From c3513e928141ad9ef0aa4e4fef641c70d0e4b4be Mon Sep 17 00:00:00 2001
From: Steve Rowe <sarowe@apache.org>
Date: Mon, 13 Nov 2017 17:57:24 -0500
Subject: [PATCH] SOLR-11231: Guard against unset fields when performing
 language detection.  This closes #232

---
 solr/CHANGES.txt                               |  3 +++
 .../TikaLanguageIdentifierUpdateProcessor.java | 18 +++++++++++++-----
 ...entifierUpdateProcessorFactoryTestCase.java | 13 +++++++++++++
 3 files changed, 29 insertions(+), 5 deletions(-)

diff --git a/solr/CHANGES.txt b/solr/CHANGES.txt
index d151353e939..f872b969a2e 100644
--- a/solr/CHANGES.txt
+++ b/solr/CHANGES.txt
@@ -105,6 +105,9 @@ Bug Fixes
 
 * SOLR-11619: V2 requests that needed to be forwarded to other nodes would get an NPE. (David Smiley)
 
+* SOLR-11231: Guard against unset fields when performing language detection.
+  (Chris Beer via Steve Rowe)
+
 Optimizations
 ----------------------
 * SOLR-11285: Refactor autoscaling framework to avoid direct references to Zookeeper and Solr
diff --git a/solr/contrib/langid/src/java/org/apache/solr/update/processor/TikaLanguageIdentifierUpdateProcessor.java b/solr/contrib/langid/src/java/org/apache/solr/update/processor/TikaLanguageIdentifierUpdateProcessor.java
index 836a3bf6795..df0e5f7fa25 100644
--- a/solr/contrib/langid/src/java/org/apache/solr/update/processor/TikaLanguageIdentifierUpdateProcessor.java
+++ b/solr/contrib/langid/src/java/org/apache/solr/update/processor/TikaLanguageIdentifierUpdateProcessor.java
@@ -112,13 +112,21 @@ public class TikaLanguageIdentifierUpdateProcessor extends LanguageIdentifierUpd
   private int getExpectedSize(SolrInputDocument doc, String[] fields) {
     int docSize = 0;
     for (String field : fields) {
-      Collection<Object> contents = doc.getFieldValues(field);
-      for (Object content : contents) {
-        if (content instanceof String) {
-          docSize += Math.min(((String) content).length(), maxFieldValueChars);
+      if (doc.containsKey(field)) {
+        Collection<Object> contents = doc.getFieldValues(field);
+        if (contents != null) {
+          for (Object content : contents) {
+            if (content instanceof String) {
+              docSize += Math.min(((String) content).length(), maxFieldValueChars);
+            }
+          }
+
+          if (docSize > maxTotalChars) {
+            docSize = maxTotalChars;
+            break;
+          }
         }
       }
-      docSize = Math.min(docSize, maxTotalChars);
     }
     return docSize;
   }
diff --git a/solr/contrib/langid/src/test/org/apache/solr/update/processor/LanguageIdentifierUpdateProcessorFactoryTestCase.java b/solr/contrib/langid/src/test/org/apache/solr/update/processor/LanguageIdentifierUpdateProcessorFactoryTestCase.java
index c7381a918e7..b90f54a4d3f 100644
--- a/solr/contrib/langid/src/test/org/apache/solr/update/processor/LanguageIdentifierUpdateProcessorFactoryTestCase.java
+++ b/solr/contrib/langid/src/test/org/apache/solr/update/processor/LanguageIdentifierUpdateProcessorFactoryTestCase.java
@@ -211,6 +211,19 @@ public abstract class LanguageIdentifierUpdateProcessorFactoryTestCase extends S
     assertEquals("", liProcessor.process(doc).getFieldValue("language"));
   }
 
+  @Test
+  public void testMissingFieldEmptyString() throws Exception {
+    SolrInputDocument doc;
+    ModifiableSolrParams parameters = new ModifiableSolrParams();
+    parameters.add("langid.fl", "no_such_field");
+    parameters.add("langid.langField", "language");
+    parameters.add("langid.enforceSchema", "false");
+    liProcessor = createLangIdProcessor(parameters);
+
+    doc = new SolrInputDocument();
+    assertEquals("", liProcessor.process(doc).getFieldValue("language"));
+  }
+
   @Test
   public void testFallback() throws Exception {
     SolrInputDocument doc;