Ignore case when we are expanding search into a phrase search

romanchyla · Apr 16, 2021 · 6a7e385 · 6a7e385
1 parent 1bf928f
commit 6a7e385
Show file tree

Hide file tree

Showing 5 changed files with 29 additions and 6 deletions.
diff --git a/...c/java/org/apache/lucene/queryparser/flexible/aqp/processors/AqpDEFOPUnfieldedTokens.java b/...c/java/org/apache/lucene/queryparser/flexible/aqp/processors/AqpDEFOPUnfieldedTokens.java
@@ -360,8 +360,14 @@ private void removeField(QueryNode n) {
 
   }
 
+
 	private QueryNode createReplacementNode(List<NodeInfo> newGroup, String tt) throws CloneNotSupportedException, QueryNodeException {
 	  String newValue = getConcatenatedValue(newGroup);
+
+	  if (newValue != newValue.toLowerCase() && newValue == newValue.toUpperCase() && newGroup.size() > getMaxAllowedAcronym()) {
+	    newValue = newValue.toLowerCase();
+	  }
+
 	  String field = "";
 	  if (newGroup.get(0).getField() != null && newGroup.get(0).getField().length() > 0) 
 	  	field = newGroup.get(0).getField() + ":";
@@ -411,7 +417,14 @@ private QueryNode createReplacementNode(List<NodeInfo> newGroup, String tt) thro
 
 
 
-	/*
+	private int getMaxAllowedAcronym() {
+	  Object obj = _getConfigVal("aqp.unfielded.max.uppercase.tokens");
+    if (obj == null)
+      return 1024; //effectively unlimited
+    return Integer.parseInt((String)obj);
+  }
+
+  /*
 	 * Ufff....this is necessary, because the QueryNodeImpl is NOT
 	 * resetting the parent. sooooo stupid....
 	 */

diff --git a/contrib/adsabs/src/java/org/apache/solr/search/AqpAdsabsQParser.java b/contrib/adsabs/src/java/org/apache/solr/search/AqpAdsabsQParser.java
@@ -349,6 +349,7 @@ public Query parse() throws SyntaxError {
 			//  QueryConfigHandler config = qParser.getQueryConfigHandler();
 			//  return qParser.parse(getString() + config.get(AqpAdsabsQueryConfigHandler.ConfigurationKeys.DUMMY_VALUE), null);
 			//}
+
 			return qParser.parse(getString(), null);
 		} catch (QueryNodeException e) {
 		  throw new SyntaxError(e);

diff --git a/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java b/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java
@@ -137,7 +137,16 @@ public void tearDown() throws Exception {
 
 
     public void testUnfieldedSearch() throws Exception {
-
+
+      // when we generate the phrase search, ignore acronyms
+      assertQueryEquals(req("defType", "aqp",
+          "q", "FOO BAR BAZ",
+          "aqp.unfielded.tokens.strategy", "disjuncts",
+          "aqp.unfielded.tokens.new.type", "simple",
+          "aqp.unfielded.max.uppercase.tokens", "2",
+          "qf", "title"),
+          "(((title:foo) (title:bar) (title:baz)) | title:\"foo bar baz\")",
+          DisjunctionMaxQuery.class);
 
       // have constant scoring work even for unfielded searches
       assertQueryEquals(req("defType", "aqp", "q", "foo bar",

diff --git a/contrib/examples/adsabs/bin/solr.in.sh b/contrib/examples/adsabs/bin/solr.in.sh
@@ -73,7 +73,7 @@ SOLR_OPTS="${SOLR_OPTS} -Xss256k \
 -Dmontysolr.load.citation_cache=${MONTYSOLR_LOAD_CITATION_CACHE:-true} \
 -Dmontysolr.autoCommit.maxDocs=${MONTYSOLR_MAX_DOCS:-40000} \
 -Dmontysolr.autoCommit.maxTime=${MONTYSOLR_MAX_TIME:-1800000} \
--Dmontysolr.autoSoftCommit.maxTime=${MONTYSOLR_SOFT_MAX_TIME:-900000} \
+-Dmontysolr.autoSoftCommit.maxTime=${MONTYSOLR_SOFT_MAX_TIME:-1} \
 -Dmontysolr.coordinate=${MONTYSOLR_COORDINATE:-true} \
 -Dmontysolr.stagger.maxDelay=${MONTYSOLR_STAGGER_MAX_DELAY:-2700} \
 -Dmontysolr.stagger.numInstances=${MONTYSOLR_STAGGER_NUM_INSTANCES:-3} \

diff --git a/contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml b/contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml
@@ -342,7 +342,7 @@
        <str name="aqp.constant_scoring">first_author^14 author^13 year^10 bibstem^10</str>
        <str name="aqp.multiphrase.keep_one">SYNONYM</str>
        <str name="aqp.multiphrase.keep_one.ignore.fields">aff_raw,aff_id,institution</str>
-
+       <str name="aqp.unfielded.max.uppercase.tokens">2</str>
        <str name="q.op">AND</str>
 
      </lst>
@@ -372,7 +372,7 @@
        <str name="aqp.constant_scoring">first_author^14 author^13 year^10 bibstem^10</str>
        <str name="aqp.multiphrase.keep_one">SYNONYM</str>
        <str name="aqp.multiphrase.keep_one.ignore.fields">aff_raw,aff_id,institution</str>
-
+       <str name="aqp.unfielded.max.uppercase.tokens">2</str>
        <str name="q.op">AND</str>
        <str name="df">unfielded_search</str>
      </lst>
@@ -426,7 +426,7 @@
        <str name="aqp.unfielded.tokens.function.name">edismax_combined_aqp</str>
        <str name="useFastVectorHighlighter">true</str>
        <str name="aqp.allow.leading_wildcard">true</str>
-
+       <str name="aqp.unfielded.max.uppercase.tokens">2</str>
        <str name="q.op">AND</str>
        <str name="df">unfielded_search</str>
      </lst>