Override bytes_to_string (#304)

Override `bytes_to_string` to validate utf-8

---------

Signed-off-by: Sri Krishna <[email protected]>

Author: srikrsna-buf

conformance/expected-failures.yaml

@@ -1,9 +0,0 @@
-standard_rules/bytes:
-  - pattern/invalid/not_utf8
-#        input: [ type.googleapis.com/buf.validate.conformance.cases.BytesPattern ]:{val:"\x99"}
-#        want: runtime error: value must be valid UTF-8 to apply regexp
-#        got: validation error (1 violation)
-#         1. rule_id: "bytes.pattern"
-#           message: "value must match regex pattern `^[\\x00-\\x7F]+$`"
-#           field: "val" elements:{field_number:1 field_name:"val" field_type:TYPE_BYTES}
-#           rule: "bytes.pattern" elements:{field_number:15 field_name:"bytes" field_type:TYPE_MESSAGE} elements:{field_number:4 field_name:"pattern" field_type:TYPE_STRING}

src/main/java/build/buf/protovalidate/CustomOverload.java

@@ -17,6 +17,8 @@
 import com.google.protobuf.ByteString;
 import com.google.protobuf.Descriptors;
 import com.google.protobuf.Message;
+import dev.cel.common.CelErrorCode;
+import dev.cel.common.CelRuntimeException;
 import dev.cel.common.types.CelType;
 import dev.cel.common.types.SimpleType;
 import dev.cel.runtime.CelEvaluationException;
@@ -47,6 +49,7 @@ static List<CelFunctionBinding> create() {
     ArrayList<CelFunctionBinding> bindings = new ArrayList<>();
     bindings.addAll(
         Arrays.asList(
+            celBytesToString(),
             celGetField(),
             celFormat(),
             celStartsWithBytes(),
@@ -70,6 +73,26 @@ static List<CelFunctionBinding> create() {
     return Collections.unmodifiableList(bindings);
   }
 
+  /**
+   * This implementes that standard {@code bytes_to_string} function. We override it because the CEL
+   * library doesn't validate that the bytes are valid utf-8.
+   *
+   * <p>Workround until https://github.com/google/cel-java/pull/717 lands.
+   */
+  private static CelFunctionBinding celBytesToString() {
+    return CelFunctionBinding.from(
+        "bytes_to_string",
+        ByteString.class,
+        v -> {
+          if (!v.isValidUtf8()) {
+            throw new CelRuntimeException(
+                new IllegalArgumentException("invalid UTF-8 in bytes, cannot convert to string"),
+                CelErrorCode.BAD_FORMAT);
+          }
+          return v.toStringUtf8();
+        });
+  }
+
   /**
    * Creates a custom function overload for the "getField" operation.
    *

src/main/java/build/buf/protovalidate/ValidateLibrary.java

@@ -22,6 +22,9 @@
 import dev.cel.parser.CelStandardMacro;
 import dev.cel.runtime.CelRuntimeBuilder;
 import dev.cel.runtime.CelRuntimeLibrary;
+import dev.cel.runtime.CelStandardFunctions;
+import dev.cel.runtime.CelStandardFunctions.StandardFunction;
+import dev.cel.runtime.CelStandardFunctions.StandardFunction.Overload.Conversions;
 
 /**
  * Custom {@link CelCompilerLibrary} and {@link CelRuntimeLibrary}. Provides all the custom
@@ -54,6 +57,17 @@ public void setCheckerOptions(CelCheckerBuilder checkerBuilder) {
 
   @Override
   public void setRuntimeOptions(CelRuntimeBuilder runtimeBuilder) {
-    runtimeBuilder.addFunctionBindings(CustomOverload.create());
+    runtimeBuilder
+        .addFunctionBindings(CustomOverload.create())
+        .setStandardEnvironmentEnabled(false)
+        .setStandardFunctions(
+            CelStandardFunctions.newBuilder()
+                .filterFunctions(
+                    // CEL doesn't validate, that the bytes are valid utf-8, so we provide out own
+                    // implementation.
+                    (function, overload) ->
+                        function != StandardFunction.STRING
+                            || !overload.equals(Conversions.BYTES_TO_STRING))
+                .build());
   }
 }