From b8c0e08068b62f49c7d44c340c00fc005c66581b Mon Sep 17 00:00:00 2001
From: Alex Groleau <agroleau@teampalmtree.com>
Date: Mon, 21 Apr 2025 17:11:24 -0400
Subject: [PATCH] more filtering

---
 src/lib.rs | 53 +++++++++++++++++++++++++++++++++++------------------
 1 file changed, 35 insertions(+), 18 deletions(-)

diff --git a/src/lib.rs b/src/lib.rs
index 0f56329..3a05d24 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -52,8 +52,8 @@ fn cache_json_schema(schema_id: &str, schema: JsonB) -> JsonB {
           // Collect leaf errors from the meta-schema validation failure
           let mut error_list = Vec::new();
           collect_leaf_errors(src, &mut error_list);
-          // Return the flat list directly
-          json!(error_list)
+          // Filter and deduplicate errors, returning as a single JSON Value (Array)
+          json!(filter_boon_errors(error_list))
         }
         _ => {
           // Keep existing handling for other compilation errors
@@ -91,16 +91,13 @@ fn validate_json_schema(schema_id: &str, instance: JsonB) -> JsonB {
       match cache.schemas.validate(&instance_value, *sch_index) {
         Ok(_) => JsonB(json!({ "success": true })),
         Err(validation_error) => {
-          // Collect all leaf errors first
-          let mut raw_error_list = Vec::new();
-          collect_leaf_errors(&validation_error, &mut raw_error_list);
-
-          // Filter the errors (e.g., deduplicate by instance_path)
-          let filtered_error_list = filter_boon_errors(raw_error_list);
-
+          // Directly use the result of format_validation_error
+          // which now includes the top-level success indicator and flat error list
+          let mut error_list = Vec::new();
+          collect_leaf_errors(&validation_error, &mut error_list);
           JsonB(json!({
             "success": false,
-            "error": filtered_error_list // Return the filtered list
+            "error": filter_boon_errors(error_list) // Filter and deduplicate errors
           }))
         }
       }
@@ -130,28 +127,48 @@ fn collect_leaf_errors(error: &ValidationError, errors_list: &mut Vec<Value>) {
   }
 }
 
-// Filters collected errors, e.g., deduplicating by instance_path
+// Filters collected errors, removing structural noise and then deduplicating by instance_path
 fn filter_boon_errors(raw_errors: Vec<Value>) -> Vec<Value> {
     use std::collections::HashMap;
     use std::collections::hash_map::Entry;
 
-    // Use a HashMap to keep only the first error for each instance_path
-    let mut unique_errors: HashMap<String, Value> = HashMap::new();
+    // Define schema keywords that indicate structural paths, not instance paths
+    let structural_path_segments = [
+        "/allOf/", "/anyOf/", "/oneOf/",
+        "/if/", "/then/", "/else/",
+        "/not/"
+        // Note: "/properties/" and "/items/" are generally valid, 
+        // but might appear spuriously in boon's paths for complex types. 
+        // We exclude only the explicitly logical/combinatorial ones for now.
+    ];
 
-    for error_value in raw_errors {
+    // 1. Filter out errors with instance_paths containing structural segments
+    let plausible_errors: Vec<Value> = raw_errors.into_iter().filter(|error_value| {
+        if let Some(instance_path_value) = error_value.get("instance_path") {
+            if let Some(instance_path_str) = instance_path_value.as_str() {
+                // Keep if NONE of the structural segments are present
+                !structural_path_segments.iter().any(|&segment| instance_path_str.contains(segment))
+            } else {
+                false // Invalid instance_path type, filter out
+            }
+        } else {
+            false // No instance_path field, filter out
+        }
+    }).collect();
+
+    // 2. Deduplicate the remaining plausible errors by instance_path
+    let mut unique_errors: HashMap<String, Value> = HashMap::new();
+    for error_value in plausible_errors {
         if let Some(instance_path_value) = error_value.get("instance_path") {
             if let Some(instance_path_str) = instance_path_value.as_str() {
-                // Use Entry API to insert only if the key is not present
                 if let Entry::Vacant(entry) = unique_errors.entry(instance_path_str.to_string()) {
                     entry.insert(error_value);
                 }
             }
         }
-        // If error doesn't have instance_path or it's not a string, we might ignore it or handle differently.
-        // For now, we implicitly ignore errors without a valid string instance_path for deduplication.
     }
 
-    // Collect the unique errors from the map values
+    // Collect the unique errors
     unique_errors.into_values().collect()
 }