test: more merge helper/merge main testing

tests/shared_data.py

@@ -0,0 +1,17 @@
+import pathlib
+
+TEST_BIN_DIR = pathlib.Path(__file__).parent / "bin"
+TEST_DATA_DIR = pathlib.Path(__file__).parent / "data"
+TEST_OUT_DIR = pathlib.Path(__file__).parent / "out"
+
+BIGWIG_TO_BEDGRAPH_BIN = TEST_BIN_DIR / "bigWigToBedGraph.linux.x86_64"
+BIGWIG_MERGE_PLUS_BIN = TEST_BIN_DIR / "bigWigMergePlus"
+
+INPUT_FILES = (
+    TEST_DATA_DIR / "25574.Blueprint.ERS487305.RNA-Seq.signal_reverse.bigWig",
+    TEST_DATA_DIR / "25575.Blueprint.ERS487305.RNA-Seq.signal_forward.bigWig",
+    TEST_DATA_DIR / "25582.Blueprint.ERS487306.RNA-Seq.signal_reverse.bigWig",
+)
+INPUT_FILES_STR = [*map(str, INPUT_FILES)]
+
+TEST_POS = "chr1:0-500000"

tests/test_against_bwmp.py

@@ -1,24 +1,9 @@
 # Test against the software this replaces: bigWigMergePlus
-import pathlib
 import subprocess
 
 from bw_merge_window.entry import entry
 
-TEST_BIN_DIR = pathlib.Path(__file__).parent / "bin"
-TEST_DATA_DIR = pathlib.Path(__file__).parent / "data"
-TEST_OUT_DIR = pathlib.Path(__file__).parent / "out"
-
-BIGWIG_TO_BEDGRAPH_BIN = TEST_BIN_DIR / "bigWigToBedGraph.linux.x86_64"
-BIGWIG_MERGE_PLUS_BIN = TEST_BIN_DIR / "bigWigMergePlus"
-
-INPUT_FILES = [
-    TEST_DATA_DIR / "25574.Blueprint.ERS487305.RNA-Seq.signal_reverse.bigWig",
-    TEST_DATA_DIR / "25575.Blueprint.ERS487305.RNA-Seq.signal_forward.bigWig",
-    TEST_DATA_DIR / "25582.Blueprint.ERS487306.RNA-Seq.signal_reverse.bigWig",
-]
-INPUT_FILES_STR = [*map(str, INPUT_FILES)]
-
-TEST_POS = "chr1:0-500000"
+from . import shared_data as sd
 
 
 def parse_bedgraph_line(line: str) -> tuple[str, int, int, float]:
@@ -27,34 +12,34 @@ def parse_bedgraph_line(line: str) -> tuple[str, int, int, float]:
 
 
 def test_against_bwmp():
-    out_new = str(TEST_OUT_DIR / "merged-new.bw")
-    out_new_bedgraph = str(TEST_OUT_DIR / "merged-new.bedGraph")
+    out_new = str(sd.TEST_OUT_DIR / "merged-new.bw")
+    out_new_bedgraph = str(sd.TEST_OUT_DIR / "merged-new.bedGraph")
     entry(
         (
-            TEST_POS,
-            *INPUT_FILES_STR,
+            sd.TEST_POS,
+            *sd.INPUT_FILES_STR,
             "--treat-missing-as-zero",
             "--range",
             "0-1000",
             "--output",
             out_new,
         )
     )
-    subprocess.run((str(BIGWIG_TO_BEDGRAPH_BIN), out_new, out_new_bedgraph))
+    subprocess.run((str(sd.BIGWIG_TO_BEDGRAPH_BIN), out_new, out_new_bedgraph))
 
-    out_old = str(TEST_OUT_DIR / "merged-old.bw")
-    out_old_bedgraph = str(TEST_OUT_DIR / "merged-old.bedGraph")
+    out_old = str(sd.TEST_OUT_DIR / "merged-old.bw")
+    out_old_bedgraph = str(sd.TEST_OUT_DIR / "merged-old.bedGraph")
     subprocess.run(
         (
-            str(BIGWIG_MERGE_PLUS_BIN),
+            str(sd.BIGWIG_MERGE_PLUS_BIN),
             "-range=0-1000",
             "-compress",
-            f"-position={TEST_POS}",
-            *INPUT_FILES_STR,
-            str(TEST_OUT_DIR / "merged-old.bw"),
+            f"-position={sd.TEST_POS}",
+            *sd.INPUT_FILES_STR,
+            str(sd.TEST_OUT_DIR / "merged-old.bw"),
         )
     )
-    subprocess.run((str(BIGWIG_TO_BEDGRAPH_BIN), out_old, out_old_bedgraph))
+    subprocess.run((str(sd.BIGWIG_TO_BEDGRAPH_BIN), out_old, out_old_bedgraph))
 
     with open(out_old_bedgraph, "r") as fho:
         fho_lines = fho.readlines()

tests/test_merge_helpers.py

@@ -1,8 +1,11 @@
 import logging
+import pyBigWig
 import pytest
 
 from bw_merge_window import merge as m
 
+from . import shared_data as sd
+
 logger = logging.getLogger(__name__)
 
 
@@ -15,12 +18,24 @@ def test_int_cast():
 def test_parse_window():
     assert m.get_window_contig_start_end("chr1:0-10000", logger) == ("chr1", 0, 10000)
     assert m.get_window_contig_start_end("chr1:-5-10000", logger) == ("chr1", 0, 10000)
+    assert m.get_window_contig_start_end("chr1:0-", logger) == ("chr1", 0, None)
+    assert m.get_window_contig_start_end("chr1", logger) == ("chr1", 0, None)
 
     with pytest.raises(ValueError):
         m.get_window_contig_start_end("chr1:x-y", logger)
 
+    with pytest.raises(ValueError):
+        m.get_window_contig_start_end("", logger)
+
+    with pytest.raises(ValueError):
+        m.get_window_contig_start_end("chr1:-", logger)
+
+    with pytest.raises(ValueError):
+        m.get_window_contig_start_end("chr1:-10000", logger)
+
 
 def test_parse_range():
+    assert m.get_range_values(None) is None
     assert m.get_range_values("0-1000") == (0, 1000)
     assert m.get_range_values("-5-1000") == (-5, 1000)
     assert m.get_range_values("500-1000") == (500, 1000)
@@ -34,3 +49,40 @@ def test_parse_range():
 
     with pytest.raises(ValueError):
         m.get_range_values("10-5")
+
+
+def test_open_handles():
+    assert m.get_bigwig_file_handles(()) == ()
+
+    fhs = m.get_bigwig_file_handles(sd.INPUT_FILES)  # TODO: fixture
+    try:
+        assert len(fhs) == len(sd.INPUT_FILES)
+    finally:
+        for fh in fhs:
+            fh.close()
+
+
+def test_contig_presence():
+    h = pyBigWig.open(str(sd.INPUT_FILES[0]))
+
+    try:
+        assert m.verify_contig_presence_and_get_length(h, 0, "chr1") == 249250621  # GRCh37 chr1 length
+
+        with pytest.raises(ValueError):
+            m.verify_contig_presence_and_get_length(h, 0, "chrDNE")
+
+    finally:
+        h.close()
+
+
+def test_contig_consensus_length():
+    with pytest.raises(ValueError) as e:
+        m.get_consensus_contig_length_or_raise((), "chr1")
+    assert "At least one bigWig file must be specified" in str(e)
+
+    fhs = m.get_bigwig_file_handles(sd.INPUT_FILES)
+    try:
+        assert m.get_consensus_contig_length_or_raise(fhs, "chr1") == 249250621  # GRCh37 chr1 length
+    finally:
+        for fh in fhs:
+            fh.close()

tests/test_merge_main.py

@@ -0,0 +1,24 @@
+import pytest
+
+from bw_merge_window.entry import entry
+
+from . import shared_data as sd
+
+ALL_TEST_ARGS = (
+    (sd.TEST_POS, *sd.INPUT_FILES_STR),
+    (sd.TEST_POS, *sd.INPUT_FILES_STR, "--range", "0-1000"),
+    (sd.TEST_POS, *sd.INPUT_FILES_STR, "--treat-missing-as-zero"),
+    (sd.TEST_POS, *sd.INPUT_FILES_STR, "--treat-missing-as-zero", "--range", "0-1000"),
+    (sd.TEST_POS, *sd.INPUT_FILES_STR, "--treat-missing-as-zero", "--range=-500-1000"),
+    (sd.TEST_POS, *sd.INPUT_FILES_STR, "--treat-missing-as-zero", "--range", "10-50"),
+    # TODO: these should be tested, but it's too slow...
+    # ("chr1", *sd.INPUT_FILES_STR, "--treat-missing-as-zero"),
+    # ("chr1:0-", *sd.INPUT_FILES_STR, "--treat-missing-as-zero"),
+    # ("chr1:5000-", *sd.INPUT_FILES_STR, "--treat-missing-as-zero"),
+)
+
+
+@pytest.mark.parametrize("args", ALL_TEST_ARGS)
+def test_merge_conditions(args):
+    out_test = str(sd.TEST_OUT_DIR / "merged-test.bw")
+    entry((*args, "--output", out_test))