From 44c24b419ba35e2eb6d93bb93b47cd7578759f56 Mon Sep 17 00:00:00 2001
From: ppinchuk <ppinchuk@nrel.gov>
Date: Mon, 18 Sep 2023 14:03:41 -0600
Subject: [PATCH 1/2] Add test for warning message

---
 tests/cli/test_cli_collect.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/tests/cli/test_cli_collect.py b/tests/cli/test_cli_collect.py
index 2601eda6..9dc877c6 100644
--- a/tests/cli/test_cli_collect.py
+++ b/tests/cli/test_cli_collect.py
@@ -86,7 +86,7 @@ def test_collect_other_inputs(
     assert out_file not in files
 
     pattern = (tmp_path / pattern).as_posix()
-    with pytest.warns(gapsWarning):
+    with pytest.warns(gapsWarning) as warning_info:
         collect(
             out_file,
             pattern,
@@ -94,6 +94,12 @@ def test_collect_other_inputs(
             datasets=["cf_profile", "dne_dataset"],
         )
 
+    expected_message = (
+        "Could not find the following datasets in the output files"
+    )
+    assert expected_message in warning_info[0].message.args[0]
+    assert "dne_dataset" in warning_info[0].message.args[0]
+
     files = list(tmp_path.glob("*"))
     assert tmp_path / "chunk_files" in files
     assert out_file in files

From 6d33d77b7872fc49e5b58094d02e530f78ebc836 Mon Sep 17 00:00:00 2001
From: ppinchuk <ppinchuk@nrel.gov>
Date: Mon, 18 Sep 2023 14:04:01 -0600
Subject: [PATCH 2/2] Add `clobber` to collect CLI

---
 gaps/cli/collect.py           | 14 ++++++++++++--
 tests/cli/test_cli_collect.py | 26 ++++++++++++++++----------
 2 files changed, 28 insertions(+), 12 deletions(-)

diff --git a/gaps/cli/collect.py b/gaps/cli/collect.py
index 19a2b3e5..7165fdce 100644
--- a/gaps/cli/collect.py
+++ b/gaps/cli/collect.py
@@ -14,7 +14,12 @@
 
 
 def collect(
-    _out_path, _pattern, project_points=None, datasets=None, purge_chunks=False
+    _out_path,
+    _pattern,
+    project_points=None,
+    datasets=None,
+    purge_chunks=False,
+    clobber=True,
 ):
     """Run collection on local worker.
 
@@ -44,6 +49,11 @@ def collect(
         input files will **not** be removed if any of the datasets they
         contain have not been collected, regardless of the value of this
         input. By default, ``False``.
+    clobber : bool, optional
+        Flag to purge all collection output HDF5 files prior to running
+        the collection step if they exist on disk. This helps avoid any
+        surprising data byproducts when re-running the collection step
+        in a project directory. By default, ``True``.
 
     Returns
     -------
@@ -62,7 +72,7 @@ def collect(
     )
 
     datasets = _find_datasets(datasets, _pattern)
-    collector = Collector(_out_path, _pattern, project_points)
+    collector = Collector(_out_path, _pattern, project_points, clobber=clobber)
     for dataset_name in datasets:
         logger.debug("Collecting %r...", dataset_name)
         collector.collect(dataset_name)
diff --git a/tests/cli/test_cli_collect.py b/tests/cli/test_cli_collect.py
index 9dc877c6..efeb9a7c 100644
--- a/tests/cli/test_cli_collect.py
+++ b/tests/cli/test_cli_collect.py
@@ -36,22 +36,28 @@ def test_collect(
     profiles = manual_collect(collect_dir / pattern, "cf_profile")
 
     assert not list(tmp_path.glob("*"))
-    for h5_file in collect_dir.glob(pattern):
+    for file_ind, h5_file in enumerate(collect_dir.glob(pattern)):
         shutil.copy(h5_file, tmp_path / h5_file.name)
+        if file_ind == 0:
+            shutil.copy(h5_file, out_file)
 
     files = list(tmp_path.glob("*"))
-    assert len(files) == 4
+    assert len(files) == 5
     assert tmp_path / "chunk_files" not in files
-    assert out_file not in files
+    assert out_file in files
 
     pattern = (tmp_path / pattern).as_posix()
-    collect(
-        out_file,
-        pattern,
-        project_points=points_path,
-        datasets=datasets,
-        purge_chunks=True,
-    )
+    with pytest.warns(gapsWarning) as warning_info:
+        collect(
+            out_file,
+            pattern,
+            project_points=points_path,
+            datasets=datasets,
+            purge_chunks=True,
+        )
+
+    expected_message = "already exists and is being replaced"
+    assert expected_message in warning_info[0].message.args[0]
 
     files = list(tmp_path.glob("*"))
     assert tmp_path / "chunk_files" not in files