Updated similar() and useful()

romanchyla · Sep 12, 2019 · 2304371 · 2304371
1 parent 8667bf6
commit 2304371
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 11 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
@@ -773,7 +773,7 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 										new SecondOrderCollectorTopN(200)),
 										new SecondOrderCollectorCitesRAM(referencesWrapper));
 
-				outerQuery.getcollector().setFinalValueType(FinalValueType.ABS_COUNT_NORM);
+				outerQuery.getcollector().setFinalValueType(FinalValueType.ABS_COUNT);
 				return outerQuery;
 			};
 		});
@@ -1096,7 +1096,7 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
 
         SolrQueryRequest req = fp.getReq();
         String input = fp.getString();
-        String filterName = input.substring(1, input.length()-1);
+        String filterName = input.substring(1, input.length()-1).trim();
 
 
         // pick the content stream (actually a filter)
@@ -1189,11 +1189,11 @@ public Query parse(FunctionQParser fp) throws SyntaxError {
       public Query parse(FunctionQParser fp) throws SyntaxError {
         String input = fp.parseId();
 
-        String toLoad = "abstract";
+        String toLoad = "abstract title";
         if (fp.hasMoreArguments())
           toLoad = fp.parseId();
 
-        int maxQueryTerms = 500;
+        int maxQueryTerms = 100;
         if (fp.hasMoreArguments())
           maxQueryTerms = Math.min(fp.parseInt(), maxQueryTerms);
 

diff --git a/contrib/adsabs/src/test/org/apache/solr/analysis/TestAdsabsTypeFulltextParsing.java b/contrib/adsabs/src/test/org/apache/solr/analysis/TestAdsabsTypeFulltextParsing.java
@@ -1222,7 +1222,8 @@ public void testOtherCases() throws Exception {
     assertQ(req("q", "title:\"GBT Survey of 50 Faint Fermi\"~2"), 
         "//*[@numFound>='4']");
 
-
+    //TODO: this test is intentionally left failing; it used to work until the scoring changes (i'd like to 
+    // investigate more how the multi-token affects recall)
     assertQ(req("q", "title:\"A 350-MHz GBT Survey of 50 Faint Fermi γ-ray Sources for Radio Millisecond Pulsars\"",
         "indent", "true",
         "debugQuery", "true"), 

diff --git a/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java b/contrib/adsabs/src/test/org/apache/solr/search/TestAqpAdsabsSolrSearch.java
@@ -426,7 +426,7 @@ public void testSpecialCases() throws Exception {
           MatchNoDocsQuery.class);
 
       assertQueryEquals(req("defType", "aqp", "q", "similar(bibcode:XXX)"),
-          "+like:foo bar baz  -BitSetQuery(1)",
+          "+like:foo bar baz title bitle  -BitSetQuery(1)",
           BooleanQuery.class);
 
       assertQueryEquals(req("defType", "aqp", "q", "similar(bibcode:XXX, title)"),
@@ -979,11 +979,11 @@ public void testSpecialCases() throws Exception {
 
         // useful() - ads classic implementation
         assertQueryEquals(req("defType", "aqp", "q", "useful(author:foo)"),
-                "SecondOrderQuery(SecondOrderQuery(SecondOrderQuery(author:foo, author:foo,*, collector=SecondOrderCollectorAdsClassicScoringFormula(cache=citations-cache, boost=float[] cite_read_boost, lucene=0.5, adsPart=0.5)), collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitesRAM(cache:citations-cache))",
+                "SecondOrderQuery(SecondOrderQuery(author:foo, author:foo,*, collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitesRAM(cache:citations-cache))",
                 SecondOrderQuery.class);
 
         assertQueryEquals(req("defType", "aqp", "q", "all:(x OR z) useful(author:foo OR title:body)"),
-                "+(all:x all:z) +SecondOrderQuery(SecondOrderQuery(SecondOrderQuery((author:foo, author:foo,*) title:body, collector=SecondOrderCollectorAdsClassicScoringFormula(cache=citations-cache, boost=float[] cite_read_boost, lucene=0.5, adsPart=0.5)), collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitesRAM(cache:citations-cache))",
+                "+(all:x all:z) +SecondOrderQuery(SecondOrderQuery((author:foo, author:foo,*) title:body, collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitesRAM(cache:citations-cache))",
                 BooleanQuery.class);
 
 
@@ -997,11 +997,11 @@ public void testSpecialCases() throws Exception {
 
         // reviews() - ADS classic impl
         assertQueryEquals(req("defType", "aqp", "q", "reviews(author:foo)"),
-                "SecondOrderQuery(SecondOrderQuery(SecondOrderQuery(author:foo, author:foo,*, collector=SecondOrderCollectorAdsClassicScoringFormula(cache=citations-cache, boost=float[] cite_read_boost, lucene=0.5, adsPart=0.5)), collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitedBy(cache:citations-cache))",
+                "SecondOrderQuery(SecondOrderQuery(author:foo, author:foo,*, collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitedBy(cache:citations-cache))",
                 SecondOrderQuery.class);
 
         assertQueryEquals(req("defType", "aqp", "q", "all:(x OR z) reviews(author:foo OR title:body)"),
-                "+(all:x all:z) +SecondOrderQuery(SecondOrderQuery(SecondOrderQuery((author:foo, author:foo,*) title:body, collector=SecondOrderCollectorAdsClassicScoringFormula(cache=citations-cache, boost=float[] cite_read_boost, lucene=0.5, adsPart=0.5)), collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitedBy(cache:citations-cache))",
+                "+(all:x all:z) +SecondOrderQuery(SecondOrderQuery((author:foo, author:foo,*) title:body, collector=SecondOrderCollectorTopN(200)), collector=SecondOrderCollectorCitedBy(cache:citations-cache))",
                 BooleanQuery.class);
 
         // reviews2() - original impl
@@ -1249,7 +1249,7 @@ public void testSubquery() throws Exception {
       streams.add(cs);
 
 
-      SolrQueryRequestBase req = (SolrQueryRequestBase) req("qt", "/bigquery", "q","(docs(fq_foo) OR docs(fq_bar)) AND bibcode:b4");
+      SolrQueryRequestBase req = (SolrQueryRequestBase) req("qt", "/bigquery", "q","(docs(fq_foo) OR docs(fq_bar)) AND bibcode:b4", "debugQuery", "true");
       req.setContentStreams(streams);
 
       assertQ(req,