SOLR-4864: RegexReplaceProcessorFactory should support pattern capture group substitution in replacement string.

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1586093 13f79535-47bb-0310-9956-ffa450edef68

solr/CHANGES.txt

@@ -175,6 +175,10 @@ New Features
   java -Durl="http://username:password@hostname:8983/solr/update" -jar post.jar sample.xml
   (Sameer Maggon via Uwe Schindler)
 
+* SOLR-4864: RegexReplaceProcessorFactory should support pattern capture group
+  substitution in replacement string.
+  (Sunil Srinivasan, Jack Krupansky via Steve Rowe)
+
 Bug Fixes
 ----------------------
 

solr/core/src/java/org/apache/solr/update/processor/RegexReplaceProcessorFactory.java

@@ -35,12 +35,22 @@ import org.slf4j.LoggerFactory;
 /**
  * An updated processor that applies a configured regex to any 
  * CharSequence values found in the selected fields, and replaces 
- * any matches with the configured replacement string
+ * any matches with the configured replacement string.
+ *
  * <p>
  * By default this processor applies itself to no fields.
  * </p>
  *
  * <p>
+ * By default, <code>literalReplacement</code> is set to true, in which
+ * case, the <code>replacement</code> string will be treated literally by
+ * quoting via {@link Matcher#quoteReplacement(String)}. And hence, '\'
+ * and '$' signs will not be processed. When <code>literalReplacement</code>
+ * is set to false, one can perform backreference operations and capture
+ * group substitutions.
+ * </p>
+ *
+ * <p>
  * For example, with the configuration listed below, any sequence of multiple 
  * whitespace characters found in values for field named <code>title</code> 
  * or <code>content</code> will be replaced by a single space character.
@@ -52,8 +62,9 @@ import org.slf4j.LoggerFactory;
  *   &lt;str name="fieldName"&gt;title&lt;/str&gt;
  *   &lt;str name="pattern"&gt;\s+&lt;/str&gt;
  *   &lt;str name="replacement"&gt; &lt;/str&gt;
+ *   &lt;bool name="literalReplacement"&gt;true&lt;/bool&gt;
  * &lt;/processor&gt;</pre>
- * 
+ *
  * @see java.util.regex.Pattern
  */
 public final class RegexReplaceProcessorFactory extends FieldMutatingUpdateProcessorFactory {
@@ -62,9 +73,12 @@ public final class RegexReplaceProcessorFactory extends FieldMutatingUpdateProce
 
   private static final String REPLACEMENT_PARAM = "replacement";
   private static final String PATTERN_PARAM = "pattern";
+  private static final String LITERAL_REPLACEMENT_PARAM = "literalReplacement";
   
   private Pattern pattern;
   private String replacement;
+  // by default, literalReplacementEnabled is set to true to allow backward compatibility
+  private boolean literalReplacementEnabled = true;
 
   @SuppressWarnings("unchecked")
   @Override
@@ -72,7 +86,7 @@ public final class RegexReplaceProcessorFactory extends FieldMutatingUpdateProce
 
     Object patternParam = args.remove(PATTERN_PARAM);
 
-    if(patternParam == null) {
+    if (patternParam == null) {
       throw new SolrException(ErrorCode.SERVER_ERROR, 
                               "Missing required init parameter: " + PATTERN_PARAM);
     }
@@ -85,11 +99,22 @@ public final class RegexReplaceProcessorFactory extends FieldMutatingUpdateProce
     }                                
 
     Object replacementParam = args.remove(REPLACEMENT_PARAM);
-    if(replacementParam == null) {
+    if (replacementParam == null) {
       throw new SolrException(ErrorCode.SERVER_ERROR, 
                               "Missing required init parameter: " + REPLACEMENT_PARAM);
     }
-    replacement = Matcher.quoteReplacement(replacementParam.toString());
+
+    Boolean literalReplacement = args.removeBooleanArg(LITERAL_REPLACEMENT_PARAM);
+
+    if (literalReplacement != null) {
+      literalReplacementEnabled = literalReplacement;
+    }
+
+    if (literalReplacementEnabled) {
+      replacement = Matcher.quoteReplacement(replacementParam.toString());
+    } else {
+      replacement = replacementParam.toString();
+    }
 
     super.init(args);
   }

solr/core/src/test-files/solr/collection1/conf/solrconfig-update-processor-chains.xml

@@ -412,6 +412,45 @@
     </processor>
   </updateRequestProcessorChain>
 
+  <updateRequestProcessorChain name="regex-replace-literal-true">
+    <processor class="solr.RegexReplaceProcessorFactory">
+      <str name="fieldName">content</str>
+      <str name="fieldName">title</str>
+      <str name="pattern">(try)</str>
+      <str name="replacement">&lt;$1&gt;</str>
+      <bool name="literalReplacement">true</bool>
+    </processor>
+  </updateRequestProcessorChain>
+
+  <updateRequestProcessorChain name="regex-replace-literal-default-true">
+    <processor class="solr.RegexReplaceProcessorFactory">
+      <str name="fieldName">content</str>
+      <str name="fieldName">title</str>
+      <str name="pattern">(try)</str>
+      <str name="replacement">&lt;$1&gt;</str>
+    </processor>
+  </updateRequestProcessorChain>
+
+  <updateRequestProcessorChain name="regex-replace-literal-false">
+    <processor class="solr.RegexReplaceProcessorFactory">
+      <str name="fieldName">content</str>
+      <str name="fieldName">title</str>
+      <str name="pattern">(try)</str>
+      <str name="replacement">&lt;$1&gt;</str>
+      <bool name="literalReplacement">false</bool>
+    </processor>
+  </updateRequestProcessorChain>
+
+  <updateRequestProcessorChain name="regex-replace-literal-str-true">
+    <processor class="solr.RegexReplaceProcessorFactory">
+      <str name="fieldName">content</str>
+      <str name="fieldName">title</str>
+      <str name="pattern">(try)</str>
+      <str name="replacement">&lt;$1&gt;</str>
+      <str name="literalReplacement">true</str>
+    </processor>
+  </updateRequestProcessorChain>
+
   <updateRequestProcessorChain name="default-values">
     <processor class="solr.DefaultValueUpdateProcessorFactory">
       <str name="fieldName">processor_default_s</str>

solr/core/src/test/org/apache/solr/update/processor/FieldMutatingUpdateProcessorTest.java

@@ -367,6 +367,59 @@ public class FieldMutatingUpdateProcessorTest extends UpdateProcessorTestBase {
                  d.getFieldValue("content"));
     assertEquals("ThisXtitleXhasXaXlotXofXspaces", 
                  d.getFieldValue("title"));
+
+    // literalReplacement = true
+    d = processAdd("regex-replace-literal-true",
+        doc(f("id", "doc2"),
+            f("content", "Let's try this one"),
+            f("title", "Let's try try this one")));
+
+    assertNotNull(d);
+
+    assertEquals("Let's <$1> this one",
+        d.getFieldValue("content"));
+    assertEquals("Let's <$1> <$1> this one",
+        d.getFieldValue("title"));
+
+    // literalReplacement is not specified, defaults to true
+    d = processAdd("regex-replace-literal-default-true",
+        doc(f("id", "doc3"),
+            f("content", "Let's try this one"),
+            f("title", "Let's try try this one")));
+
+    assertNotNull(d);
+
+    assertEquals("Let's <$1> this one",
+        d.getFieldValue("content"));
+    assertEquals("Let's <$1> <$1> this one",
+        d.getFieldValue("title"));
+
+    // if user passes literalReplacement as a string param instead of boolean
+    d = processAdd("regex-replace-literal-str-true",
+        doc(f("id", "doc4"),
+            f("content", "Let's try this one"),
+            f("title", "Let's try try this one")));
+
+    assertNotNull(d);
+
+    assertEquals("Let's <$1> this one",
+        d.getFieldValue("content"));
+    assertEquals("Let's <$1> <$1> this one",
+        d.getFieldValue("title"));
+
+    // This is with literalReplacement = false
+    d = processAdd("regex-replace-literal-false",
+        doc(f("id", "doc5"),
+            f("content", "Let's try this one"),
+            f("title", "Let's try try this one")));
+
+    assertNotNull(d);
+
+    assertEquals("Let's <try> this one",
+        d.getFieldValue("content"));
+    assertEquals("Let's <try> <try> this one",
+        d.getFieldValue("title"));
+
   }
  
   public void testFirstValue() throws Exception {