aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--CMakeLists.txt10
-rw-r--r--signature.cpp3
-rw-r--r--signature.hpp1
-rw-r--r--signature_db.cpp247
-rw-r--r--signature_db.hpp60
5 files changed, 320 insertions, 1 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 0c20a0d..815e901 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -17,6 +17,14 @@ add_compile_definitions(PATH_VALSIZE=${PATH_VALSIZE})
include_directories(.)
-add_library(xsig STATIC imageutil.cpp signature.cpp subslice_signature.cpp base64.cpp)
+add_library(xsig STATIC
+ imageutil.cpp
+ signature.cpp
+ subslice_signature.cpp
+ signature_db.cpp
+ base64.cpp
+)
+
+target_compile_options(xsig PRIVATE -Werror=return-type)
add_subdirectory(tests)
diff --git a/signature.cpp b/signature.cpp
index 74c7590..b912198 100644
--- a/signature.cpp
+++ b/signature.cpp
@@ -216,6 +216,9 @@ void signature::dump() const
if (p) p->dump();
}
+bool signature::valid() const
+{return (bool)p;}
+
signature signature::clone() const
{
return signature(*this);
diff --git a/signature.hpp b/signature.hpp
index ba342fa..b655e73 100644
--- a/signature.hpp
+++ b/signature.hpp
@@ -40,6 +40,7 @@ public:
signature& operator=(signature&&)=default;
signature clone() const;//do not use unless absolutely needed
void dump() const;
+ bool valid() const;
double length() const;
double distance(const signature &o) const;
bool operator ==(const signature &o) const;
diff --git a/signature_db.cpp b/signature_db.cpp
new file mode 100644
index 0000000..db2d645
--- /dev/null
+++ b/signature_db.cpp
@@ -0,0 +1,247 @@
+//Chris Xiong 2022
+//License: MPL-2.0
+#include <sqlite3.h>
+
+#include "signature_db.hpp"
+
+enum batch_status
+{
+ single = 0,
+ putsub,
+ findsub
+};
+
+struct signature_db_priv
+{
+ sqlite3 *db;
+ sqlite3_mutex *mtx;
+ sqlite3_stmt *bst;
+ batch_status batch_mode;
+};
+
+signature_db::signature_db()
+{
+ p = new signature_db_priv();
+ sqlite3_open(":memory:", &p->db);
+ p->mtx = sqlite3_db_mutex(p->db);
+ sqlite3_exec(p->db, R"sql(
+ create table images(
+ id int primary key,
+ path text,
+ signature text
+ );
+ )sql", nullptr, nullptr, nullptr);
+ sqlite3_exec(p->db, R"sql(
+ create table subslices(
+ image int,
+ slice int,
+ slicesig text
+ );
+ )sql", nullptr, nullptr, nullptr);
+ sqlite3_exec(p->db, R"sql(
+ create index ssidx on subslices(slicesig);
+ )sql", nullptr, nullptr, nullptr);
+ sqlite3_exec(p->db, R"sql(
+ create table dupes(
+ id1 int,
+ id2 int,
+ dist real,
+ primary key (id1, id2)
+ );
+ )sql", nullptr, nullptr, nullptr);
+ p->bst = nullptr;
+ p->batch_mode = batch_status::single;
+}
+
+signature_db::~signature_db()
+{
+ if (p->bst)
+ sqlite3_finalize(p->bst);
+ sqlite3_close(p->db);
+ delete p;
+}
+
+void signature_db::put_signature(size_t id, const fs::path &path, const signature &sig)
+{
+ sqlite3_stmt *st;
+ std::string sigs = sig.to_string();
+ sqlite3_prepare_v2(p->db, "insert into signatures (id, path, signature) values(?, ?, ?);", -1, &st, 0);
+ sqlite3_bind_int(st, 1, id);
+#if PATH_VALSIZE == 2
+ sqlite3_bind_text16(st, 2, path.c_str(), -1, nullptr);
+#else
+ sqlite3_bind_text(st, 2, path.c_str(), -1, nullptr);
+#endif
+ sqlite3_bind_text(st, 3, sigs.c_str(), -1, nullptr);
+ sqlite3_step(st);
+ sqlite3_finalize(st);
+}
+
+std::pair<fs::path, signature> signature_db::get_signature(size_t id)
+{
+ sqlite3_stmt *st;
+ sqlite3_prepare_v2(p->db, "select path, signature from signatures where id = ?;", -1, &st, 0);
+ sqlite3_bind_int(st, 1, id);
+ int rr = sqlite3_step(st);
+ if (rr == SQLITE_ROW)
+ {
+#if PATH_VALSIZE == 2
+ fs::path path((wchar_t*)sqlite3_column_text16(st, 0));
+#else
+ fs::path path((char*)sqlite3_column_text(st, 0));
+#endif
+ std::string sigs((char*)sqlite3_column_text(st, 1));
+ sqlite3_finalize(st);
+ return std::make_pair(path, signature::from_string(std::move(sigs)));
+ }
+ else
+ {
+ sqlite3_finalize(st);
+ return std::make_pair(fs::path(), signature());
+ }
+}
+
+void signature_db::batch_put_subslice_begin()
+{
+ p->batch_mode = batch_status::putsub;
+ sqlite3_prepare_v2(p->db, "insert into subslices (image, slice, slicesig) values(?, ?, ?);", -1, &p->bst, 0);
+}
+
+void signature_db::put_subslice(size_t id, size_t slice, const signature &slicesig)
+{
+ sqlite3_stmt *st = nullptr;
+ if (p->batch_mode == batch_status::putsub)
+ st = p->bst;
+ else
+ sqlite3_prepare_v2(p->db, "insert into subslices (image, slice, slicesig) values(?, ?, ?);", -1, &st, 0);
+ sqlite3_bind_int(st, 1, id);
+ sqlite3_bind_int(st, 2, slice);
+ std::string slicesigs = slicesig.to_string();
+ sqlite3_bind_text(st, 3, slicesigs.c_str(), -1, nullptr);
+ sqlite3_step(st);
+ if (p->batch_mode == batch_status::putsub)
+ sqlite3_reset(st);
+ else
+ sqlite3_finalize(st);
+}
+
+void signature_db::batch_find_subslice_begin()
+{
+ p->batch_mode = batch_status::findsub;
+ sqlite3_prepare_v2(p->db, "select image, slice from subslices where slicesig = ?;", -1, &p->bst, 0);
+}
+
+std::vector<subslice_t> signature_db::find_subslice(const signature &slicesig)
+{
+ sqlite3_stmt *st = nullptr;
+ if (p->batch_mode == batch_status::findsub)
+ st = p->bst;
+ else
+ sqlite3_prepare_v2(p->db, "select image, slice from subslices where slicesig = ?;", -1, &st, 0);
+
+ std::string slicesigs = slicesig.to_string();
+ sqlite3_bind_text(st, 1, slicesigs.c_str(), -1, nullptr);
+
+ std::vector<subslice_t> ret;
+ while (1)
+ {
+ int r = sqlite3_step(st);
+ if (r != SQLITE_ROW) break;
+ size_t im = sqlite3_column_int(st, 0);
+ size_t sl = sqlite3_column_int(st, 1);
+ ret.push_back({im, sl});
+ }
+ if (p->batch_mode == batch_status::findsub)
+ sqlite3_reset(st);
+ else
+ sqlite3_finalize(st);
+ return ret;
+}
+
+void signature_db::batch_end()
+{
+ p->batch_mode = batch_status::single;
+ sqlite3_finalize(p->bst);
+ p->bst = nullptr;
+}
+
+void signature_db::put_dupe_pair(size_t ida, size_t idb, double dist)
+{
+ sqlite3_stmt *st = nullptr;
+ sqlite3_prepare_v2(p->db, "insert into dupes (id1, id2, dist) values(?, ?, ?);", -1, &st, 0);
+ sqlite3_bind_int(st, 1, ida);
+ sqlite3_bind_int(st, 2, idb);
+ sqlite3_bind_double(st, 3, dist);
+ sqlite3_step(st);
+ sqlite3_finalize(st);
+}
+std::vector<dupe_t> signature_db::dupe_pairs()
+{
+ sqlite3_stmt *st = nullptr;
+ sqlite3_prepare_v2(p->db, "select id1, id2, dist from dupes;", -1, &st, 0);
+ std::vector<dupe_t> ret;
+ while (1)
+ {
+ int r = sqlite3_step(st);
+ if (r != SQLITE_ROW) break;
+ ret.push_back({
+ (size_t)sqlite3_column_int(st, 0),
+ (size_t)sqlite3_column_int(st, 1),
+ sqlite3_column_double(st, 2)
+ });
+ }
+ sqlite3_finalize(st);
+ return ret;
+}
+
+void signature_db::lock()
+{sqlite3_mutex_enter(p->mtx);}
+void signature_db::unlock()
+{sqlite3_mutex_leave(p->mtx);}
+
+bool signature_db::to_db_file(const fs::path &path)
+{
+ sqlite3 *dest;
+ int r;
+#if PATH_VALSIZE == 2
+ r = sqlite3_open16(path.c_str(), &dest);
+#else
+ r = sqlite3_open(path.c_str(), &dest);
+#endif
+ if (r != SQLITE_OK) return false;
+ sqlite3_backup *bk = sqlite3_backup_init(dest, "main", p->db, "main");
+ bool ret = (bk != nullptr);
+ while (ret)
+ {
+ r = sqlite3_backup_step(bk, -1);
+ if (r == SQLITE_DONE) break;
+ else if (r != SQLITE_OK)
+ ret = false;
+ }
+ ret &= (SQLITE_OK == sqlite3_backup_finish(bk));
+ ret &= (SQLITE_OK == sqlite3_close(dest));
+ return ret;
+}
+bool signature_db::from_db_file(const fs::path &path)
+{
+ sqlite3 *src;
+ int r;
+#if PATH_VALSIZE == 2
+ r = sqlite3_open16(path.c_str(), &src);
+#else
+ r = sqlite3_open(path.c_str(), &src);
+#endif
+ if (r != SQLITE_OK) return false;
+ sqlite3_backup *bk = sqlite3_backup_init(p->db, "main", src, "main");
+ bool ret = (bk != nullptr);
+ while (ret)
+ {
+ r = sqlite3_backup_step(bk, -1);
+ if (r == SQLITE_DONE) break;
+ else if (r != SQLITE_OK)
+ ret = false;
+ }
+ ret &= (SQLITE_OK == sqlite3_backup_finish(bk));
+ ret &= (SQLITE_OK == sqlite3_close(src));
+ return ret;
+}
diff --git a/signature_db.hpp b/signature_db.hpp
new file mode 100644
index 0000000..f7259bb
--- /dev/null
+++ b/signature_db.hpp
@@ -0,0 +1,60 @@
+//Chris Xiong 2022
+//License: MPL-2.0
+#ifndef SIGNATURE_DB_HPP
+#define SIGNATURE_DB_HPP
+
+#include <filesystem>
+#include <vector>
+
+#include "signature.hpp"
+
+namespace fs = std::filesystem;
+
+struct subslice_t {size_t id; size_t slice;};
+
+struct dupe_t {size_t id1, id2; double distance;};
+
+struct signature_db_priv;
+
+class signature_db
+{
+private:
+ signature_db_priv *p;
+public:
+ signature_db();
+ ~signature_db();
+
+ //insert image signature into database
+ //id must be unique
+ void put_signature(size_t id, const fs::path &path, const signature &sig);
+ //get image signature from database
+ std::pair<fs::path, signature> get_signature(size_t id);
+
+ //place batch_put_subslice_begin() and batch_end() around a group of
+ //put_subslice() calls to improve performance
+ void batch_put_subslice_begin();
+ //insert subslice into database
+ //(id, slice) must be unique
+ //calling put_subslice_begin() before this is NOT required, but
+ //will improve performance
+ void put_subslice(size_t id, size_t slice, const signature &slicesig);
+
+ //same thing as put_subslice_begin()
+ void batch_find_subslice_begin();
+ //find identical subslices from database
+ std::vector<subslice_t> find_subslice(const signature &slicesig);
+
+ //call this to finish a batch
+ void batch_end();
+
+ void put_dupe_pair(size_t ida, size_t idb, double dist);
+ std::vector<dupe_t> dupe_pairs();
+
+ void lock();
+ void unlock();
+
+ bool to_db_file(const fs::path &path);
+ bool from_db_file(const fs::path &path);
+};
+
+#endif
generated by cgit v1.2.3 (git 2.39.1) at 2024-06-29 08:14:53 +0000