Move stuff around to accommodate new family members.

3 files changed, 228 insertions, 0 deletions

diff --git a/xsig/include/signature.hpp b/xsig/include/signature.hpp
new file mode 100644
index 0000000..b655e73
--- /dev/null
+++ b/xsig/include/signature.hpp
@@ -0,0 +1,83 @@
+//Chris Xiong 2022
+//License: MPL-2.0
+#ifndef SIGNATURE_HPP
+#define SIGNATURE_HPP
+
+#include <memory>
+#include <filesystem>
+#include <string>
+
+struct signature_config
+{
+    int slices;
+    int blur_window;
+    int min_window;
+    bool crop;
+    bool compress;
+    double pr;
+    double noise_threshold;
+    double contrast_threshold;
+    double max_cropping;
+};
+
+namespace cv
+{
+    class Mat;
+};
+
+class signature_priv;
+class signature
+{
+private:
+    std::shared_ptr<signature_priv> p;
+    signature(signature_priv* _p);
+    signature(const signature&)=default;
+    signature& operator=(const signature&)=default;
+public:
+    signature();
+    ~signature();
+    signature(signature&&)=default;
+    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;
+    std::string to_string() const;
+
+    static signature from_string(std::string &&s);
+
+    static signature from_path(const std::filesystem::path &path, const signature_config &cfg);
+
+    static signature from_file(const char *fn, const signature_config &cfg);
+
+    /*
+     * Input will be stripped of alpha channel (by blending with white),
+     * converted to single channel (rgb2gray).
+     * Then it will be passed to from_preprocessed_matrix.
+     * The matrix doesn't have to be continuous.
+     */
+    static signature from_cvmatrix(cv::Mat *m, const signature_config &cfg);
+
+    /*
+     * Input must be a single channel, floating point matrix
+     * (values clamped to 0-1)
+     * The matrix must be continuous if cropping is used
+     * STILL *Will* be cropped if config().crop == true
+     * STILL *Will* be blurred if config().blur_window > 1
+     */
+    static signature from_preprocessed_matrix(cv::Mat *m, const signature_config &cfg);
+
+    static signature_config default_cfg();
+
+    friend class signature_priv;
+    friend struct signature_hash;
+};
+
+struct signature_hash
+{
+    size_t operator()(signature const& sig) const noexcept;
+};
+
+#endif
diff --git a/xsig/include/signature_db.hpp b/xsig/include/signature_db.hpp
new file mode 100644
index 0000000..b37cf0a
--- /dev/null
+++ b/xsig/include/signature_db.hpp
@@ -0,0 +1,109 @@
+//Chris Xiong 2022
+//License: MPL-2.0
+#ifndef SIGNATURE_DB_HPP
+#define SIGNATURE_DB_HPP
+
+#include <functional>
+#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 populate_cfg_t
+{
+    size_t nsliceh;
+    size_t nslicev;
+    signature_config scfg_full;
+    signature_config scfg_subslice;
+    double threshold;
+    std::function<void(size_t, int)> callback;
+    int njobs;
+};
+
+struct signature_db_priv;
+
+class signature_db
+{
+private:
+    signature_db_priv *p;
+public:
+    //open a signature database
+    //if dbpath is an empty path (default), the database will reside in RAM
+    //and will be automatically initialized
+    //otherwise it opens the database specified by dbpath
+    //if the database specified by dbpath doesn't exist, it will be created
+    //and initialized
+    //if the database file exists but is not a valid signature database, it
+    //will be immediately closed and any subsequent calls to this signature db
+    //object will do nothing. The object will be marked invalid.
+    signature_db(const fs::path &dbpath = fs::path());
+    ~signature_db();
+
+    bool valid();
+
+    //insert image signature into database
+    //if id is omitted, it's assigned automatically and returned
+    //if specificted, id must be unique
+    //treat automatically assigned id as arbitrary opaque integers
+    size_t put_signature(const fs::path &path, const signature &sig, size_t id = ~size_t(0));
+    void batch_get_signature_begin();
+    //get image signature from database
+    std::pair<fs::path, signature> get_signature(size_t id);
+    void batch_get_signature_end();
+
+    //place batch_put_subslice_begin() and batch_put_subslice_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);
+    void batch_put_subslice_end();
+
+    //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);
+    void batch_find_subslice_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);
+
+    void populate(const std::vector<fs::path> &paths, const populate_cfg_t &cfg);
+
+    //disjoint set for keeping similar images in the same group
+    //some of these probably shouldn't be public. TBD...
+    void ds_init();
+
+    void batch_ds_get_parent_begin();
+    size_t ds_get_parent(size_t id);
+    void batch_ds_get_parent_end();
+
+    void batch_ds_set_parent_begin();
+    size_t ds_set_parent(size_t id, size_t par);
+    void batch_ds_set_parent_end();
+
+    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();
+};
+
+#endif
diff --git a/xsig/include/subslice_signature.hpp b/xsig/include/subslice_signature.hpp
new file mode 100644
index 0000000..928d396
--- /dev/null
+++ b/xsig/include/subslice_signature.hpp
@@ -0,0 +1,36 @@
+//Chris Xiong 2022
+//License: MPL-2.0
+#ifndef SUBSLICE_SIGNATURE_HPP
+#define SUBSLICE_SIGNATURE_HPP
+#include "signature.hpp"
+
+#include <vector>
+#include <filesystem>
+
+namespace cv
+{
+    class Mat;
+};
+
+class subsliced_signature
+{
+public:
+    signature full;
+    std::vector<signature> subslices;
+    size_t nhslices, nvslices;
+
+    static subsliced_signature from_path(const std::filesystem::path &path,
+                                         size_t nhslices, size_t nvslices,
+                                         const signature_config &fcfg,
+                                         const signature_config &scfg);
+    static subsliced_signature from_file(const char *fn,
+                                         size_t nhslices, size_t nvslices,
+                                         const signature_config &fcfg,
+                                         const signature_config &scfg);
+    static subsliced_signature from_cvmatrix(cv::Mat *m,
+                                             size_t nhslices, size_t nvslices,
+                                             const signature_config &fcfg,
+                                             const signature_config &scfg);
+};
+
+#endif