From 73c1cf7102455c426fa9a048ccabf318d99207ba Mon Sep 17 00:00:00 2001
From: Bede Constantinides <bedeabc@gmail.com>
Date: Mon, 16 Dec 2024 13:42:30 +0000
Subject: [PATCH] Error msg for mismatched read counts, update log output for
 stdout, test

---
 src/hostile/lib.py  | 26 ++++++++++++++++++++------
 src/hostile/util.py |  4 +++-
 tests/test_all.py   | 10 +++++-----
 3 files changed, 28 insertions(+), 12 deletions(-)

diff --git a/src/hostile/lib.py b/src/hostile/lib.py
index 48b6fc8..b40fb88 100644
--- a/src/hostile/lib.py
+++ b/src/hostile/lib.py
@@ -46,6 +46,7 @@ def gather_stats(
     aligner: str,
     invert: bool,
     index: str,
+    stdout: bool,
 ) -> list[dict[str, str | int | float | list[str]]]:
     stats = []
     for fastq1 in fastqs:
@@ -64,7 +65,12 @@ def gather_stats(
             proportion_removed = float(0)
         options = [
             k
-            for k, v in {"rename": rename, "reorder": reorder, "invert": invert}.items()
+            for k, v in {
+                "invert": invert,
+                "rename": rename,
+                "reorder": reorder,
+                "stdout": stdout,
+            }.items()
             if v
         ]
         report = SampleReport(
@@ -93,6 +99,7 @@ def gather_stats_paired(
     aligner: str,
     index: str,
     invert: bool,
+    stdout: bool,
 ) -> list[dict[str, str | int | float]]:
     stats = []
     for fastq1, fastq2 in fastqs:
@@ -113,7 +120,12 @@ def gather_stats_paired(
             proportion_removed = float(0)
         options = [
             k
-            for k, v in {"rename": rename, "reorder": reorder, "invert": invert}.items()
+            for k, v in {
+                "invert": invert,
+                "rename": rename,
+                "reorder": reorder,
+                "stdout": stdout,
+            }.items()
             if v
         ]
         stats.append(
@@ -126,10 +138,10 @@ def gather_stats_paired(
                 fastq2_in_name=fastq2.name,
                 fastq1_in_path=str(fastq1),
                 fastq2_in_path=str(fastq2),
-                fastq1_out_name=fastq1_out_path.name,
-                fastq2_out_name=fastq2_out_path.name,
-                fastq1_out_path=str(fastq1_out_path),
-                fastq2_out_path=str(fastq2_out_path),
+                fastq1_out_name=fastq1_out_path.name if not stdout else None,
+                fastq2_out_name=fastq2_out_path.name if not stdout else None,
+                fastq1_out_path=str(fastq1_out_path) if not stdout else None,
+                fastq2_out_path=str(fastq2_out_path) if not stdout else None,
                 reads_in=n_reads_in,
                 reads_out=n_reads_out,
                 reads_removed=n_reads_removed,
@@ -191,6 +203,7 @@ def clean_fastqs(
         aligner=aligner.name,
         index=index,
         invert=invert,
+        stdout=stdout,
     )
     util.fix_empty_fastqs(stats)
     logging.info("Cleaning complete")
@@ -256,6 +269,7 @@ def clean_paired_fastqs(
         aligner=aligner.name,
         index=index,
         invert=invert,
+        stdout=stdout,
     )
     util.fix_empty_fastqs(stats)
     logging.info("Cleaning complete")
diff --git a/src/hostile/util.py b/src/hostile/util.py
index 4b44146..9c364ae 100644
--- a/src/hostile/util.py
+++ b/src/hostile/util.py
@@ -72,6 +72,8 @@ def handle_alignment_exceptions(exception: subprocess.CalledProcessError) -> Non
     logging.debug(f"stderr: {exception.stderr}")
     alignment_successful = False
     stream_empty = False
+    if "Error, fewer reads in file specified" in exception.stderr:  # Bowtie2
+        raise RuntimeError("fastq1 and fastq2 contain different numbers of reads")
     if 'Failed to read header for "-"' in exception.stderr:
         stream_empty = True
     if "overall alignment rate" in exception.stderr:  # Bowtie2
@@ -83,7 +85,7 @@ def handle_alignment_exceptions(exception: subprocess.CalledProcessError) -> Non
         pass
     else:
         logging.error(
-            f"Hostile encountered a problem. Check available RAM and storage\n"
+            f"Hostile encountered a problem. Details below\n"
             f"pipeline stdout:\n{exception.stdout}\n"
             f"pipeline stderr:\n{exception.stderr}\n"
         )
diff --git a/tests/test_all.py b/tests/test_all.py
index b99c419..dd4cb6a 100644
--- a/tests/test_all.py
+++ b/tests/test_all.py
@@ -656,8 +656,9 @@ def test_stats_options(tmp_path):
         rename=True,
         reorder=True,
         force=True,
+        stdout=True,
     )
-    assert ["rename", "reorder", "invert"] == stats[0]["options"]
+    assert {"rename", "reorder", "invert", "stdout"} == set(stats[0]["options"])
 
 
 def test_fixing_empty_fastqs_single(tmp_path):
@@ -690,18 +691,17 @@ def test_fixing_empty_fastqs_paired(tmp_path):
 
 def test_mismatched_number_of_reads_bowtie2(tmp_path):
     """This has caused sinister errors in the wild, yet is handled gracefully here"""
-    with pytest.raises(subprocess.CalledProcessError):
+    with pytest.raises(RuntimeError):
         lib.clean_paired_fastqs(
             fastqs=[
                 (
                     data_dir / "sars-cov-2_100_1.fastq.gz",
-                    data_dir / "sars-cov-2_50_2.fastq.gz",
+                    data_dir / "sars-cov-2_1_2.fastq",
                 ),
             ],
             aligner=lib.ALIGNER.bowtie2,
             index=data_dir / "sars-cov-2/sars-cov-2",
-            out_dir=tmp_path,
-            force=True,
+            stdout=True,
         )