fix(web): improve number formatting and history validation

- Use significant figures for very small decimals (< 0.0001)
  to prevent values like 0.0000366 from displaying as "0"
- Add entity ID/URL path validation to filter corrupted history
  entries where entityId doesn't match the URL destination

Author: Mearman

apps/web/src/utils/format-number.ts

@@ -26,6 +26,12 @@ const DECIMAL_PRECISION_FIELD_PATTERNS = ["score", "share", "percentile", "fwci"
  */
 const DECIMAL_PRECISION_DIGITS = 4;
 
+/**
+ * Threshold below which we use significant figures instead of fixed decimals
+ * Values smaller than this would round to 0 with DECIMAL_PRECISION_DIGITS
+ */
+const SMALL_VALUE_THRESHOLD = 0.0001;
+
 /**
  * Check if a field name indicates a year value
  * @param fieldName - The field name to check
@@ -83,6 +89,10 @@ export const formatNumber = (value: number, fieldName?: string): string => {
 
 	// For score/share/percentile fields, preserve decimal precision
 	if (fieldName && isDecimalPrecisionField(fieldName) && !Number.isInteger(value)) {
+		// For very small values, use significant figures to avoid showing "0"
+		if (Math.abs(value) > 0 && Math.abs(value) < SMALL_VALUE_THRESHOLD) {
+			return value.toPrecision(2);
+		}
 		return value.toLocaleString(undefined, {
 			minimumFractionDigits: 0,
 			maximumFractionDigits: DECIMAL_PRECISION_DIGITS,

packages/utils/src/storage/catalogue-db.ts

@@ -1084,13 +1084,40 @@ export class CatalogueService {
     // Runtime cleanup: filter out corrupted entries that may have been created after migration
     // Check both entityId and notes (which contains URL) for corruption patterns
     const urlEncodedPattern = "[object%20Object]";
+
+    // Entity ID prefix to URL path mapping for validation
+    const entityPrefixToPath: Record<string, string> = {
+      W: "/works/",
+      A: "/authors/",
+      I: "/institutions/",
+      S: "/sources/",
+      P: "/publishers/",
+      F: "/funders/",
+      T: "/topics/",
+      C: "/concepts/",
+    };
+
     return entities.filter((entity) => {
       if (entity.entityId.length === 0) return false;
       if (entity.entityId.includes(CORRUPTED_ENTITY_ID_PATTERN)) return false;
       if (entity.entityId.includes(urlEncodedPattern)) return false;
       // Also check notes for corrupted URLs
       if (entity.notes?.includes(CORRUPTED_ENTITY_ID_PATTERN)) return false;
       if (entity.notes?.includes(urlEncodedPattern)) return false;
+
+      // Validate entityId matches URL path (detect mismatch like Work ID with Author URL)
+      const urlMatch = entity.notes?.match(/URL: ([^\n]+)/);
+      if (urlMatch) {
+        const url = urlMatch[1];
+        const entityPrefix = entity.entityId.charAt(0).toUpperCase();
+        const expectedPath = entityPrefixToPath[entityPrefix];
+        // If we have a known prefix and URL doesn't match expected path, filter it out
+        // Skip validation for non-entity pages like /about, /settings
+        if (expectedPath && !url.includes(expectedPath) && !url.startsWith("/about") && !url.startsWith("/settings")) {
+          return false;
+        }
+      }
+
       return true;
     });
   }