From 6c61445b3ff24204fde41867f231fa17df04d05d Mon Sep 17 00:00:00 2001
From: Anand Krishnamoorthi <anakrish@microsoft.com>
Date: Thu, 8 Aug 2024 09:39:12 -0700
Subject: [PATCH] fix: Match OPA behavior for split

In case of empty delimiter, Rust's split returns leading and trailing
empty strings whereas Golang's doesn't.
Change behavior to match Golang/OPA.

fixes #291

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>
---
 src/builtins/strings.rs                       | 31 ++++++++++++++++---
 .../cases/builtins/strings/split.yaml         | 17 ++++++++++
 2 files changed, 43 insertions(+), 5 deletions(-)
 create mode 100644 tests/interpreter/cases/builtins/strings/split.yaml

diff --git a/src/builtins/strings.rs b/src/builtins/strings.rs
index 611d0028..15eb0ca1 100644
--- a/src/builtins/strings.rs
+++ b/src/builtins/strings.rs
@@ -146,11 +146,32 @@ fn split(span: &Span, params: &[Ref<Expr>], args: &[Value], _strict: bool) -> Re
     let s = ensure_string(name, &params[0], &args[0])?;
     let delimiter = ensure_string(name, &params[1], &args[1])?;
 
-    Ok(Value::from_array(
-        s.split(delimiter.as_ref())
-            .map(|s| Value::String(s.into()))
-            .collect(),
-    ))
+    let mut parts: Vec<Value> = s
+        .split(delimiter.as_ref())
+        .map(|s| Value::String(s.into()))
+        .collect();
+
+    // Handle https://github.com/microsoft/regorus/issues/291
+    // If delimiter is "", rust returns a leading and trailing "", Golang doesn't.
+    if delimiter.as_ref() == "" {
+        if parts.len() >= 2 {
+            if let ("", "") = (
+                parts[0].as_string()?.as_ref(),
+                parts[parts.len() - 1].as_string()?.as_ref(),
+            ) {
+                // Remove trailing and leading "".
+                parts.pop();
+                parts.remove(0);
+                return Ok(Value::from_array(parts));
+            }
+        }
+
+        bail!(span.error(
+            "split with empty delimiter did not return leading and trailing empty delimiters"
+        ));
+    }
+
+    Ok(Value::from_array(parts))
 }
 
 fn to_string(v: &Value, unescape: bool) -> String {
diff --git a/tests/interpreter/cases/builtins/strings/split.yaml b/tests/interpreter/cases/builtins/strings/split.yaml
new file mode 100644
index 00000000..a71444a3
--- /dev/null
+++ b/tests/interpreter/cases/builtins/strings/split.yaml
@@ -0,0 +1,17 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+
+cases:
+  - note: empty separator
+    data: {}
+    modules: []
+    query: "x := split(\"test\", \"\")"
+    want_result:
+      x: ["t", "e", "s", "t"]
+      
+  - note: empty separator, empty string
+    data: {}
+    modules: []
+    query: "x := split(\"\", \"\")"
+    want_result:
+      x: []