From 66e0ba4fa912d58e4150f4f675f34113db79625e Mon Sep 17 00:00:00 2001
From: Y Ethan Guo <ethan.guoyihua@gmail.com>
Date: Fri, 10 Jan 2025 18:26:46 -0800
Subject: [PATCH] [HUDI-8857] Log shortened list of files instead of all files
 in MDT validator to avoid huge log message

---
 .../apache/hudi/common/util/StringUtils.java  | 48 +++++++++++++++++++
 1 file changed, 48 insertions(+)

diff --git a/hudi-io/src/main/java/org/apache/hudi/common/util/StringUtils.java b/hudi-io/src/main/java/org/apache/hudi/common/util/StringUtils.java
index 2b9f1af11b433..63836a788c4a4 100644
--- a/hudi-io/src/main/java/org/apache/hudi/common/util/StringUtils.java
+++ b/hudi-io/src/main/java/org/apache/hudi/common/util/StringUtils.java
@@ -27,6 +27,7 @@
 import java.util.List;
 import java.util.Map;
 import java.util.stream.Collectors;
+import java.util.stream.IntStream;
 import java.util.stream.Stream;
 
 /**
@@ -369,6 +370,53 @@ public static String concatenateWithThreshold(String a, String b, int byteLength
     return a.substring(0, bestLength) + b;
   }
 
+  /**
+   * Concatenates the string representation of each object in the list
+   * and returns a single string. The result is capped at "lengthThreshold" characters.
+   *
+   * @param objectList      object list
+   * @param lengthThreshold number of characters to cap the string
+   * @param <T>             type of the object
+   *
+   * @return capped string representation of the object list
+   */
+  public static <T> String toStringWithThreshold(List<T> objectList, int lengthThreshold) {
+    if (objectList == null || objectList.isEmpty()) {
+      return "";
+    }
+    StringBuilder sb = new StringBuilder();
+
+    for (Object obj : objectList) {
+      if (sb.length() >= lengthThreshold) {
+        setLastThreeDots(sb);
+        break;
+      }
+
+      String objString = (sb.length() > 0 ? "," : "") + obj;
+
+      // Check if appending this string would exceed the limit
+      if (sb.length() + objString.length() > lengthThreshold) {
+        int remaining = lengthThreshold - sb.length();
+        sb.append(objString, 0, remaining);
+        setLastThreeDots(sb);
+        break;
+      } else {
+        sb.append(objString);
+      }
+    }
+
+    return sb.toString();
+  }
+
+  private static void setLastThreeDots(StringBuilder sb) {
+    IntStream.range(1, 4).forEach(i -> {
+      int loc = sb.length() - i;
+      if (loc >= 0) {
+        sb.setCharAt(loc, '.');
+      }
+    });
+  }
+
   private static int getBestLength(String a, int threshold) {
     // Binary search to find the maximum length of substring of 'a' that fits within maxBytesForA
     int low = 0;