feat(py): add origin argument to Compiler.add_source method and i…

…mplement `Compiler.errors` method

py/Cargo.toml

@@ -15,6 +15,7 @@ crate-type = ["cdylib"]
 [dependencies]
 pyo3 = { version = "0.21.2", features = ["abi3", "abi3-py38", "extension-module"] }
 pyo3-file = "0.8.0"
+serde_json = { workspace = true }
 
 protobuf-json-mapping = { workspace = true }
 yara-x = { workspace = true }

py/src/lib.rs

@@ -99,11 +99,36 @@ impl Compiler {
 
     /// Adds a YARA source code to be compiled.
     ///
-    /// This function can be used multiple times before calling [`Compiler::build`].
-    fn add_source(&mut self, src: &str) -> PyResult<()> {
+    /// This function may be invoked multiple times to add several sets of YARA
+    /// rules before calling [`Compiler::build`]. If the rules provided in
+    /// `src` contain errors that prevent compilation, the function will raise
+    /// an exception with the first error encountered. Additionally, the
+    /// compiler will store this error, along with any others discovered during
+    /// compilation, which can be accessed using [`Compiler::errors`].
+    ///
+    /// Even if a previous invocation resulted in a compilation error, you can
+    /// continue calling this function. In such cases, any rules that failed to
+    /// compile will not be included in the final compiled set.
+    ///
+    /// The optional parameter `origin` allows to specify the origin of the
+    /// source code. This usually receives the path of the file from where the
+    /// code was read, but it can be any arbitrary string that conveys information
+    /// about the source code's origin.
+    fn add_source(
+        &mut self,
+        src: &str,
+        origin: Option<String>,
+    ) -> PyResult<()> {
+        let mut src = yrx::SourceCode::from(src);
+
+        if let Some(origin) = origin.as_ref() {
+            src = src.with_origin(origin)
+        }
+
         self.inner
             .add_source(src)
             .map_err(|err| CompileError::new_err(err.to_string()))?;
+
         Ok(())
     }
 
@@ -181,6 +206,20 @@ impl Compiler {
         );
         Rules::new(compiler.build())
     }
+
+    /// Retrieves all errors generated by the compiler.
+    ///
+    /// This method returns every error encountered during the compilation,
+    /// across all invocations of [`Compiler::add_source`].
+    fn errors<'py>(&'py self, py: Python<'py>) -> PyResult<Bound<'py, PyAny>> {
+        let json = PyModule::import_bound(py, "json")?;
+        let json_loads = json.getattr("loads")?;
+        let errors_json = serde_json::to_string_pretty(&self.inner.errors());
+        let errors_json = errors_json
+            .map_err(|err| PyValueError::new_err(err.to_string()))?;
+
+        json_loads.call((errors_json,), None)
+    }
 }
 
 /// Scans data with already compiled YARA rules.

py/tests/test_api.py

@@ -207,6 +207,20 @@ def test_serialization():
   assert len(rules.scan(b'').matching_rules) == 1
 
 
+def tests_compiler_errors():
+  compiler = yara_x.Compiler()
+
+  with pytest.raises(yara_x.CompileError):
+    compiler.add_source('rule foo { condition: bar }')
+
+  errors = compiler.errors()
+
+  assert len(errors) == 1
+  assert errors[0]['type'] == "UnknownIdentifier"
+  assert errors[0]['code'] == "E009"
+  assert errors[0]['title'] == "unknown identifier `bar`"
+
+
 def test_console_log():
   ok = False