From 79470ec36768dae6c17922dfe340ab59e021aa46 Mon Sep 17 00:00:00 2001
From: Paul <pinchukpaul@gmail.com>
Date: Fri, 8 May 2026 16:00:56 -0600
Subject: [PATCH 1/2] Fix regex pattern in preprocess_collect_config and add
 test for untagged files

Co-authored-by: Copilot <copilot@github.com>
---
 gaps/cli/preprocessing.py          |  7 +++++--
 tests/cli/test_cli_preprocesing.py | 32 ++++++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/gaps/cli/preprocessing.py b/gaps/cli/preprocessing.py
index a125754f..9bd81ea9 100644
--- a/gaps/cli/preprocessing.py
+++ b/gaps/cli/preprocessing.py
@@ -150,13 +150,16 @@ def preprocess_collect_config(
     files = collect_pattern
     if files == "PIPELINE":
         files = parse_previous_status(project_dir, command_name)
-        files = [re.sub(f"{TAG}\\d+", "*", fname) for fname in files]
+        files = [re.sub(f"{TAG}\\d+", f"{TAG}*", fname) for fname in files]
 
     if isinstance(files, str):
         files = [files]
 
     if isinstance(files, abc.Sequence):
-        files = {pattern.replace("*", ""): pattern for pattern in files}
+        files = {
+            pattern.replace(f"{TAG}*", "").replace("*", ""): pattern
+            for pattern in files
+        }
 
     files = [
         (
diff --git a/tests/cli/test_cli_preprocesing.py b/tests/cli/test_cli_preprocesing.py
index d4e10c69..69eaa96c 100644
--- a/tests/cli/test_cli_preprocesing.py
+++ b/tests/cli/test_cli_preprocesing.py
@@ -6,6 +6,7 @@
 """
 
 import json
+import glob
 from pathlib import Path
 
 import pytest
@@ -114,6 +115,37 @@ def test_preprocess_collect_config_pipeline_input(tmp_path):
         assert out_fp == pattern.replace("*", "")
 
 
+def test_preprocess_collect_config_pipeline_input_ignores_untagged_file(
+    tmp_path,
+):
+    """Test that PIPELINE collection patterns do not match untagged files."""
+    config_fp = tmp_path / "pipe_config.json"
+    with open(config_fp, "w") as file_:
+        json.dump(SAMPLE_CONFIG, file_)
+
+    (tmp_path / "config.json").touch()
+    (tmp_path / "collect_config.json").touch()
+
+    Pipeline(config_fp)
+
+    job_file = tmp_path / "output_file_j0.h5"
+    job_file.touch()
+    (tmp_path / "output_file.h5").touch()
+    Status.make_single_job_file(
+        tmp_path,
+        pipeline_step="run",
+        job_name="test_0",
+        attrs={StatusField.OUT_FILE: job_file.as_posix()},
+    )
+
+    config = preprocess_collect_config({}, tmp_path, "collect-run")
+
+    matched_files = sorted(
+        Path(path) for path in glob.glob(config["_pattern"][0])
+    )
+    assert matched_files == [job_file]
+
+
 def test_split_project_points_into_ranges():
     """Test the `split_project_points_into_ranges` function."""
 

From 37127aef2e3176bd8c3b1fdaf345e5fa6ab3af74 Mon Sep 17 00:00:00 2001
From: Paul <pinchukpaul@gmail.com>
Date: Sat, 9 May 2026 14:47:24 -0600
Subject: [PATCH 2/2] Fix tests

---
 tests/cli/test_cli.py              | 4 ++--
 tests/cli/test_cli_preprocesing.py | 3 ++-
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/tests/cli/test_cli.py b/tests/cli/test_cli.py
index 8077349a..a8412dd5 100644
--- a/tests/cli/test_cli.py
+++ b/tests/cli/test_cli.py
@@ -370,7 +370,7 @@ def test_cli_monitor(
         assert "collect-run_collect_pattern" in collected_outputs.attrs
         assert (
             Path(collected_outputs.attrs["collect-run_collect_pattern"])
-            == tmp_cwd / file_pattern
+            == tmp_cwd / file_pattern.replace("*", f"{TAG}*")
         )
 
     profiles = manual_collect(data_dir / file_pattern, "cf_profile")
@@ -487,7 +487,7 @@ def test_cli_background(
         assert "collect-run_collect_pattern" in collected_outputs.attrs
         assert (
             Path(collected_outputs.attrs["collect-run_collect_pattern"])
-            == tmp_cwd / file_pattern
+            == tmp_cwd / file_pattern.replace("*", f"{TAG}*")
         )
 
     profiles = manual_collect(data_dir / file_pattern, "cf_profile")
diff --git a/tests/cli/test_cli_preprocesing.py b/tests/cli/test_cli_preprocesing.py
index 69eaa96c..f72e5aab 100644
--- a/tests/cli/test_cli_preprocesing.py
+++ b/tests/cli/test_cli_preprocesing.py
@@ -17,6 +17,7 @@
     preprocess_collect_config,
     split_project_points_into_ranges,
 )
+from gaps.cli.config import TAG
 from gaps.exceptions import gapsConfigError
 from gaps.warn import gapsWarning
 
@@ -112,7 +113,7 @@ def test_preprocess_collect_config_pipeline_input(tmp_path):
     assert len(config["_pattern"]) == 2
     for out_fp, pattern in zip(config["_out_path"], config["_pattern"]):
         assert any(name in out_fp for name in allowed_out_fn)
-        assert out_fp == pattern.replace("*", "")
+        assert out_fp == pattern.replace(f"{TAG}*", "")
 
 
 def test_preprocess_collect_config_pipeline_input_ignores_untagged_file(