From 04f509fd948c03c8b8732ddb6787eea800bdc7f6 Mon Sep 17 00:00:00 2001
From: Chris Xiong <chirs241097@gmail.com>
Date: Sat, 17 Sep 2022 23:48:38 -0400
Subject: Finish similar image grouping.

---
 signature_db.cpp           | 35 ++++++++++++++++++++++++++++++++++-
 signature_db.hpp           |  3 +++
 tests/testdrive_sqlite.cpp | 21 +++++++++++++++++++++
 3 files changed, 58 insertions(+), 1 deletion(-)

diff --git a/signature_db.cpp b/signature_db.cpp
index 71f2142..429581d 100644
--- a/signature_db.cpp
+++ b/signature_db.cpp
@@ -491,9 +491,42 @@ void signature_db::ds_merge(size_t id1, size_t id2)
 
 void signature_db::group_similar()
 {
+    ds_init();
+    batch_ds_get_parent_begin();
+    batch_ds_set_parent_begin();
+    auto pairs = this->dupe_pairs();
+    for (auto &p : pairs)
+        ds_merge(p.id1, p.id2);
+    batch_ds_get_parent_end();
+    batch_ds_set_parent_end();
 }
 
 std::vector<std::vector<size_t>> signature_db::groups_get()
 {
-    return {};
+    sqlite3_stmt *sto = nullptr;
+    sqlite3_stmt *sti = nullptr;
+    sqlite3_prepare_v2(p->db, "select distinct parent from dspar;", -1, &sto, 0);
+    sqlite3_prepare_v2(p->db, "select id from dspar where parent = ?;", -1, &sti, 0);
+    std::vector<std::vector<size_t>> ret;
+
+    while (1)
+    {
+        int r = sqlite3_step(sto);
+        if (r != SQLITE_ROW) break;
+        size_t dpar = (size_t)sqlite3_column_int(sto, 0);
+        sqlite3_bind_int(sti, 1, dpar);
+        std::vector<size_t> v;
+        while (1)
+        {
+            int ri = sqlite3_step(sti);
+            if (ri != SQLITE_ROW) break;
+            size_t id = (size_t)sqlite3_column_int(sti, 0);
+            v.push_back(id);
+        }
+        ret.push_back(v);
+        sqlite3_reset(sti);
+    }
+    sqlite3_finalize(sto);
+    sqlite3_finalize(sti);
+    return ret;
 }
diff --git a/signature_db.hpp b/signature_db.hpp
index c7e3997..a56ae1f 100644
--- a/signature_db.hpp
+++ b/signature_db.hpp
@@ -97,7 +97,10 @@ public:
     size_t ds_find(size_t id);
     void ds_merge(size_t id1, size_t id2);
 
+    //group similar images together using results from dupe_pairs()
+    //usually very fast, unless you have a crack ton of duplicates...
     void group_similar();
+    //get all groups, each countained in their own lists.
     std::vector<std::vector<size_t>> groups_get();
 };
 
diff --git a/tests/testdrive_sqlite.cpp b/tests/testdrive_sqlite.cpp
index 3f1fe40..c9e9aad 100644
--- a/tests/testdrive_sqlite.cpp
+++ b/tests/testdrive_sqlite.cpp
@@ -226,6 +226,9 @@ int main(int argc,char** argv)
     };
     sdb->populate(files, pcfg);
 
+    puts("grouping similar images...");
+    sdb->group_similar();
+
     std::vector<dupe_t> dupes = sdb->dupe_pairs();
     for (auto &p : dupes)
     {
@@ -238,6 +241,24 @@ int main(int argc,char** argv)
         printf("%s %s %f\n", p1.c_str(), p2.c_str(), p.distance);
 #endif
     }
+
+    std::vector<std::vector<size_t>> gp = sdb->groups_get();
+    for (auto gi = gp.begin(); gi != gp.end(); ++gi)
+    {
+        if (gi->size() < 2) continue;
+        printf("group #%lu:\n", gi - gp.begin());
+        for (auto &id : *gi)
+        {
+            fs::path p;
+            std::tie(p, std::ignore) = sdb->get_signature(id);
+#if PATH_VALSIZE == 2
+            wprintf(L"\t%ls\n", p.c_str());
+#else
+            printf("\t%s\n", p.c_str());
+#endif
+        }
+    }
+
     sdb->to_db_file("test.sigdb");
     delete sdb;
     return 0;
-- 
cgit v1.2.3