From ae40547464425dc1d278938a28d4d698c0b12c0b Mon Sep 17 00:00:00 2001
From: Markus Klein <markus-klein@live.de>
Date: Tue, 19 Dec 2023 17:04:11 +0100
Subject: [PATCH] do not map empty string sto null

---
 src/reader/text.rs   |  6 ++++++
 tests/integration.rs | 15 +++++++++++----
 2 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/src/reader/text.rs b/src/reader/text.rs
index c8e62e5..7fbdc54 100644
--- a/src/reader/text.rs
+++ b/src/reader/text.rs
@@ -1,6 +1,7 @@
 use std::{char::decode_utf16, cmp::min, ffi::CStr, num::NonZeroUsize, sync::Arc};
 
 use arrow::array::{ArrayRef, StringBuilder};
+use log::warn;
 use odbc_api::{
     buffers::{AnySlice, BufferDesc},
     DataType as OdbcDataType,
@@ -59,6 +60,11 @@ fn narrow_text_strategy(
     assume_indicators_are_memory_garbage: bool,
 ) -> Box<dyn ReadStrategy> {
     if assume_indicators_are_memory_garbage {
+        warn!(
+            "Ignoring indicators, because we expect the ODBC driver of your database to return \
+            garbage memory. We can not distiguish between empty strings and NULL. Everything is \
+            empty."
+        );
         Box::new(NarrowUseTerminatingZero::new(octet_len))
     } else {
         Box::new(NarrowText::new(octet_len))
diff --git a/tests/integration.rs b/tests/integration.rs
index 3c7b459..d07cab1 100644
--- a/tests/integration.rs
+++ b/tests/integration.rs
@@ -308,7 +308,11 @@ fn fetch_varchar() {
 #[test]
 fn fetch_varchar_using_terminating_zeroes_to_indicate_string_length() {
     let table_name = function_name!().rsplit_once(':').unwrap().1;
-    let cursor = cursor_over(table_name, "VARCHAR(50)", "('Hello'),('Bonjour'),(NULL)");
+    let cursor = cursor_over(
+        table_name,
+        "VARCHAR(50)",
+        "('Hello'),('Bonjour'),(NULL)",
+    );
 
     let mut quirks = Quirks::new();
     quirks.indicators_returned_from_bulk_fetch_are_memory_garbage = true;
@@ -331,9 +335,12 @@ fn fetch_varchar_using_terminating_zeroes_to_indicate_string_length() {
     assert_eq!("Bonjour", array_vals.value(1));
 
     // This workaround is currently only active for UTF-8. Which in turn is only active on Linux
-    #[cfg(target_os="windows")]
+    #[cfg(target_os = "windows")]
     assert!(array_vals.is_null(2));
-    #[cfg(not(target_os="windows"))]
+    // Due to the ambiguity between empty and NULL we map everything to empty. This can not be
+    // mapped to NULL, due to the fact, that the schema might be a mandatory column. The
+    // representation is ambigious, because we need to ignore the indicator buffer.
+    #[cfg(not(target_os = "windows"))]
     assert_eq!("", array_vals.value(0));
 }
 
@@ -345,7 +352,7 @@ fn fetch_varchar_using_terminating_zeroes_to_indicate_string_length() {
 #[test]
 fn fetch_empty_string_from_non_null_varchar_using_terminating_zeroes_to_indicate_string_length() {
     let table_name = function_name!().rsplit_once(':').unwrap().1;
-    let cursor = cursor_over(table_name, "VARCHAR(50)", "('')");
+    let cursor = cursor_over(table_name, "VARCHAR(50) NOT NULL", "('')");
 
     let mut quirks = Quirks::new();
     quirks.indicators_returned_from_bulk_fetch_are_memory_garbage = true;