Merge branch 'branch-24.10' into pq-writer-dict-benchmark

cpp/src/io/parquet/compact_protocol_reader.cpp

@@ -140,6 +140,7 @@ struct parquet_field_bool_list : public parquet_field_list<bool, FieldType::BOOL
     auto const read_value = [&val = v](uint32_t i, CompactProtocolReader* cpr) {
       auto const current_byte = cpr->getb();
       assert_bool_field_type(current_byte);
+      CUDF_EXPECTS(i < val.size(), "Index out of bounds");
       val[i] = current_byte == static_cast<int>(FieldType::BOOLEAN_TRUE);
     };
     bind_read_func(read_value);
@@ -189,6 +190,7 @@ struct parquet_field_int_list : public parquet_field_list<T, EXPECTED_TYPE> {
   parquet_field_int_list(int f, std::vector<T>& v) : parquet_field_list<T, EXPECTED_TYPE>(f, v)
   {
     auto const read_value = [&val = v](uint32_t i, CompactProtocolReader* cpr) {
+      CUDF_EXPECTS(i < val.size(), "Index out of bounds");
       val[i] = cpr->get_zigzag<T>();
     };
     this->bind_read_func(read_value);
@@ -233,6 +235,7 @@ struct parquet_field_string_list : public parquet_field_list<std::string, FieldT
       auto const l = cpr->get_u32();
       CUDF_EXPECTS(l < static_cast<size_t>(cpr->m_end - cpr->m_cur), "string length mismatch");
 
+      CUDF_EXPECTS(i < val.size(), "Index out of bounds");
       val[i].assign(reinterpret_cast<char const*>(cpr->m_cur), l);
       cpr->m_cur += l;
     };
@@ -270,6 +273,7 @@ struct parquet_field_enum_list : public parquet_field_list<Enum, FieldType::I32>
     : parquet_field_list<Enum, FieldType::I32>(f, v)
   {
     auto const read_value = [&val = v](uint32_t i, CompactProtocolReader* cpr) {
+      CUDF_EXPECTS(i < val.size(), "Index out of bounds");
       val[i] = static_cast<Enum>(cpr->get_i32());
     };
     this->bind_read_func(read_value);
@@ -355,6 +359,7 @@ struct parquet_field_struct_list : public parquet_field_list<T, FieldType::STRUC
     : parquet_field_list<T, FieldType::STRUCT>(f, v)
   {
     auto const read_value = [&val = v](uint32_t i, CompactProtocolReader* cpr) {
+      CUDF_EXPECTS(i < val.size(), "Index out of bounds");
       cpr->read(&val[i]);
     };
     this->bind_read_func(read_value);
@@ -399,6 +404,7 @@ struct parquet_field_binary_list
       auto const l = cpr->get_u32();
       CUDF_EXPECTS(l <= static_cast<size_t>(cpr->m_end - cpr->m_cur), "binary length mismatch");
 
+      CUDF_EXPECTS(i < val.size(), "Index out of bounds");
       val[i].resize(l);
       val[i].assign(cpr->m_cur, cpr->m_cur + l);
       cpr->m_cur += l;

python/cudf/cudf/_lib/csv.pyx

@@ -276,8 +276,10 @@ def read_csv(
                     col_name = df._data.names[index]
                     df._data[col_name] = df._data[col_name].astype(col_dtype)
 
-    if names is not None and len(names) and isinstance(names[0], (int)):
+    if names is not None and len(names) and isinstance(names[0], int):
         df.columns = [int(x) for x in df._data]
+    elif names is None and header == -1 and cudf.get_option("mode.pandas_compatible"):
+        df.columns = [int(x) for x in df._column_names]
 
     # Set index if the index_col parameter is passed
     if index_col is not None and index_col is not False:

python/cudf/cudf/tests/test_csv.py

@@ -2269,3 +2269,11 @@ def test_read_compressed_BOM(tmpdir):
         f.write(buffer)
 
     assert_eq(pd.read_csv(fname), cudf.read_csv(fname))
+
+
+def test_read_header_none_pandas_compat_column_type():
+    data = "1\n2\n"
+    with cudf.option_context("mode.pandas_compatible", True):
+        result = cudf.read_csv(StringIO(data), header=None).columns
+    expected = pd.read_csv(StringIO(data), header=None).columns
+    pd.testing.assert_index_equal(result, expected, exact=True)