From c13dc1d1817ad3870ae66022f995056820fae6f1 Mon Sep 17 00:00:00 2001
From: Joe Gilvary <ihavequestions@gilvary.net>
Date: Fri, 17 May 2024 18:45:14 -0400
Subject: [PATCH] NUTCH-3057 - Fix for index-arbitrary plugin improper
 retention and use of calculated value for arbitrary field after an exception

---
 .../arbitrary/ArbitraryIndexingFilter.java    |  3 +
 .../TestArbitraryIndexingFilter.java          | 58 +++++++++++++++++++
 2 files changed, 61 insertions(+)

diff --git a/src/plugin/index-arbitrary/src/java/org/apache/nutch/indexer/arbitrary/ArbitraryIndexingFilter.java b/src/plugin/index-arbitrary/src/java/org/apache/nutch/indexer/arbitrary/ArbitraryIndexingFilter.java
index 7677ef7f8..21a4537bf 100644
--- a/src/plugin/index-arbitrary/src/java/org/apache/nutch/indexer/arbitrary/ArbitraryIndexingFilter.java
+++ b/src/plugin/index-arbitrary/src/java/org/apache/nutch/indexer/arbitrary/ArbitraryIndexingFilter.java
@@ -170,6 +170,7 @@ public NutchDocument filter(NutchDocument doc, Parse parse, Text url,
 
     int cfgCounter = 0;
     while (cfgCounter <  arbitraryAddsCount) {
+      result = null;
       setIndexedConf(conf,cfgCounter);
       cfgCounter++;
       try {
@@ -184,6 +185,7 @@ public NutchDocument filter(NutchDocument doc, Parse parse, Text url,
         LOG.error("Exception preparing reflection tasks. className was {}",
 		 String.valueOf(className));
         e.printStackTrace();
+        continue;
       }
       try {
         constrArgs = new String[userConstrArgs.length + 1];
@@ -207,6 +209,7 @@ public NutchDocument filter(NutchDocument doc, Parse parse, Text url,
           LOG.error("methodArgs[0] was {}", String.valueOf(methodArgs[0]));
         }
         e.printStackTrace();
+	continue;
       }
 
       LOG.debug("{}.{}() returned {} for field {}.", className,
diff --git a/src/plugin/index-arbitrary/src/test/org/apache/nutch/indexer/arbitrary/TestArbitraryIndexingFilter.java b/src/plugin/index-arbitrary/src/test/org/apache/nutch/indexer/arbitrary/TestArbitraryIndexingFilter.java
index 17f31b183..adaecf55c 100644
--- a/src/plugin/index-arbitrary/src/test/org/apache/nutch/indexer/arbitrary/TestArbitraryIndexingFilter.java
+++ b/src/plugin/index-arbitrary/src/test/org/apache/nutch/indexer/arbitrary/TestArbitraryIndexingFilter.java
@@ -219,4 +219,62 @@ public void testOverwritingExistingField() throws Exception {
     Assert.assertTrue("field philosopher does not have new value 'Popeye'", doc.getField("philosopher")
                     .getValues().contains("Popeye"));
   }
+
+  /**
+   * Test processing a field after exception processing earlier field
+   *
+   * @throws Exception
+   */
+  @Test
+  public void testProcessingFieldAfterException() throws Exception {
+    conf = NutchConfiguration.create();
+    conf.set("index.arbitrary.function.count","3");
+    conf.set("index.arbitrary.fieldName.0","foo");
+    conf.set("index.arbitrary.className.0","org.apache.nutch.indexer.arbitrary.Echo");
+    conf.set("index.arbitrary.constructorArgs.0","first added value");
+    conf.set("index.arbitrary.methodName.0","getText");
+
+    conf.set("index.arbitrary.fieldName.1","mangled");
+    conf.set("index.arbitrary.className.1","java.lang.String");
+    conf.set("index.arbitrary.constructorArgs.1","bar");
+    conf.set("index.arbitrary.methodName.1","noExistingMethod");
+    conf.set("index.arbitrary.methodArgs.1","100");
+    conf.set("index.arbitrary.overwrite.1","true");
+
+    conf.set("index.arbitrary.fieldName.2","philosopher");
+    conf.set("index.arbitrary.className.2","org.apache.nutch.indexer.arbitrary.Echo");
+    conf.set("index.arbitrary.constructorArgs.2","last added value");
+    conf.set("index.arbitrary.methodName.2","getText");
+    conf.set("index.arbitrary.overwrite.2","true");
+
+    filter = new ArbitraryIndexingFilter();
+    Assert.assertNotNull("No filter exists for testProcessingFieldAfterException",filter);
+
+    filter.setConf(conf);
+    Assert.assertNotNull("conf does not exist",conf);
+
+    doc = new NutchDocument();
+
+    Assert.assertNotNull("doc does not exist",doc);
+
+    try {
+      filter.filter(doc, parse, url, crawlDatum, inlinks);
+    } catch (Exception e) {
+      e.printStackTrace(System.out);
+      Assert.fail(e.getMessage());
+    }
+
+    Assert.assertNotNull(doc);
+
+    Assert.assertTrue("field foo does not have 'first added value'", doc.getField("foo")
+                      .getValues().contains("first added value"));
+
+    Assert.assertNull("field mangled has a value", doc.getField("mangled"));
+
+    Assert.assertFalse("Value 'first added value' has leaked into field philospoher", doc.getField("philosopher")
+                      .getValues().contains("first added value"));
+
+    Assert.assertTrue("field philosopher does not have new value 'last added value'", doc.getField("philosopher")
+                    .getValues().contains("last added value"));
+  }
 }