chore: release (#298)

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

CHANGELOG.md

@@ -6,6 +6,17 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.2.3](https://github.com/microsoft/regorus/compare/regorus-v0.2.2...regorus-v0.2.3) - 2024-08-16
+
+### Fixed
+- Match OPA behavior for split ([#295](https://github.com/microsoft/regorus/pull/295))
+- Merge data to init document ([#293](https://github.com/microsoft/regorus/pull/293))
+
+### Other
+- Update cbindgen requirement from 0.26.0 to 0.27.0 ([#296](https://github.com/microsoft/regorus/pull/296))
+- Bump rexml in /bindings/ruby in the bundler group across 1 directory ([#294](https://github.com/microsoft/regorus/pull/294))
+- Update csbindgen requirement from =1.9.0 to =1.9.3 ([#292](https://github.com/microsoft/regorus/pull/292))
+
 ## [0.2.2](https://github.com/microsoft/regorus/compare/regorus-v0.2.1...regorus-v0.2.2) - 2024-07-28
 
 ### Added

Cargo.toml

@@ -12,7 +12,7 @@ members = [
 [package]
 name = "regorus"
 description = "A fast, lightweight Rego (OPA policy language) interpreter"
-version = "0.2.2"
+version = "0.2.3"
 edition = "2021"
 license-file = "LICENSE"
 repository = "https://github.com/microsoft/regorus"

bindings/ffi/Cargo.toml

@@ -20,5 +20,5 @@ coverage = ["regorus/coverage"]
 custom_allocator = []
 
 [build-dependencies]
-cbindgen = "0.26.0"
-csbindgen = "=1.9.0"
+cbindgen = "0.27.0"
+csbindgen = "=1.9.3"

bindings/ruby/Gemfile.lock

@@ -23,7 +23,7 @@ GEM
     rake-compiler-dock (1.5.1)
     rb_sys (0.9.99)
     regexp_parser (2.9.2)
-    rexml (3.3.2)
+    rexml (3.3.3)
       strscan
     rubocop (1.65.0)
       json (~> 2.3)

src/builtins/strings.rs

@@ -146,11 +146,18 @@ 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(
+    // Handle https://github.com/microsoft/regorus/issues/291
+    let parts: Vec<Value> = if delimiter.as_ref() == "" {
+        // If delimiter is "", str::split returns a leading and trailing "" whereas Golang's split doesn't.
+        // Therefore avoid str::split and instead return each char as a Value::String.
+        s.chars().map(|c| Value::from(c.to_string())).collect()
+    } else {
         s.split(delimiter.as_ref())
             .map(|s| Value::String(s.into()))
-            .collect(),
-    ))
+            .collect()
+    };
+
+    Ok(Value::from(parts))
 }
 
 fn to_string(v: &Value, unescape: bool) -> String {

src/engine.rs

@@ -239,7 +239,7 @@ impl Engine {
     /// # }
     /// ```
     pub fn clear_data(&mut self) {
-        self.interpreter.set_data(Value::new_object());
+        self.interpreter.set_init_data(Value::new_object());
         self.prepared = false;
     }
 
@@ -276,7 +276,40 @@ impl Engine {
             bail!("data must be object");
         }
         self.prepared = false;
-        self.interpreter.get_data_mut().merge(data)
+        self.interpreter.get_init_data_mut().merge(data)
+    }
+
+    /// Get the data document.
+    ///
+    /// The returned value is the data document that has been constructed using
+    /// one or more calls to [`Engine::add_data`]. The values of policy rules are
+    /// not included in the returned document.
+    ///
+    ///
+    /// ```
+    /// # use regorus::*;
+    /// # fn main() -> anyhow::Result<()> {
+    /// let mut engine = Engine::new();
+    ///
+    /// // If not set, data document is empty.
+    /// assert_eq!(engine.get_data(), Value::new_object());
+    ///
+    /// // Merge { "x" : 1, "y" : {} }
+    /// assert!(engine.add_data(Value::from_json_str(r#"{ "x" : 1, "y" : {}}"#)?).is_ok());
+    ///
+    /// // Merge { "z" : 2 }
+    /// assert!(engine.add_data(Value::from_json_str(r#"{ "z" : 2 }"#)?).is_ok());
+    ///
+    /// let data = engine.get_data();
+    /// assert_eq!(data["x"], Value::from(1));
+    /// assert_eq!(data["y"], Value::new_object());
+    /// assert_eq!(data["z"], Value::from(2));
+    ///
+    /// # Ok(())
+    /// # }
+    /// ```
+    pub fn get_data(&self) -> Value {
+        self.interpreter.get_init_data().clone()
     }
 
     pub fn add_data_json(&mut self, data_json: &str) -> Result<()> {
@@ -537,11 +570,7 @@ impl Engine {
             self.interpreter.set_modules(&self.modules);
 
             self.interpreter.clear_builtins_cache();
-            // when the interpreter is prepared the initial data is saved
-            // the data will be reset to init_data each time clean_internal_evaluation_state is called
-            let init_data = self.interpreter.get_data_mut().clone();
-            self.interpreter.set_init_data(init_data);
-
+            // clean_internal_evaluation_state will set data to an efficient clont of use supplied init_data
             // Initialize the with-document with initial data values.
             // with-modifiers will be applied to this document.
             self.interpreter.init_with_document()?;

src/interpreter.rs

@@ -216,18 +216,22 @@ impl Interpreter {
         self.modules = modules.to_vec();
     }
 
-    pub fn set_init_data(&mut self, init_data: Value) {
-        self.init_data = init_data;
-    }
-
-    pub fn set_data(&mut self, data: Value) {
-        self.data = data;
-    }
-
     pub fn get_data_mut(&mut self) -> &mut Value {
         &mut self.data
     }
 
+    pub fn set_init_data(&mut self, data: Value) {
+        self.init_data = data;
+    }
+
+    pub fn get_init_data(&self) -> &Value {
+        &self.init_data
+    }
+
+    pub fn get_init_data_mut(&mut self) -> &mut Value {
+        &mut self.init_data
+    }
+
     pub fn set_traces(&mut self, enable_tracing: bool) {
         self.traces = match enable_tracing {
             true => Some(vec![]),

src/tests/interpreter/mod.rs

@@ -422,3 +422,32 @@ fn one_yaml() -> Result<()> {
 fn run(path: &str) {
     yaml_test(path).unwrap()
 }
+
+#[test]
+fn test_get_data() -> Result<()> {
+    let mut engine = Engine::new();
+
+    // Merge { "x" : 1, "y" : {} }
+    engine.add_data(Value::from_json_str(r#"{ "x" : 1, "y" : {}}"#)?)?;
+
+    // Merge { "z" : 2 }
+    engine.add_data(Value::from_json_str(r#"{ "z" : 2 }"#)?)?;
+
+    // Add a policy
+    engine.add_policy("policy.rego".to_string(), "package a".to_string())?;
+
+    // Evaluate virtual data document. The virtual document includes all rules as well.
+    let v_data = engine.eval_query("data".to_string(), false)?.result[0].expressions[0]
+        .value
+        .clone();
+    // There must be an empty package.
+    assert_eq!(v_data["a"], Value::new_object());
+
+    // Get the data document.
+    let data = engine.get_data();
+
+    // There must NOT be any value of `a`.
+    assert_eq!(data["a"], Value::Undefined);
+
+    Ok(())
+}

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: []

tests/kata/main.rs

@@ -172,6 +172,64 @@ struct Cli {
     generate: bool,
 }
 
+fn stateful_policy_test() -> Result<()> {
+    // Create an engine for evaluating Rego policies.
+    let mut engine = regorus::Engine::new();
+
+    let policy = String::from(
+        r#"
+       package example
+       import rego.v1
+
+       default allow := false
+
+       allow if {
+         print("data.allowed_actions = ", data.allowed_actions)
+         input.action in data.allowed_actions["user1"]
+         print("This rule should be allowed")
+       }
+	"#,
+    );
+
+    // Add policy to the engine.
+    engine.add_policy(String::from("policy.rego"), policy)?;
+
+    // Evaluate first input. Expect to evaluate to false, since state is not set
+    engine.set_input(regorus::Value::from_json_str(
+        r#"{
+      "action": "write"
+    }"#,
+    )?);
+
+    let r = engine.eval_bool_query(String::from("data.example.allow"), false)?;
+    println!("Received result: {:?}", r);
+    assert_eq!(r, false);
+
+    // Add data to engine. Set state
+    engine.add_data(regorus::Value::from_json_str(
+        r#"{
+     "allowed_actions": {
+        "user1" : ["read", "write"]
+     }}"#,
+    )?)?;
+
+    // Evaluate second input. Expect to evaluate to true, since state has been set now
+    engine.set_input(regorus::Value::from_json_str(
+        r#"{
+      "action": "write"
+    }"#,
+    )?);
+
+    let r = engine.eval_bool_query(String::from("data.example.allow"), false)?;
+    println!("Received result: {:?}", r);
+    assert_eq!(
+        r, true,
+        "expect result to be true since rule evaluates to true after state has been updated, per rego logs"
+    );
+
+    Ok(())
+}
+
 fn main() -> Result<()> {
     let cli = Cli::parse();
     run_kata_tests(
@@ -179,5 +237,6 @@ fn main() -> Result<()> {
         &cli.name,
         cli.coverage,
         cli.generate,
-    )
+    )?;
+    stateful_policy_test()
 }