Add "reverse image search".

Fixed a stupid performance degradation in the signature library
in the process.

8 files changed, 150 insertions, 47 deletions

diff --git a/qdeduper/imageitem.cpp b/qdeduper/imageitem.cpp
index f18d32c..9ee7c6f 100644
--- a/qdeduper/imageitem.cpp
+++ b/qdeduper/imageitem.cpp
@@ -9,7 +9,6 @@
 #include <QFontMetrics>
 #include <QPainter>
 #include <QLocale>
-#include <qnamespace.h>
 
 #define DEBUGPAINT 0
 
diff --git a/qdeduper/mingui.cpp b/qdeduper/mingui.cpp
index ba802c7..f93fda3 100644
--- a/qdeduper/mingui.cpp
+++ b/qdeduper/mingui.cpp
@@ -126,11 +126,11 @@ DeduperMainWindow::DeduperMainWindow()
         auto &k = keys[i];
         QAction *a = new QAction();
         a->setShortcut(QKeySequence(k));
-        QObject::connect(a, &QAction::triggered, [this, i](){this->mark_toggle(i);});
+        QObject::connect(a, &QAction::triggered, std::bind(&DeduperMainWindow::mark_toggle, this, i));
         selhk.push_back(a);
         QAction *sa = new QAction();
         sa->setShortcut(QKeySequence(Qt::Modifier::SHIFT | k));
-        QObject::connect(sa, &QAction::triggered, [this, i](){this->mark_all_but(i);});
+        QObject::connect(sa, &QAction::triggered, std::bind(&DeduperMainWindow::mark_all_but, this, i));
         selhk.push_back(a);
     }
     this->addActions(selhk);
@@ -220,7 +220,31 @@ void DeduperMainWindow::setup_menu()
     this->addAction(loadlist);
 
     file->addSeparator();
-    file->addAction("Search for Image...");
+    QAction *search_img = file->addAction("Search for Image...");
+    QObject::connect(search_img, &QAction::triggered, [this]{
+        QString fpath = QFileDialog::getOpenFileName(this, "Select Image", QString(), "Image file");
+        if (fpath.isNull()) return;
+        auto sim = this->sdb->search_file(qstring_to_path(fpath));
+        if (sim.empty())
+        {
+            this->sb->showMessage("No similar image found.", 2000);
+            return;
+        }
+        this->vm = ViewMode::view_searchresult;
+        std::vector<fs::path> ps;
+        std::map<std::pair<size_t, size_t>, double> dm;
+        for (size_t i = 0; i < sim.size(); ++i)
+        {
+            auto &s = sim[i];
+            ps.push_back(this->sdb->get_image_path(s.first));
+            dm[std::make_pair(0, i)] = s.second;
+        }
+        this->show_images(ps);
+        this->update_distances(dm);
+        this->sb->showMessage("Use next group / previous group to go back.");
+        this->permamsg->setText("Viewing image search result");
+    });
+    menuact["search_image"] = search_img;
     file->addSeparator();
     file->addAction("Preferences...");
     file->addAction("Exit");
@@ -230,6 +254,7 @@ void DeduperMainWindow::setup_menu()
     menuact["next_group"] = nxtgrp;
     nxtgrp->setShortcut(QKeySequence(Qt::Key::Key_M));
     QObject::connect(nxtgrp, &QAction::triggered, [this] {
+        if (this->vm == ViewMode::view_searchresult) { this->show_group(curgroup); return; }
         if (this->sdb && curgroup + 1 < this->sdb->num_groups())
             this->show_group(++curgroup);
     });
@@ -240,6 +265,7 @@ void DeduperMainWindow::setup_menu()
     menuact["prev_group"] = prvgrp;
     prvgrp->setShortcut(QKeySequence(Qt::Key::Key_Z));
     QObject::connect(prvgrp, &QAction::triggered, [this] {
+        if (this->vm == ViewMode::view_searchresult) { this->show_group(curgroup); return; }
         if (this->sdb && curgroup > 0)
             this->show_group(--curgroup);
     });
@@ -306,14 +332,21 @@ void DeduperMainWindow::update_actions()
         menuact["save_db"]->setEnabled(false);
         menuact["load_list"]->setEnabled(false);
         menuact["save_list"]->setEnabled(false);
+        menuact["search_image"]->setEnabled(false);
         return;
     }
     menuact["skip_group"]->setEnabled(true);
     menuact["prev_group"]->setEnabled(curgroup > 0);
     menuact["next_group"]->setEnabled(curgroup + 1 < sdb->num_groups());
+    menuact["search_image"]->setEnabled(true);
     menuact["save_db"]->setEnabled(true);
     menuact["load_list"]->setEnabled(true);
     menuact["save_list"]->setEnabled(true);
+    if (vm == ViewMode::view_searchresult)
+    {
+        menuact["next_group"]->setEnabled(true);
+        menuact["prev_group"]->setEnabled(true);
+    }
 }
 
 void DeduperMainWindow::show_images(const std::vector<fs::path> &fns)
@@ -350,7 +383,10 @@ void DeduperMainWindow::update_distances(const std::map<std::pair<size_t, size_t
             ka = QKeySequence(keys[p.first.first]).toString();
         if (p.first.second < keys.size())
             kb = QKeySequence(keys[p.first.second]).toString();
-        r += QString("%1 <-> %2: %3\n").arg(ka).arg(kb).arg(QString::number(p.second));
+        if (vm == ViewMode::view_normal)
+            r += QString("%1 <-> %2: %3\n").arg(ka).arg(kb).arg(QString::number(p.second));
+        else
+            r += QString("%1 : %2\n").arg(kb).arg(QString::number(p.second));;
     }
     infopanel->setText(r);
 }
@@ -444,6 +480,7 @@ void DeduperMainWindow::scan_dirs(std::vector<std::pair<fs::path, bool>> paths)
         }
         int flsize = fs->file_list().size() - 1;
         QMetaObject::invokeMethod(this->pd, [flsize, this] {this->pd->setMaximum(flsize);}, Qt::ConnectionType::QueuedConnection);
+        if (this->sdb) delete this->sdb;
         this->sdb = new SignatureDB();
         QObject::connect(this->sdb, &SignatureDB::image_scanned, this, [this](size_t n) {
             static auto lt = std::chrono::steady_clock::now();
@@ -470,6 +507,7 @@ void DeduperMainWindow::scan_dirs(std::vector<std::pair<fs::path, bool>> paths)
         this->pd->reset();
         this->pd->close();
         this->curgroup = 0;
+        this->vm = ViewMode::view_normal;
         this->show_group(this->curgroup);
     }, Qt::ConnectionType::QueuedConnection);
     QObject::connect(pd, &QProgressDialog::canceled, [this] {
@@ -482,6 +520,7 @@ void DeduperMainWindow::show_group(size_t gid)
 {
     if (!sdb || gid >= sdb->num_groups())
         return;
+    this->vm = ViewMode::view_normal;
     auto g = sdb->get_group(gid);
     current_set.clear();
     std::for_each(g.begin(), g.end(), [this](auto id){current_set.push_back(sdb->get_image_path(id));});
@@ -493,6 +532,12 @@ void DeduperMainWindow::show_group(size_t gid)
 
 void DeduperMainWindow::mark_toggle(size_t x)
 {
+    if (vm == ViewMode::view_searchresult)
+    {
+        mark_none(false);
+        sb->showMessage("Marking images in search result is disabled.", 2000);
+        return;
+    }
     if (x < marks.size())
     {
         marks[x] = !marks[x];
@@ -507,6 +552,12 @@ void DeduperMainWindow::mark_toggle(size_t x)
 
 void DeduperMainWindow::mark_all_but(size_t x)
 {
+    if (vm == ViewMode::view_searchresult)
+    {
+        mark_none(false);
+        sb->showMessage("Marking images in search result is disabled.", 2000);
+        return;
+    }
     if (x < marks.size())
     {
         for (size_t i = 0; i < marks.size(); ++i)
@@ -524,6 +575,12 @@ void DeduperMainWindow::mark_all_but(size_t x)
 
 void DeduperMainWindow::mark_all()
 {
+    if (vm == ViewMode::view_searchresult)
+    {
+        mark_none(false);
+        sb->showMessage("Marking images in search result is disabled.", 2000);
+        return;
+    }
     for (size_t i = 0; i < marks.size(); ++i)
     {
         marks[i] = true;
@@ -532,7 +589,7 @@ void DeduperMainWindow::mark_all()
     mark_view_update();
 }
 
-void DeduperMainWindow::mark_none()
+void DeduperMainWindow::mark_none(bool msg)
 {
     for (size_t i = 0; i < marks.size(); ++i)
     {
@@ -540,7 +597,7 @@ void DeduperMainWindow::mark_none()
         if (marked.find(current_set[i]) != marked.end())
             marked.erase(marked.find(current_set[i]));
     }
-    mark_view_update();
+    mark_view_update(msg);
 }
 
 void DeduperMainWindow::mark_view_update(bool update_msg)
diff --git a/qdeduper/mingui.hpp b/qdeduper/mingui.hpp
index e4b1c3f..9f3ff2a 100644
--- a/qdeduper/mingui.hpp
+++ b/qdeduper/mingui.hpp
@@ -28,6 +28,12 @@ class ImageItemDelegate;
 
 namespace fs = std::filesystem;
 
+enum ViewMode
+{
+    view_normal,
+    view_searchresult
+};
+
 class DeduperMainWindow : public QMainWindow
 {
     Q_OBJECT
@@ -46,12 +52,13 @@ private:
     SignatureDB *sdb = nullptr;
     FileScanner *fsc = nullptr;
 
-    std::size_t curgroup;
+    size_t curgroup;
+    ViewMode vm;
     bool nohotkeywarn;
-    void mark_toggle(std::size_t x);
-    void mark_all_but(std::size_t x);
+    void mark_toggle(size_t x);
+    void mark_all_but(size_t x);
     void mark_all();
-    void mark_none();
+    void mark_none(bool msg = true);
     void mark_view_update(bool update_msg = true);
     fs::path::string_type common_prefix(const std::vector<fs::path> &fns);
     std::vector<bool> marks;
@@ -65,8 +72,8 @@ public:
 
     void setup_menu();
     void show_images(const std::vector<fs::path> &fns);
-    void update_distances(const std::map<std::pair<std::size_t, std::size_t>, double> &d);
-    void update_viewstatus(std::size_t cur, std::size_t size);
+    void update_distances(const std::map<std::pair<size_t, size_t>, double> &d);
+    void update_viewstatus(size_t cur, size_t size);
     void save_list();
     void load_list();
 
@@ -78,7 +85,7 @@ public Q_SLOTS:
 Q_SIGNALS:
     void next();
     void prev();
-    void switch_group(std::size_t group);
+    void switch_group(size_t group);
 };
 
 #endif
diff --git a/qdeduper/sigdb_qt.cpp b/qdeduper/sigdb_qt.cpp
index 67cc3e6..733a198 100644
--- a/qdeduper/sigdb_qt.cpp
+++ b/qdeduper/sigdb_qt.cpp
@@ -97,6 +97,22 @@ void SignatureDB::interrupt_scan()
         sdb->populate_interrupt();
 }
 
+std::vector<std::pair<size_t, double>> SignatureDB::search_file(const fs::path &files)
+{
+    populate_cfg_t pcfg = {
+        3,
+        3,
+        cfg_full,
+        cfg_subslice,
+        0.3,
+        nullptr,
+        0
+    };
+    if(sdb)
+        return sdb->search_image(files, pcfg, false);
+    return {};
+}
+
 size_t SignatureDB::num_groups()
 {
     return groups.size();
diff --git a/qdeduper/sigdb_qt.hpp b/qdeduper/sigdb_qt.hpp
index 112ffa9..772c264 100644
--- a/qdeduper/sigdb_qt.hpp
+++ b/qdeduper/sigdb_qt.hpp
@@ -33,6 +33,9 @@ public:
 
     void scan_files(const std::vector<fs::path> &files, int njobs);
     void interrupt_scan();
+
+    std::vector<std::pair<size_t, double>> search_file(const fs::path &files);
+
     size_t num_groups();
     std::vector<size_t> get_group(size_t gid);
     std::map<std::pair<size_t, size_t>, double> group_distances(size_t gid);
diff --git a/xsig/include/signature_db.hpp b/xsig/include/signature_db.hpp
index a74e90b..9b14fbb 100644
--- a/xsig/include/signature_db.hpp
+++ b/xsig/include/signature_db.hpp
@@ -87,6 +87,8 @@ public:
     void populate(const std::vector<fs::path> &paths, const populate_cfg_t &cfg);
     void populate_interrupt();
 
+    std::vector<std::pair<size_t, double>> search_image(const fs::path &path, const populate_cfg_t &cfg, bool insert = false);
+
     //disjoint set for keeping similar images in the same group
     //some of these probably shouldn't be public. TBD...
     void ds_init();
diff --git a/xsig/src/signature.cpp b/xsig/src/signature.cpp
index 1f0ec28..0f0b2e9 100644
--- a/xsig/src/signature.cpp
+++ b/xsig/src/signature.cpp
@@ -226,19 +226,19 @@ signature signature::clone() const
 
 double signature::length() const
 {
-    if (!p) {fprintf(stderr, "length: null signature"); return -1;}
+    if (!p) {fprintf(stderr, "length: null signature\n"); return -1;}
     return p->length();
 }
 
 double signature::distance(const signature &o) const
 {
-    if (!p || !o.p) {fprintf(stderr, "distance: null signature"); return -1;}
+    if (!p || !o.p) {fprintf(stderr, "distance: null signature\n"); return -1;}
     return p->distance(*o.p);
 }
 
 bool signature::operator==(const signature &o) const
 {
-    if (!p || !o.p) {fprintf(stderr, "eq: null signature"); return false;}
+    if (!p || !o.p) {fprintf(stderr, "eq: null signature\n"); return false;}
     return *p == *o.p;
 }
 
diff --git a/xsig/src/signature_db.cpp b/xsig/src/signature_db.cpp
index 6b328d6..5396d1d 100644
--- a/xsig/src/signature_db.cpp
+++ b/xsig/src/signature_db.cpp
@@ -398,36 +398,7 @@ void signature_db::populate(const std::vector<fs::path> &paths, const populate_c
     std::atomic<size_t> count(0);
     auto job_func = [&, this](int thid, const fs::path& path)
     {
-        subsliced_signature ss = subsliced_signature::from_path(path, cfg.nsliceh, cfg.nslicev, cfg.scfg_full, cfg.scfg_subslice);
-
-        this->lock();
-        std::set<size_t> v;
-        size_t dbid = this->put_signature(path, ss.full);
-
-        this->batch_find_subslice_begin();
-        for (size_t i = 0; i < cfg.nsliceh * cfg.nslicev; ++i)
-        {
-            std::vector<subslice_t> ssmatches = this->find_subslice(ss.subslices[i]);
-            for (auto &match : ssmatches)
-            {
-                if (match.slice == i && v.find(match.id) == v.end())
-                {
-                    signature othersig;
-                    std::tie(std::ignore, othersig) = this->get_signature(match.id);
-                    double dist = ss.full.distance(othersig);
-                    if (dist < cfg.threshold)
-                        this->put_dupe_pair(dbid, match.id, dist);
-                }
-            }
-        }
-        this->batch_find_subslice_end();
-
-        this->batch_put_subslice_begin();
-        for (size_t i = 0; i < cfg.nsliceh * cfg.nslicev; ++i)
-            this->put_subslice(dbid, i, ss.subslices[i]);
-        this->batch_put_subslice_end();
-
-        this->unlock();
+        this->search_image(path, cfg, true);
         ++count;
         cfg.callback(count.load(), thid);
     };
@@ -441,12 +412,60 @@ void signature_db::populate(const std::vector<fs::path> &paths, const populate_c
     delete p->tp;
     p->tp = nullptr;
 }
+
 void signature_db::populate_interrupt()
 {
     if (p->tp)
         p->tp->terminate();
 }
 
+std::vector<std::pair<size_t, double>> signature_db::search_image(const fs::path &path, const populate_cfg_t &cfg, bool insert)
+{
+    subsliced_signature ss = subsliced_signature::from_path(path, cfg.nsliceh, cfg.nslicev, cfg.scfg_full, cfg.scfg_subslice);
+    if (!ss.full.valid()) return {};
+
+    this->lock();
+    std::set<size_t> v;
+    std::vector<std::pair<size_t, double>> ret;
+    size_t dbid = 0;
+    if (insert) dbid = this->put_signature(path, ss.full);
+
+    this->batch_find_subslice_begin();
+    for (size_t i = 0; i < cfg.nsliceh * cfg.nslicev; ++i)
+    {
+        std::vector<subslice_t> ssmatches = this->find_subslice(ss.subslices[i]);
+        for (auto &match : ssmatches)
+        {
+            if (match.slice == i && v.find(match.id) == v.end())
+            {
+                signature othersig;
+                std::tie(std::ignore, othersig) = this->get_signature(match.id);
+                double dist = ss.full.distance(othersig);
+                if (dist < cfg.threshold)
+                {
+                    if (insert)
+                        this->put_dupe_pair(dbid, match.id, dist);
+                    else
+                        ret.emplace_back(match.id, dist);
+                    v.insert(match.id);
+                }
+            }
+        }
+    }
+    this->batch_find_subslice_end();
+
+    if (insert)
+    {
+        this->batch_put_subslice_begin();
+        for (size_t i = 0; i < cfg.nsliceh * cfg.nslicev; ++i)
+            this->put_subslice(dbid, i, ss.subslices[i]);
+        this->batch_put_subslice_end();
+    }
+
+    this->unlock();
+    return ret;
+}
+
 void signature_db::ds_init()
 {
     sqlite3_exec(p->db, R"sql(