- Use unique_ptr to hide FuzzyMatcher to make most common SearchContext smaller.

- GC unused stringattribute files.

14 files changed, 34 insertions, 60 deletions

diff --git a/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp b/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp
index 3fa74b78d2a..9b6187bcc77 100644
--- a/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp
+++ b/searchlib/src/tests/attribute/attributemanager/attributemanager_test.cpp
@@ -7,7 +7,6 @@
 #include <vespa/searchlib/attribute/configconverter.h>
 #include <vespa/searchlib/attribute/multinumericattribute.h>
 #include <vespa/searchlib/attribute/multinumericattribute.hpp>
-#include <vespa/searchlib/attribute/stringattribute.h>
 #include <vespa/vespalib/testkit/testapp.h>
 
 #include <vespa/log/log.h>
@@ -31,7 +30,7 @@ using TestAttributeBase = MultiValueNumericAttribute< IntegerAttributeTemplate<i
 class TestAttribute : public TestAttributeBase
 {
 public:
-    TestAttribute(const std::string &name)
+    explicit TestAttribute(const std::string &name)
         : TestAttributeBase(name)
     {}
 
@@ -44,18 +43,17 @@ public:
 
 TEST("Test attribute guards")
 {
-    AttributeVector::SP vec(new TestAttribute("mvint") );
-    TestAttribute * v = static_cast<TestAttribute *> (vec.get());
+    auto v = std::make_shared<TestAttribute>("mvint");
     EXPECT_EQUAL(v->getGen(), unsigned(0));
     EXPECT_EQUAL(v->getRefCount(0), unsigned(0));
     EXPECT_EQUAL(v->oldest_used_gen(), unsigned(0));
     {
-        AttributeGuard g0(vec);
+        AttributeGuard g0(v);
         EXPECT_EQUAL(v->getGen(), unsigned(0));
         EXPECT_EQUAL(v->getRefCount(0), unsigned(1));
         EXPECT_EQUAL(v->oldest_used_gen(), unsigned(0));
         {
-            AttributeGuard g1(vec);
+            AttributeGuard g1(v);
             EXPECT_EQUAL(v->getGen(), unsigned(0));
             EXPECT_EQUAL(v->getRefCount(0), unsigned(2));
             EXPECT_EQUAL(v->oldest_used_gen(), unsigned(0));
@@ -72,14 +70,14 @@ TEST("Test attribute guards")
     EXPECT_EQUAL(v->getRefCount(1), unsigned(0));
     EXPECT_EQUAL(v->oldest_used_gen(), unsigned(1));
     {
-        AttributeGuard g0(vec);
+        AttributeGuard g0(v);
         EXPECT_EQUAL(v->getGen(), unsigned(1));
         EXPECT_EQUAL(v->getRefCount(0), unsigned(0));
         EXPECT_EQUAL(v->getRefCount(1), unsigned(1));
         EXPECT_EQUAL(v->oldest_used_gen(), unsigned(1));
         {
             v->incGen();
-            AttributeGuard g1(vec);
+            AttributeGuard g1(v);
             EXPECT_EQUAL(v->getGen(), unsigned(2));
             EXPECT_EQUAL(v->getRefCount(0), unsigned(0));
             EXPECT_EQUAL(v->getRefCount(1), unsigned(1));
diff --git a/searchlib/src/tests/attribute/stringattribute/stringattribute_test.cpp b/searchlib/src/tests/attribute/stringattribute/stringattribute_test.cpp
index 96039bee15b..2289000a273 100644
--- a/searchlib/src/tests/attribute/stringattribute/stringattribute_test.cpp
+++ b/searchlib/src/tests/attribute/stringattribute/stringattribute_test.cpp
@@ -386,8 +386,8 @@ testSingleValue(Attribute & svsa, Config &cfg)
 TEST("testSingleValue")
 {
     EXPECT_EQUAL(24u, sizeof(SearchContext));
-    EXPECT_EQUAL(96u, sizeof(StringSearchHelper));
-    EXPECT_EQUAL(144u, sizeof(attribute::SingleStringEnumSearchContext));
+    EXPECT_EQUAL(32u, sizeof(StringSearchHelper));
+    EXPECT_EQUAL(80u, sizeof(attribute::SingleStringEnumSearchContext));
     {
         Config cfg(BasicType::STRING, CollectionType::SINGLE);
         SingleValueStringAttribute svsa("svsa", cfg);
diff --git a/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt b/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt
index 704db67aa03..72c7efe3094 100644
--- a/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt
+++ b/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt
@@ -127,7 +127,6 @@ vespa_add_library(searchlib_attribute OBJECT
     single_string_enum_search_context.cpp
     single_string_enum_hint_search_context.cpp
     sourceselector.cpp
-    stringattribute.cpp
     stringbase.cpp
     string_matcher.cpp
     string_search_context.cpp
diff --git a/searchlib/src/vespa/searchlib/attribute/multistringattribute.hpp b/searchlib/src/vespa/searchlib/attribute/multistringattribute.hpp
index 5e2d279a756..c34f64a7da4 100644
--- a/searchlib/src/vespa/searchlib/attribute/multistringattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/multistringattribute.hpp
@@ -3,7 +3,6 @@
 #pragma once
 
 #include "multistringattribute.h"
-#include "stringattribute.h"
 #include "enumattribute.hpp"
 #include "enumerated_multi_value_read_view.h"
 #include "multienumattribute.hpp"
diff --git a/searchlib/src/vespa/searchlib/attribute/multistringpostattribute.hpp b/searchlib/src/vespa/searchlib/attribute/multistringpostattribute.hpp
index cfd00f84636..1c1ad909da6 100644
--- a/searchlib/src/vespa/searchlib/attribute/multistringpostattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/multistringpostattribute.hpp
@@ -3,7 +3,6 @@
 #pragma once
 
 #include "multistringpostattribute.h"
-#include "stringattribute.h"
 #include "multistringattribute.hpp"
 #include "multi_string_enum_search_context.h"
 #include <vespa/searchlib/query/query_term_simple.h>
diff --git a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h
index 61bd8e2e2bb..40682002818 100644
--- a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h
+++ b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h
@@ -10,6 +10,7 @@
 #include <vespa/searchcommon/attribute/search_context_params.h>
 #include <vespa/searchcommon/common/range.h>
 #include <vespa/vespalib/util/regexp.h>
+#include <vespa/vespalib/fuzzy/fuzzy_matcher.h>
 #include <regex>
 
 namespace search::attribute {
@@ -52,7 +53,7 @@ protected:
     PostingListSearchContext(const IEnumStoreDictionary& dictionary, uint32_t docIdLimit, uint64_t numValues, bool hasWeight,
                              uint32_t minBvDocFreq, bool useBitVector, const ISearchContext &baseSearchCtx);
 
-    ~PostingListSearchContext();
+    ~PostingListSearchContext() override;
 
     void lookupTerm(const vespalib::datastore::EntryComparator &comp);
     void lookupRange(const vespalib::datastore::EntryComparator &low, const vespalib::datastore::EntryComparator &high);
diff --git a/searchlib/src/vespa/searchlib/attribute/single_enum_search_context.h b/searchlib/src/vespa/searchlib/attribute/single_enum_search_context.h
index f0e6908fc14..83d6c696117 100644
--- a/searchlib/src/vespa/searchlib/attribute/single_enum_search_context.h
+++ b/searchlib/src/vespa/searchlib/attribute/single_enum_search_context.h
@@ -4,7 +4,6 @@
 
 #include "search_context.h"
 #include "enumstore.h"
-#include "multi_value_mapping.h"
 
 namespace search::attribute {
 
@@ -21,11 +20,11 @@ protected:
     const vespalib::datastore::AtomicEntryRef* _enum_indices;
     const EnumStoreT<T>&        _enum_store;
 
-    int32_t onFind(DocId docId, int32_t elemId, int32_t & weight) const override final {
+    int32_t onFind(DocId docId, int32_t elemId, int32_t & weight) const final {
         return find(docId, elemId, weight);
     }
 
-    int32_t onFind(DocId docId, int32_t elemId) const override final {
+    int32_t onFind(DocId docId, int32_t elemId) const final {
         return find(docId, elemId);
     }
 
diff --git a/searchlib/src/vespa/searchlib/attribute/singlestringattribute.hpp b/searchlib/src/vespa/searchlib/attribute/singlestringattribute.hpp
index 64abbf86108..82a4393fc91 100644
--- a/searchlib/src/vespa/searchlib/attribute/singlestringattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/singlestringattribute.hpp
@@ -3,7 +3,6 @@
 #pragma once
 
 #include "singlestringattribute.h"
-#include "stringattribute.h"
 #include "singleenumattribute.hpp"
 #include "attributevector.hpp"
 #include "single_string_enum_hint_search_context.h"
diff --git a/searchlib/src/vespa/searchlib/attribute/string_search_context.cpp b/searchlib/src/vespa/searchlib/attribute/string_search_context.cpp
index aa7aa016720..e548ab8078c 100644
--- a/searchlib/src/vespa/searchlib/attribute/string_search_context.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/string_search_context.cpp
@@ -3,8 +3,9 @@
 #include "string_search_context.h"
 #include "enumhintsearchcontext.h"
 #include "enumstore.h"
-#include <vespa/vespalib/util/regexp.h>
 #include <vespa/searchlib/query/query_term_ucs4.h>
+#include <vespa/vespalib/util/regexp.h>
+#include <vespa/vespalib/fuzzy/fuzzy_matcher.h>
 
 namespace search::attribute {
 
diff --git a/searchlib/src/vespa/searchlib/attribute/string_search_helper.cpp b/searchlib/src/vespa/searchlib/attribute/string_search_helper.cpp
index 5df0efe256e..206c2bcbd69 100644
--- a/searchlib/src/vespa/searchlib/attribute/string_search_helper.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/string_search_helper.cpp
@@ -4,6 +4,8 @@
 #include <vespa/searchlib/query/query_term_ucs4.h>
 #include <vespa/vespalib/text/lowercase.h>
 #include <vespa/vespalib/text/utf8.h>
+#include <vespa/vespalib/fuzzy/fuzzy_matcher.h>
+
 
 namespace search::attribute {
 
@@ -18,17 +20,14 @@ StringSearchHelper::StringSearchHelper(QueryTermUCS4 & term, bool cased)
       _isFuzzy(term.isFuzzy())
 {
     if (isRegex()) {
-        if (isCased()) {
-            _regex = vespalib::Regex::from_pattern(term.getTerm(), vespalib::Regex::Options::None);
-        } else {
-            _regex = vespalib::Regex::from_pattern(term.getTerm(), vespalib::Regex::Options::IgnoreCase);
-        }
+        _regex = (isCased())
+                ? vespalib::Regex::from_pattern(term.getTerm(), vespalib::Regex::Options::None)
+                : vespalib::Regex::from_pattern(term.getTerm(), vespalib::Regex::Options::IgnoreCase);
     } else if (isFuzzy()) {
-        _fuzzyMatcher = vespalib::FuzzyMatcher(
-                term.getTerm(),
-                term.getFuzzyMaxEditDistance(),
-                term.getFuzzyPrefixLength(),
-                isCased());
+        _fuzzyMatcher = std::make_unique<vespalib::FuzzyMatcher>(term.getTerm(),
+                                                                 term.getFuzzyMaxEditDistance(),
+                                                                 term.getFuzzyPrefixLength(),
+                                                                 isCased());
     } else if (isCased()) {
         _term._char = term.getTerm();
         _termLen = term.getTermLen();
@@ -44,7 +43,7 @@ StringSearchHelper::~StringSearchHelper() = default;
 bool
 StringSearchHelper::isMatch(const char *src) const {
     if (__builtin_expect(isRegex(), false)) {
-        return getRegex().valid() ? getRegex().partial_match(std::string_view(src)) : false;
+        return getRegex().valid() && getRegex().partial_match(std::string_view(src));
     }
     if (__builtin_expect(isFuzzy(), false)) {
         return getFuzzyMatcher().isMatch(src);
diff --git a/searchlib/src/vespa/searchlib/attribute/string_search_helper.h b/searchlib/src/vespa/searchlib/attribute/string_search_helper.h
index 5947f397359..146d1653a3e 100644
--- a/searchlib/src/vespa/searchlib/attribute/string_search_helper.h
+++ b/searchlib/src/vespa/searchlib/attribute/string_search_helper.h
@@ -4,8 +4,8 @@
 
 #include <vespa/fastlib/text/unicodeutil.h>
 #include <vespa/vespalib/regex/regex.h>
-#include <vespa/vespalib/fuzzy/fuzzy_matcher.h>
 
+namespace vespalib { class FuzzyMatcher; }
 namespace search { class QueryTermUCS4; }
 
 namespace search::attribute {
@@ -18,17 +18,19 @@ class StringSearchHelper {
 public:
     StringSearchHelper(QueryTermUCS4 & qTerm, bool cased);
     StringSearchHelper(StringSearchHelper&&) noexcept;
+    StringSearchHelper(const StringSearchHelper &) = delete;
+    StringSearchHelper & operator =(const StringSearchHelper &) = delete;
     ~StringSearchHelper();
     bool isMatch(const char *src) const;
-    bool isPrefix() const { return _isPrefix; }
-    bool isRegex() const { return _isRegex; }
-    bool isCased() const { return _isCased; }
-    bool isFuzzy() const { return _isFuzzy; }
-    const vespalib::Regex & getRegex() const { return _regex; }
-    const vespalib::FuzzyMatcher & getFuzzyMatcher() const { return _fuzzyMatcher; }
+    bool isPrefix() const noexcept{ return _isPrefix; }
+    bool isRegex() const noexcept{ return _isRegex; }
+    bool isCased() const noexcept{ return _isCased; }
+    bool isFuzzy() const noexcept{ return _isFuzzy; }
+    const vespalib::Regex & getRegex() const noexcept { return _regex; }
+    const vespalib::FuzzyMatcher & getFuzzyMatcher() const noexcept { return *_fuzzyMatcher; }
 private:
     vespalib::Regex                _regex;
-    vespalib::FuzzyMatcher         _fuzzyMatcher;
+    std::unique_ptr<vespalib::FuzzyMatcher> _fuzzyMatcher;
     union {
         const ucs4_t *_ucs4;
         const char   *_char;
diff --git a/searchlib/src/vespa/searchlib/attribute/stringattribute.cpp b/searchlib/src/vespa/searchlib/attribute/stringattribute.cpp
deleted file mode 100644
index 2f47f2f9862..00000000000
--- a/searchlib/src/vespa/searchlib/attribute/stringattribute.cpp
+++ /dev/null
@@ -1,8 +0,0 @@
-// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
-
-
-#include "stringattribute.h"
-
-namespace search {
-
-} // namespace search
diff --git a/searchlib/src/vespa/searchlib/attribute/stringattribute.h b/searchlib/src/vespa/searchlib/attribute/stringattribute.h
deleted file mode 100644
index 70508f14168..00000000000
--- a/searchlib/src/vespa/searchlib/attribute/stringattribute.h
+++ /dev/null
@@ -1,11 +0,0 @@
-// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
-
-#pragma once
-
-#include "stringbase.h"
-#include "enumstore.h"
-
-namespace search {
-
-} // namespace search
-
diff --git a/searchlib/src/vespa/searchlib/attribute/stringbase.h b/searchlib/src/vespa/searchlib/attribute/stringbase.h
index b1e55c3d640..e20a40d2df3 100644
--- a/searchlib/src/vespa/searchlib/attribute/stringbase.h
+++ b/searchlib/src/vespa/searchlib/attribute/stringbase.h
@@ -4,11 +4,8 @@
 
 #include "no_loaded_vector.h"
 #include "attributevector.h"
-#include "changevector.h"
 #include "i_enum_store.h"
 #include "loadedenumvalue.h"
-#include "search_context.h"
-#include "string_matcher.h"
 #include "string_search_context.h"
 
 namespace search {