From c4eb78dc57d8bb6c78505545810cf4f335f2f265 Mon Sep 17 00:00:00 2001
From: uw4 <uli.walther@gmail.com>
Date: Thu, 18 Jan 2024 12:38:57 +0100
Subject: [PATCH] Fix $split issues. #29 #31

- don't treat string separator as regexp. #29
- preserve trailing empty result strings. #31
- correctly $split("", "") and only split limit characters in first place. #31
- added new tests covering these cases
---
 .../java/com/dashjoin/jsonata/Functions.java  | 14 ++++++-
 .../java/com/dashjoin/jsonata/StringTest.java | 38 +++++++++++++++++++
 2 files changed, 50 insertions(+), 2 deletions(-)

diff --git a/src/main/java/com/dashjoin/jsonata/Functions.java b/src/main/java/com/dashjoin/jsonata/Functions.java
index ec76f2a..b57f739 100644
--- a/src/main/java/com/dashjoin/jsonata/Functions.java
+++ b/src/main/java/com/dashjoin/jsonata/Functions.java
@@ -1061,9 +1061,19 @@ public static List<String> split(String str, Object pattern, Number limit) {
             return result;
 
         if (pattern instanceof String) {
-            result = Arrays.asList( str.split((String)pattern) );
+            String sep = (String)pattern;
+            if (sep.isEmpty()) {
+                // $split("str", ""): Split string into characters
+                int l = limit!=null ? limit.intValue() : Integer.MAX_VALUE;
+                for (int i=0; i<str.length() && i<l; i++) {
+                    result.add( ""+str.charAt(i) );
+                }
+            } else {
+                // Quote separator string + preserve trailing empty strings (-1)
+                result = Arrays.asList( str.split( Pattern.quote(sep), -1) );
+            }
         } else {
-            result = Arrays.asList( str.split(((Pattern)pattern).pattern() ) );
+            result = Arrays.asList( ((Pattern)pattern).split(str, -1) );
         }
         if (limit!=null && limit.intValue()<result.size()) {
             result = result.subList(0, limit.intValue());
diff --git a/src/test/java/com/dashjoin/jsonata/StringTest.java b/src/test/java/com/dashjoin/jsonata/StringTest.java
index dd92401..7756f86 100644
--- a/src/test/java/com/dashjoin/jsonata/StringTest.java
+++ b/src/test/java/com/dashjoin/jsonata/StringTest.java
@@ -55,4 +55,42 @@ public void escapeTest() {
     Assertions.assertEquals("{\"a\":\"</\"}",
         jsonata("$string($)").evaluate(Map.of("a", "</")));
   }
+
+  /**
+   * Additional $split tests
+   */
+  @Test
+  public void splitTest() {
+    Object res;
+
+    // Splitting empty string with empty separator must return empty list
+    res = jsonata("$split('', '')").evaluate(null);
+    Assertions.assertEquals(Arrays.asList(), res);
+
+    // Split characters with limit
+    res = jsonata("$split('a1b2c3d4', '', 4)").evaluate(null);
+    Assertions.assertEquals(Arrays.asList("a", "1", "b", "2"), res);
+
+    // Check string is not treated as regexp
+    res = jsonata("$split('this..is.a.test', '.')").evaluate(null);
+    //System.out.println( Functions.string(res, false));
+    Assertions.assertEquals(Arrays.asList("this","","is","a","test"), res);
+    
+    // Check trailing empty strings
+    res = jsonata("$split('this..is.a.test...', '.')").evaluate(null);
+    //System.out.println( Functions.string(res, false));
+    Assertions.assertEquals(Arrays.asList("this","","is","a","test","","",""), res);
+  
+    // Check trailing empty strings
+    res = jsonata("$split('this..is.a.test...', /\\./)").evaluate(null);
+    Assertions.assertEquals(Arrays.asList("this","","is","a","test","","",""), res);
+
+    // Check string is not treated as regexp, trailing empty strings, and limit
+    res = jsonata("$split('this.*.*is.*a.*test.*.*.*.*.*.*', '.*', 8)").evaluate(null);
+    Assertions.assertEquals(Arrays.asList("this","","is","a","test","","",""), res);
+
+    // Escaped regexp, trailing empty strings, and limit
+    res = jsonata("$split('this.*.*is.*a.*test.*.*.*.*.*.*', /\\.\\*/, 8)").evaluate(null);
+    Assertions.assertEquals(Arrays.asList("this","","is","a","test","","",""), res);
+  }
 }