_config.py

@@ -8,7 +8,6 @@
 import polars as pl
 
 from ert.config._reservoir_data_utils import SeismicData
-from ert.substitutions import substitute_runpath_name
 
 from .parsing import (
     ConfigDict,
@@ -64,15 +63,21 @@ def validate_distance_between_responses(df: pl.DataFrame) -> None:
                 f"{SeismicData.TOLERANCE * 2} m apart."
             )
 
+    def _collect_response_filepaths(self, run_path: str) -> list[Path]:
+        filepaths = []
+        for file in self.expected_input_files:
+            filepaths.extend(
+                SeismicData.resolve_pattern_filepaths(
+                    run_path, file, on_error=InvalidResponseFile
+                )
+            )
+        return list(dict.fromkeys(filepaths))
+
     def read_from_file(self, run_path: str, iens: int, iter_: int) -> pl.DataFrame:
         responses = pl.DataFrame(schema=self.response_schema())
-        for key, file in zip(self.keys, self.expected_input_files, strict=True):
-            filepath_runpath_relative = substitute_runpath_name(file, iens, iter_)
-            filepath = Path(run_path) / filepath_runpath_relative
-            if not filepath.exists():
-                raise InvalidResponseFile(
-                    f"Expected seismic response file {filepath} does not exist."
-                )
+        filepaths = self._collect_response_filepaths(run_path)
+        keys = [f.stem for f in filepaths]
+        for key, filepath in zip(keys, filepaths, strict=True):
             suffix = filepath.suffix.lower()
             if suffix == ".parquet":
                 data = pl.read_parquet(filepath)
@@ -114,4 +119,5 @@ def from_config_dict(cls, config_dict: ConfigDict) -> Self | None:
             name="seismic",
             input_files=files,
             keys=[Path(f).stem for f in files],
+            has_finalized_keys=False,
         )

tests/ert/unit_tests/config/test_seismic_config.py

@@ -1,5 +1,6 @@
 from io import BytesIO, StringIO
 from pathlib import Path
+from textwrap import dedent
 from typing import cast
 
 import polars as pl
@@ -108,19 +109,18 @@ def test_that_unsupported_seismic_response_file_extension_raises_invalid_respons
 ):
     key = "horizon--amplitude_full_min_depth--20250101_20240101"
     name = f"{key}.txt"
-    runpath = "/runpath"
-    mocked_files[f"{runpath}/{name}"] = "irrelevant content"
+    mocked_files[f"{name}"] = "irrelevant content"
 
     seismic_config = SeismicConfig(
         input_files=[name],
         keys=[key],
     )
 
     with pytest.raises(InvalidResponseFile) as err:
-        seismic_config.read_from_file(runpath, 1, 1)
+        seismic_config.read_from_file("", 1, 1)
 
     assert (
-        f"Unsupported seismic response file extension '.txt' for {runpath}/{name}. "
+        f"Unsupported seismic response file extension '.txt' for {name}. "
         "Expected '.csv' or '.parquet'." in str(err.value)
     )
 
@@ -140,7 +140,6 @@ def test_that_seismic_config_reads_from_all_input_files(mocked_files, suffix):
     key2 = "horizon--amplitude_full_mean_depth--20260101_20240101"
     name1 = f"{key1}{suffix}"
     name2 = f"{key2}{suffix}"
-    runpath = "/runpath"
 
     frame1 = pl.DataFrame(
         {
@@ -160,27 +159,67 @@ def test_that_seismic_config_reads_from_all_input_files(mocked_files, suffix):
             "REGION": [1.0, 1.0],
         }
     )
-    _mock_seismic_response(mocked_files, f"{runpath}/{name1}", frame1, suffix)
-    _mock_seismic_response(mocked_files, f"{runpath}/{name2}", frame2, suffix)
+    _mock_seismic_response(mocked_files, f"{name1}", frame1, suffix)
+    _mock_seismic_response(mocked_files, f"{name2}", frame2, suffix)
 
     seismic_config = SeismicConfig(
         input_files=[name1, name2],
         keys=[key1, key2],
     )
 
-    data = seismic_config.read_from_file(runpath, 1, 1)
+    data = seismic_config.read_from_file("", 1, 1)
     assert data.shape == (4, 4)
     assert data["response_key"].to_list() == [key1, key1, key2, key2]
     assert data["east"].to_list() == [100.0, 105.0, 100.0, 105.0]
     assert data["north"].to_list() == [200.0, 205.0, 200.0, 205.0]
     assert data["values"].to_list() == [1.0, 2.0, 3.0, 4.0]
 
 
+@pytest.mark.usefixtures("use_tmpdir")
+def test_that_seismic_config_supports_glob_pattern():
+    key1 = "horizon1--amplitude_full_mean_depth--20260101_20240101"
+    key2 = "horizon1--amplitude_full_min_depth--20250101_20240101"
+    key3 = "horizon2--amplitude_far_min_depth--20250101_20240101"
+    key4 = "other_horizon--amplitude_full_mean_depth--20260101_20240101"
+
+    runpath = "runpath"
+    Path(runpath).mkdir(parents=True, exist_ok=True)
+
+    for key in [key1, key2, key3, key4]:
+        name = f"{key}.csv"
+        simulated_path = runpath + "/" + name
+        content = dedent(
+            """
+            X_UTME,Y_UTMN,OBS,OBS_ERROR,REGION
+            100.00,200.00,1.0,0.005,1.0
+            """
+        )
+        Path(simulated_path).write_text(
+            content,
+            encoding="utf8",
+        )
+
+    pattern1 = "horizon1--amplitude_full*"
+    pattern2 = "horizon2--amplitude_far*"
+    duplicate_pattern = "horizon*"
+    config = ErtConfig.from_dict(
+        {
+            "SEISMIC": [pattern1, pattern2, duplicate_pattern],
+        }
+    )
+
+    seismic_config = cast(
+        SeismicConfig, config.ensemble_config.response_configs["seismic"]
+    )
+
+    data = seismic_config.read_from_file(runpath, 1, 1)
+    assert sorted(data["response_key"].to_list()) == [key1, key2, key3]
+
+
 @pytest.mark.parametrize("suffix", [".csv", ".parquet"])
 def test_that_empty_seismic_response_file_does_not_raise(mocked_files, suffix):
     key = "horizon--amplitude_full_min_depth--20250101_20240101"
     name = f"{key}{suffix}"
-    runpath = "/runpath"
 
     empty = pl.DataFrame(
         schema={
@@ -191,14 +230,14 @@ def test_that_empty_seismic_response_file_does_not_raise(mocked_files, suffix):
             "REGION": pl.Float64,
         }
     )
-    _mock_seismic_response(mocked_files, f"{runpath}/{name}", empty, suffix)
+    _mock_seismic_response(mocked_files, f"{name}", empty, suffix)
 
     seismic_config = SeismicConfig(
         input_files=[name],
         keys=[key],
     )
 
-    data = seismic_config.read_from_file(runpath, 1, 1)
+    data = seismic_config.read_from_file("", 1, 1)
     assert data.is_empty()
 
 
@@ -209,32 +248,27 @@ def test_that_empty_seismic_response_file_does_not_raise(mocked_files, suffix):
         pytest.param([0.0, 0.0], [0.1953125, 0.0], id="less than double tolerance"),
     ],
 )
-@pytest.mark.parametrize("suffix", [".csv", ".parquet"])
 def test_that_seismic_response_coordinate_distance_below_tolerance_raises(
-    mocked_files, east, north, suffix
+    mocked_files, east, north
 ):
     key = "horizon--amplitude_full_min_depth--20250101_20240101"
-    name = f"{key}{suffix}"
-    runpath = "/runpath"
-
-    frame = pl.DataFrame(
-        {
-            "X_UTME": east,
-            "Y_UTMN": north,
-            "OBS": [1.0, 2.0],
-            "OBS_ERROR": [0.005, 0.005],
-            "REGION": [1.0, 1.0],
-        }
+    name = f"{key}.csv"
+
+    mocked_files[name] = dedent(
+        f"""
+        X_UTME,Y_UTMN,OBS,OBS_ERROR,REGION
+        {east[0]},{north[0]},1.0,0.005,1.0
+        {east[1]},{north[1]},2.0,0.005,1.0
+        """
     )
-    _mock_seismic_response(mocked_files, f"{runpath}/{name}", frame, suffix)
 
     seismic_config = SeismicConfig(
         input_files=[name],
         keys=[key],
     )
 
     with pytest.raises(InvalidResponseFile) as err:
-        seismic_config.read_from_file(runpath, 1, 1)
+        seismic_config.read_from_file("", 1, 1)
 
     assert (
         "Seismic response coordinates with approximate locations "