Properly propagate exceptions in sas.pyx (pandas-dev#47149)

Author: jonashaag

pandas/io/sas/sas.pyx

@@ -13,7 +13,7 @@ ctypedef unsigned short     uint16_t
 # algorithm.  It is partially documented here:
 #
 # https://cran.r-project.org/package=sas7bdat/vignettes/sas7bdat.pdf
-cdef const uint8_t[:] rle_decompress(int result_length, const uint8_t[:] inbuff):
+cdef const uint8_t[:] rle_decompress(int result_length, const uint8_t[:] inbuff) except *:
 
     cdef:
         uint8_t control_byte, x
@@ -116,7 +116,7 @@ cdef const uint8_t[:] rle_decompress(int result_length, const uint8_t[:] inbuff)
 # rdc_decompress decompresses data using the Ross Data Compression algorithm:
 #
 # http://collaboration.cmc.ec.gc.ca/science/rpn/biblio/ddj/Website/articles/CUJ/1992/9210/ross/ross.htm
-cdef const uint8_t[:] rdc_decompress(int result_length, const uint8_t[:] inbuff):
+cdef const uint8_t[:] rdc_decompress(int result_length, const uint8_t[:] inbuff) except *:
 
     cdef:
         uint8_t cmd
@@ -177,17 +177,14 @@ cdef const uint8_t[:] rdc_decompress(int result_length, const uint8_t[:] inbuff)
             rpos += cnt
 
         # short pattern
-        elif (cmd >= 3) & (cmd <= 15):
+        else:
             ofs = cnt + 3
             ofs += <uint16_t>inbuff[ipos] << 4
             ipos += 1
             for k in range(cmd):
                 outbuff[rpos + k] = outbuff[rpos - <int>ofs + k]
             rpos += cmd
 
-        else:
-            raise ValueError("unknown RDC command")
-
     # In py37 cython/clang sees `len(outbuff)` as size_t and not Py_ssize_t
     if <Py_ssize_t>len(outbuff) != <Py_ssize_t>result_length:
         raise ValueError(f"RDC: {len(outbuff)} != {result_length}\n")
@@ -231,7 +228,7 @@ cdef class Parser:
         int subheader_pointer_length
         int current_page_type
         bint is_little_endian
-        const uint8_t[:] (*decompress)(int result_length, const uint8_t[:] inbuff)
+        const uint8_t[:] (*decompress)(int result_length, const uint8_t[:] inbuff) except *
         object parser
 
     def __init__(self, object parser):
@@ -294,8 +291,8 @@ cdef class Parser:
         self.parser._current_row_in_chunk_index = self.current_row_in_chunk_index
         self.parser._current_row_in_file_index = self.current_row_in_file_index
 
-    cdef bint read_next_page(self):
-        cdef done
+    cdef bint read_next_page(self) except? True:
+        cdef bint done
 
         done = self.parser._read_next_page()
         if done:
@@ -316,7 +313,7 @@ cdef class Parser:
         )
         self.current_page_subheaders_count = self.parser._current_page_subheaders_count
 
-    cdef readline(self):
+    cdef bint readline(self) except? True:
 
         cdef:
             int offset, bit_offset, align_correction
@@ -385,7 +382,7 @@ cdef class Parser:
             else:
                 raise ValueError(f"unknown page type: {self.current_page_type}")
 
-    cdef void process_byte_array_with_data(self, int offset, int length):
+    cdef void process_byte_array_with_data(self, int offset, int length) except *:
 
         cdef:
             Py_ssize_t j

pandas/tests/io/sas/test_sas7bdat.py

@@ -341,3 +341,36 @@ def test_null_date(datapath):
         },
     )
     tm.assert_frame_equal(df, expected)
+
+
+@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:
+        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)