effspm 0.3.2__cp39-cp39-macosx_10_9_x86_64.whl → 0.3.3__cp39-cp39-macosx_10_9_x86_64.whl

effspm/_effspm.cpp

@@ -166,7 +166,7 @@ PYBIND11_MODULE(_effspm, m) {
    // ─────────────────────────────────────────────────────────────
 // BTMiner  (always uses professor's Load_instance)
 // ─────────────────────────────────────────────────────────────
-m.def("BTMiner",
+/*m.def("BTMiner",
     [](py::object data,
        double minsup,
        unsigned int time_limit,
@@ -250,8 +250,100 @@ m.def("BTMiner",
     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)
@@ -351,49 +443,68 @@ m.def("HTMiner",
     // ─────────────────────────────────────────────────────────────
     // LargePrefixProjection  (already has its own Load_py)
     // ─────────────────────────────────────────────────────────────
-    m.def("LargePrefixProjection",
-        [](py::object data,
-           double minsup,
-           unsigned int time_limit,
-           bool preproc,
-           bool use_dic,
-           bool verbose,
-           const std::string &out_file)
-        {
-            largepp::time_limit = time_limit;
-            largepp::pre_pro    = preproc;
-            largepp::use_dic    = use_dic;
-            largepp::use_list   = true;        // large prefix uses list-based mining
-            largepp::b_disp     = verbose;
-            largepp::b_write    = !out_file.empty();
-            largepp::out_file   = out_file;
-            largepp::just_build = false;
-
-            largepp::ClearCollected();
-            largepp::start_time = std::clock();
+   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;
 
-            if (py::isinstance<py::str>(data)) {
-                std::string fname = data.cast<std::string>();
-                largepp::Load_instance(fname, minsup);
-            } else {
-                largepp::Load_py(data, minsup);
-            }
+        // 2) HARD RESET of largepp global state
+        //    (only touch symbols that actually exist in largepp)
+        largepp::ClearCollected();         // clear previously collected patterns
 
-            largepp::Freq_miner();
+        // 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
 
-            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")   = ""
-    );
+        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)

effspm/btminer/src/load_inst.cpp

@@ -31,7 +31,7 @@ map<int, string>  item_map_rev;
 
 std::vector<int>  freq;
 std::vector<int>  item_dic;
-
+std::vector<std::vector<int>> items; 
 // ✅ REAL DEFINITION lives here:
 std::vector<Pattern> DFS;
 
@@ -103,6 +103,15 @@ bool Load_instance(string &items_file, double thresh) {
 // preprocessing pass
 // ---------------------------------------------------------------------
 bool Preprocess(string &inst, double thresh) {
+    N = 0;
+    L = 0;
+    freq.clear();
+    item_dic.clear();
+    item_map.clear();
+    item_map_rev.clear();
+    // (E is usually for entries during Build_MDD, so we can leave it
+    //  for the load phase; it’s already reset in the binding)
+
     ifstream file(inst);
 
     if (file.good()) {

{effspm-0.3.2.dist-info → effspm-0.3.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: effspm
-Version: 0.3.2
+Version: 0.3.3
 Summary: Prefix‑Projection and other sequential pattern mining algorithms
 Author: Yeswanth Vootla
 Author-email: yeshu999 <vootlayeswanth20@gmail.com>

{effspm-0.3.2.dist-info → effspm-0.3.3.dist-info}/RECORD

@@ -1,10 +1,10 @@
 effspm/utility.hpp,sha256=Y_MQVk9AmJWjguKyuyk0LraTi_6VzZFg0StsmVOCkNc,744
 effspm/_core.cpp,sha256=S6UsUcl0HQLMQUTy2tMJxcrbLGJo8tjkLskyHqESYyI,3675
 effspm/load_inst.hpp,sha256=VF0o_Ur-mO5xhdWWn1KHrFLu9_bTbd8wcURCVF0A0dM,907
-effspm/_effspm.cpython-39-darwin.so,sha256=tYoNx3LOrqFnoHElyzxd1ufWI0y0aZs9MWrFZy_9fpo,561304
+effspm/_effspm.cpython-39-darwin.so,sha256=YBHCR5TZgsCI951TUZC1kIrHqNF2E9wculJI-C-RHxI,561376
 effspm/__init__.py,sha256=_BkZ8cFlB_l1haxTzxJMgFHhO6LEmF3pIY-nmPTPMj8,246
 effspm/freq_miner.cpp,sha256=9K-nO_c-vVzbFlmh511vmvypA2fu1TXpiJhyx03owDU,4642
-effspm/_effspm.cpp,sha256=JC4CPqjV4nyyHRv7EhQPQvXOb7Pj0Ge8Prxo0WCsCAg,47255
+effspm/_effspm.cpp,sha256=NkxFvVVbR5CAsSUZO5Z_Z-UwFVCQRhy6_st8_cD-2Us,51027
 effspm/load_inst.cpp,sha256=h9pjtvRCIrtqIAt8W1r8SxOb1YViW1pCR52tbxbPdh0,4591
 effspm/freq_miner.hpp,sha256=XDhydwUFmute6slWwZcFuiBjGZs4Kv5-UvAmcI_IMwI,786
 effspm/utility.cpp,sha256=luWwBNy7OVWRmam9gz2RjhtrRmx6c37oOobFJaDWqcA,1403
@@ -48,13 +48,13 @@ effspm/btminer/src/build_mdd.cpp,sha256=7J5KMNNB3DngB4Bkn1h-zMfAgRvRiXP4MEkPbd3C
 effspm/btminer/src/utility.hpp,sha256=5m34T2ATvYY_32Sz6h0Zro4kLolosPwASJF0TiDKhhI,312
 effspm/btminer/src/load_inst.hpp,sha256=Fj72XjlR7ppGi76vc0PMZZAkj_-JklTVuIqzbsZcTAw,987
 effspm/btminer/src/freq_miner.cpp,sha256=sTA-bkUZqgFU9tVtSsUUfaVj7BU8MiFZm3scwRsxMeg,9192
-effspm/btminer/src/load_inst.cpp,sha256=yRHso99ban7sQNe2SpZpXeNUXu0NDVNq4-rt1yZVoso,8083
+effspm/btminer/src/load_inst.cpp,sha256=fquloXEJlmSXYbwstgX_lAO5-ar0CS6Gx1cfHQRQbS0,8366
 effspm/btminer/src/freq_miner.hpp,sha256=sS2CGdNT0zSGtSz0evZZlUkuiKyWlSvrR3-M2YXP7-Q,1055
 effspm/btminer/src/build_mdd.hpp,sha256=p0pEcNZMD-WqV0UB1WtOn-Soe0105gEL071Ht5okgJM,627
 effspm/btminer/src/utility.cpp,sha256=YmwdNPCUHFjydwrUAyBLEkFqUecyg7r4HbPdgoD-j3s,1194
 effspm/btminer/src/main.cpp,sha256=Jh9M5nsZUL3i-iENn7Jh--TOY0F5dnu6CvqZ7udWU3A,2678
-effspm-0.3.2.dist-info/RECORD,,
-effspm-0.3.2.dist-info/WHEEL,sha256=Ui60MbQStN7UWmlHrvRSzB8m_B5-okiNGeb8CWQReOs,108
-effspm-0.3.2.dist-info/top_level.txt,sha256=2O-AuI0nw0pDmJMo2jzM1wvV2rj48AmkjskkAnsuuQk,7
-effspm-0.3.2.dist-info/METADATA,sha256=JfNcmGZRF3qJbQKDXloeRbOHMw0R8tQNb-80p9KNjW0,14227
-effspm-0.3.2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+effspm-0.3.3.dist-info/RECORD,,
+effspm-0.3.3.dist-info/WHEEL,sha256=Ui60MbQStN7UWmlHrvRSzB8m_B5-okiNGeb8CWQReOs,108
+effspm-0.3.3.dist-info/top_level.txt,sha256=2O-AuI0nw0pDmJMo2jzM1wvV2rj48AmkjskkAnsuuQk,7
+effspm-0.3.3.dist-info/METADATA,sha256=0V27Cms7UBjdK62_SOAdqiAhq0bOIByG6oUTTALZEn8,14227
+effspm-0.3.3.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357