(avro-rs) branch main updated: fix: Strip the field `order` attribute from the Parsing Canonical Form (#716)

[kriskras99]

This is an automated email from the ASF dual-hosted git repository.

Kriskras99 pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/avro-rs.git


The following commit(s) were added to refs/heads/main by this push:
     new 771363d  fix: Strip the field `order` attribute from the Parsing 
Canonical Form (#716)
771363d is described below

commit 771363dc8d5b57a113f89ba539109f43202968b7
Author: Wleeaf <ts.solidar...@gmail.com>
AuthorDate: Sun Oct 25 19:30:10 2026 +0300

    fix: Strip the field `order` attribute from the Parsing Canonical Form 
(#716)
    
    The [STRIP] rule keeps only type, name, fields, symbols, items, values
    and size. `order` was kept for record fields, so the canonical form and
    fingerprint of such a schema differed from test vector 026 in the
    specification's schema-tests.txt and from the Java and Python SDKs.
    
    Co-authored-by: Wleeaf <179125857+wle...@users.noreply.github.com>
---
 avro/src/schema/mod.rs | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/avro/src/schema/mod.rs b/avro/src/schema/mod.rs
index cf97bd5..0f4b129 100644
--- a/avro/src/schema/mod.rs
+++ b/avro/src/schema/mod.rs
@@ -1130,6 +1130,7 @@ fn pcf_map(schema: &Map<String, JsonValue>, 
defined_names: &mut HashSet<String>)
             || k == "doc"
             || k == "aliases"
             || k == "logicalType"
+            || k == "order"
         {
             continue;
         }
@@ -2207,6 +2208,32 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn test_canonical_form_strips_field_order() -> TestResult {
+        // Test vector 026 from the specification's 
share/test/data/schema-tests.txt
+        let schema = Schema::parse_str(
+            r#"{
+                "fields": [
+                    {"type": "boolean", "aliases": [], "name": "f1", 
"default": true},
+                    {"order": "descending", "name": "f2", "doc": "Hello", 
"type": "int"}
+                ],
+                "type": "record",
+                "name": "foo"
+            }"#,
+        )?;
+        assert_eq!(
+            schema.canonical_form(),
+            
r#"{"name":"foo","type":"record","fields":[{"name":"f1","type":"boolean"},{"name":"f2","type":"int"}]}"#
+        );
+        // -4860222112080293046 in schema-tests.txt, as little-endian bytes
+        assert_eq!(
+            format!("{}", schema.fingerprint::<Rabin>()),
+            "4a93f457bd058dbc"
+        );
+
+        Ok(())
+    }
+
     #[test]
     fn test_logical_types() -> TestResult {
         let schema = Schema::parse_str(r#"{"type": "int", "logicalType": 
"date"}"#)?;