Merge pull request #125 from paulsengroup/testing

Improve tests

src/file.cpp

@@ -190,7 +190,7 @@ static nb::dict get_hic_attrs(const hictk::hic::File &hf) {
 
   py_attrs["bin_size"] = hf.resolution();
   py_attrs["format"] = "HIC";
-  py_attrs["format_version"] = hf.version();
+  py_attrs["format-version"] = hf.version();
   py_attrs["assembly"] = hf.assembly();
   py_attrs["format-url"] = "https://github.com/aidenlab/hic-format";
   py_attrs["nbins"] = hf.bins().size();
@@ -301,7 +301,8 @@ void declare_file_class(nb::module_ &m) {
 
   file.def("resolution", &hictk::File::resolution, "Get the bin size in bp.");
   file.def("nbins", &hictk::File::nbins, "Get the total number of bins.");
-  file.def("nchroms", &hictk::File::nchroms, "Get the total number of chromosomes.");
+  file.def("nchroms", &hictk::File::nchroms, nb::arg("include_ALL") = false,
+           "Get the total number of chromosomes.");
 
   file.def("attributes", &file::attributes, "Get file attributes as a dictionary.",
            nb::rv_policy::take_ownership);

test/test_bin_table.py

@@ -61,17 +61,16 @@ def test_getters(self):
             bins.get_id("abc", 100)
 
     @pytest.mark.skipif(
-        not numpy_avail() or not pandas_avail() or not pyarrow_avail(),
-        reason="numpy, pandas, or pyarrow are not available",
+        not pandas_avail() or not pyarrow_avail(),
+        reason="pandas or pyarrow are not available",
     )
     def test_vectorized_getters(self):
-        import numpy as np
 
         chroms = {"chr1": 1000, "chr2": 500}
         bins = hictkpy.BinTable(chroms, 100)
 
-        assert len(bins.get(np.array([1, 1]))) == 2
-        assert len(bins.get_ids(np.array(["chr1", "chr1"]), np.array([1, 1]))) == 2
+        assert len(bins.get([1, 1])) == 2
+        assert len(bins.get_ids(["chr1", "chr1"], [1, 1])) == 2
 
     @pytest.mark.skipif(not pandas_avail() or not pyarrow_avail(), reason="pandas is not available")
     def test_merge(self):

test/test_file_accessors.py

@@ -27,7 +27,7 @@ class TestClass:
     def test_attributes(self, file, resolution):
         f = hictkpy.File(file, resolution)
         assert f.resolution() == 100_000
-        # assert f.nchroms() == 8  # TODO enable after merging https://github.com/paulsengroup/hictk/pull/294
+        assert f.nchroms() == 8
         assert f.nbins() == 1380
 
         assert "chr2L" in f.chromosomes()

test/test_file_creation_cool.py

@@ -31,6 +31,20 @@ def setup_method():
         logging.basicConfig(level="INFO", force=True)
         logging.getLogger().setLevel("INFO")
 
+    def test_accessors(self, file, resolution, tmpdir):
+        bins = hictkpy.File(file, resolution).bins()
+
+        path = tmpdir / "test.cool"
+        w = hictkpy.cooler.FileWriter(path, bins)
+
+        assert str(w).startswith("CoolFileWriter(")
+        assert w.path() == path
+        if resolution is None:
+            assert w.resolution() == 0
+        else:
+            assert w.resolution() == resolution
+        assert w.chromosomes() == bins.chromosomes()
+
     def test_file_creation_thin_pixel(self, file, resolution, tmpdir):
         f = hictkpy.File(file, resolution)
         if f.bins().type() != "fixed":
@@ -39,7 +53,7 @@ def test_file_creation_thin_pixel(self, file, resolution, tmpdir):
         df = f.fetch(join=False).to_df()
         expected_sum = df["count"].sum()
 
-        path = tmpdir / "test1.cool"
+        path = tmpdir / "test.cool"
         w = hictkpy.cooler.FileWriter(path, f.chromosomes(), f.resolution())
 
         chunk_size = 1000
@@ -67,7 +81,7 @@ def test_file_creation(self, file, resolution, tmpdir):
         df = f.fetch(join=True).to_df()
         expected_sum = df["count"].sum()
 
-        path = tmpdir / "test2.cool"
+        path = tmpdir / "test.cool"
         w = hictkpy.cooler.FileWriter(path, f.chromosomes(), f.resolution())
 
         chunk_size = 1000
@@ -93,7 +107,7 @@ def test_file_creation_bin_table(self, file, resolution, tmpdir):
         df = f.fetch(join=True).to_df()
         expected_sum = df["count"].sum()
 
-        path = tmpdir / "test2.cool"
+        path = tmpdir / "test.cool"
         w = hictkpy.cooler.FileWriter(path, f.bins())
 
         chunk_size = 1000
@@ -122,7 +136,7 @@ def test_file_creation_float_counts(self, file, resolution, tmpdir):
         df["count"] += 0.12345
         expected_sum = df["count"].sum()
 
-        path = tmpdir / "test3.cool"
+        path = tmpdir / "test.cool"
         w = hictkpy.cooler.FileWriter(path, f.chromosomes(), f.resolution())
 
         chunk_size = 1000

test/test_file_validators.py

@@ -4,8 +4,6 @@
 
 import pathlib
 
-import pytest
-
 import hictkpy
 
 testdir = pathlib.Path(__file__).resolve().parent
@@ -17,7 +15,7 @@
 
 
 class TestClass:
-    def test_validators(self):
+    def test_valid_formats(self):
         assert hictkpy.is_cooler(cool_file)
         assert not hictkpy.is_cooler(hic_file)
 
@@ -26,3 +24,31 @@ def test_validators(self):
 
         assert hictkpy.is_scool_file(scool_file)
         assert not hictkpy.is_scool_file(cool_file)
+
+        assert hictkpy.is_hic(hic_file)
+        assert not hictkpy.is_hic(cool_file)
+
+    def test_invalid_formats(self):
+        path = pathlib.Path(__file__).resolve()
+
+        assert not hictkpy.is_cooler(path)
+        assert not hictkpy.is_mcool_file(path)
+        assert not hictkpy.is_scool_file(path)
+        assert not hictkpy.is_hic(path)
+
+    def test_invalid_files(self):
+        non_existing_file = testdir / "foobar.123"
+        assert not non_existing_file.exists()
+
+        assert not hictkpy.is_cooler(non_existing_file)
+        assert not hictkpy.is_mcool_file(non_existing_file)
+        assert not hictkpy.is_scool_file(non_existing_file)
+        assert not hictkpy.is_hic(non_existing_file)
+
+        folder = testdir
+        assert folder.is_dir()
+
+        assert not hictkpy.is_cooler(folder)
+        assert not hictkpy.is_mcool_file(folder)
+        assert not hictkpy.is_scool_file(folder)
+        assert not hictkpy.is_hic(folder)

test/test_singlecell_file_accessors.py

@@ -19,13 +19,19 @@
 
 
 class TestClass:
-    def test_attributes(self, file):
+    def test_accessors(self, file):
         f = hictkpy.cooler.SingleCellFile(file)
 
+        assert str(f).startswith("SingleCellFile(")
+
         assert f.path() == file
         assert f.resolution() == 100_000
         assert len(f.chromosomes()) == 20
+        assert len(f.bins()) == 26398
         assert len(f.cells()) == 5
 
         assert f.attributes()["format"] == "HDF5::SCOOL"
         assert f["GSM2687248_41669_ACAGTG-R1-DpnII.100000.cool"].resolution() == 100_000
+
+        with pytest.raises(Exception):
+            f["ABC"]  # noqa