Merge pull request #4345 from vespa-engine/vekterli/add-sameelement-f…

…uzzy-test

Add system test for sameElement + fuzzy combination

tests/search/fuzzy_matching/docs.json

@@ -6,7 +6,11 @@
       "array_slow": ["ThisIsAFox", "Bear"],
       "array_fast": ["ThisIsAFox", "Bear"],
       "wset_slow": {"ThisIsAFox": 100, "Bear": 100},
-      "wset_fast": {"ThisIsAFox": 100, "Bear": 100}
+      "wset_fast": {"ThisIsAFox": 100, "Bear": 100},
+      "my_struct_array": [
+        {"f1": "aaa", "f2": "bbb"},
+        {"f1": "ccc", "f2": "ddd"}
+      ]
     }
   },
   {
@@ -16,7 +20,11 @@
       "array_slow": ["ThisIsAFox", "Beaver"],
       "array_fast": ["ThisIsAFox", "Beaver"],
       "wset_slow": {"ThisIsAFox": 100, "Beaver": 100},
-      "wset_fast": {"ThisIsAFox": 100, "Beaver": 100}
+      "wset_fast": {"ThisIsAFox": 100, "Beaver": 100},
+      "my_struct_array": [
+        {"f1": "xaa", "f2": "bbb"},
+        {"f1": "aaa", "f2": "ddd"}
+      ]
     }
   },
   {

tests/search/fuzzy_matching/fuzzy_same_element.rb

@@ -0,0 +1,52 @@
+# Copyright Vespa.ai. All rights reserved.
+require 'indexed_search_test'
+require 'cgi'
+
+# sameElement+fuzzy is in a dedicated test that does _not_ check streaming mode,
+# as this combination does not currently work.
+# TODO make it work with both modes and move to other test.
+class FuzzySameElementSearch < SearchTest
+
+  def setup
+    set_owner('vekterli')
+  end
+
+  def teardown
+    stop
+  end
+
+  def make_query(a)
+    yql_query = CGI::escape("select * from sources * where #{a}")
+    my_query = "query=" + yql_query + "&type=yql"
+    my_query
+  end
+
+  def assert_documents(query, exp_docids)
+    result = search(query)
+    assert_hitcount(result, exp_docids.size)
+    result.sort_results_by("documentid")
+    for i in 0...exp_docids.size do
+      exp_docid = "id:test:test::#{exp_docids[i]}"
+      assert_field_value(result, "documentid", exp_docid, i)
+    end
+  end
+
+  def assert_fuzzy_same_element(max_edits, term1, term2, expected_docs)
+    q = "my_struct_array contains sameElement(" +
+        "f1 contains ({maxEditDistance:#{max_edits}}fuzzy(\"#{term1}\")), " +
+        "f2 contains ({maxEditDistance:#{max_edits}}fuzzy(\"#{term2}\")))"
+    assert_documents(make_query(q), expected_docs)
+  end
+
+  def test_fuzzysearch_with_sameelement
+    set_description('Test that the fuzzy() operator can be used as part of sameElement')
+    deploy_app(SearchApp.new.sd(selfdir+'test.sd'))
+    start
+    feed_and_wait_for_docs('test', 6, :file => selfdir + 'docs.json')
+
+    assert_fuzzy_same_element(1, 'aax', 'bbx', [1])
+    assert_fuzzy_same_element(2, 'aax', 'bbx', [1, 2])
+    assert_fuzzy_same_element(2, 'aax', 'ddx', [2])
+  end
+
+end

tests/search/fuzzy_matching/test.sd

@@ -18,6 +18,20 @@ schema test {
       indexing: attribute
       attribute: fast-search
     }
+    struct my_struct {
+      field f1 type string {}
+      field f2 type string {}
+    }
+    field my_struct_array type array<my_struct> {
+      struct-field f1 {
+        indexing: attribute
+        attribute: fast-search
+      }
+      struct-field f2 {
+        indexing: attribute
+        attribute: fast-search
+      }
+    }
   }