Merge branch 'main' into v2

acoustid · Feb 25, 2024 · 42cad35 · 42cad35
2 parents d6a68c5 + f4f5a0a
commit 42cad35
Show file tree

Hide file tree

Showing 6 changed files with 55 additions and 19 deletions.
diff --git a/src/index/index_reader.cpp b/src/index/index_reader.cpp
@@ -71,7 +71,7 @@ std::vector<SearchResult> IndexReader::search(const std::vector<uint32_t> &terms
 		const SegmentInfo& s = segments.at(i);
 		SegmentSearcher searcher(s.index(), segmentDataReader(s), s.lastKey());
         segmentHits.clear();
-		searcher.search(sortedTerms.data(), sortedTerms.size(), segmentHits);
+		searcher.search(sortedTerms, segmentHits);
         auto segmentId = s.id();
         for (auto hit : segmentHits) {
             auto docId = hit.first;

diff --git a/src/index/index_reader_test.cpp b/src/index/index_reader_test.cpp
@@ -0,0 +1,42 @@
+// Copyright (C) 2011  Lukas Lalinsky
+// Distributed under the MIT license, see the LICENSE file for details.
+
+#include <gtest/gtest.h>
+#include "util/test_utils.h"
+#include "store/ram_directory.h"
+#include "store/input_stream.h"
+#include "store/output_stream.h"
+#include "top_hits_collector.h"
+#include "index.h"
+#include "index_writer.h"
+#include "index_reader.h"
+
+using namespace Acoustid;
+
+TEST(IndexReaderTest, Search)
+{
+	DirectorySharedPtr dir(new RAMDirectory());
+	IndexSharedPtr index(new Index(dir, true));
+
+	std::vector<uint32_t> fp1 = { 7, 9, 12 };
+	std::vector<uint32_t> fp2 = { 7, 9, 11 };
+
+	{
+		auto writer = index->openWriter();
+		writer->addDocument(1, fp1.data(), fp1.size());
+		writer->commit();
+		writer->addDocument(2, fp2.data(), fp2.size());
+		writer->commit();
+	}
+
+	{
+		IndexReader reader(index);
+		auto results = reader.search(fp1);
+		ASSERT_EQ(2, results.size());
+		ASSERT_EQ(1, results.at(0).docId());
+		ASSERT_EQ(3, results.at(0).score());
+		ASSERT_EQ(2, results.at(1).docId());
+		ASSERT_EQ(2, results.at(1).score());
+	}
+}
+
diff --git a/src/index/search_result.h b/src/index/search_result.h
@@ -29,13 +29,7 @@ class SearchResult {
 inline void sortSearchResults(std::vector<SearchResult> &results)
 {
     std::sort(results.begin(), results.end(), [](const SearchResult &a, const SearchResult &b) {
-        if (a.score() > b.score()) {
-            return true;
-        } else if (a.score() < b.score()) {
-            return false;
-        } else {
-            return a.docId() < b.docId();
-        }
+	return a.score() > b.score() || (a.score() == b.score() && a.docId() < b.docId());
     });
 }
 

diff --git a/src/index/segment_searcher.cpp b/src/index/segment_searcher.cpp
@@ -16,17 +16,17 @@ SegmentSearcher::~SegmentSearcher()
 {
 }
 
-void SegmentSearcher::search(uint32_t *fingerprint, size_t length, std::unordered_map<uint32_t, int> &hits)
+void SegmentSearcher::search(const std::vector<uint32_t> &hashes, std::unordered_map<uint32_t, int> &hits)
 {
 	size_t i = 0, block = 0, lastBlock = SIZE_MAX;
-	while (i < length) {
+	while (i < hashes.size()) {
 		if (block > lastBlock || lastBlock == SIZE_MAX) {
 			size_t localFirstBlock, localLastBlock;
-			if (fingerprint[i] > m_lastKey) {
+			if (hashes[i] > m_lastKey) {
 				// All following items are larger than the last segment's key.
 				return;
 			}
-			if (m_index->search(fingerprint[i], &localFirstBlock, &localLastBlock)) {
+			if (m_index->search(hashes[i], &localFirstBlock, &localLastBlock)) {
 				if (block > localLastBlock) {
 					// We already searched this block and the fingerprint item was not found.
 					i++;
@@ -47,18 +47,18 @@ void SegmentSearcher::search(uint32_t *fingerprint, size_t length, std::unordere
 		std::unique_ptr<BlockDataIterator> blockData(m_dataReader->readBlock(block, firstKey));
 		while (blockData->next()) {
 			uint32_t key = blockData->key();
-			if (key >= fingerprint[i]) {
-				while (key > fingerprint[i]) {
+			if (key >= hashes[i]) {
+				while (key > hashes[i]) {
 					i++;
-					if (i >= length) {
+					if (i >= hashes.size()) {
 						return;
 					}
-					else if (lastKey < fingerprint[i]) {
+					else if (lastKey < hashes[i]) {
 						// There are no longer any items in this block that we could match.
 						goto nextBlock;
 					}
 				}
-				if (key == fingerprint[i]) {
+				if (key == hashes[i]) {
                     auto docId = blockData->value();
                     hits[docId]++;
 				}

diff --git a/src/index/segment_searcher.h b/src/index/segment_searcher.h
@@ -17,7 +17,7 @@ class SegmentSearcher
 	SegmentSearcher(SegmentIndexSharedPtr index, SegmentDataReader *dataReader, uint32_t lastKey = UINT32_MAX);
 	virtual ~SegmentSearcher();
 
-	void search(uint32_t *fingerprint, size_t length, std::unordered_map<uint32_t, int> &hits);
+	void search(const std::vector<uint32_t> &hashes, std::unordered_map<uint32_t, int> &hits);
 
 private:
 	SegmentIndexSharedPtr m_index;

diff --git a/src/server/session.h b/src/server/session.h
@@ -7,12 +7,12 @@
 #include <QMutex>
 #include <QSharedPointer>
 #include <memory>
+#include "index/search_result.h"
 
 namespace Acoustid {
 
 class Index;
 class IndexWriter;
-class SearchResult;
 class OpBatch;
 
 namespace Server {