Normalize NUL to \0 in sqllogictests

[Jefffrey]

Which issue does this PR close?

• Closes Escape NUL characters in sqllogictests #17180

Rationale for this change

What changes are included in this PR?

In the SLT files, change NUL character to \0 string (apparently we already had this for char.slt

Are these changes tested?

Are there any user-facing changes?

[alamb]

It is strange that Github still renders this as binary for me

🤷

Looks like git still treats them as binary too 🤔

diff --git a/datafusion/sqllogictest/src/engines/conversion.rs b/datafusion/sqllogictest/src/engines/conversion.rs
index 92ab64059..de3acbee9 100644
--- a/datafusion/sqllogictest/src/engines/conversion.rs
+++ b/datafusion/sqllogictest/src/engines/conversion.rs
@@ -35,7 +35,8 @@ pub(crate) fn varchar_to_str(value: &str) -> String {
     if value.is_empty() {
         "(empty)".to_string()
     } else {
-        value.trim_end_matches('\n').to_string()
+        // Escape nulls so that github renders them correctly in the webui
+        value.trim_end_matches('\n').replace("\u{0000}", "\\0")
     }
 }

diff --git a/datafusion/sqllogictest/test_files/expr.slt b/datafusion/sqllogictest/test_files/expr.slt
index f0bf5d76e..eeea3cd39 100644
Binary files a/datafusion/sqllogictest/test_files/expr.slt and b/datafusion/sqllogictest/test_files/expr.slt differ
diff --git a/datafusion/sqllogictest/test_files/spark/string/char.slt b/datafusion/sqllogictest/test_files/spark/string/char.slt
index abf0c7076..299e2a041 100644
Binary files a/datafusion/sqllogictest/test_files/spark/string/char.slt and b/datafusion/sqllogictest/test_files/spark/string/char.slt differ

[Jefffrey]

It is strange that Github still renders this as binary for me

🤷

Looks like git still treats them as binary too 🤔

diff --git a/datafusion/sqllogictest/src/engines/conversion.rs b/datafusion/sqllogictest/src/engines/conversion.rs
index 92ab64059..de3acbee9 100644
--- a/datafusion/sqllogictest/src/engines/conversion.rs
+++ b/datafusion/sqllogictest/src/engines/conversion.rs
@@ -35,7 +35,8 @@ pub(crate) fn varchar_to_str(value: &str) -> String {
     if value.is_empty() {
         "(empty)".to_string()
     } else {
-        value.trim_end_matches('\n').to_string()
+        // Escape nulls so that github renders them correctly in the webui
+        value.trim_end_matches('\n').replace("\u{0000}", "\\0")
     }
 }

diff --git a/datafusion/sqllogictest/test_files/expr.slt b/datafusion/sqllogictest/test_files/expr.slt
index f0bf5d76e..eeea3cd39 100644
Binary files a/datafusion/sqllogictest/test_files/expr.slt and b/datafusion/sqllogictest/test_files/expr.slt differ
diff --git a/datafusion/sqllogictest/test_files/spark/string/char.slt b/datafusion/sqllogictest/test_files/spark/string/char.slt
index abf0c7076..299e2a041 100644
Binary files a/datafusion/sqllogictest/test_files/spark/string/char.slt and b/datafusion/sqllogictest/test_files/spark/string/char.slt differ

I think only for this diff, since it still has the NUL character. If I check the file after the change, its back to text:

https://github.com/apache/datafusion/blob/dd224c4a6db4d6e2e68f74160c7577a3b122983f/datafusion/sqllogictest/test_files/expr.slt

[alamb]

Thank you @Jefffrey

In case anyone else is interested, here is the actua diff of the files, as expected:

(venv) andrewlamb@Andrews-MacBook-Pro-3:~/Software/datafusion$ git diff --text 3b8fabdefd0ad0087fc93f727a8f582e1d74f0c5
diff --git a/datafusion/sqllogictest/src/engines/conversion.rs b/datafusion/sqllogictest/src/engines/conversion.rs
index 92ab64059..de3acbee9 100644
--- a/datafusion/sqllogictest/src/engines/conversion.rs
+++ b/datafusion/sqllogictest/src/engines/conversion.rs
@@ -35,7 +35,8 @@ pub(crate) fn varchar_to_str(value: &str) -> String {
     if value.is_empty() {
         "(empty)".to_string()
     } else {
-        value.trim_end_matches('\n').to_string()
+        // Escape nulls so that github renders them correctly in the webui
+        value.trim_end_matches('\n').replace("\u{0000}", "\\0")
     }
 }

diff --git a/datafusion/sqllogictest/test_files/expr.slt b/datafusion/sqllogictest/test_files/expr.slt
index f0bf5d76e..eeea3cd39 100644
--- a/datafusion/sqllogictest/test_files/expr.slt
+++ b/datafusion/sqllogictest/test_files/expr.slt
@@ -427,7 +427,7 @@ NULL
 query T
 SELECT chr(CAST(0 AS int))
 ----
-^@
+\0

 statement error DataFusion error: Execution error: invalid Unicode scalar value: 9223372036854775807
 SELECT chr(CAST(9223372036854775807 AS bigint))
diff --git a/datafusion/sqllogictest/test_files/spark/string/char.slt b/datafusion/sqllogictest/test_files/spark/string/char.slt
index abf0c7076..299e2a041 100644
--- a/datafusion/sqllogictest/test_files/spark/string/char.slt
+++ b/datafusion/sqllogictest/test_files/spark/string/char.slt
@@ -29,6 +29,6 @@ query T
 SELECT char(a) FROM (VALUES (-1::INT), (0::INT), (65::INT), (321::INT)) AS t(a);
 ----
 (empty)
-^@
+\0
 A
 A