Merge pull request #26836 from vespa-engine/geirst/streaming-field-searcher-refactorv8.157.14

Streaming field searcher refactor

14 files changed, 49 insertions, 60 deletions

diff --git a/streamingvisitors/src/vespa/vsm/common/document.h b/streamingvisitors/src/vespa/vsm/common/document.h
index de9ab052aa1..365d0e33ed0 100644
--- a/streamingvisitors/src/vespa/vsm/common/document.h
+++ b/streamingvisitors/src/vespa/vsm/common/document.h
@@ -13,7 +13,7 @@ namespace vespalib {
 namespace vsm {
 
 /// Type to identify fields in documents.
-using FieldIdT = unsigned int;
+using FieldIdT = uint32_t;
 /// A type to represent a list of FieldIds.
 using FieldIdTList = std::vector<FieldIdT>;
 /// A type to represent all the fields contained in all the indexs.
diff --git a/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.cpp
index cbf8903caab..9a89d0bebae 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.cpp
@@ -61,13 +61,12 @@ void FieldSearcherBase::prepare(const QueryTermList & qtl)
     _qtlFastBuffer.resize(sizeof(*_qtlFast)*(_qtl.size()+1), 0x13);
     _qtlFast = reinterpret_cast<v16qi *>(reinterpret_cast<unsigned long>(&_qtlFastBuffer[0]+15) & ~0xf);
     _qtlFastSize = 0;
-    for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-        const QueryTerm & qt = **it;
-        memcpy(&_qtlFast[_qtlFastSize++], qt.getTerm(), std::min(size_t(16), qt.termLen()));
+    for (auto qt : _qtl) {
+        memcpy(&_qtlFast[_qtlFastSize++], qt->getTerm(), std::min(size_t(16), qt->termLen()));
     }
 }
 
-FieldSearcher::FieldSearcher(const FieldIdT & fId, bool defaultPrefix) :
+FieldSearcher::FieldSearcher(FieldIdT fId, bool defaultPrefix) :
     FieldSearcherBase(),
     _field(fId),
     _matchType(defaultPrefix ? PREFIX : REGULAR),
@@ -89,16 +88,14 @@ FieldSearcher::~FieldSearcher() = default;
 
 bool FieldSearcher::search(const StorageDocument & doc)
 {
-    for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-        QueryTerm & qt = **it;
-        QueryTerm::FieldInfo & fInfo = qt.getFieldInfo(field());
-        fInfo.setHitOffset(qt.getHitList().size());
+    for (auto qt : _qtl) {
+        QueryTerm::FieldInfo & fInfo = qt->getFieldInfo(field());
+        fInfo.setHitOffset(qt->getHitList().size());
     }
     onSearch(doc);
-    for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-        QueryTerm & qt = **it;
-        QueryTerm::FieldInfo & fInfo = qt.getFieldInfo(field());
-        fInfo.setHitCount(qt.getHitList().size() - fInfo.getHitOffset());
+    for(auto qt : _qtl) {
+        QueryTerm::FieldInfo & fInfo = qt->getFieldInfo(field());
+        fInfo.setHitCount(qt->getHitList().size() - fInfo.getHitOffset());
         fInfo.setFieldLength(_words);
     }
     _words = 0;
@@ -132,9 +129,8 @@ size_t FieldSearcher::countWords(const FieldRef & f)
 
 void FieldSearcher::prepareFieldId()
 {
-    for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-        QueryTerm & qt = **it;
-        qt.resizeFieldId(field());
+    for(auto qt : _qtl) {
+        qt->resizeFieldId(field());
     }
 }
 
@@ -232,26 +228,26 @@ void FieldIdTSearcherMap::prepare(const DocumentTypeIndexFieldMapT& difm,
     QueryTermList qtl;
     query.getLeafs(qtl);
     vespalib::string tmp;
-    for (FieldIdTSearcherMap::iterator it = begin(), mt = end(); it != mt; it++) {
+    for (auto& searcher : *this) {
         QueryTermList onlyInIndex;
-        FieldIdT fid = (*it)->field();
-        for (QueryTermList::iterator qt = qtl.begin(), mqt = qtl.end(); qt != mqt; qt++) {
-            QueryTerm * q = *qt;
-            for (DocumentTypeIndexFieldMapT::const_iterator dt(difm.begin()), dmt(difm.end()); dt != dmt; dt++) {
-                const IndexFieldMapT & fim = dt->second;
-                IndexFieldMapT::const_iterator found = fim.find(FieldSearchSpecMap::stripNonFields(q->index()));
+        FieldIdT fid = searcher->field();
+        for (auto qt : qtl) {
+            for (const auto& doc_type_elem : difm) {
+                const IndexFieldMapT & fim = doc_type_elem.second;
+                auto found = fim.find(FieldSearchSpecMap::stripNonFields(qt->index()));
                 if (found != fim.end()) {
                     const FieldIdTList & index = found->second;
-                    if ((find(index.begin(), index.end(), fid) != index.end()) && (find(onlyInIndex.begin(), onlyInIndex.end(), q) == onlyInIndex.end())) {
-                        onlyInIndex.push_back(q);
+                    if ((find(index.begin(), index.end(), fid) != index.end()) && (find(onlyInIndex.begin(), onlyInIndex.end(), qt) == onlyInIndex.end())) {
+                        onlyInIndex.push_back(qt);
                     }
                 } else {
-                    LOG(debug, "Could not find the requested index=%s in the index config map. Query does not fit search definition.", q->index().c_str());
+                    LOG(debug, "Could not find the requested index=%s in the index config map. Query does not fit search definition.",
+                        qt->index().c_str());
                 }
             }
         }
         /// Should perhaps do a unique on onlyInIndex
-        (*it)->prepare(onlyInIndex, searcherBuf, field_paths, query_env);
+        searcher->prepare(onlyInIndex, searcherBuf, field_paths, query_env);
         if (LOG_WOULD_LOG(spam)) {
             char tmpBuf[16];
             snprintf(tmpBuf, sizeof(tmpBuf), "%d", fid);
diff --git a/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.h b/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.h
index 879902ca514..abc2bc9d870 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.h
+++ b/streamingvisitors/src/vespa/vsm/searcher/fieldsearcher.h
@@ -52,7 +52,7 @@ public:
         EXACT
     };
 
-    FieldSearcher(const FieldIdT & fId, bool defaultPrefix=false);
+    FieldSearcher(FieldIdT fId, bool defaultPrefix=false);
     ~FieldSearcher() override;
     virtual std::unique_ptr<FieldSearcher> duplicate() const = 0;
     bool search(const StorageDocument & doc);
@@ -61,8 +61,8 @@ public:
                          const vsm::FieldPathMapT& field_paths,
                          search::fef::IQueryEnvironment& query_env);
 
-    const FieldIdT & field()         const { return _field; }
-    void field(const FieldIdT & v)         { _field = v; prepareFieldId(); }
+    FieldIdT field()                 const { return _field; }
+    void field(FieldIdT v)                 { _field = v; prepareFieldId(); }
     bool prefix()                    const { return _matchType == PREFIX; }
     bool substring()                 const { return _matchType == SUBSTRING; }
     bool suffix()                    const { return _matchType == SUFFIX; }
diff --git a/streamingvisitors/src/vespa/vsm/searcher/floatfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/floatfieldsearcher.cpp
index 8585975ca3c..578fc9fe0e5 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/floatfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/floatfieldsearcher.cpp
@@ -36,8 +36,7 @@ void FloatFieldSearcherT<T>::prepare(search::streaming::QueryTermList& qtl,
 {
     _floatTerm.clear();
     FieldSearcher::prepare(qtl, buf, field_paths, query_env);
-    for (QueryTermList::const_iterator it=qtl.begin(); it < qtl.end(); it++) {
-    const QueryTerm * qt = *it;
+    for (auto qt : qtl) {
     size_t sz(qt->termLen());
         if (sz) {
             double low;
diff --git a/streamingvisitors/src/vespa/vsm/searcher/geo_pos_field_searcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/geo_pos_field_searcher.cpp
index 6b0bbbb368d..43ecba29b33 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/geo_pos_field_searcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/geo_pos_field_searcher.cpp
@@ -35,7 +35,7 @@ void GeoPosFieldSearcher::prepare(search::streaming::QueryTermList& qtl,
 {
     _geoPosTerm.clear();
     FieldSearcher::prepare(qtl, buf, field_paths, query_env);
-    for (const QueryTerm * qt : qtl) {
+    for (auto qt : qtl) {
         const vespalib::string & str = qt->getTermString();
         GeoLocationParser parser;
         bool valid = parser.parseNoField(str);
diff --git a/streamingvisitors/src/vespa/vsm/searcher/intfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/intfieldsearcher.cpp
index 18b286946f7..0fb71a3c3c6 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/intfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/intfieldsearcher.cpp
@@ -26,8 +26,7 @@ void IntFieldSearcher::prepare(search::streaming::QueryTermList& qtl,
 {
     _intTerm.clear();
     FieldSearcher::prepare(qtl, buf, field_paths, query_env);
-    for (QueryTermList::const_iterator it=qtl.begin(); it < qtl.end(); it++) {
-        const QueryTerm * qt = *it;
+    for (auto qt : qtl) {
         size_t sz(qt->termLen());
         if (sz) {
             int64_t low;
diff --git a/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.cpp
index 045ec9b04a3..f064760e55d 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.cpp
@@ -52,7 +52,7 @@ NearestNeighborFieldSearcher::NodeAndCalc::NodeAndCalc(search::streaming::Neares
 {
 }
 
-NearestNeighborFieldSearcher::NearestNeighborFieldSearcher(const FieldIdT& fid,
+NearestNeighborFieldSearcher::NearestNeighborFieldSearcher(FieldIdT fid,
                                                            search::attribute::DistanceMetric metric)
     : FieldSearcher(fid),
       _metric(metric),
diff --git a/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.h b/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.h
index 83f2c444e5a..ba39b91c677 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.h
+++ b/streamingvisitors/src/vespa/vsm/searcher/nearest_neighbor_field_searcher.h
@@ -38,7 +38,7 @@ private:
     std::vector<NodeAndCalc> _calcs;
 
 public:
-    NearestNeighborFieldSearcher(const FieldIdT& fid,
+    NearestNeighborFieldSearcher(FieldIdT fid,
                                  search::attribute::DistanceMetric metric);
     ~NearestNeighborFieldSearcher();
 
diff --git a/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp
index f15290526d9..6a46e4604be 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp
@@ -34,10 +34,9 @@ bool StrChrFieldSearcher::matchDoc(const FieldRef & fieldRef)
       _words += countWords(fieldRef);
     }
   } else {
-    for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-      QueryTerm & qt = **it;
-      if (fieldRef.size() >= qt.termLen()) {
-        _words += matchTerm(fieldRef, qt);
+    for (auto qt : _qtl) {
+      if (fieldRef.size() >= qt->termLen()) {
+        _words += matchTerm(fieldRef, *qt);
       } else {
         _words += countWords(fieldRef);
       }
@@ -49,7 +48,7 @@ bool StrChrFieldSearcher::matchDoc(const FieldRef & fieldRef)
 size_t StrChrFieldSearcher::shortestTerm() const
 {
   size_t mintsz(_qtl.front()->termLen());
-  for(QueryTermList::const_iterator it=_qtl.begin()+1, mt=_qtl.end(); it != mt; it++) {
+  for (auto it=_qtl.begin()+1, mt=_qtl.end(); it != mt; it++) {
     const QueryTerm & qt = **it;
     mintsz = std::min(mintsz, qt.termLen());
   }
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8exactstringfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/utf8exactstringfieldsearcher.cpp
index 977602a691c..a7ad02fa9d9 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8exactstringfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8exactstringfieldsearcher.cpp
@@ -17,9 +17,8 @@ size_t
 UTF8ExactStringFieldSearcher::matchTerms(const FieldRef & f, const size_t mintsz)
 {
     (void) mintsz;
-    for (QueryTermList::iterator it = _qtl.begin(), mt = _qtl.end(); it != mt; ++it) {
-        QueryTerm & qt = **it;
-        matchTermExact(f, qt);
+    for (auto qt : _qtl) {
+        matchTermExact(f, *qt);
     }
     return 1;
 }
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp
index 9aef99f9fa1..5809738456f 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp
@@ -20,8 +20,8 @@ UTF8FlexibleStringFieldSearcher::matchTerms(const FieldRef & f, const size_t min
 {
     (void) mintsz;
     size_t words = 0;
-    for (QueryTermList::iterator it = _qtl.begin(); it != _qtl.end(); ++it) {
-        words = matchTerm(f, **it);
+    for (auto qt : _qtl) {
+        words = matchTerm(f, *qt);
     }
     return words;
 }
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8strchrfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/utf8strchrfieldsearcher.cpp
index 0d93009655c..e8ac87b836b 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8strchrfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8strchrfieldsearcher.cpp
@@ -29,15 +29,14 @@ UTF8StrChrFieldSearcher::matchTerms(const FieldRef & f, const size_t mintsz)
     for( ; n < e; ) {
         if (!*n) { _zeroCount++; n++; }
         n = tokenize(n, _buf->capacity(), fn, fl);
-        for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-            QueryTerm & qt = **it;
+        for (auto qt : _qtl) {
             const cmptype_t * term;
-            termsize_t tsz = qt.term(term);
-            if ((tsz <= fl) && (prefix() || qt.isPrefix() || (tsz == fl))) {
+            termsize_t tsz = qt->term(term);
+            if ((tsz <= fl) && (prefix() || qt->isPrefix() || (tsz == fl))) {
                 const cmptype_t *tt=term, *et=term+tsz;
                 for (const cmptype_t *fnt=fn; (tt < et) && (*tt == *fnt); tt++, fnt++);
                 if (tt == et) {
-                    addHit(qt, words);
+                    addHit(*qt, words);
                 }
             }
         }
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8substringsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/utf8substringsearcher.cpp
index fd327d3a3df..adcf7a937c1 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8substringsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8substringsearcher.cpp
@@ -29,15 +29,14 @@ UTF8SubStringFieldSearcher::matchTerms(const FieldRef & f, const size_t mintsz)
     const cmptype_t * fre = fe - mintsz;
     termcount_t words(0);
     for(words = 0; fn <= fre; ) {
-        for(QueryTermList::iterator it=_qtl.begin(), mt=_qtl.end(); it != mt; it++) {
-            QueryTerm & qt = **it;
+        for (auto qt : _qtl) {
             const cmptype_t * term;
-            termsize_t tsz = qt.term(term);
+            termsize_t tsz = qt->term(term);
 
             const cmptype_t *tt=term, *et=term+tsz, *fnt=fn;
             for (; (tt < et) && (*tt == *fnt); tt++, fnt++);
             if (tt == et) {
-                addHit(qt, words);
+                addHit(*qt, words);
             }
         }
         if ( ! Fast_UnicodeUtil::IsWordChar(*fn++) ) {
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8substringsnippetmodifier.cpp b/streamingvisitors/src/vespa/vsm/searcher/utf8substringsnippetmodifier.cpp
index 9046c0063d5..89388c01354 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8substringsnippetmodifier.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8substringsnippetmodifier.cpp
@@ -41,10 +41,9 @@ UTF8SubstringSnippetModifier::matchTerms(const FieldRef & f, const size_t mintsz
     const cmptype_t * drend = dend - mintsz;
     termcount_t words = 0;
     for(; ditr <= drend; ) {
-        for (QueryTermList::iterator itr = _qtl.begin(); itr != _qtl.end(); ++itr) {
-            QueryTerm & qt = **itr;
+        for (auto qt : _qtl) {
             const cmptype_t * term;
-            termsize_t tsz = qt.term(term);
+            termsize_t tsz = qt->term(term);
 
             const cmptype_t * titr = term;
             const cmptype_t * tend = term + tsz;
@@ -58,7 +57,7 @@ UTF8SubstringSnippetModifier::matchTerms(const FieldRef & f, const size_t mintsz
                     // If we have overlapping matches only the first one will be considered.
                     insertSeparators(mbegin, mend);
                 }
-                addHit(qt, words);
+                addHit(*qt, words);
             }
         }
         if ( ! Fast_UnicodeUtil::IsWordChar(*ditr++) ) {