Added more scoring methods; book_author and editor; and constant() me…

…thod
adsabs · May 18, 2018 · 64b0e4e · 64b0e4e
1 parent 6820d0e
commit 64b0e4e
Show file tree

Hide file tree

Showing 6 changed files with 90 additions and 20 deletions.
diff --git a/...c/java/org/apache/lucene/queryparser/flexible/aqp/builders/AqpAdsabsSubQueryProvider.java b/...c/java/org/apache/lucene/queryparser/flexible/aqp/builders/AqpAdsabsSubQueryProvider.java
@@ -29,6 +29,7 @@
 import org.apache.lucene.search.MatchNoDocsQuery;
 import org.apache.lucene.search.Query;
 import org.apache.lucene.search.Scorer;
+import org.apache.lucene.search.ConstantScoreQuery;
 import org.apache.lucene.search.SecondOrderCollector;
 import org.apache.lucene.search.SecondOrderCollector.FinalValueType;
 import org.apache.lucene.search.SecondOrderCollectorAdsClassicScoringFormula;
@@ -640,11 +641,11 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 				Query innerQuery = fp.parseNestedQuery();
 
 				@SuppressWarnings("unchecked")
-        SolrCacheWrapper<CitationCache<Object, Integer>> referencesWrapper = new SolrCacheWrapper.ReferencesCache(
+				SolrCacheWrapper<CitationCache<Object, Integer>> referencesWrapper = new SolrCacheWrapper.ReferencesCache(
 						(CitationCache<Object, Integer>) fp.getReq().getSearcher().getCache("citations-cache"));
 
 
-				return new SecondOrderQuery(innerQuery, 
+					return new SecondOrderQuery(innerQuery, 
 						new SecondOrderCollectorCitesRAM(referencesWrapper), false);
 			}
 		});
@@ -1017,7 +1018,7 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 		// so we have to do it ourselves
 		parsers.put("warm_cache", new AqpSubqueryParserFull() {
 			@SuppressWarnings("unchecked")
-      public Query parse(FunctionQParser fp) throws SyntaxError {
+				public Query parse(FunctionQParser fp) throws SyntaxError {
 
 				final SolrQueryRequest req = fp.getReq();
 				@SuppressWarnings("rawtypes")
@@ -1028,6 +1029,30 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 				return new MatchNoDocsQuery();
 			}
 		});
+
+		/* @api.doc
+		 * 
+		 * def constant(query):
+		 * 		"""
+		 *    Applies constant score (that can be set by boost factor)
+		 *    
+		 *    Example: 
+		 *    
+		 *    	```constant(title:hubble^2)```
+		 *    
+		 *    
+		 *    @since 63.1.0.24
+		 *    """
+		 *    return "constant(%s)" % (query,)
+		 *    
+		 */
+		parsers.put("constant", new AqpSubqueryParserFull() {
+			public Query parse(FunctionQParser fp) throws SyntaxError {    		  
+				Query innerQuery = fp.parseNestedQuery();
+
+				return new ConstantScoreQuery(innerQuery);
+			}
+		});
 
 	};
 

diff --git a/...rg/apache/lucene/queryparser/flexible/aqp/processors/AqpChangeRewriteMethodProcessor.java b/...rg/apache/lucene/queryparser/flexible/aqp/processors/AqpChangeRewriteMethodProcessor.java
@@ -72,6 +72,15 @@ else if ("constant_boolean".equals(method)) {
     else if ("boolean".equals(method)) {
       node.setTag(MultiTermRewriteMethodProcessor.TAG_ID, MultiTermQuery.SCORING_BOOLEAN_REWRITE);
     }
+    else if ("topterms_blended".equals(method)) {
+      node.setTag(MultiTermRewriteMethodProcessor.TAG_ID, new MultiTermQuery.TopTermsBlendedFreqScoringRewrite(1024));
+    }
+    else if ("topterms".equals(method)) {
+        node.setTag(MultiTermRewriteMethodProcessor.TAG_ID, new MultiTermQuery.TopTermsScoringBooleanQueryRewrite(1024));
+    }
+    else if ("topterms_boosted".equals(method)) {
+        node.setTag(MultiTermRewriteMethodProcessor.TAG_ID, new MultiTermQuery.TopTermsBoostOnlyBooleanQueryRewrite(1024));
+    }
     else {
       throw new QueryNodeException(new MessageImpl(
           QueryParserMessages.PARAMETER_VALUE_NOT_SUPPORTED, "Unknown rewrite method: \"" + method + "\""));

diff --git a/contrib/adsabs/src/test/org/adsabs/TestAdsAllFields.java b/contrib/adsabs/src/test/org/adsabs/TestAdsAllFields.java
@@ -155,6 +155,7 @@ public void test() throws Exception {
 				 */
 				 ", \"bibstem\": [\"JNuM\", \"JNuM..455\"]" +
 				 ", \"body\": \"Some fulltext hashimoto added\"" +
+				 ", \"book_author\": [\"book, author\", \"book, fauthor\"]" +
 				 ", \"caption\": [\"caption1 captionFoo\", \"caption2\"]" +
 				 ", \"citation\": [\"2014JNuM..455...10C\", \"2014JNuM..455...10D\"]" +
 				 ", \"cite_read_boost\": 0.52" +
@@ -164,18 +165,19 @@ public void test() throws Exception {
 				 ", \"comment\": [\"comment1 commentFoo\", \"comment2\"]" +
 				 ", \"database\": [\"ASTRONOMY\", \"PHYSICS\"]" +
 
-        ", \"data\": [\"NED:15\", \"CDS:5\"]" +
-        ", \"data_count\": 20" +
+		        ", \"data\": [\"NED:15\", \"CDS:5\"]" +
+		        ", \"data_count\": 20" +
 				 // it is solr format for the pubdate, must be in the right format
 				 // we need to add 30 minutes to every day; this allows us to search
 				 // for ranges effectively; thus:
 				 // 2013-08-5 -> 2013-08-05T00:30:00Z
 				 // 2013-08   -> 2013-08-01T00:30:00Z
 				 // 2013      -> 2013-01-01T00:30:00Z
-        ", \"date\": \"2013-08-05T00:30:00Z\"" +
+				 ", \"date\": \"2013-08-05T00:30:00Z\"" +
 				", \"doctype\": \"article\"" +
 				", \"doctype_facet_hier\": [\"0/Article\", \"1/Article/Book chapter\"]" +
 			  ", \"doi\": \"doi:ŽŠČŘĎŤŇ:123456789\"" +
+			  ", \"editor\": [\"t' Hooft, van X\"]" +
 			  ", \"eid\": \"00001\"" +
 			  ", \"email\": [\"-\", \"[email protected]\", \"-\"]" +
 			  // entry_date --> see below
@@ -345,14 +347,14 @@ public void test() throws Exception {
 
 
 		/*
-     * caption
-     */
-    assertQ(req("q", "caption:caption1"),
-        "//*[@numFound='1']",
-        "//doc/int[@name='recid'][.='100']");
-    assertQ(req("q", "caption:captionfoo"),
-        "//*[@numFound='1']",
-        "//doc/int[@name='recid'][.='100']");
+	     * caption
+	     */
+	    assertQ(req("q", "caption:caption1"),
+	        "//*[@numFound='1']",
+	        "//doc/int[@name='recid'][.='100']");
+	    assertQ(req("q", "caption:captionfoo"),
+	        "//*[@numFound='1']",
+	        "//doc/int[@name='recid'][.='100']");
 
 
 		/*
@@ -399,6 +401,13 @@ public void test() throws Exception {
 				"<str>Anders, John Michael</str>" +
 				"<str>Einstein, A</str></arr>");
 
+		/*
+		 * book_author
+		 */
+		assertQ(req("q", "book_author:\"book, fauthor\""),
+				"//*[@numFound='1']",
+				"//doc/int[@name='recid'][.='100']");
+
 		/*
 		 * author_count
 		 */
@@ -477,6 +486,13 @@ public void test() throws Exception {
 		);
 
 
+		/*
+		 * editor
+		 */
+		assertQ(req("q", "editor:\"'t hooft, v x\""),
+				"//*[@numFound='1']",
+				"//doc/int[@name='recid'][.='100']");
+
 
 		/*
 		 * email

diff --git a/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java b/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java
@@ -10,6 +10,7 @@
 import org.apache.lucene.queryparser.flexible.aqp.TestAqpAdsabs;
 import org.apache.lucene.search.BooleanQuery;
 import org.apache.lucene.search.BoostQuery;
+import org.apache.lucene.search.ConstantScoreQuery;
 import org.apache.lucene.search.DisjunctionMaxQuery;
 import org.apache.lucene.search.FuzzyQuery;
 import org.apache.lucene.search.MatchAllDocsQuery;
@@ -319,6 +320,12 @@ public void testUnfieldedSearch() throws Exception {
 
     public void testSpecialCases() throws Exception {
 
+    	// constant() score
+	    assertQueryEquals(req("defType", "aqp", "q", "constant(title:foo)"),
+	            "ConstantScore(title:foo)", ConstantScoreQuery.class);
+	    assertQueryEquals(req("defType", "aqp", "q", "constant(title:foo^2)"),
+	            "ConstantScore((title:foo)^2.0)", ConstantScoreQuery.class);
+
       // allow to set scoring method for a given field
       BooleanQuery q = (BooleanQuery) assertQueryEquals(req("defType", "aqp", 
           "q", "author:riess", "aqp.qprefix.scoring.author", "boolean"),
@@ -338,6 +345,12 @@ public void testSpecialCases() throws Exception {
             BooleanQuery.class);
       assertEquals(((PrefixQuery) q.clauses().get(1).getQuery()).getRewriteMethod(), MultiTermQuery.CONSTANT_SCORE_REWRITE);
 
+
+      q = (BooleanQuery) assertQueryEquals(req("defType", "aqp", 
+          "q", "author:riess", "aqp.qprefix.scoring.author", "topterms"),
+            "author:riess, author:riess,*",
+            BooleanQuery.class);
+      assertEquals(((PrefixQuery) q.clauses().get(1).getQuery()).getRewriteMethod().getClass(), MultiTermQuery.TopTermsScoringBooleanQueryRewrite.class);
 
 	  //verification for https://github.com/romanchyla/montysolr/issues/45
       // expansion of synonyms inside a virtual field together with nested boolean query
@@ -831,7 +844,8 @@ public void testSpecialCases() throws Exception {
                 "SecondOrderQuery(title:foo, collector=SecondOrderCollectorAdsClassicScoringFormula(cache=citations-cache, boost=float[] cite_read_boost, lucene=0.5, adsPart=0.5))", SecondOrderQuery.class);
         assertQueryEquals(req("defType", "aqp", "q", "cr(title:foo, 0.4)"),
                 "SecondOrderQuery(title:foo, collector=SecondOrderCollectorAdsClassicScoringFormula(cache=citations-cache, boost=float[] cite_read_boost, lucene=0.4, adsPart=0.6))", SecondOrderQuery.class);
-
+
+
     }
 
 

diff --git a/contrib/examples/adsabs/server/solr/collection1/conf/schema.xml b/contrib/examples/adsabs/server/solr/collection1/conf/schema.xml
@@ -1159,7 +1159,13 @@
 
 		<field name="first_author_norm" type="normalized_text_ascii_notokenization"
 			indexed="true" stored="true" />
-
+
+		<field name="book_author" type="author" indexed="true" stored="true"
+            multiValued="true" omitNorms="true" useDocValuesAsStored="false"/>
+
+        <field name="editor" type="author" indexed="true" stored="true"
+            multiValued="true" omitNorms="true" useDocValuesAsStored="false"/>
+
 		<field name="aff" type="affiliation_text" indexed="true" stored="true"
 			multiValued="true" omitNorms="true"/>
 

diff --git a/contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml b/contrib/examples/adsabs/server/solr/collection1/conf/solrconfig.xml
@@ -696,9 +696,9 @@
        <str name="aqp.timestampFields">indexstamp,update_timestamp,entry_date,metadata_ctime,metadata_mtime,fulltext_ctime,fulltext_mtime,nonbib_ctime,nonbib_mtime,metrics_ctime,metrics_mtime,orcid_ctime,orcid_mtime</str>
        <str name="aqp.floatFields">cite_read_boost,citation_count_norm</str>
        <str name="aqp.intFields">recid,pubdate_sort,citation_count,classic_factor,simbid,read_count,author_count,page_count,data_count</str>
-       <str name="aqp.authorFields">author,first_author</str>
+       <str name="aqp.authorFields">author,first_author,book_author,editor</str>
        <str name="aqp.humanized.dates">pubdate:date,entdate:entry_date</str>
-       <str name="aqp.force.fuzzy.phrases">author,first_author</str>
+       <str name="aqp.force.fuzzy.phrases">author,first_author,book_author,editor</str>
     </lst>
   </queryParser>
 
@@ -726,9 +726,9 @@
        <str name="aqp.timestampFields">indexstamp,update_timestamp,entry_date,metadata_ctime,metadata_mtime,fulltext_ctime,fulltext_mtime,nonbib_ctime,nonbib_mtime,metrics_ctime,metrics_mtime,orcid_ctime,orcid_mtime</str>
        <str name="aqp.floatFields">cite_read_boost,citation_count_norm</str>
        <str name="aqp.intFields">recid,pubdate_sort,citation_count,classic_factor,simbid,read_count,author_count,page_count,data_count</str>
-       <str name="aqp.authorFields">author,first_author</str>
+       <str name="aqp.authorFields">author,first_author,book_author,editor</str>
        <str name="aqp.humanized.dates">pubdate:date,entdate:entry_date:timestamp</str>
-       <str name="aqp.force.fuzzy.phrases">author,first_author</str>
+       <str name="aqp.force.fuzzy.phrases">author,first_author,book_author,editor</str>
     </lst>
   </queryParser>