Merge branch 'master' of github.com:romanchyla/montysolr

contrib/adsabs/src/java/org/apache/lucene/queryparser/flexible/aqp/config/AqpAdsabsQueryConfigHandler.java

@@ -77,7 +77,7 @@ public AqpAdsabsQueryConfigHandler() {
 		set(ConfigurationKeys.DEFAULT_DATE_RANGE_FIELD, "date");
 		set(ConfigurationKeys.DEFAULT_IDENTIFIER_FIELD, "identifier");
 
-		set(StandardQueryConfigHandler.ConfigurationKeys.ALLOW_LEADING_WILDCARD, true);
+		set(StandardQueryConfigHandler.ConfigurationKeys.ALLOW_LEADING_WILDCARD, false);
 
 
 

contrib/adsabs/src/java/org/apache/lucene/queryparser/flexible/aqp/processors/AqpDEFOPUnfieldedTokens.java

@@ -482,7 +482,7 @@ private Float getDisjunctTieBreaker() {
     Object obj = _getConfigVal("aqp.unfielded.tokens.tiebreaker");
     if (obj == null)
       return 0.0f;
-    return (Float) obj;
+    return Float.valueOf((String) obj);
   }
 
 	private Set<String> aqpIgnorableFields = null;

contrib/adsabs/src/java/org/apache/solr/search/AqpAdsabsQParser.java

@@ -244,6 +244,11 @@ public AqpAdsabsQParser(AqpQueryParser parser, String qstr, SolrParams localPara
 
 		config.get(AqpAdsabsQueryConfigHandler.ConfigurationKeys.VIRTUAL_FIELDS).putAll(defaultConfig.virtualFields);
 
+		if (params.get("aqp.allow.leading_wildcard", null) != null) {
+		  config.set(StandardQueryConfigHandler.ConfigurationKeys.ALLOW_LEADING_WILDCARD, 
+		      params.getBool("aqp.allow.leading_wildcard", false));
+		}
+
 	}
 
 

contrib/adsabs/src/java/perf/CreatePerformanceQueriesHandler.java

@@ -359,6 +359,7 @@ private void resolveIfNecessary(String fieldName, Set<FormattedQuery> seen) thro
 		SolrParams oldParams = req.getParams();
 		ModifiableSolrParams newParams = new ModifiableSolrParams(oldParams);
 		newParams.set(CommonParams.DF, fieldName);
+		newParams.set("aqp.allow.leading_wildcard", "true");
 		req.setParams(newParams);
 
 		String defType = req.getParams().get(QueryParsing.DEFTYPE, QParserPlugin.DEFAULT_QTYPE);

contrib/adsabs/src/test/org/adsabs/TestAdsAllFields.java

@@ -708,7 +708,8 @@ public void test() throws Exception {
 		assertQ(req("q", "bibgroup:CFA"), "//*[@numFound='1']");
 		assertQ(req("q", "bibgroup:cf*"), "//*[@numFound='1']");
 		assertQ(req("q", "bibgroup:CF*"), "//*[@numFound='1']");
-		assertQ(req("q", "bibgroup:?FA"), "//*[@numFound='1']");
+		assertQ(req("q", "bibgroup:?FA", "qt", "/query", "aqp.allow.leading_wildcard", "true"), 
+		    "//*[@numFound='1']");
 
 		// facets are case sensitive and  you must get the exact wording
 		// TODO: shall we be consistent and turn *everything* to lowercase?
@@ -1332,10 +1333,10 @@ public void test() throws Exception {
 
 
 		// test we can search for all docs that have certain field
-		assertQ(req("q", "reference:*"),
+		assertQ(req("q", "reference:*", "qt", "/query", "aqp.allow.leading_wildcard", "true"),
 				"//doc[1]/int[@name='recid'][.='100']"
 		);
-		assertQ(req("q", "id:?"), // but works only for text fields
+		assertQ(req("q", "id:?", "qt", "/query", "aqp.allow.leading_wildcard", "true"), // but works only for text fields
 				"//*[@numFound='8']"
 		);
 

contrib/adsabs/src/test/org/apache/lucene/queryparser/flexible/aqp/TestAqpAdsabs.java

@@ -311,9 +311,9 @@ public void testWildCards() throws Exception {
 	    assertEquals(MultiTermQuery.CONSTANT_SCORE_REWRITE, ((MultiTermQuery) q).getRewriteMethod());
 
 		assertQueryEquals("te*t", null, "te*t", WildcardQuery.class);
-		assertQueryEquals("*te*t", null, "*te*t", WildcardQuery.class);
-		assertQueryEquals("*te*t*", null, "*te*t*", WildcardQuery.class);
-		assertQueryEquals("?te*t?", null, "?te*t?", WildcardQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("*te*t", null, "*te*t");
+		assertQueryEqualsAllowLeadingWildcard("*te*t*", null, "*te*t*");
+		assertQueryEqualsAllowLeadingWildcard("?te*t?", null, "?te*t?");
 		assertQueryEquals("te?t", null, "te?t", WildcardQuery.class);
 		assertQueryEquals("te??t", null, "te??t", WildcardQuery.class);
 
@@ -328,9 +328,9 @@ public void testWildCards() throws Exception {
 		assertQueryEquals("\"te*t phrase\"", null, "te*t phrase", WildcardQuery.class);
 		assertQueryEquals("\"test* phrase\"", null, "test* phrase", WildcardQuery.class);
 		assertQueryEquals("\"te*t phrase\"", null, "te*t phrase", WildcardQuery.class);
-		assertQueryEquals("\"*te*t phrase\"", null, "*te*t phrase", WildcardQuery.class);
-		assertQueryEquals("\"*te*t* phrase\"", null, "*te*t* phrase", WildcardQuery.class);
-		assertQueryEquals("\"?te*t? phrase\"", null, "?te*t? phrase", WildcardQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("\"*te*t phrase\"", null, "*te*t phrase");
+		assertQueryEqualsAllowLeadingWildcard("\"*te*t* phrase\"", null, "*te*t* phrase");
+		assertQueryEqualsAllowLeadingWildcard("\"?te*t? phrase\"", null, "?te*t? phrase");
 		assertQueryEquals("\"te?t phrase\"", null, "te?t phrase", WildcardQuery.class);
 		assertQueryEquals("\"te??t phrase\"", null, "te??t phrase", WildcardQuery.class);
 		assertQueryEquals("\"te*?t phrase\"", null, "te*?t phrase", WildcardQuery.class);
@@ -339,22 +339,22 @@ public void testWildCards() throws Exception {
 		assertQueryEquals("*", null, "*:*", MatchAllDocsQuery.class);
 		assertQueryEquals("*:*", null, "*:*", MatchAllDocsQuery.class);
 
-		assertQueryEquals("?", null, "?", WildcardQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("?", null, "?");
 
 
 		// XXX: in fact, in the WildcardQuery, even escaped start \* will become *
 		// so it is not possible to search for words that contain * as a literal 
 		// character, to have it differently, WildcardTermEnum class would have 
 		// to think of skipping \* and \?
 
-		q = assertQueryEquals("*t\\*a", null, "*t*a", WildcardQuery.class); 
+		assertQueryEqualsAllowLeadingWildcard("*t\\*a", null, "*t*a"); 
 
-		assertQueryEquals("*t*a\\*", null, "*t*a*", WildcardQuery.class);
-		assertQueryEquals("*t*a\\?", null, "*t*a?", WildcardQuery.class);
-		assertQueryEquals("*t*\\a", null, "*t*a", WildcardQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("*t*a\\*", null, "*t*a*");
+		assertQueryEqualsAllowLeadingWildcard("*t*a\\?", null, "*t*a?");
+		assertQueryEqualsAllowLeadingWildcard("*t*\\a", null, "*t*a");
 
-		assertQueryEquals("title:*", null, "title:*", PrefixQuery.class);
-		assertQueryEquals("doi:*", null, "doi:*", PrefixQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("title:*", null, "title:*", PrefixQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("doi:*", null, "doi:*", PrefixQuery.class);
 
 	}
 
@@ -379,7 +379,7 @@ public void testBasics() throws Exception{
 		KeywordAnalyzer kwa = new KeywordAnalyzer();
 		assertQueryEquals("keyword:\"planets and satellites\"", wsa, "keyword:\"planets and satellites\"", PhraseQuery.class);
 
-		assertQueryEquals("full:*", null, "full:*", PrefixQuery.class);
+		assertQueryEqualsAllowLeadingWildcard("full:*", null, "full:*", PrefixQuery.class);
 
 		assertQueryEquals("weak lensing", null, "+weak +lensing");
 		assertQueryEquals("+contact +binaries -eclipsing", null, "+contact +binaries -eclipsing");

contrib/adsabs/src/test/org/apache/solr/analysis/TestAdsabsTypeFulltextParsing.java

@@ -965,7 +965,9 @@ public void testOtherCases() throws Exception {
         "//doc/str[@name='id'][.='11']"
         );
 
-    assertQ(req("q", "title" + ":*sky"), "//*[@numFound='4']", 
+    assertQ(req("qt", "/query",
+        "aqp.allow.leading_wildcard", "true",
+        "q", "title" + ":*sky"), "//*[@numFound='4']", 
         "//doc/str[@name='id'][.='10']",
         "//doc/str[@name='id'][.='11']",
         "//doc/str[@name='id'][.='12']",

contrib/adsabs/src/test/org/apache/solr/analysis/TestAdsabsTypeNormalizedTextAscii.java

@@ -116,10 +116,10 @@ public void test() throws Exception {
       assertQ(req("q", f + ":\"fivejets\""), "//*[@numFound='0']");
       assertQ(req("q", f + ":\"onejets\""), "//*[@numFound='1']", "//doc[1]/str[@name='id'][.='6']");
       assertQ(req("q", f + ":\"one-jets\""), "//*[@numFound='1']", "//doc[1]/str[@name='id'][.='6']");
-      assertQ(req("q", f + ":\"*jets\""), "//*[@numFound='2']", "//doc[1]/str[@name='id'][.='6']"); // curiously also finds 'jets' (id 5.)
+      assertQ(req("q", f + ":\"*jets\"", "qt", "/query", "aqp.allow.leading_wildcard", "true"), "//*[@numFound='2']", "//doc[1]/str[@name='id'][.='6']"); // curiously also finds 'jets' (id 5.)
       assertQ(req("q", f + ":\"jets\""), "//*[@numFound='1']", "//doc[1]/str[@name='id'][.='9']");
       assertQ(req("q", f + ":\"jets*\""), "//*[@numFound='2']", "//doc[1]/str[@name='id'][.='7']"); // also 'jets' (id 5.)
-      assertQ(req("q", f + ":\"*jets*\""), "//*[@numFound='4']");
+      assertQ(req("q", f + ":\"*jets*\"", "qt", "/query", "aqp.allow.leading_wildcard", "true"), "//*[@numFound='4']");
 
       // find only 'jets' (where the word stood alone)
       assertQ(req("q", f + ":\"jets\""), "//*[@numFound='1']", "//doc[1]/str[@name='id'][.='9']");

contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java

@@ -292,6 +292,17 @@ public void testUnfieldedSearch() throws Exception {
             "((+((keyword:r)^0.7 | (title:r)^0.9) +((keyword:s)^0.7 | (title:s)^0.9) +((keyword:t)^0.7 | (title:t)^0.9)) "
             + "| (((keyword:\"r s t\" | Synonym(keyword:acr::rst keyword:syn::r s t)))^0.7 | ((title:\"r s t\" | Synonym(title:acr::rst title:syn::r s t)))^0.9))",
             DisjunctionMaxQuery.class);
+        assertQueryEquals(req("defType", "aqp",
+            "q", "r s t",
+            "aqp.unfielded.tokens.strategy", "disjuncts",
+            "aqp.unfielded.tokens.tiebreaker", "0.5",
+            "aqp.unfielded.tokens.new.type", "simple",
+            "aqp.unfielded.phrase.edismax.synonym.workaround", "false",
+            "qf", "title^0.9 keyword^0.7"),
+            "((+((keyword:r)^0.7 | (title:r)^0.9) +((keyword:s)^0.7 | (title:s)^0.9) +((keyword:t)^0.7 | (title:t)^0.9)) "
+            + "| (((keyword:\"r s t\" | Synonym(keyword:acr::rst keyword:syn::r s t)))^0.7 | ((title:\"r s t\" | Synonym(title:acr::rst title:syn::r s t)))^0.9))~0.5",
+            DisjunctionMaxQuery.class);
+
         assertQueryEquals(req("defType", "aqp",
                 "q", "r s t",
                 "aqp.unfielded.tokens.strategy", "multiply",
@@ -972,10 +983,12 @@ public void testSearch() throws Exception {
 
 
         // search for all docs with a field
-        assertQueryEquals(req("defType", "aqp", "q", "title:*"),
+        assertQueryEquals(req("defType", "aqp", "q", "title:*", 
+            "aqp.allow.leading_wildcard", "true"),
                 "title:*",
                 PrefixQuery.class);
-        assertQueryEquals(req("defType", "aqp", "q", "title:?"),
+        assertQueryEquals(req("defType", "aqp", "q", "title:?",
+            "aqp.allow.leading_wildcard", "true"),
                 "title:?",
                 WildcardQuery.class);
 

contrib/antlrqueryparser/src/test/org/apache/lucene/queryparser/flexible/aqp/AqpTestAbstractCase.java

@@ -167,6 +167,22 @@ public Query assertQueryEquals(String query, Analyzer a, String result,
     return q;
   }
 
+  public void assertQueryEqualsAllowLeadingWildcard(String query, Analyzer a,
+      String result, Class<?> clazz) throws Exception {
+    Query q = getQueryAllowLeadingWildcard(query, a);
+    String s = q.toString("field");
+    if (!s.equals(result)) {
+      fail("Query /" + query + "/ yielded /" + s + "/, expecting /" + result
+          + "/");
+    }
+    if (clazz != null) {
+      if (!q.getClass().isAssignableFrom(clazz)) {
+        debugFail(q.toString(), result,
+            "Query is not: " + clazz + " but: " + q.getClass());
+      }      
+    }
+  }
+
   public void assertQueryEqualsAllowLeadingWildcard(String query, Analyzer a,
       String result) throws Exception {
     Query q = getQueryAllowLeadingWildcard(query, a);

contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml

@@ -318,6 +318,7 @@
      </lst>
      <lst name="invariants">
        <str name="echoParams">explicit</str>
+       <str name="aqp.allow.leading_wildcard">false</str>
      </lst>
      <arr name="last-components">
       <str>wordcloud</str>
@@ -339,6 +340,7 @@
      </lst>
     <lst name="invariants">
        <str name="echoParams">explicit</str>
+       <str name="aqp.allow.leading_wildcard">false</str>
     </lst>
     <arr name="last-components">
       <str>tvComponent</str>
@@ -356,6 +358,7 @@
 		</lst>
 		<lst name="invariants">
 	       <str name="echoParams">explicit</str>
+	       <str name="aqp.allow.leading_wildcard">false</str>
 	     </lst>
 	</requestHandler> 
 
@@ -381,6 +384,8 @@
        <str name="aqp.unfielded.ignore.fields">entdate pubdate year</str>
        <str name="aqp.unfielded.tokens.function.name">edismax_combined_aqp</str>
        <str name="useFastVectorHighlighter">true</str>
+       <str name="aqp.allow.leading_wildcard">true</str>
+
      </lst>
      <lst name="invariants">
        <str name="echoParams">explicit</str>