Update tests

Author: jonashaag

pandas/io/sas/sas.pyx

@@ -23,17 +23,17 @@ cdef struct Buffer:
 
 
 cdef inline uint8_t buf_get(Buffer buf, size_t offset) except? 0:
-    assert offset < buf.length
+    assert offset < buf.length, f"Out of bounds read"
     return buf.data[offset]
 
 
 cdef inline void buf_set(Buffer buf, size_t offset, uint8_t value) except *:
-    assert offset < buf.length
+    assert offset < buf.length, "Out of bounds write"
     buf.data[offset] = value
 
 
 cdef inline bytes buf_as_bytes(Buffer buf, size_t offset, size_t length):
-    assert offset + length <= buf.length
+    assert offset + length <= buf.length, "Out of bounds read"
     return buf.data[offset:offset+length]
 
 
@@ -428,7 +428,7 @@ cdef class Parser:
             object[:, :] string_chunk
             bint compressed
 
-        assert offset + length <= self.cached_page_len
+        assert offset + length <= self.cached_page_len, "Out of bounds read"
         source = Buffer(&self.cached_page[offset], length)
 
         compressed = self.decompress != NULL and length < self.row_length

pandas/tests/io/sas/test_sas7bdat.py

@@ -350,34 +350,20 @@ def test_meta2_page(datapath):
     assert len(df) == 1000
 
 
-@pytest.mark.parametrize("test_file", ["test2.sas7bdat", "test3.sas7bdat"])
-def test_exception_propagation_rdc_rle_decompress(datapath, monkeypatch, test_file):
-    """Errors in RLE/RDC decompression should propagate the same error."""
-    orig_np_zeros = np.zeros
-
-    def _patched_zeros(size, dtype):
-        if isinstance(size, int):
-            # np.zeros() call in {rdc,rle}_decompress
-            raise Exception("Test exception")
-        else:
-            # Other calls to np.zeros
-            return orig_np_zeros(size, dtype)
-
-    monkeypatch.setattr(np, "zeros", _patched_zeros)
-
-    with pytest.raises(Exception, match="^Test exception$"):
-        pd.read_sas(datapath("io", "sas", "data", test_file))
-
-
-def test_exception_propagation_rle_decompress(tmp_path, datapath):
-    """Illegal control byte in RLE decompressor should raise the correct ValueError."""
-    with open(datapath("io", "sas", "data", "test2.sas7bdat"), "rb") as f:
+@pytest.mark.parametrize(
+    "test_file, override_offset, override_value, expected_msg",
+    [
+        ("test2.sas7bdat", 0x10000 + 55229, 0x80 | 0x0F, "Out of bounds"),
+        ("test2.sas7bdat", 0x10000 + 55229, 0x10, "unknown control byte"),
+        ("test3.sas7bdat", 118170, 184, "Out of bounds"),
+    ],
+)
+def test_rle_rdc_exceptions(
+    datapath, test_file, override_offset, override_value, expected_msg
+):
+    """Errors in RLE/RDC decompression should propagate."""
+    with open(datapath("io", "sas", "data", test_file), "rb") as f:
         data = bytearray(f.read())
-    invalid_control_byte = 0x10
-    page_offset = 0x10000
-    control_byte_pos = 55229
-    data[page_offset + control_byte_pos] = invalid_control_byte
-    tmp_file = tmp_path / "test2.sas7bdat"
-    tmp_file.write_bytes(data)
-    with pytest.raises(ValueError, match="unknown control byte"):
-        pd.read_sas(tmp_file)
+    data[override_offset] = override_value
+    with pytest.raises(Exception, match=expected_msg):
+        pd.read_sas(io.BytesIO(data), format="sas7bdat")