Merge pull request #30316 from vespa-engine/balder/use-highlight-termsv8.307.19

Balder/use highlight terms

5 files changed, 50 insertions, 28 deletions

diff --git a/searchlib/src/vespa/searchlib/common/packets.cpp b/searchlib/src/vespa/searchlib/common/packets.cpp
index 3f2b5750631..af5dc5fc110 100644
--- a/searchlib/src/vespa/searchlib/common/packets.cpp
+++ b/searchlib/src/vespa/searchlib/common/packets.cpp
@@ -35,14 +35,14 @@ FS4Properties::FS4Properties()
       _backing()
 { }
 
-FS4Properties::FS4Properties(FS4Properties && rhs)
+FS4Properties::FS4Properties(FS4Properties && rhs) noexcept
     : _entries(std::move(rhs._entries)),
       _name(std::move(rhs._name)),
       _backing(std::move(rhs._backing))
 { }
 
 FS4Properties &
-FS4Properties::operator=(FS4Properties && rhs)
+FS4Properties::operator=(FS4Properties && rhs) noexcept
 {
     _entries = std::move(rhs._entries);
     _name = std::move(rhs._name);
@@ -72,14 +72,25 @@ FS4Properties::setValue(uint32_t entry, const char *value, uint32_t valueSize)
 }
 
 uint32_t
-FS4Properties::getLength()
+FS4Properties::getLength() const noexcept
 {
-    uint32_t len = sizeof(uint32_t) * 2 + getNameLen();
+    uint32_t len = sizeof(uint32_t) * 2 + name().size();
     len += _backing.size();
     len += _entries.size() * sizeof(uint32_t) * 2;
     return len;
 }
 
+vespalib::stringref
+FS4Properties::key(uint32_t entry) const noexcept {
+    auto pair = _entries[entry].first;
+    return {c_str(pair.first), pair.second};
+}
+vespalib::stringref
+FS4Properties::value(uint32_t entry) const noexcept {
+    auto pair = _entries[entry].second;
+    return {c_str(pair.first), pair.second};
+}
+
 vespalib::string
 FS4Properties::toString(uint32_t indent) const
 {
@@ -90,8 +101,8 @@ FS4Properties::toString(uint32_t indent) const
     s += "\n";
     for (uint32_t i = 0; i < size(); ++i) {
         s += make_string("%*s  Entry[%d] {\n", indent, "", i);
-        s += make_string("%*s    key  : %s\n", indent, "", vespalib::string(getKey(i), getKeyLen(i)).c_str());
-        s += make_string("%*s    value: %s\n", indent, "", vespalib::string(getValue(i), getValueLen(i)).c_str());
+        s += make_string("%*s    key  : %s\n", indent, "", string(key(i)).c_str());
+        s += make_string("%*s    value: %s\n", indent, "", string(value(i)).c_str());
         s += make_string("%*s  }\n", indent, "");
     }
     s += make_string("%*s}\n", indent, "");
diff --git a/searchlib/src/vespa/searchlib/common/packets.h b/searchlib/src/vespa/searchlib/common/packets.h
index b04e1fb44d0..fbdae8b51ec 100644
--- a/searchlib/src/vespa/searchlib/common/packets.h
+++ b/searchlib/src/vespa/searchlib/common/packets.h
@@ -50,8 +50,10 @@ private:
     void set(StringRef & e, vespalib::stringref s);
     void allocEntries(uint32_t cnt);
 public:
-    FS4Properties(FS4Properties &&);
-    FS4Properties &operator=(FS4Properties &&);
+    FS4Properties(FS4Properties &&) noexcept;
+    FS4Properties &operator=(FS4Properties &&) noexcept;
+    FS4Properties(const FS4Properties &) = delete;
+    FS4Properties &operator=(const FS4Properties &) = delete;
 
     FS4Properties();
     ~FS4Properties();
@@ -67,16 +69,13 @@ public:
     void setValue(uint32_t entry, vespalib::stringref val) {
         setValue(entry, val.data(), val.size());
     }
-    uint32_t size() const { return _entries.size(); }
-    const char *getName() const { return _name.c_str(); }
-    uint32_t getNameLen() const { return _name.size(); }
-    const char *getKey(uint32_t entry) const { return c_str(_entries[entry].first.first); }
-    uint32_t getKeyLen(uint32_t entry) const { return _entries[entry].first.second; }
-    const char *getValue(uint32_t entry) const { return c_str(_entries[entry].second.first); }
-    uint32_t getValueLen(uint32_t entry) const { return _entries[entry].second.second; }
+    uint32_t size() const noexcept { return _entries.size(); }
+    const vespalib::string & name() const noexcept { return _name; }
+    vespalib::stringref key(uint32_t entry) const noexcept;
+    vespalib::stringref value(uint32_t entry) const noexcept;
 
     // sub-packet methods below
-    uint32_t getLength();
+    uint32_t getLength() const noexcept;
 
     void encode(FNET_DataBuffer &dst);
     bool decode(FNET_DataBuffer &src, uint32_t &len);
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/getdocsumargs.h b/searchsummary/src/vespa/searchsummary/docsummary/getdocsumargs.h
index 0051af29c21..6e0d610da97 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/getdocsumargs.h
+++ b/searchsummary/src/vespa/searchsummary/docsummary/getdocsumargs.h
@@ -46,6 +46,7 @@ public:
     bool dumpFeatures() const { return _dumpFeatures; }
 
     const fef::Properties &highlightTerms() const { return _highlightTerms; }
+    void highlightTerms(fef::Properties & terms) { _highlightTerms = terms; }
     void set_fields(const FieldSet& fields_in) { _fields = fields_in; }
     const FieldSet& get_fields() const { return _fields; }
     bool need_field(vespalib::stringref field) const;
diff --git a/streamingvisitors/src/vespa/searchvisitor/searchvisitor.cpp b/streamingvisitors/src/vespa/searchvisitor/searchvisitor.cpp
index 1875664f6c4..e2eb3535223 100644
--- a/streamingvisitors/src/vespa/searchvisitor/searchvisitor.cpp
+++ b/streamingvisitors/src/vespa/searchvisitor/searchvisitor.cpp
@@ -51,6 +51,7 @@ using vsm::DocsumFilter;
 using vsm::FieldPath;
 using vsm::StorageDocument;
 using vsm::StringFieldIdTMap;
+using vespalib::string;
 
 namespace {
 
@@ -186,6 +187,7 @@ SearchVisitor::SummaryGenerator::SummaryGenerator(const search::IAttributeManage
       _dump_features(),
       _location(),
       _stack_dump(),
+      _highlight_terms(),
       _attr_manager(attr_manager),
       _query_normalization(query_normalization)
 {
@@ -218,6 +220,7 @@ SearchVisitor::SummaryGenerator::get_streaming_docsums_state(vespalib::stringref
     if (_stack_dump.has_value()) {
         ds._args.setStackDump(_stack_dump.value().size(), _stack_dump.value().data());
     }
+    ds._args.highlightTerms(_highlight_terms);
     _docsumWriter->initState(_attr_manager, ds, state->get_resolve_class_info());
     auto insres = _docsum_states.insert(std::make_pair(summary_class, std::move(state)));
     return *insres.first->second;
@@ -429,21 +432,27 @@ SearchVisitor::init(const Parameters & params)
             if (!prop.decode(src, len)) {
                 LOG(warning, "Could not decode rank properties");
             } else {
-                LOG(debug, "Properties[%u]: name '%s', size '%u'", i, prop.getName(), prop.size());
-                if (strcmp(prop.getName(), "rank") == 0) { // pick up rank properties
+                LOG(debug, "Properties[%u]: name '%s', size '%u'", i, prop.name().c_str(), prop.size());
+                if (prop.name() == "rank") { // pick up rank properties
                     for (uint32_t j = 0; j < prop.size(); ++j) {
-                        vespalib::string k{prop.getKey(j), prop.getKeyLen(j)};
-                        vespalib::string v{prop.getValue(j), prop.getValueLen(j)};
-                        LOG(debug, "Properties[%u][%u]: key '%s' -> value '%s'", i, j, k.c_str(), v.c_str());
-                        _rankController.getQueryProperties().add(k, v);
+                        LOG(debug, "Properties[%u][%u]: key '%s' -> value '%s'",
+                            i, j, string(prop.key(j)).c_str(), string(prop.value(j)).c_str());
+                        _rankController.getQueryProperties().add(prop.key(j), prop.value(j));
                     }
-                }
-                if (strcmp(prop.getName(), "feature") == 0) { // pick up feature overrides
+                } else if (prop.name() == "feature") { // pick up feature overrides
+                    for (uint32_t j = 0; j < prop.size(); ++j) {
+                        LOG(debug, "Feature override[%u][%u]: key '%s' -> value '%s'",
+                            i, j, string(prop.key(j)).c_str(), string(prop.value(j)).c_str());
+                        _rankController.getFeatureOverrides().add(prop.key(j), prop.value(j));
+                    }
+                } else if (prop.name() == "highlightterms") {
                     for (uint32_t j = 0; j < prop.size(); ++j) {
-                        vespalib::string k{prop.getKey(j), prop.getKeyLen(j)};
-                        vespalib::string v{prop.getValue(j), prop.getValueLen(j)};
-                        LOG(debug, "Feature override[%u][%u]: key '%s' -> value '%s'", i, j, k.c_str(), v.c_str());
-                        _rankController.getFeatureOverrides().add(k, v);
+                        LOG(debug, "Hightligthterms[%u][%u]: key '%s' -> value '%s'",
+                            i, j, string(prop.key(j)).c_str(), string(prop.value(j)).c_str());
+                        vespalib::stringref index = prop.key(j);
+                        vespalib::stringref term = prop.value(j);
+                        vespalib::string norm_term = QueryNormalization::optional_fold(term, search::TermType::WORD, normalizing_mode(index));
+                        _summaryGenerator.highlightTerms().add(index, norm_term);
                     }
                 }
             }
diff --git a/streamingvisitors/src/vespa/searchvisitor/searchvisitor.h b/streamingvisitors/src/vespa/searchvisitor/searchvisitor.h
index 567cb1b1f2f..7d73a159f6f 100644
--- a/streamingvisitors/src/vespa/searchvisitor/searchvisitor.h
+++ b/streamingvisitors/src/vespa/searchvisitor/searchvisitor.h
@@ -423,6 +423,7 @@ private:
         void set_location(const vespalib::string& location) { _location = location; }
         void set_stack_dump(std::vector<char> stack_dump) { _stack_dump = std::move(stack_dump); }
         void add_summary_field(vespalib::stringref field) { _summaryFields.emplace_back(field); }
+        search::fef::Properties & highlightTerms() { return _highlight_terms;}
     private:
         StreamingDocsumsState& get_streaming_docsums_state(vespalib::stringref summary_class);
         vsm::GetDocsumsStateCallback            _callback;
@@ -434,6 +435,7 @@ private:
         std::optional<bool>                     _dump_features;
         std::optional<vespalib::string>         _location;
         std::optional<std::vector<char>>        _stack_dump;
+        search::fef::Properties                 _highlight_terms;
         const search::IAttributeManager&        _attr_manager;
         const search::QueryNormalization &      _query_normalization;
     };