Skip to content

Commit

Permalink
[Search] Normalize Image Metadata
Browse files Browse the repository at this point in the history 
This CL splits annotation table into three tables to more
efficiently store image/file metadata.

Test: "passes unittests. Works on DUT."
Bug: b:295960328
Change-Id: Iacca5988a20bc040d1599fd72b9f1d29fd619a4f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4787468
Reviewed-by: CJ Huang <chenjih@google.com>
Commit-Queue: Dmitry Grebenyuk <dgrebenyuk@google.com>
Cr-Commit-Position: refs/heads/main@{#1185040}
  • Loading branch information
@Rendok
Rendok authored and Chromium LUCI CQ committed Aug 18, 2023
1 parent 3506afb commit c556cd3
Show file tree
Hide file tree
Showing 15 changed files with 667 additions and 159 deletions.
184 changes: 56 additions & 128 deletions chrome/browser/ash/app_list/search/local_image_search/annotation_storage.cc
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,10 @@
#include "base/logging.h"
#include "base/strings/strcat.h"
#include "base/strings/utf_string_conversions.h"
#include "chrome/browser/ash/app_list/search/local_image_search/annotations_table.h"
#include "chrome/browser/ash/app_list/search/local_image_search/documents_table.h"
#include "chrome/browser/ash/app_list/search/local_image_search/image_annotation_worker.h"
#include "chrome/browser/ash/app_list/search/local_image_search/inverted_index_table.h"
#include "chrome/browser/ash/app_list/search/local_image_search/search_utils.h"
#include "chrome/browser/ash/app_list/search/local_image_search/sql_database.h"
#include "chromeos/ash/components/string_matching/fuzzy_tokenized_string_match.h"
Expand All @@ -24,97 +27,37 @@ using TokenizedString = ::ash::string_matching::TokenizedString;
using Mode = ::ash::string_matching::TokenizedString::Mode;

constexpr double kRelevanceThreshold = 0.79;
constexpr int kVersionNumber = 3;
constexpr int kVersionNumber = 4;

// Initializes a new annotation table, returning a schema version number
// on success. The table can be searched by label and image path.
// The map between label and image is many-to-one.
// on success. The database implements inverted index.
// The table cannot exist when calling this function.
int CreateNewSchema(SqlDatabase* db) {
DVLOG(1) << "Making a table";
if (!db) {
return 0;
}

static constexpr char kQuery[] =
// clang-format off
"CREATE TABLE annotations("
"label TEXT NOT NULL,"
"image_path TEXT NOT NULL,"
"last_modified_time INTEGER NOT NULL,"
"is_ignored INTEGER NOT NULL)";
// clang-format on
std::unique_ptr<sql::Statement> statement =
db->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement || !statement->Run()) {
return 0;
}

static constexpr char kQuery1[] =
"CREATE INDEX ind_annotations_label ON annotations(label)";

std::unique_ptr<sql::Statement> statement1 =
db->GetStatementForQuery(SQL_FROM_HERE, kQuery1);
if (!statement1 || !statement1->Run()) {
return 0;
}

static constexpr char kQuery2[] =
"CREATE INDEX ind_annotations_image_path ON annotations(image_path)";

std::unique_ptr<sql::Statement> statement2 =
db->GetStatementForQuery(SQL_FROM_HERE, kQuery2);
if (!statement2 || !statement2->Run()) {
if (!db || !AnnotationsTable::Create(db) || !DocumentsTable::Create(db) ||
!InvertedIndexTable::Create(db)) {
LOG(ERROR) << "Failed to create schema.";
return 0;
}

return kVersionNumber;
}

int MigrateSchema(SqlDatabase* db, int current_version_number) {
if (!db) {
return 0;
}

if (current_version_number == kVersionNumber) {
return current_version_number;
}

static constexpr char kQuery[] = "DROP TABLE IF EXISTS annotations";
std::unique_ptr<sql::Statement> statement =
db->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement || !statement->Run()) {
if (!db || !AnnotationsTable::Drop(db) || !DocumentsTable::Drop(db) ||
!InvertedIndexTable::Drop(db)) {
LOG(ERROR) << "Failed to drop schema.";
return 0;
}

return CreateNewSchema(db);
}

// Returns sorted `FileSearchResult`s contained in both sorted arrays.
std::vector<FileSearchResult> FindIntersection(
const std::vector<FileSearchResult>& vec1,
const std::vector<FileSearchResult>& vec2) {
std::vector<FileSearchResult> result;

auto it1 = vec1.begin();
auto it2 = vec2.begin();

while (it1 != vec1.end() && it2 != vec2.end()) {
if (it1->file_path < it2->file_path) {
++it1;
} else if (it2->file_path < it1->file_path) {
++it2;
} else {
result.emplace_back(FileSearchResult(it1->file_path, it1->last_modified,
it1->relevance + it2->relevance));
++it1;
++it2;
}
}

return result;
}

} // namespace

ImageInfo::ImageInfo(const std::set<std::string>& annotations,
Expand All @@ -129,18 +72,6 @@ ImageInfo::ImageInfo(const std::set<std::string>& annotations,
ImageInfo::~ImageInfo() = default;
ImageInfo::ImageInfo(const ImageInfo&) = default;

FileSearchResult::FileSearchResult(const base::FilePath& file_path,
const base::Time& last_modified,
double relevance)
: file_path(file_path),
last_modified(last_modified),
relevance(relevance) {}

FileSearchResult::~FileSearchResult() = default;
FileSearchResult::FileSearchResult(const FileSearchResult&) = default;
FileSearchResult& FileSearchResult::operator=(const FileSearchResult&) =
default;

AnnotationStorage::AnnotationStorage(
const base::FilePath& path_to_db,
const std::string& histogram_tag,
Expand Down Expand Up @@ -181,49 +112,41 @@ void AnnotationStorage::Initialize() {

void AnnotationStorage::Insert(const ImageInfo& image_info) {
DCHECK_CALLED_ON_VALID_SEQUENCE(sequence_checker_);
DVLOG(1) << "Insert";

static constexpr char kQuery[] =
// clang-format off
"INSERT INTO annotations(label,image_path,last_modified_time,is_ignored) "
"VALUES(?,?,?,?)";
// clang-format on
DVLOG(1) << "Insert " << image_info.path;

int64_t document_id;
if (!DocumentsTable::InsertOrIgnore(sql_database_.get(), image_info.path,
image_info.last_modified,
DocumentType::kImage) ||
!DocumentsTable::GetDocumentId(sql_database_.get(), image_info.path,
document_id)) {
LOG(ERROR) << "Failed to insert into the db.";
return;
}

for (const auto& annotation : image_info.annotations) {
std::unique_ptr<sql::Statement> statement =
sql_database_->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement) {
return;
}
DVLOG(1) << annotation;
statement->BindString(0, annotation);
statement->BindString(1, image_info.path.value());
statement->BindTime(2, image_info.last_modified);
statement->BindInt(3, image_info.is_ignored);

if (!statement->Run()) {
// TODO(b/260646344): log to UMA instead.
int64_t annotation_id;
if (!AnnotationsTable::InsertOrIgnore(sql_database_.get(), annotation) ||
!AnnotationsTable::GetTermId(sql_database_.get(), annotation,
annotation_id) ||
!InvertedIndexTable::Insert(sql_database_.get(), annotation_id,
document_id)) {
LOG(ERROR) << "Failed to insert into the db.";
return;
}
}
return;
}

void AnnotationStorage::Remove(const base::FilePath& image_path) {
DCHECK_CALLED_ON_VALID_SEQUENCE(sequence_checker_);
DVLOG(1) << "Remove";

static constexpr char kQuery[] = "DELETE FROM annotations WHERE image_path=?";
DVLOG(1) << "Remove " << image_path;

std::unique_ptr<sql::Statement> statement =
sql_database_->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement) {
return;
if (!InvertedIndexTable::Remove(sql_database_.get(), image_path) ||
!DocumentsTable::Remove(sql_database_.get(), image_path) ||
!AnnotationsTable::Prune(sql_database_.get())) {
LOG(ERROR) << "Failed to remove from the db.";
}

statement->BindString(0, image_path.value());

statement->Run();
}

std::vector<ImageInfo> AnnotationStorage::GetAllAnnotations() {
Expand All @@ -232,28 +155,30 @@ std::vector<ImageInfo> AnnotationStorage::GetAllAnnotations() {

static constexpr char kQuery[] =
// clang-format off
"SELECT label,image_path,last_modified_time,is_ignored "
"FROM annotations "
"ORDER BY label";
"SELECT a.term, d.file_path, d.last_modified_time "
"FROM annotations AS a "
"JOIN inverted_index AS ii ON a.term_id = ii.term_id "
"JOIN documents AS d ON ii.document_id = d.document_id "
"ORDER BY a.term, d.file_path";
// clang-format on

std::unique_ptr<sql::Statement> statement =
sql_database_->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement) {
LOG(ERROR) << "Couldn't create the statement";
return {};
}

std::vector<ImageInfo> matched_paths;
while (statement->Step()) {
const base::FilePath path = base::FilePath(statement->ColumnString(1));
const base::Time time = statement->ColumnTime(2);
const bool is_ignored = statement->ColumnBool(3);
DVLOG(1) << "Select find: " << statement->ColumnString(0) << ", " << path
<< ", " << time;
matched_paths.push_back({{statement->ColumnString(0)},
std::move(path),
std::move(time),
is_ignored});
false});
}

return matched_paths;
Expand All @@ -267,15 +192,18 @@ std::vector<ImageInfo> AnnotationStorage::FindImagePath(

static constexpr char kQuery[] =
// clang-format off
"SELECT label,image_path,last_modified_time,is_ignored "
"FROM annotations "
"WHERE image_path=? "
"ORDER BY label";
"SELECT a.term, d.file_path, d.last_modified_time "
"FROM annotations AS a "
"JOIN inverted_index AS ii ON a.term_id = ii.term_id "
"JOIN documents AS d ON ii.document_id = d.document_id "
"WHERE d.file_path=? "
"ORDER BY a.term";
// clang-format on

std::unique_ptr<sql::Statement> statement =
sql_database_->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement) {
LOG(ERROR) << "Couldn't create the statement";
return {};
}
statement->BindString(0, image_path.value());
Expand All @@ -284,13 +212,12 @@ std::vector<ImageInfo> AnnotationStorage::FindImagePath(
while (statement->Step()) {
const base::FilePath path = base::FilePath(statement->ColumnString(1));
const base::Time time = statement->ColumnTime(2);
const bool is_ignored = statement->ColumnBool(3);
DVLOG(1) << "Select find: " << statement->ColumnString(0) << ", " << path
<< ", " << time;
matched_paths.push_back({{statement->ColumnString(0)},
std::move(path),
std::move(time),
is_ignored});
false});
}

return matched_paths;
Expand All @@ -301,19 +228,20 @@ std::vector<FileSearchResult> AnnotationStorage::PrefixSearch(
DCHECK_CALLED_ON_VALID_SEQUENCE(sequence_checker_);
DVLOG(1) << "PrefixSearch " << query_term;

// LIKE is 10 times faster than the linear search.
static constexpr char kQuery[] =
// clang-format off
"SELECT label,image_path,last_modified_time,is_ignored "
"FROM annotations "
"WHERE is_ignored=0 "
"AND label LIKE ? "
"ORDER BY image_path";
"SELECT a.term, d.file_path, d.last_modified_time "
"FROM annotations AS a "
"JOIN inverted_index AS ii ON a.term_id = ii.term_id "
"JOIN documents AS d ON ii.document_id = d.document_id "
"WHERE a.term LIKE ? "
"ORDER BY d.file_path";
// clang-format on

std::unique_ptr<sql::Statement> statement =
sql_database_->GetStatementForQuery(SQL_FROM_HERE, kQuery);
if (!statement) {
LOG(ERROR) << "Couldn't create the statement";
return {};
}
statement->BindString(0, base::StrCat({base::UTF16ToUTF8(query_term), "%"}));
Expand Down
20 changes: 1 addition & 19 deletions chrome/browser/ash/app_list/search/local_image_search/annotation_storage.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
#include "base/files/file_path.h"
#include "base/sequence_checker.h"
#include "base/time/time.h"
#include "chrome/browser/ash/app_list/search/local_image_search/file_search_result.h"

namespace app_list {

Expand Down Expand Up @@ -40,25 +41,6 @@ struct ImageInfo {
ImageInfo& operator=(const ImageInfo&) = delete;
};

// A search result with `relevance` to the supplied query.
struct FileSearchResult {
// The full path to the file.
base::FilePath file_path;
// The file's last modified time.
base::Time last_modified;
// The file's relevance on the scale from 0-1. It represents how closely a
// query matches the file's annotation.
double relevance;

FileSearchResult(const base::FilePath& file_path,
const base::Time& last_modified,
double relevance);

~FileSearchResult();
FileSearchResult(const FileSearchResult&);
FileSearchResult& operator=(const FileSearchResult&);
};

// A persistent storage to efficiently store, retrieve and search annotations.
// Creates or opens a database under `path_to_db`. If `annotation_worker` is
// not null, it updates the database on file changes.
Expand Down

0 comments on commit c556cd3

Please sign in to comment.