Add regular expression support to streaming search

Introduces an explicit regex query term node (which wraps an
RE2 regex instance internally) and extends the existing
UTF-8 flexible string searcher to use this query node.

Regex matching is optionally case (in)sensitive depending
on the normalization mode used.

Note on `searcher/searcher_test.cpp`: this adds a magic
sentinel `#` char prefix to query term parsing in the test
to let a query term be interpreted as a regex rather than
exact/prefix/suffix/substring match.

3 files changed, 26 insertions, 6 deletions

diff --git a/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp
index c1fa6090021..c0a0249125f 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/strchrfieldsearcher.cpp
@@ -17,7 +17,7 @@ void StrChrFieldSearcher::prepare(search::streaming::QueryTermList& qtl,
 
 void StrChrFieldSearcher::onValue(const document::FieldValue & fv)
 {
-    const document::LiteralFieldValueB & sfv = static_cast<const document::LiteralFieldValueB &>(fv);
+    const auto & sfv = static_cast<const document::LiteralFieldValueB &>(fv);
     vespalib::stringref val = sfv.getValueRef();
     FieldRef fr(val.data(), std::min(maxFieldLength(), val.size()));
     matchDoc(fr);
@@ -25,7 +25,6 @@ void StrChrFieldSearcher::onValue(const document::FieldValue & fv)
 
 bool StrChrFieldSearcher::matchDoc(const FieldRef & fieldRef)
 {
-  bool retval(true);
   if (_qtl.size() > 1) {
     size_t mintsz = shortestTerm();
     if (fieldRef.size() >= mintsz) {
@@ -35,14 +34,14 @@ bool StrChrFieldSearcher::matchDoc(const FieldRef & fieldRef)
     }
   } else {
     for (auto qt : _qtl) {
-      if (fieldRef.size() >= qt->termLen()) {
+      if (fieldRef.size() >= qt->termLen() || qt->isRegex()) {
         _words += matchTerm(fieldRef, *qt);
       } else {
         _words += countWords(fieldRef);
       }
     }
   }
-  return retval;
+  return true;
 }
 
 size_t StrChrFieldSearcher::shortestTerm() const
@@ -50,6 +49,9 @@ size_t StrChrFieldSearcher::shortestTerm() const
   size_t mintsz(_qtl.front()->termLen());
   for (auto it=_qtl.begin()+1, mt=_qtl.end(); it != mt; it++) {
     const QueryTerm & qt = **it;
+    if (qt.isRegex()) {
+        return 0; // Must avoid "too short query term" optimization when using regex
+    }
     mintsz = std::min(mintsz, qt.termLen());
   }
   return mintsz;
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp b/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp
index 78f491198ad..c6deb6eacd1 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.cpp
@@ -1,5 +1,7 @@
 // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 #include "utf8flexiblestringfieldsearcher.h"
+#include <vespa/searchlib/query/streaming/regexp_term.h>
+#include <cassert>
 
 #include <vespa/log/log.h>
 LOG_SETUP(".vsm.searcher.utf8flexiblestringfieldsearcher");
@@ -27,6 +29,17 @@ UTF8FlexibleStringFieldSearcher::matchTerms(const FieldRef & f, const size_t min
 }
 
 size_t
+UTF8FlexibleStringFieldSearcher::match_regexp(const FieldRef & f, search::streaming::QueryTerm & qt)
+{
+    auto* regexp_term = qt.as_regexp_term();
+    assert(regexp_term != nullptr);
+    if (regexp_term->regexp().partial_match({f.data(), f.size()})) {
+        addHit(qt, 0);
+    }
+    return countWords(f);
+}
+
+size_t
 UTF8FlexibleStringFieldSearcher::matchTerm(const FieldRef & f, QueryTerm & qt)
 {
     if (qt.isPrefix()) {
@@ -41,6 +54,9 @@ UTF8FlexibleStringFieldSearcher::matchTerm(const FieldRef & f, QueryTerm & qt)
     } else if (qt.isExactstring()) {
         LOG(debug, "Use exact match for exact term '%s:%s'", qt.index().c_str(), qt.getTerm());
         return matchTermExact(f, qt);
+    } else if (qt.isRegex()) {
+        LOG(debug, "Use regexp match for term '%s:%s'", qt.index().c_str(), qt.getTerm());
+        return match_regexp(f, qt);
     } else {
         if (substring()) {
             LOG(debug, "Use substring match for term '%s:%s'", qt.index().c_str(), qt.getTerm());
diff --git a/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.h b/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.h
index bb1b55dffe4..cd1715ad158 100644
--- a/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.h
+++ b/streamingvisitors/src/vespa/vsm/searcher/utf8flexiblestringfieldsearcher.h
@@ -14,16 +14,18 @@ class UTF8FlexibleStringFieldSearcher : public UTF8StringFieldSearcherBase
 private:
     /**
      * Tries to match the given query term against the content of the given field reference.
-     * Search strategy is choosen based on the query term type.
+     * Search strategy is chosen based on the query term type.
      **/
     size_t matchTerm(const FieldRef & f, search::streaming::QueryTerm & qt) override;
 
     /**
      * Tries to match each query term in the underlying query against the content of the given field reference.
-     * Search strategy is choosen based on the query term type.
+     * Search strategy is chosen based on the query term type.
      **/
     size_t matchTerms(const FieldRef & f, size_t shortestTerm) override;
 
+    size_t match_regexp(const FieldRef & f, search::streaming::QueryTerm & qt);
+
 public:
     std::unique_ptr<FieldSearcher> duplicate() const override;
     explicit UTF8FlexibleStringFieldSearcher(FieldIdT fId);