effspm 0.2.7__cp312-cp312-win_amd64.whl → 0.3.3__cp312-cp312-win_amd64.whl

effspm/_effspm.cpp

@@ -2,49 +2,97 @@
 
 #include <pybind11/pybind11.h>
 #include <pybind11/stl.h>
-namespace py = pybind11;
+
 #include <iostream>
+#include <fstream>
+#include <cstdio>   // std::remove
+#include <vector>
+#include <string>
+#include <ctime>
+#include <cmath>
 
+namespace py = pybind11;
 
-// PrefixProjection headers
+// PrefixProjection headers (global namespace)
 #include "freq_miner.hpp"
 #include "load_inst.hpp"
 #include "utility.hpp"
 
-// BTMiner (wrapped in its own namespace in source files)
+// BTMiner (namespaced)
 #include "btminer/src/freq_miner.hpp"
 #include "btminer/src/load_inst.hpp"
 #include "btminer/src/utility.hpp"
 #include "btminer/src/build_mdd.hpp"
 
-// HTMiner (wrapped in its own namespace in source files)
-#include "htminer/src/build_mdd.hpp"    // ← ensure HTMiner MDD builder is available
+// HTMiner (namespaced)
+#include "htminer/src/build_mdd.hpp"
 #include "htminer/src/freq_miner.hpp"
 #include "htminer/src/load_inst.hpp"
 #include "htminer/src/utility.hpp"
 
-
+// LargePrefixProjection
 #include "largepp/src/freq_miner.hpp"
 #include "largepp/src/load_inst.hpp"
 #include "largepp/src/utility.hpp"
 
+// LargeBTMiner
 #include "largebm/src/freq_miner.hpp"
 #include "largebm/src/load_inst.hpp"
 #include "largebm/src/utility.hpp"
 #include "largebm/src/build_mdd.hpp"
 
+// LargeHTMiner
 #include "largehm/src/freq_miner.hpp"
 #include "largehm/src/load_inst.hpp"
 #include "largehm/src/utility.hpp"
 #include "largehm/src/build_mdd.hpp"
 
+namespace {
+
+// RAII helper for temp file
+struct TempFile {
+    std::string path;
+    ~TempFile() {
+        if (!path.empty()) {
+            std::remove(path.c_str());
+        }
+    }
+};
+
+// Write Python list[list[int]] to a temp file in professor’s format:
+// one sequence per line, items separated by spaces.
+std::string write_temp_seq_file(const std::vector<std::vector<int>>& seqs) {
+    char tmp_name[L_tmpnam];
+    if (!std::tmpnam(tmp_name)) {
+        throw std::runtime_error("Failed to create temporary file name");
+    }
+    std::string path = std::string(tmp_name) + ".txt";
+
+    std::ofstream ofs(path);
+    if (!ofs) {
+        throw std::runtime_error("Failed to open temporary file for writing: " + path);
+    }
+
+    for (const auto& seq : seqs) {
+        for (size_t i = 0; i < seq.size(); ++i) {
+            if (i) ofs << ' ';
+            ofs << seq[i];
+        }
+        ofs << '\n';
+    }
+
+    ofs.close();
+    return path;
+}
+
+} // anonymous namespace
 
 
 PYBIND11_MODULE(_effspm, m) {
-    m.doc() = "Unified SPM library: PrefixProjection, BTMiner, HTMiner";
+    m.doc() = "Unified SPM library: PrefixProjection, BTMiner, HTMiner, Large* variants";
 
     // ─────────────────────────────────────────────────────────────
-    // PrefixProjection
+    // PrefixProjection  (works directly on Python lists or files)
     // ─────────────────────────────────────────────────────────────
     m.def("PrefixProjection",
         [](py::object data,
@@ -59,7 +107,7 @@ PYBIND11_MODULE(_effspm, m) {
             ::pre_pro    = preproc;
             ::use_dic    = use_dic;
             ::use_list   = false;
-            ::b_disp     = verbose;
+            ::b_disp     = verbose;                  // controls prints in original code
             ::b_write    = !out_file.empty();
             ::out_file   = out_file;
 
@@ -69,7 +117,7 @@ PYBIND11_MODULE(_effspm, m) {
             if (py::isinstance<py::str>(data)) {
                 std::string path = data.cast<std::string>();
                 if (!Load_instance(path, minsup))
-                    throw std::runtime_error("Failed to load file: " + path);
+                    throw std::runtime_error("PrefixProjection: failed to load file: " + path);
             } else {
                 auto seqs = data.cast<std::vector<std::vector<int>>>();
                 items = std::move(seqs);
@@ -113,9 +161,355 @@ PYBIND11_MODULE(_effspm, m) {
     );
 
     // ─────────────────────────────────────────────────────────────
-    // BTMiner
+    // BTMiner  (always uses professor's Load_instance)
+    // ─────────────────────────────────────────────────────────────
+   // ─────────────────────────────────────────────────────────────
+// BTMiner  (always uses professor's Load_instance)
+// ─────────────────────────────────────────────────────────────
+/*m.def("BTMiner",
+    [](py::object data,
+       double minsup,
+       unsigned int time_limit,
+       bool preproc,
+       bool use_dic,
+       bool verbose,
+       const std::string &out_file)
+    {
+        // 1) Configure professor globals
+        btminer::time_limit = static_cast<int>(time_limit);
+        btminer::pre_pro    = preproc;
+        btminer::use_dic    = use_dic;
+        btminer::b_disp     = verbose;
+        btminer::b_write    = !out_file.empty();
+        btminer::out_file   = out_file;
+        btminer::N_mult     = 1;
+        btminer::M_mult     = 1;
+        btminer::just_build = false;
+
+        // 2) HARD RESET of *known* global state for BTMiner
+        // (Only touch what we know exists in btminer namespace)
+        btminer::ClearCollected();   // clear collected patterns
+        btminer::Tree.clear();       // clear MDD tree
+        btminer::DFS.clear();        // clear DFS patterns
+
+        btminer::M      = 0;
+        btminer::L      = 0;
+        btminer::N      = 0;
+        btminer::theta  = 0;
+        btminer::E      = 0;
+        btminer::num_patt = 0;       // reset pattern counter if defined
+
+        // NOTE: we do NOT reinsert root here; btminer::Load_instance()
+        // is responsible for calling Tree.emplace_back(0,0,0) as needed.
+
+        btminer::start_time = std::clock();
+
+        // 3) Handle input (path or list-of-lists)
+        TempFile tmp;
+        std::string path;
+
+        if (py::isinstance<py::str>(data)) {
+            // File path: use directly
+            path = data.cast<std::string>();
+        } else {
+            // Python list → write to a temp file in professor’s format
+            auto seqs = data.cast<std::vector<std::vector<int>>>();
+            tmp.path  = write_temp_seq_file(seqs);
+            path      = tmp.path;
+        }
+
+        if (verbose) {
+            std::cerr << "[BTMiner] path=" << path
+                      << " minsup=" << minsup
+                      << " preproc=" << preproc
+                      << " use_dic=" << use_dic
+                      << std::endl;
+        }
+
+        // 4) Build MDD + run miner
+        if (!btminer::Load_instance(path, minsup)) {
+            throw std::runtime_error("BTMiner: failed to load instance from: " + path);
+        }
+
+        btminer::Freq_miner();
+
+        // 5) Return results
+        py::dict out;
+        out["patterns"]     = btminer::GetCollected();
+        out["num_patterns"] = btminer::num_patt;
+        out["time"]         = btminer::give_time(std::clock() - btminer::start_time);
+        out["N"]            = btminer::N;
+        out["L"]            = btminer::L;
+        out["theta"]        = btminer::theta;
+        return out;
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+); */
+m.def("BTMiner",
+    [](py::object data,
+       double minsup,
+       unsigned int time_limit,
+       bool preproc,
+       bool use_dic,
+       bool verbose,
+       const std::string &out_file)
+    {
+        // 1) Configure professor globals
+        btminer::time_limit = static_cast<int>(time_limit);
+        btminer::pre_pro    = preproc;
+        btminer::use_dic    = use_dic;
+        btminer::b_disp     = verbose;
+        btminer::b_write    = !out_file.empty();
+        btminer::out_file   = out_file;
+        btminer::N_mult     = 1;
+        btminer::M_mult     = 1;
+        btminer::just_build = false;
+
+                // 2) HARD RESET of *known* global state for BTMiner
+        btminer::ClearCollected();   // clear collected patterns
+        btminer::Tree.clear();       // clear MDD tree
+        btminer::DFS.clear();        // clear DFS patterns
+
+        // clear all frequency / mapping / item structures
+        btminer::freq.clear();
+        btminer::item_dic.clear();
+        btminer::item_map.clear();
+        btminer::item_map_rev.clear();
+        btminer::items.clear();      // if you have this defined anywhere
+
+        // reset scalar globals
+        btminer::M         = 0;
+        btminer::L         = 0;
+        btminer::N         = 0;
+        btminer::theta     = 0;
+        btminer::E         = 0;
+        btminer::num_patt  = 0;
+        btminer::num_nodes = 0;
+        btminer::cur_node  = 0;
+        // N_mult, M_mult, flags are set just above
+
+        btminer::start_time = std::clock();
+
+
+        // 3) Handle input (path or list-of-lists)
+        TempFile tmp;
+        std::string path;
+
+        if (py::isinstance<py::str>(data)) {
+            // File path: use directly
+            path = data.cast<std::string>();
+        } else {
+            // Python list → write to a temp file in professor’s format
+            auto seqs = data.cast<std::vector<std::vector<int>>>();
+            tmp.path  = write_temp_seq_file(seqs);
+            path      = tmp.path;
+        }
+
+        if (verbose) {
+            std::cerr << "[BTMiner] path=" << path
+                      << " minsup=" << minsup
+                      << " preproc=" << preproc
+                      << " use_dic=" << use_dic
+                      << std::endl;
+        }
+
+        // 4) Build MDD + run miner
+        if (!btminer::Load_instance(path, minsup)) {
+            throw std::runtime_error("BTMiner: failed to load instance from: " + path);
+        }
+
+        btminer::Freq_miner();
+
+        // 5) Return results
+        py::dict out;
+        out["patterns"]     = btminer::GetCollected();
+        out["num_patterns"] = btminer::num_patt;
+        out["time"]         = btminer::give_time(std::clock() - btminer::start_time);
+        out["N"]            = btminer::N;
+        out["L"]            = btminer::L;
+        out["theta"]        = btminer::theta;
+        return out;
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+);
+
+        // ─────────────────────────────────────────────────────────────
+    // HTMiner  (works on files; we use a temp file for in-memory data)
+    // ─────────────────────────────────────────────────────────────
+   // ─────────────────────────────────────────────────────────────
+// HTMiner  (always uses professor's Load_instance; pre_pro forced ON)
+// ─────────────────────────────────────────────────────────────
+m.def("HTMiner",
+    [](py::object data,
+       double minsup,
+       unsigned int time_limit,
+       bool /*preproc*/,   // Python arg is ignored internally
+       bool use_dic,
+       bool verbose,
+       const std::string &out_file)
+    {
+        using namespace htminer;
+
+        // ───────── Global parameter setup ─────────
+        htminer::time_limit = time_limit;
+
+        // IMPORTANT: always run with preprocessing ON,
+        // regardless of the Python `preproc` flag.
+        htminer::pre_pro    = true;
+        htminer::use_dic    = use_dic;
+        htminer::just_build = false;
+        htminer::b_disp     = verbose;
+        htminer::b_write    = !out_file.empty();
+        htminer::out_file   = out_file;
+
+        // ───────── HARD RESET of HTMiner globals ─────────
+        htminer::ClearCollected();
+        htminer::Tree.clear();
+        htminer::VTree.clear();
+        htminer::CTree.clear();
+        htminer::DFS.clear();
+        htminer::VDFS.clear();
+        htminer::item_dic.clear();
+
+        htminer::M             = 0;
+        htminer::N             = 0;
+        htminer::L             = 0;
+        htminer::E             = 0;
+        htminer::theta         = 0;
+        htminer::mlim          = 0;
+        htminer::itmset_exists = false;
+
+        // NOTE: do NOT add a root arc here;
+        // htminer::Load_instance() already does Tree.emplace_back(0,0,0)
+        htminer::start_time = std::clock();
+
+        // ───────── Handle input (path or in-memory sequences) ─────────
+        TempFile    tmp;
+        std::string path;
+
+        if (py::isinstance<py::str>(data)) {
+            // data is a file path
+            path = data.cast<std::string>();
+        } else {
+            // data is a list[list[int]] → write a temp file in the same text format
+            auto seqs = data.cast<std::vector<std::vector<int>>>();
+            tmp.path  = write_temp_seq_file(seqs);
+            path      = tmp.path;
+        }
+
+        if (verbose) {
+            std::cerr << "[HTMiner] path=" << path
+                      << " minsup=" << minsup
+                      << " preproc(always)=true"
+                      << " use_dic=" << use_dic
+                      << std::endl;
+        }
+
+        // ───────── Build MDD via professor's loader ─────────
+        if (!htminer::Load_instance(path, minsup)) {
+            throw std::runtime_error("HTMiner: failed to load instance from: " + path);
+        }
+
+        // ───────── Run miner ─────────
+        htminer::Freq_miner();
+
+        // ───────── Return results ─────────
+        py::dict out;
+        out["patterns"] = htminer::GetCollected();
+        out["time"]     = htminer::give_time(std::clock() - htminer::start_time);
+        return out;
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,   // kept for API symmetry, but IGNORED
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+);
+
+    // ─────────────────────────────────────────────────────────────
+    // LargePrefixProjection  (already has its own Load_py)
     // ─────────────────────────────────────────────────────────────
-    m.def("BTMiner",
+   m.def("LargePrefixProjection",
+    [](py::object data,
+       double minsup,
+       unsigned int time_limit,
+       bool preproc,
+       bool use_dic,
+       bool verbose,
+       const std::string &out_file)
+    {
+        // 1) Configure global flags
+        largepp::time_limit = time_limit;
+        largepp::pre_pro    = preproc;
+        largepp::use_dic    = use_dic;
+        largepp::use_list   = true;        // LargePrefixProjection is list-based
+        largepp::b_disp     = verbose;
+        largepp::b_write    = !out_file.empty();
+        largepp::out_file   = out_file;
+        largepp::just_build = false;
+
+        // 2) HARD RESET of largepp global state
+        //    (only touch symbols that actually exist in largepp)
+        largepp::ClearCollected();         // clear previously collected patterns
+
+        // If these exist in largepp::load_inst.hpp / utility.hpp they’ll compile;
+        // if the compiler complains about any of them, just comment that line out.
+        largepp::items.clear();            // transaction DB
+        largepp::DFS.clear();              // DFS pattern stack, if list-based miner uses it
+
+        largepp::M      = 0;
+        largepp::L      = 0;
+        largepp::N      = 0;
+        largepp::theta  = 0;
+        largepp::E      = 0;
+        largepp::num_patt = 0;
+
+        largepp::start_time = std::clock();
+
+        // 3) Handle input (path or Python list)
+        if (py::isinstance<py::str>(data)) {
+            std::string fname = data.cast<std::string>();
+            largepp::Load_instance(fname, minsup);
+        } else {
+            largepp::Load_py(data, minsup);
+        }
+
+        // 4) Run miner
+        largepp::Freq_miner();
+
+        // 5) Return results
+        py::dict out;
+        out["patterns"] = largepp::GetCollected();
+        out["time"]     = largepp::give_time(std::clock() - largepp::start_time);
+        return out;
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+);
+
+    // ─────────────────────────────────────────────────────────────
+    // LargeBTMiner  (always uses professor's largebm::Load_instance)
+    // ─────────────────────────────────────────────────────────────
+    m.def("LargeBTMiner",
         [](py::object data,
            double minsup,
            unsigned int time_limit,
@@ -124,140 +518,270 @@ PYBIND11_MODULE(_effspm, m) {
            bool verbose,
            const std::string &out_file)
         {
-            btminer::time_limit = time_limit;
-            btminer::pre_pro    = preproc;
-            btminer::use_dic    = use_dic;
-            btminer::use_list   = false;
-            btminer::b_disp     = verbose;
-            btminer::b_write    = !out_file.empty();
-            btminer::out_file   = out_file;
+            using namespace largebm;
+
+            largebm::time_limit = time_limit;
+            largebm::pre_pro    = preproc;
+            largebm::use_dic    = use_dic;
+            largebm::use_list   = false;   // MDD-based
+            largebm::b_disp     = verbose;
+            largebm::b_write    = !out_file.empty();
+            largebm::out_file   = out_file;
+            largebm::just_build = false;
+
+            largebm::ClearCollected();
+            largebm::items.clear();
+            largebm::item_dic.clear();
+            largebm::inv_item_dic.clear();
+            largebm::Tree.clear();
+            largebm::DFS.clear();
+
+            largebm::start_time = std::clock();
 
-            btminer::ClearCollected();
-            btminer::start_time = std::clock();
+            TempFile tmp;
+            std::string path;
 
             if (py::isinstance<py::str>(data)) {
-                std::string path = data.cast<std::string>();
-                if (!btminer::Load_instance(path, minsup))
-                    throw std::runtime_error("Failed to load file: " + path);
+                path = data.cast<std::string>();
             } else {
                 auto seqs = data.cast<std::vector<std::vector<int>>>();
-                btminer::items = std::move(seqs);
-                btminer::N = btminer::items.size();
+                tmp.path  = write_temp_seq_file(seqs);
+                path      = tmp.path;
+            }
 
-                int max_id = 0;
-                for (auto &seq : btminer::items)
-                    for (int x : seq)
-                        max_id = std::max(max_id, std::abs(x));
-                btminer::L = max_id;
+            if (verbose) {
+                std::cerr << "[LargeBTMiner] path=" << path
+                          << " minsup=" << minsup
+                          << " preproc=" << preproc
+                          << " use_dic=" << use_dic
+                          << std::endl;
+            }
 
-                btminer::theta = (minsup < 1.0) ? std::ceil(minsup * btminer::N) : minsup;
+            if (!largebm::Load_instance(path, minsup)) {
+                throw std::runtime_error("LargeBTMiner: failed to load instance from: " + path);
+            }
 
-                btminer::DFS.clear();
-                btminer::DFS.reserve(btminer::L);
-                for (unsigned int i = 0; i < btminer::L; ++i)
-                    btminer::DFS.emplace_back(-static_cast<int>(i) - 1);
+            largebm::Freq_miner();
 
-                btminer::M = 0;
-                btminer::E = 0;
-                for (auto &seq : btminer::items) {
-                    btminer::M = std::max<unsigned int>(btminer::M, seq.size());
-                    btminer::E += seq.size();
-                }
+            py::dict out;
+            out["patterns"] = largebm::GetCollected();
+            out["time"]     = largebm::give_time(std::clock() - largebm::start_time);
+            return out;
+        },
+        py::arg("data"),
+        py::arg("minsup")     = 0.01,
+        py::arg("time_limit") = 36000,
+        py::arg("preproc")    = false,
+        py::arg("use_dic")    = false,
+        py::arg("verbose")    = false,
+        py::arg("out_file")   = ""
+    );
+
+    // ─────────────────────────────────────────────────────────────
+    // LargeHTMiner  (always uses professor's largehm::Load_instance; pre_pro forced ON)
+    // ─────────────────────────────────────────────────────────────
+      // ─────────────────────────────────────────────────────────────
+    // LargeHTMiner  (professor's Large HTMiner, namespaced as largehm)
+    // ─────────────────────────────────────────────────────────────
+    m.def("LargeHTMiner",
+        [](py::object data,
+           double minsup,
+           unsigned int time_limit,
+           bool /*preproc*/,   // kept for API symmetry; ignored
+           bool use_dic,
+           bool verbose,
+           const std::string &out_file)
+        {
+            using namespace largehm;
+
+            // 1) Global configuration (mirror professor's style)
+            largehm::time_limit = time_limit;
+            largehm::pre_pro    = true;       // always preprocess
+            largehm::use_dic    = use_dic;
+            largehm::just_build = false;
+            largehm::b_disp     = verbose;
+            largehm::b_write    = !out_file.empty();
+            largehm::out_file   = out_file;
+
+            // 2) HARD RESET of all global state for a fresh run
+            largehm::ClearCollected();   // our helper in largehm::utility.cpp
+
+            largehm::M             = 0;
+            largehm::L             = 0;
+            largehm::mlim          = 0;
+            largehm::N             = 0;
+            largehm::theta         = 0;
+            largehm::E             = 0;
+            largehm::itmset_exists = false;
+
+            // containers
+            // (item_dic reset is optional and not strictly needed here)
+            largehm::DFS.clear();
+            largehm::VDFS.clear();
+            largehm::Tree.clear();
+            largehm::VTree.clear();
+            largehm::CTree.clear();
+
+            largehm::start_time = std::clock();
+
+            // 3) Handle input (file path or Python list)
+            TempFile tmp;
+            std::string path;
+
+            if (py::isinstance<py::str>(data)) {
+                path = data.cast<std::string>();
+            } else {
+                auto seqs = data.cast<std::vector<std::vector<int>>>();
+                tmp.path  = write_temp_seq_file(seqs);
+                path      = tmp.path;
+            }
+
+            if (verbose) {
+                std::cerr << "[LargeHTMiner] path=" << path
+                          << " minsup=" << minsup
+                          << " preproc(always)=true"
+                          << " use_dic=" << use_dic
+                          << std::endl;
             }
 
-            btminer::Freq_miner();
+            // 4) Build MDD / load instance.
+            //    NOTE: Load_instance() itself does Tree.emplace_back(0,0,0),
+            //    so we DO NOT create a root node here.
+            if (!largehm::Load_instance(path, minsup)) {
+                throw std::runtime_error("LargeHTMiner: failed to load instance from: " + path);
+            }
 
+            // 5) Run miner (same timing logic as original main)
+            if (!largehm::just_build &&
+                largehm::give_time(std::clock() - largehm::start_time) < largehm::time_limit)
+            {
+                largehm::Freq_miner();
+                if (largehm::give_time(std::clock() - largehm::start_time) >= largehm::time_limit) {
+                    std::cout << "TIME LIMIT REACHED\n";
+                }
+            }
+
+            // 6) Return collected patterns + runtime
             py::dict out;
-            out["patterns"] = btminer::GetCollected();
-            out["time"]     = btminer::give_time(std::clock() - btminer::start_time);
+            out["patterns"] = largehm::GetCollected();
+            out["time"]     = largehm::give_time(std::clock() - largehm::start_time);
             return out;
         },
         py::arg("data"),
         py::arg("minsup")     = 0.01,
         py::arg("time_limit") = 36000,
-        py::arg("preproc")    = false,
+        py::arg("preproc")    = false,   // kept for API symmetry
         py::arg("use_dic")    = false,
         py::arg("verbose")    = false,
         py::arg("out_file")   = ""
     );
 
-// ─────────────────────────────────────────────────────────────
-// HTMiner
-// ─────────────────────────────────────────────────────────────
-m.def("HTMiner",
+
+}
+
+
+/*#include <pybind11/pybind11.h>
+#include <pybind11/stl.h>
+namespace py = pybind11;
+#include <iostream>
+
+
+// PrefixProjection headers
+#include "freq_miner.hpp"
+#include "load_inst.hpp"
+#include "utility.hpp"
+
+// BTMiner (wrapped in its own namespace in source files)
+#include "btminer/src/freq_miner.hpp"
+#include "btminer/src/load_inst.hpp"
+#include "btminer/src/utility.hpp"
+#include "btminer/src/build_mdd.hpp"
+
+// HTMiner (wrapped in its own namespace in source files)
+#include "htminer/src/build_mdd.hpp"    // ← ensure HTMiner MDD builder is available
+#include "htminer/src/freq_miner.hpp"
+#include "htminer/src/load_inst.hpp"
+#include "htminer/src/utility.hpp"
+
+
+#include "largepp/src/freq_miner.hpp"
+#include "largepp/src/load_inst.hpp"
+#include "largepp/src/utility.hpp"
+
+
+#include "largebm/src/freq_miner.hpp"
+#include "largebm/src/load_inst.hpp"
+#include "largebm/src/utility.hpp"
+#include "largebm/src/build_mdd.hpp"
+
+#include "largehm/src/freq_miner.hpp"
+#include "largehm/src/load_inst.hpp"
+#include "largehm/src/utility.hpp"
+#include "largehm/src/build_mdd.hpp"
+
+
+
+PYBIND11_MODULE(_effspm, m) {
+    m.doc() = "Unified SPM library: PrefixProjection, BTMiner, HTMiner";
+
+    // ─────────────────────────────────────────────────────────────
+    // PrefixProjection
+    // ─────────────────────────────────────────────────────────────
+    m.def("PrefixProjection",
         [](py::object data,
-           double minsup, unsigned int time_limit,
-           bool preproc, bool use_dic,
-           bool verbose, const std::string &out_file)
+           double minsup,
+           unsigned int time_limit,
+           bool preproc,
+           bool use_dic,
+           bool verbose,
+           const std::string &out_file)
         {
-            // 1) set HTMiner globals (declared in htminer/src/utility.hpp)
-            htminer::time_limit = time_limit;
-            htminer::pre_pro    = preproc;
-            htminer::use_dic    = use_dic;
-            htminer::just_build = false;         // or true if you want “build only”
-            htminer::use_list   = false;         // HTMiner always uses MDD‐based mode
-            htminer::b_disp     = verbose;
-            htminer::b_write    = !out_file.empty();
-            htminer::out_file   = out_file;
-            htminer::ClearCollected();           // clear any leftover patterns
-            htminer::start_time = std::clock();
-
-            // 2) load sequences (either from filename or from Python list)
+            ::time_limit = time_limit;
+            ::pre_pro    = preproc;
+            ::use_dic    = use_dic;
+            ::use_list   = false;
+            ::b_disp     = verbose;
+            ::b_write    = !out_file.empty();
+            ::out_file   = out_file;
+
+            ClearCollected();
+            start_time = std::clock();
+
             if (py::isinstance<py::str>(data)) {
                 std::string path = data.cast<std::string>();
-                if (!htminer::Load_instance(path, minsup))
+                if (!Load_instance(path, minsup))
                     throw std::runtime_error("Failed to load file: " + path);
             } else {
                 auto seqs = data.cast<std::vector<std::vector<int>>>();
-                htminer::items = std::move(seqs);
-                htminer::N = htminer::items.size();
+                items = std::move(seqs);
+                N = items.size();
 
-                // compute L (max item ID), M (max sequence length), E (total entries)
                 int max_id = 0;
-                htminer::M = 0;
-                htminer::E = 0;
-                for (auto &seq : htminer::items) {
-                    htminer::M = std::max<unsigned int>(htminer::M, seq.size());
+                for (auto &seq : items)
                     for (int x : seq)
                         max_id = std::max(max_id, std::abs(x));
-                    htminer::E += seq.size();
+                L = max_id;
+
+                theta = (minsup < 1.0) ? std::ceil(minsup * N) : minsup;
+
+                DFS.clear();
+                DFS.reserve(L);
+                for (unsigned int i = 0; i < L; ++i)
+                    DFS.emplace_back(-static_cast<int>(i) - 1);
+
+                M = 0;
+                E = 0;
+                for (auto &seq : items) {
+                    M = std::max<unsigned int>(M, seq.size());
+                    E += seq.size();
                 }
-                htminer::L = max_id;
-                htminer::theta = (minsup < 1.0) 
-                                  ? static_cast<unsigned long long>(std::ceil(minsup * htminer::N))
-                                  : static_cast<unsigned long long>(minsup);
-
-                // build empty DFS stack (size L) as HTMiner expects
-                htminer::DFS.clear();
-                htminer::DFS.reserve(htminer::L);
-                for (unsigned int i = 0; i < static_cast<unsigned int>(htminer::L); ++i)
-                    htminer::DFS.emplace_back(-static_cast<int>(i) - 1);
-
-                // initialize VDFS if HTMiner needs it
-                htminer::VDFS.clear();
-                htminer::VDFS.resize(htminer::L);
             }
 
-            // 3) run the mining algorithm
-            htminer::Freq_miner();
-
-            // std::cout << "[HTMiner] dumping all collected patterns:\n";
-            // for (size_t i = 0; i < htminer::collectedPatterns.size(); ++i) {
-            //     const auto &seq = htminer::collectedPatterns[i];
-            //     std::cout << "Pattern " << i << ": { ";
-            //     for (int x : seq) {
-            //          std::cout << x << " ";
-            //         }
-            //         std::cout << "}\n";
-//}
-std::cout << " total patterns = "
-          << htminer::collectedPatterns.size() << "\n";
-// ─────────────────────────────────────────────────
-
-            // 4) return patterns + elapsed time
+            Freq_miner();
+
             py::dict out;
-            out["patterns"] = htminer::GetCollected();
-            out["time"]     = htminer::give_time(std::clock() - htminer::start_time);
+            out["patterns"] = GetCollected();
+            out["time"]     = give_time(std::clock() - start_time);
             return out;
         },
         py::arg("data"),
@@ -268,8 +792,223 @@ std::cout << " total patterns = "
         py::arg("verbose")    = false,
         py::arg("out_file")   = ""
     );
+m.def("BTMiner",
+    [](py::object data,
+       double minsup,
+       unsigned int time_limit,
+       bool preproc,
+       bool use_dic,
+       bool verbose,
+       const std::string &out_file)
+    {
+        // We are calling the *professor* BTMiner, now namespaced as btminer::.
+        // So we only set the globals the professor code actually has.
+
+        // 1) configure professor globals
+        btminer::time_limit = static_cast<int>(time_limit);
+        btminer::pre_pro    = preproc;
+        btminer::use_dic    = use_dic;
+        btminer::b_disp     = verbose;
+        btminer::b_write    = !out_file.empty();
+        btminer::out_file   = out_file;
+        btminer::N_mult     = 1;         // professor uses these too
+        btminer::M_mult     = 1;
+        btminer::just_build = false;     // we want full mining
+
+        btminer::start_time = std::clock();
+
+        // 2) load data
+        //
+        // Professor’s code is primarily file-based (Load_instance(const string&, double)).
+        // So: if user passes a file path → use the professor loader directly.
+        // If user passes a Python list-of-lists → we will build the MDD the same
+        // way professor’s loader does, but without changing his logic.
+        if (py::isinstance<py::str>(data)) {
+            // ----- FILE MODE -----
+            std::string path = data.cast<std::string>();
+
+            if (verbose) {
+                std::cerr << "[BT][binding] file=" << path
+                          << " minsup=" << minsup
+                          << " preproc=" << preproc << std::endl;
+            }
+
+            if (!btminer::Load_instance(path, minsup)) {
+                throw std::runtime_error("BTMiner: failed to load file: " + path);
+            }
+        } else {
+            // ----- PYTHON LIST MODE -----
+            //
+            // We mimic professor’s loader:
+            //   - create root in Tree
+            //   - compute N, M, L
+            //   - compute theta from minsup
+            //   - seed DFS (one Pattern per item, as in Preprocess branch)
+            //   - call Build_MDD(...) for each sequence
+            //
+            // This DOES NOT change his mining logic; it just drives it from memory.
+
+            auto seqs = data.cast<std::vector<std::vector<int>>>();
+
+            // clear MDD and globals to a known state
+            btminer::Tree.clear();
+            btminer::Tree.emplace_back(0, 0, 0);      // root (exactly like professor)
+
+            // compute basic stats
+            int max_id = 0;
+            int max_len = 0;
+            int seq_count = 0;
+            long long entries = 0;
+
+            for (const auto &s : seqs) {
+                if (s.empty()) continue;
+                ++seq_count;
+                max_len = std::max<int>(max_len, static_cast<int>(s.size()));
+                for (int x : s) {
+                    max_id = std::max(max_id, std::abs(x));
+                    ++entries;
+                }
+            }
+
+            btminer::N = seq_count;
+            btminer::M = max_len;
+            btminer::L = max_id;
+            btminer::E = static_cast<int>(entries);
+
+            // theta = abs support
+            if (minsup < 1.0)
+                btminer::theta = static_cast<int>(std::ceil(minsup * btminer::N * btminer::N_mult));
+            else
+                btminer::theta = static_cast<int>(minsup);
+
+            // seed DFS exactly like professor does in the preprocessed branch:
+            btminer::DFS.clear();
+            btminer::DFS.reserve(btminer::L);
+            for (int i = 0; i < btminer::L; ++i)
+                btminer::DFS.emplace_back(-i - 1);
+
+            // now build the MDD, sequence by sequence
+            for (const auto &s : seqs) {
+                if (s.empty()) continue;
+                // professor’s Build_MDD takes a vector<int> by non-const ref
+                std::vector<int> tmp = s;
+                btminer::Build_MDD(tmp);
+            }
+
+            if (verbose) {
+                std::cerr << "[BT][binding] PY mode: N=" << btminer::N
+                          << " L=" << btminer::L
+                          << " M=" << btminer::M
+                          << " E=" << btminer::E
+                          << " theta=" << btminer::theta
+                          << " Tree.size()=" << btminer::Tree.size()
+                          << std::endl;
+            }
+        }
+
+        // 3) run professor’s miner
+        btminer::Freq_miner();
+
+        // 4) build python result
+        // 4) build python result
+            py::dict out;
+            out["patterns"]     = btminer::GetCollected();    // ← NEW
+            out["num_patterns"] = btminer::num_patt;
+            out["time"]         = btminer::give_time(std::clock() - btminer::start_time);
+            out["N"]            = btminer::N;
+            out["L"]            = btminer::L;
+            out["theta"]        = btminer::theta;
+            return out;
+
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+);
+
+
+
+       
+// HTMiner
+// ─────────────────────────────────────────────────────────────
+// HTMiner
+m.def("HTMiner",
+    [](py::object data,
+       double minsup, unsigned int time_limit,
+       bool preproc, bool use_dic,
+       bool verbose, const std::string &out_file)
+    {
+        htminer::time_limit = time_limit;
+        htminer::pre_pro    = preproc;
+        htminer::use_dic    = use_dic;
+        htminer::just_build = false;
+        htminer::use_list   = false;
+        htminer::b_disp     = verbose;
+        htminer::b_write    = !out_file.empty();
+        htminer::out_file   = out_file;
+        htminer::ClearCollected();
+        htminer::start_time = std::clock();
+
+        if (py::isinstance<py::str>(data)) {
+            std::string path = data.cast<std::string>();
+            if (!htminer::Load_instance(path, minsup))
+                throw std::runtime_error("Failed to load file: " + path);
+        } else {
+            auto seqs = data.cast<std::vector<std::vector<int>>>();
+            htminer::items = std::move(seqs);
+            htminer::N = htminer::items.size();
+
+            int max_id = 0;
+            htminer::M = 0;
+            htminer::E = 0;
+            for (auto &seq : htminer::items) {
+                htminer::M = std::max<unsigned int>(htminer::M, seq.size());
+                for (int x : seq)
+                    max_id = std::max(max_id, std::abs(x));
+                htminer::E += seq.size();
+            }
+            htminer::L = max_id;
+            htminer::theta = (minsup < 1.0)
+                              ? static_cast<unsigned long long>(std::ceil(minsup * htminer::N))
+                              : static_cast<unsigned long long>(minsup);
+
+            htminer::DFS.clear();
+            htminer::DFS.reserve(htminer::L);
+            for (unsigned int i = 0; i < static_cast<unsigned int>(htminer::L); ++i)
+                htminer::DFS.emplace_back(-static_cast<int>(i) - 1);
+
+            htminer::VDFS.clear();
+            htminer::VDFS.resize(htminer::L);
+        }
+
+        htminer::Freq_miner();
+
+        // 👇 now really respects verbose
+        if (verbose) {
+            std::cout << " total patterns = "
+                      << htminer::collectedPatterns.size() << "\n";
+        }
+
+        py::dict out;
+        out["patterns"] = htminer::GetCollected();
+        out["time"]     = htminer::give_time(std::clock() - htminer::start_time);
+        return out;
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+);
+
 
-    m.def("LargePrefixProjection",
+ m.def("LargePrefixProjection",
     [](py::object data,
        double minsup,
        unsigned int time_limit,
@@ -281,28 +1020,30 @@ std::cout << " total patterns = "
         largepp::time_limit = time_limit;
         largepp::pre_pro    = preproc;
         largepp::use_dic    = use_dic;
-        largepp::use_list   = true;          // ← key difference
+        largepp::use_list   = true;
         largepp::b_disp     = verbose;
         largepp::b_write    = !out_file.empty();
         largepp::out_file   = out_file;
-        largepp::just_build = false;  
+        largepp::just_build = false;
 
         largepp::ClearCollected();
         largepp::start_time = std::clock();
-        std::string fname = data.cast<std::string>();
-        /* 1) load instance (py list or filename) */
-        if (py::isinstance<py::str>(data))
-            
-            largepp::Load_instance(fname, minsup);
-        else
-            largepp::Load_py(data, minsup);          // helper you’ll expose
-        
-        std::vector<unsigned long long> dbg;
-
-
-        
 
+        // 👇 this was the last noisy one
+        if (verbose) {
+            std::cerr << " minsup=" << minsup
+                      << " preproc=" << preproc
+                      << " verbose=" << verbose
+                      << " out_file=" << (out_file.empty() ? "(none)" : out_file)
+                      << " use_dic=" << use_dic << "\n";
+        }
 
+        if (py::isinstance<py::str>(data)) {
+            std::string fname = data.cast<std::string>();
+            largepp::Load_instance(fname, minsup);
+        } else {
+            largepp::Load_py(data, minsup);
+        }
 
         largepp::Freq_miner();
 
@@ -320,6 +1061,8 @@ std::cout << " total patterns = "
     py::arg("out_file")   = ""
 );
 
+
+
 // ─────────────────────────────────────────────────────────────
 // LargeBTMiner  -- Python wrapper for the largebm implementation
 // ─────────────────────────────────────────────────────────────
@@ -404,100 +1147,108 @@ std::cout << " total patterns = "
 
 
 
-m.def("LargeBTMiner",
-        [](py::object data,
-           double minsup,
-           unsigned int time_limit,
-           bool preproc,
-           bool use_dic,
-           bool verbose,
-           const std::string &out_file)
-        {
-            // 0) Set global flags and timers
-            largebm::time_limit = time_limit;
-            largebm::pre_pro    = preproc;
-            largebm::use_dic    = use_dic;
-            largebm::use_list   = false;         // large‑mode → always MDD
-            largebm::b_disp     = verbose;
-            largebm::b_write    = !out_file.empty();
-            largebm::out_file   = out_file;
-            largebm::just_build = false;
+    // ─────────────────────────────────────────────────────────────────────────
+    // LargeBTMiner (MDD-based)
+    // ─────────────────────────────────────────────────────────────────────────
+ /*m.def("LargeBTMiner",
+    [](py::object data,
+       double minsup,
+       unsigned int time_limit,
+       bool preproc,
+       bool use_dic,
+       bool verbose,
+       const std::string &out_file)
+    {
+        using namespace largebm;
+
+        // 0) Set global flags and timers
+        largebm::time_limit = time_limit;
+        largebm::pre_pro    = preproc;
+        largebm::use_dic    = use_dic;
+        largebm::use_list   = false;         // large-mode → always MDD
+        largebm::b_disp     = verbose;
+        largebm::b_write    = !out_file.empty();
+        largebm::out_file   = out_file;
+        largebm::just_build = false;
+
+        // 0.1) Clear any leftover data/state from previous runs
+        largebm::items.clear();
+        largebm::item_dic.clear();
+        largebm::inv_item_dic.clear();
+        largebm::Tree.clear();
+        largebm::DFS.clear();
+        largebm::ClearCollected();
+
+        // 1) Load sequences (either from filename or from Python list)
+        if (py::isinstance<py::str>(data)) {
+            // ─────────── FILE-BASED MODE ───────────
+            std::string path = data.cast<std::string>();
+            if (!largebm::Load_instance(path, minsup))
+                throw std::runtime_error("Failed to load file: " + path);
 
-            // 0.1) Clear any leftover data/state from previous runs
-            largebm::items.clear();
-            largebm::item_dic.clear();
-            largebm::inv_item_dic.clear();
-            largebm::Tree.clear();
-            largebm::DFS.clear();
-            largebm::ClearCollected();
+        } else {
+            // ────────── IN-MEMORY MODE ──────────
+            auto seqs = data.cast<std::vector<std::vector<int>>>();
+            largebm::items = std::move(seqs);
+            largebm::N     = largebm::items.size();
 
-            // 1) Load sequences (either from filename or from Python list)
-            if (py::isinstance<py::str>(data)) {
-                // ─────────── FILE‑BASED MODE ───────────
-                std::string path = data.cast<std::string>();
-                if (!largebm::Load_instance(path, minsup))
-                    throw std::runtime_error("Failed to load file: " + path);
+            // 1.1) Compute basic DB statistics (M, E, L) and absolute support θ
+            int max_id = 0;
+            largebm::M = 0;
+            largebm::E = 0;
+            for (auto &seq : largebm::items) {
+                largebm::M = std::max<unsigned int>(largebm::M, static_cast<unsigned int>(seq.size()));
+                largebm::E += static_cast<unsigned long long>(seq.size());
+                for (int x : seq) max_id = std::max(max_id, std::abs(x));
+            }
+            largebm::L = static_cast<unsigned int>(max_id);
+            largebm::theta = (minsup < 1.0)
+                               ? static_cast<unsigned long long>(std::ceil(minsup * largebm::N))
+                               : static_cast<unsigned long long>(minsup);
 
-            } else {
-                // ────────── IN‑MEMORY MODE ──────────
-                auto seqs = data.cast<std::vector<std::vector<int>>>();
-                largebm::items = std::move(seqs);
-                largebm::N     = largebm::items.size();
+            // 1.2) Initialize DFS buffer (size = L)
+            largebm::DFS.reserve(largebm::L);
+            for (unsigned int i = 0; i < largebm::L; ++i)
+                largebm::DFS.emplace_back(-static_cast<int>(i) - 1);
 
-                // 1.1) Compute basic DB statistics (M, E, L) and absolute support θ
-                int max_id = 0;
-                largebm::M = 0;
-                largebm::E = 0;
-                for (auto &seq : largebm::items) {
-                    largebm::M = std::max<unsigned int>(largebm::M, static_cast<unsigned int>(seq.size()));
-                    largebm::E += static_cast<unsigned long long>(seq.size());
-                    for (int x : seq) max_id = std::max(max_id, std::abs(x));
-                }
-                largebm::L = static_cast<unsigned int>(max_id);
-                largebm::theta = (minsup < 1.0)
-                                   ? static_cast<unsigned long long>(std::ceil(minsup * largebm::N))
-                                   : static_cast<unsigned long long>(minsup);
-
-                // 1.2) Initialize DFS buffer (size = L)
-                largebm::DFS.reserve(largebm::L);
-                for (unsigned int i = 0; i < largebm::L; ++i)
-                    largebm::DFS.emplace_back(-static_cast<int>(i) - 1);
-
-                // 1.3) Build the MDD “Tree”
-                // Insert one dummy root node (item=0, freq=0, anct=0)
-                largebm::Tree.emplace_back(0, 0, 0);
-                for (auto &seq : largebm::items)
-                    largebm::Build_MDD(const_cast<std::vector<int>&>(seq));
-            }
+            // 1.3) Build the MDD “Tree”
+            // Insert one dummy root node (item=0, freq=0, anct=0)
+            largebm::Tree.emplace_back(0, 0, 0);
+            for (auto &seq : largebm::items)
+                largebm::Build_MDD(const_cast<std::vector<int>&>(seq));
+        }
 
-            // 2) Rebuild inverse‑dictionary from fresh item_dic
-            {
-                std::vector<int> inv(largebm::item_dic.size() + 1);
-                for (int old = 1; old <= static_cast<int>(largebm::item_dic.size()); ++old) {
-                    int cid = largebm::item_dic[old - 1];
-                    if (cid > 0) inv[cid] = old;
-                }
-                largebm::inv_item_dic = std::move(inv);
+        // 2) Rebuild inverse-dictionary from fresh item_dic
+        {
+            std::vector<int> inv(largebm::item_dic.size() + 1);
+            for (int old = 1; old <= static_cast<int>(largebm::item_dic.size()); ++old) {
+                int cid = largebm::item_dic[old - 1];
+                if (cid > 0) inv[cid] = old;
             }
+            largebm::inv_item_dic = std::move(inv);
+        }
 
-            // 3) Start timing and run the miner
-            largebm::start_time = std::clock();
-            largebm::Freq_miner();
+        // 3) Start timing and run the miner
+        largebm::start_time = std::clock();
+        largebm::Freq_miner();
+
+        // 4) Collect results and elapsed time
+        const auto& pats = largebm::GetCollected();
+
+        py::dict out;
+        out["patterns"] = pats;
+        out["time"]     = largebm::give_time(std::clock() - largebm::start_time);
+        return out;
+    },
+    py::arg("data"),
+    py::arg("minsup")     = 0.01,
+    py::arg("time_limit") = 36000,
+    py::arg("preproc")    = false,
+    py::arg("use_dic")    = false,
+    py::arg("verbose")    = false,
+    py::arg("out_file")   = ""
+);
 
-            // 4) Collect results and elapsed time
-            py::dict out;
-            out["patterns"] = largebm::GetCollected();
-            out["time"]     = largebm::give_time(std::clock() - largebm::start_time);
-            return out;
-        },
-        py::arg("data"),
-        py::arg("minsup")     = 0.01,
-        py::arg("time_limit") = 36000,
-        py::arg("preproc")    = false,
-        py::arg("use_dic")    = false,
-        py::arg("verbose")    = false,
-        py::arg("out_file")   = ""
-    );
 
    
 m.def("LargeHTMiner",
@@ -606,4 +1357,4 @@ m.def("LargeHTMiner",
 
 
 
-}
+} */