Split tests related to virtualfiles into multiple test files (#3512)

seisman · seisman · commit e49347366086 · 2024-10-15T07:50:15.000+08:00
diff --git a/pygmt/tests/test_clib_inquire_virtualfile.py b/pygmt/tests/test_clib_inquire_virtualfile.py
@@ -0,0 +1,33 @@
+"""
+Test the Session.inquire_virtualfile method.
+"""
+
+from pygmt import clib
+
+
+def test_inquire_virtualfile():
+    """
+    Test that the inquire_virtualfile method returns the correct family.
+
+    Currently, only output virtual files are tested.
+    """
+    with clib.Session() as lib:
+        for family in [
+            "GMT_IS_DATASET",
+            "GMT_IS_DATASET|GMT_VIA_MATRIX",
+            "GMT_IS_DATASET|GMT_VIA_VECTOR",
+        ]:
+            with lib.open_virtualfile(
+                family, "GMT_IS_PLP", "GMT_OUT|GMT_IS_REFERENCE", None
+            ) as vfile:
+                assert lib.inquire_virtualfile(vfile) == lib["GMT_IS_DATASET"]
+
+        for family, geometry in [
+            ("GMT_IS_GRID", "GMT_IS_SURFACE"),
+            ("GMT_IS_IMAGE", "GMT_IS_SURFACE"),
+            ("GMT_IS_CUBE", "GMT_IS_VOLUME"),
+            ("GMT_IS_PALETTE", "GMT_IS_NONE"),
+            ("GMT_IS_POSTSCRIPT", "GMT_IS_NONE"),
+        ]:
+            with lib.open_virtualfile(family, geometry, "GMT_OUT", None) as vfile:
+                assert lib.inquire_virtualfile(vfile) == lib[family]
diff --git a/pygmt/tests/test_clib_virtualfile_from_matrix.py b/pygmt/tests/test_clib_virtualfile_from_matrix.py
@@ -0,0 +1,54 @@
+"""
+Test the Session.virtualfile_from_matrix method.
+"""
+
+import numpy as np
+import pytest
+from pygmt import clib
+from pygmt.helpers import GMTTempFile
+
+
+@pytest.fixture(scope="module", name="dtypes")
+def fixture_dtypes():
+    """
+    List of supported numpy dtypes.
+    """
+    return "int8 int16 int32 int64 uint8 uint16 uint32 uint64 float32 float64".split()
+
+
+@pytest.mark.benchmark
+def test_virtualfile_from_matrix(dtypes):
+    """
+    Test transforming a matrix to virtual file dataset.
+    """
+    shape = (7, 5)
+    for dtype in dtypes:
+        data = np.arange(shape[0] * shape[1], dtype=dtype).reshape(shape)
+        with clib.Session() as lib:
+            with lib.virtualfile_from_matrix(data) as vfile:
+                with GMTTempFile() as outfile:
+                    lib.call_module("info", [vfile, f"->{outfile.name}"])
+                    output = outfile.read(keep_tabs=True)
+            bounds = "\t".join([f"<{col.min():.0f}/{col.max():.0f}>" for col in data.T])
+            expected = f"<matrix memory>: N = {shape[0]}\t{bounds}\n"
+            assert output == expected
+
+
+def test_virtualfile_from_matrix_slice(dtypes):
+    """
+    Test transforming a slice of a larger array to virtual file dataset.
+    """
+    shape = (10, 6)
+    for dtype in dtypes:
+        full_data = np.arange(shape[0] * shape[1], dtype=dtype).reshape(shape)
+        rows = 5
+        cols = 3
+        data = full_data[:rows, :cols]
+        with clib.Session() as lib:
+            with lib.virtualfile_from_matrix(data) as vfile:
+                with GMTTempFile() as outfile:
+                    lib.call_module("info", [vfile, f"->{outfile.name}"])
+                    output = outfile.read(keep_tabs=True)
+            bounds = "\t".join([f"<{col.min():.0f}/{col.max():.0f}>" for col in data.T])
+            expected = f"<matrix memory>: N = {rows}\t{bounds}\n"
+            assert output == expected
diff --git a/pygmt/tests/test_clib_virtualfile_from_stringio.py b/pygmt/tests/test_clib_virtualfile_from_stringio.py
@@ -0,0 +1,106 @@
+"""
+Test the Session.virtualfile_from_stringio method.
+"""
+
+import io
+
+import numpy as np
+from pygmt import clib
+
+
+def _stringio_to_dataset(data: io.StringIO):
+    """
+    A helper function for check the virtualfile_from_stringio method.
+
+    The function does the following:
+
+    1. Creates a virtual file from the input StringIO object.
+    2. Pass the virtual file to the ``read`` module, which reads the virtual file and
+       writes it to another virtual file.
+    3. Reads the output virtual file as a GMT_DATASET object.
+    4. Extracts the header and the trailing text from the dataset and returns it as a
+       string.
+    """
+    with clib.Session() as lib:
+        with (
+            lib.virtualfile_from_stringio(data) as vintbl,
+            lib.virtualfile_out(kind="dataset") as vouttbl,
+        ):
+            lib.call_module("read", args=[vintbl, vouttbl, "-Td"])
+            ds = lib.read_virtualfile(vouttbl, kind="dataset").contents
+
+            output = []
+            table = ds.table[0].contents
+            for segment in table.segment[: table.n_segments]:
+                seg = segment.contents
+                output.append(f"> {seg.header.decode()}" if seg.header else ">")
+                output.extend(np.char.decode(seg.text[: seg.n_rows]))
+        return "\n".join(output) + "\n"
+
+
+def test_virtualfile_from_stringio():
+    """
+    Test the virtualfile_from_stringio method.
+    """
+    data = io.StringIO(
+        "# Comment\n"
+        "H 24p Legend\n"
+        "N 2\n"
+        "S 0.1i c 0.15i p300/12 0.25p 0.3i My circle\n"
+    )
+    expected = (
+        ">\n" "H 24p Legend\n" "N 2\n" "S 0.1i c 0.15i p300/12 0.25p 0.3i My circle\n"
+    )
+    assert _stringio_to_dataset(data) == expected
+
+
+def test_one_segment():
+    """
+    Test the virtualfile_from_stringio method with one segment.
+    """
+    data = io.StringIO(
+        "# Comment\n"
+        "> Segment 1\n"
+        "1 2 3 ABC\n"
+        "4 5 DE\n"
+        "6 7 8   9  FGHIJK LMN OPQ\n"
+        "RSTUVWXYZ\n"
+    )
+    expected = (
+        "> Segment 1\n"
+        "1 2 3 ABC\n"
+        "4 5 DE\n"
+        "6 7 8   9  FGHIJK LMN OPQ\n"
+        "RSTUVWXYZ\n"
+    )
+    assert _stringio_to_dataset(data) == expected
+
+
+def test_multiple_segments():
+    """
+    Test the virtualfile_from_stringio method with multiple segments.
+    """
+    data = io.StringIO(
+        "# Comment line 1\n"
+        "# Comment line 2\n"
+        "> Segment 1\n"
+        "1 2 3 ABC\n"
+        "4 5 DE\n"
+        "6 7 8   9  FG\n"
+        "# Comment line 3\n"
+        "> Segment 2\n"
+        "1 2 3 ABC\n"
+        "4 5 DE\n"
+        "6 7 8   9  FG\n"
+    )
+    expected = (
+        "> Segment 1\n"
+        "1 2 3 ABC\n"
+        "4 5 DE\n"
+        "6 7 8   9  FG\n"
+        "> Segment 2\n"
+        "1 2 3 ABC\n"
+        "4 5 DE\n"
+        "6 7 8   9  FG\n"
+    )
+    assert _stringio_to_dataset(data) == expected
diff --git a/pygmt/tests/test_clib_virtualfile_from_vectors.py b/pygmt/tests/test_clib_virtualfile_from_vectors.py
@@ -0,0 +1,172 @@
+"""
+Test the Session.virtualfile_from_vectors method.
+"""
+
+from importlib.util import find_spec
+
+import numpy as np
+import pandas as pd
+import pytest
+from pygmt import clib
+from pygmt.exceptions import GMTInvalidInput
+from pygmt.helpers import GMTTempFile
+
+
+@pytest.fixture(scope="module", name="dtypes")
+def fixture_dtypes():
+    """
+    List of supported numpy dtypes.
+    """
+    return "int8 int16 int32 int64 uint8 uint16 uint32 uint64 float32 float64".split()
+
+
+@pytest.fixture(scope="module", name="dtypes_pandas")
+def fixture_dtypes_pandas(dtypes):
+    """
+    List of supported pandas dtypes.
+    """
+    dtypes_pandas = dtypes.copy()
+
+    if find_spec("pyarrow") is not None:
+        dtypes_pandas.extend([f"{dtype}[pyarrow]" for dtype in dtypes_pandas])
+
+    return tuple(dtypes_pandas)
+
+
+@pytest.mark.benchmark
+def test_virtualfile_from_vectors(dtypes):
+    """
+    Test the automation for transforming vectors to virtual file dataset.
+    """
+    size = 10
+    for dtype in dtypes:
+        x = np.arange(size, dtype=dtype)
+        y = np.arange(size, size * 2, 1, dtype=dtype)
+        z = np.arange(size * 2, size * 3, 1, dtype=dtype)
+        with clib.Session() as lib:
+            with lib.virtualfile_from_vectors(x, y, z) as vfile:
+                with GMTTempFile() as outfile:
+                    lib.call_module("info", [vfile, f"->{outfile.name}"])
+                    output = outfile.read(keep_tabs=True)
+            bounds = "\t".join([f"<{i.min():.0f}/{i.max():.0f}>" for i in (x, y, z)])
+            expected = f"<vector memory>: N = {size}\t{bounds}\n"
+            assert output == expected
+
+
+@pytest.mark.benchmark
+@pytest.mark.parametrize("dtype", [str, object])
+def test_virtualfile_from_vectors_one_string_or_object_column(dtype):
+    """
+    Test passing in one column with string or object dtype into virtual file dataset.
+    """
+    size = 5
+    x = np.arange(size, dtype=np.int32)
+    y = np.arange(size, size * 2, 1, dtype=np.int32)
+    strings = np.array(["a", "bc", "defg", "hijklmn", "opqrst"], dtype=dtype)
+    with clib.Session() as lib:
+        with lib.virtualfile_from_vectors(x, y, strings) as vfile:
+            with GMTTempFile() as outfile:
+                lib.call_module("convert", [vfile, f"->{outfile.name}"])
+                output = outfile.read(keep_tabs=True)
+        expected = "".join(
+            f"{i}\t{j}\t{k}\n" for i, j, k in zip(x, y, strings, strict=True)
+        )
+        assert output == expected
+
+
+@pytest.mark.parametrize("dtype", [str, object])
+def test_virtualfile_from_vectors_two_string_or_object_columns(dtype):
+    """
+    Test passing in two columns of string or object dtype into virtual file dataset.
+    """
+    size = 5
+    x = np.arange(size, dtype=np.int32)
+    y = np.arange(size, size * 2, 1, dtype=np.int32)
+    # Catch bug in https://github.com/GenericMappingTools/pygmt/pull/2719
+    strings1 = np.array(["a", "bc", "def", "ghij", "klmnolooong"], dtype=dtype)
+    strings2 = np.array(["pqrst", "uvwx", "yz!", "@#", "$"], dtype=dtype)
+    with clib.Session() as lib:
+        with lib.virtualfile_from_vectors(x, y, strings1, strings2) as vfile:
+            with GMTTempFile() as outfile:
+                lib.call_module("convert", [vfile, f"->{outfile.name}"])
+                output = outfile.read(keep_tabs=True)
+        expected = "".join(
+            f"{h}\t{i}\t{j} {k}\n"
+            for h, i, j, k in zip(x, y, strings1, strings2, strict=True)
+        )
+        assert output == expected
+
+
+def test_virtualfile_from_vectors_transpose(dtypes):
+    """
+    Test transforming matrix columns to virtual file dataset.
+    """
+    shape = (7, 5)
+    for dtype in dtypes:
+        data = np.arange(shape[0] * shape[1], dtype=dtype).reshape(shape)
+        with clib.Session() as lib:
+            with lib.virtualfile_from_vectors(*data.T) as vfile:
+                with GMTTempFile() as outfile:
+                    lib.call_module("info", [vfile, "-C", f"->{outfile.name}"])
+                    output = outfile.read(keep_tabs=True)
+            bounds = "\t".join([f"{col.min():.0f}\t{col.max():.0f}" for col in data.T])
+            expected = f"{bounds}\n"
+            assert output == expected
+
+
+def test_virtualfile_from_vectors_diff_size():
+    """
+    Test the function fails for arrays of different sizes.
+    """
+    x = np.arange(5)
+    y = np.arange(6)
+    with clib.Session() as lib:
+        with pytest.raises(GMTInvalidInput):
+            with lib.virtualfile_from_vectors(x, y):
+                pass
+
+
+def test_virtualfile_from_vectors_pandas(dtypes_pandas):
+    """
+    Pass vectors to a dataset using pandas.Series, checking both numpy and pyarrow
+    dtypes.
+    """
+    size = 13
+
+    for dtype in dtypes_pandas:
+        data = pd.DataFrame(
+            data={
+                "x": np.arange(size),
+                "y": np.arange(size, size * 2, 1),
+                "z": np.arange(size * 2, size * 3, 1),
+            },
+            dtype=dtype,
+        )
+        with clib.Session() as lib:
+            with lib.virtualfile_from_vectors(data.x, data.y, data.z) as vfile:
+                with GMTTempFile() as outfile:
+                    lib.call_module("info", [vfile, f"->{outfile.name}"])
+                    output = outfile.read(keep_tabs=True)
+            bounds = "\t".join(
+                [f"<{i.min():.0f}/{i.max():.0f}>" for i in (data.x, data.y, data.z)]
+            )
+            expected = f"<vector memory>: N = {size}\t{bounds}\n"
+            assert output == expected
+
+
+def test_virtualfile_from_vectors_arraylike():
+    """
+    Pass array-like vectors to a dataset.
+    """
+    size = 13
+    x = list(range(0, size, 1))
+    y = tuple(range(size, size * 2, 1))
+    z = range(size * 2, size * 3, 1)
+    with clib.Session() as lib:
+        with lib.virtualfile_from_vectors(x, y, z) as vfile:
+            with GMTTempFile() as outfile:
+                lib.call_module("info", [vfile, f"->{outfile.name}"])
+                output = outfile.read(keep_tabs=True)
+        bounds = "\t".join([f"<{min(i):.0f}/{max(i):.0f}>" for i in (x, y, z)])
+        expected = f"<vector memory>: N = {size}\t{bounds}\n"
+        assert output == expected
diff --git a/pygmt/tests/test_clib_virtualfile_in.py b/pygmt/tests/test_clib_virtualfile_in.py
diff --git a/pygmt/tests/test_clib_virtualfiles.py b/pygmt/tests/test_clib_virtualfiles.py
