Limit fields used for match feature.

2 files changed, 60 insertions, 16 deletions

diff --git a/searchlib/src/tests/features/prod_features.cpp b/searchlib/src/tests/features/prod_features.cpp
index dc64c3328e4..c22d3b3abb8 100644
--- a/searchlib/src/tests/features/prod_features.cpp
+++ b/searchlib/src/tests/features/prod_features.cpp
@@ -67,6 +67,7 @@ using search::attribute::WeightedEnumContent;
 using search::attribute::test::AttributeBuilder;
 using search::common::GeoLocation;
 using search::common::GeoLocationSpec;
+using vespalib::eval::ValueType;
 
 using AttributePtr = AttributeVector::SP;
 using AVC = search::attribute::Config;
@@ -391,6 +392,14 @@ Test::setupForAttributeTest(FtFeatureTest &ft, bool setup_env)
     avs.push_back(AttributeFactory::createAttribute("sbool",   AVC(AVBT::BOOL,  AVCT::SINGLE))); // 14
     avs.push_back(AttributeFactory::createAttribute("sebool",   AVC(AVBT::BOOL,  AVCT::SINGLE))); // 15
     avs.push_back(AttributeFactory::createAttribute("sdouble",   AVC(AVBT::DOUBLE,  AVCT::SINGLE))); // 16
+    {
+        AVC cfg(AVBT::TENSOR, AVCT::SINGLE);
+        cfg.setTensorType(ValueType::from_spec("tensor(x[2])"));
+        avs.push_back(AttributeFactory::createAttribute("tensor", cfg));
+    }
+    avs.push_back(AttributeFactory::createAttribute("predicate", AVC(AVBT::PREDICATE, AVCT::SINGLE))); // 18
+    avs.push_back(AttributeFactory::createAttribute("reference", AVC(AVBT::REFERENCE, AVCT::SINGLE))); // 19
+    avs.push_back(AttributeFactory::createAttribute("raw", AVC(AVBT::RAW, AVCT::SINGLE))); // 20
 
     // simulate a unique only attribute as specified in sd
     AVC cfg(AVBT::INT32, AVCT::SINGLE);
@@ -417,7 +426,11 @@ Test::setupForAttributeTest(FtFeatureTest &ft, bool setup_env)
             .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "sdouble")
             .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "sbyte")
             .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::BOOL,"sbool")
-            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::BOOL,"sebool");
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::BOOL,"sebool")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::TENSOR, "tensor")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::BOOLEANTREE, "predicate")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::REFERENCE, "reference")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::RAW, "raw");
     }
 
     for (const auto & attr : avs) {
@@ -1499,6 +1512,10 @@ Test::testMatch()
          ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "sint");
          ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::ARRAY, "aint");
          ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::WEIGHTEDSET, "wsint");
+         ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "tensor");
+         ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "predicate");
+         ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "reference");
+         ft.getIndexEnv().getBuilder().addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "raw");
 
         FtIndexEnvironment idx_env;
         idx_env.getBuilder()
@@ -1507,7 +1524,11 @@ Test::testMatch()
             .addField(FieldType::INDEX, CollectionType::WEIGHTEDSET, "baz")
             .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, "sint")
             .addField(FieldType::ATTRIBUTE, CollectionType::ARRAY, "aint")
-            .addField(FieldType::ATTRIBUTE, CollectionType::WEIGHTEDSET, "wsint");
+            .addField(FieldType::ATTRIBUTE, CollectionType::WEIGHTEDSET, "wsint")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::TENSOR, "tensor")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::BOOLEANTREE, "predicate")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::REFERENCE, "reference")
+            .addField(FieldType::ATTRIBUTE, CollectionType::SINGLE, DataType::RAW, "raw");
 
         StringList params, in, out;
         FT_SETUP_OK(pt, params, in, out.add("score").add("totalWeight"));
diff --git a/searchlib/src/vespa/searchlib/features/matchfeature.cpp b/searchlib/src/vespa/searchlib/features/matchfeature.cpp
index 53a7ce6e108..7a2148510d4 100644
--- a/searchlib/src/vespa/searchlib/features/matchfeature.cpp
+++ b/searchlib/src/vespa/searchlib/features/matchfeature.cpp
@@ -10,9 +10,34 @@
 
 using namespace search::fef;
 using CollectionType = FieldInfo::CollectionType;
+using DataType = FieldInfo::DataType;
 
 namespace search::features {
 
+namespace {
+
+auto attribute_match_data_types = ParameterDataTypeSet::normalTypeSet();
+
+bool matchable_field(const FieldInfo& info)
+{
+    auto field_type = info.type();
+    if (field_type != FieldType::INDEX && field_type != FieldType::ATTRIBUTE) {
+        return false;
+    }
+    auto data_type = info.get_data_type();
+    if (data_type == DataType::TENSOR || data_type == DataType::RAW) {
+        // not matchable
+        return false;
+    }
+    if (field_type == FieldType::ATTRIBUTE && !attribute_match_data_types.allowedType(data_type)) {
+        // bad data type for attributeMatch feature
+        return false;
+    }
+    return true;
+}
+
+}
+
 MatchExecutor::MatchExecutor(const MatchParams & params) :
     FeatureExecutor(),
     _params(params)
@@ -67,30 +92,28 @@ MatchBlueprint::setup(const IIndexEnvironment & env,
 {
     for (uint32_t i = 0; i < env.getNumFields(); ++i) {
         const FieldInfo * info = env.getField(i);
-        if (info->get_data_type() == FieldInfo::DataType::TENSOR) {
-            // not matchable
+        if (!matchable_field(*info)) {
             continue;
         }
-        if ((info->type() == FieldType::INDEX) || (info->type() == FieldType::ATTRIBUTE)) {
-            _params.weights.push_back(indexproperties::FieldWeight::lookup(env.getProperties(), info->name()));
-            if (info->type() == FieldType::INDEX) {
-                if (info->collection() == CollectionType::SINGLE) {
-                    defineInput("fieldMatch(" + info->name() + ")");
-                } else {
-                    defineInput("elementCompleteness(" + info->name() + ")");
-                }
-            } else if (info->type() == FieldType::ATTRIBUTE) {
-                defineInput("attributeMatch(" + info->name() + ")");
+        _params.weights.push_back(indexproperties::FieldWeight::lookup(env.getProperties(), info->name()));
+        if (info->type() == FieldType::INDEX) {
+            if (info->collection() == CollectionType::SINGLE) {
+                defineInput("fieldMatch(" + info->name() + ")");
+            } else {
+                defineInput("elementCompleteness(" + info->name() + ")");
             }
+        } else if (info->type() == FieldType::ATTRIBUTE) {
+            defineInput("attributeMatch(" + info->name() + ")");
         }
     }
     describeOutput("score", "Normalized sum over all matched fields");
     describeOutput("totalWeight", "Sum of rank weights for all matched fields");
     for (uint32_t i = 0; i < env.getNumFields(); ++i) {
         const FieldInfo * info = env.getField(i);
-        if ((info->type() == FieldType::INDEX) || (info->type() == FieldType::ATTRIBUTE)) {
-            describeOutput("weight." + info->name(), "The rank weight value for field '" + info->name() + "'");
+        if (!matchable_field(*info)) {
+            continue;
         }
+        describeOutput("weight." + info->name(), "The rank weight value for field '" + info->name() + "'");
     }
     return true;
 }