effspm 0.2.7__cp39-cp39-win_amd64.whl → 0.3.3__cp39-cp39-win_amd64.whl

effspm/largebm/src/freq_miner.hpp

@@ -1,48 +1,37 @@
 #pragma once
-
-#include "load_inst.hpp"
-#include "build_mdd.hpp"
+#include <vector>
 
 namespace largebm {
-	
-void Freq_miner();
-// recursive helper for the list‐based mode
- void Freq_miner_list(const std::vector<std::vector<int>>& db,
-                            std::vector<int>& prefix,
-                            unsigned long long theta,
-                            std::vector<std::vector<int>>& out);
+
 class Pattern {
 public:
-
-	vector<int> seq;
-	vector<unsigned long long int> str_pnt;
-	vector<int> list;
-
-	unsigned long long int freq;
-
-	Pattern(vector<int>& _seq, int item) {
-		seq.swap(_seq);
-		seq.push_back(item);
-		freq = 0;
-	}
-
-	Pattern(int item) {
-		seq.push_back(item);
-		freq = 0;
-	}
-
-	Pattern() {
-		freq = 0;
-	}
-
-
+    std::vector<int> seq;
+    std::vector<unsigned long long> str_pnt;
+    std::vector<int> list;
+    unsigned long long freq = 0;
+
+    Pattern() = default;
+    Pattern(std::vector<int>& _seq, int item) {
+        seq.swap(_seq);
+        seq.push_back(item);
+        freq = 0;
+    }
+    Pattern(int item) {
+        seq.push_back(item);
+        freq = 0;
+    }
 };
 
+void Freq_miner();
+void Freq_miner_list(const std::vector<std::vector<int>>& db,
+                     std::vector<int>& prefix,
+                     unsigned long long theta,
+                     std::vector<std::vector<int>>& out);
+
 extern unsigned long long int num_patt;
 extern std::vector<bool> ilist;
 extern std::vector<bool> slist;
 extern std::vector<int>  DFS_numfound;
-extern Pattern          _patt;
-
+extern Pattern            _patt;
 
-}
+} // namespace largebm

effspm/largebm/src/load_inst.cpp

@@ -1,13 +1,8 @@
-
-#include <sstream>
 #include <algorithm>
 #include <cmath>
-#include <ctime>
-#include <iostream>      // for std::cout, std::endl
 #include <fstream>
-#include <vector>
-#include <string>
-
+#include <iostream>
+#include <sstream>
 #include "load_inst.hpp"
 #include "build_mdd.hpp"
 #include "utility.hpp"
@@ -15,7 +10,7 @@
 
 namespace largebm {
 
-// ── global definitions ────────────────────────────────────────────
+// ── single definitions of globals ─────────────────────────────────
 bool use_list      = false;
 bool b_disp        = false;
 bool b_write       = false;
@@ -51,11 +46,10 @@ static void Load_items_list(const std::string& fname) {
             if (item_dic[a - 1] == -1) continue;
             seq.push_back(x);
         }
-        if (!seq.empty()) items.push_back(std::move(seq));
+        if (!seq.empty()) items.push_back(seq);
     }
 }
 
-// ─────────────── main loader ─────────────────────────────────────
 bool Load_instance(const std::string& items_file, double minsup) {
     // reset state
     N = L = num_nodes = theta = M = E = 0;
@@ -90,9 +84,6 @@ bool Load_instance(const std::string& items_file, double minsup) {
     // MDD build mode
     if (pre_pro) {
         if (!Preprocess(items_file, minsup)) return false;
-        std::cout << "\nPreprocess done in "
-                  << give_time(std::clock() - kk)
-                  << " seconds\n\n";
         DFS.clear();
         DFS.reserve(L);
         for (unsigned int i = 0; i < L; ++i)
@@ -100,23 +91,29 @@ bool Load_instance(const std::string& items_file, double minsup) {
         kk = std::clock();
         Load_items_pre(items_file);
     } else {
-        if (!Preprocess(items_file, 0.0)) return false;
+        if (!Preprocess(items_file, minsup)) return false;
         kk = std::clock();
         Load_items(items_file);
     }
 
-    std::cout << "\nMDD Database built in "
-              << give_time(std::clock() - kk)
-              << " seconds\n\n";
-    std::cout << "Found " << N
-              << " sequences, with max line len " << M
-              << ", and " << L << " items, and " << E << " entries\n";
-    std::cout << "Total MDD nodes: " << Tree.size() << std::endl;
+    // ensure DFS size
+    if (DFS.size() < L) {
+        DFS.reserve(L);
+        while (DFS.size() < L) {
+            DFS.emplace_back(-int(DFS.size()) - 1);
+        }
+    }
+
+    // SAFETY — seed any zeroed singletons from their str_pnt list
+    for (unsigned int i = 0; i < L && i < DFS.size(); ++i) {
+        if (DFS[i].freq == 0 && !DFS[i].str_pnt.empty()) {
+            DFS[i].freq = static_cast<unsigned long long>(DFS[i].str_pnt.size());
+        }
+    }
 
     return true;
 }
 
-// ────────────── Preprocess (list mode) ───────────────────────────
 bool Preprocess(const std::string& inst, double thresh) {
     std::ifstream file(inst);
     if (!file.good()) return false;
@@ -158,7 +155,6 @@ bool Preprocess(const std::string& inst, double thresh) {
     return true;
 }
 
-// Load_items_pre: MDD insert from file
 void Load_items_pre(const std::string& inst_name) {
     std::ifstream file(inst_name);
     if (!file.good()) return;
@@ -182,7 +178,7 @@ void Load_items_pre(const std::string& inst_name) {
                 continue;
             }
             if (ditem > 0) { ditem = item_dic[ditem - 1]; itmset_exists = true; }
-            else { ditem = -item_dic[-ditem - 1]; }
+            else           { ditem = -item_dic[-ditem - 1]; }
             if (sgn) { if (ditem > 0) ditem = -ditem; sgn = false; }
             temp_vec.push_back(ditem);
         }
@@ -193,7 +189,6 @@ void Load_items_pre(const std::string& inst_name) {
     }
 }
 
-// Load_items: full MDD build
 bool Load_items(const std::string& inst_name) {
     std::ifstream file(inst_name);
     if (!file.good()) return false;
@@ -227,4 +222,3 @@ void ClearCollected()   { collected.clear(); }
 const std::vector<std::vector<int>>& GetCollected() { return collected; }
 
 } // namespace largebm
-

effspm/largebm/src/load_inst.hpp

@@ -1,45 +1,35 @@
 #pragma once
-
 #include <vector>
 #include <string>
-#include <fstream>
-#include <cmath>
 #include <ctime>
 
-
 namespace largebm {
-    using namespace std;
 
-    // forward-declare Pattern (defined in freq_miner.hpp)
-    class Pattern;
-    
-     struct Arc; 
-    extern std::vector<Arc> Tree;                   // <-- add this
-    void Build_MDD(const std::vector<int>& seq);   
-    // ─── Entry points ─────────────────────────────────────────────────────────
-    bool Load_instance(const string& items_file, double thresh);
-    bool Preprocess(const string& fname, double thresh);
-    void Load_items_pre(const string& fname);
-    bool Load_items(const string& fname);
-    extern std::vector<int> inv_item_dic;
-    // Called by the Python‐wrapper when passing a Python list of lists
-    
+class Pattern; // forward
+// [2025-10-25 NEW]: match the real definition (class, not struct) to avoid ABI warnings
+class Arc;     // was: struct Arc;
+
+// Config & state (single definitions in load_inst.cpp)
+extern std::string        out_file, folder;
+extern bool               use_list;
+extern bool               b_disp, b_write, use_dic, just_build, pre_pro, itmset_exists;
+extern unsigned int       M, L, time_limit;
+extern unsigned long long N, num_nodes, theta, E;
+extern std::clock_t       start_time;
+
+extern std::vector<std::vector<int>> items;
+extern std::vector<int>              item_dic;
+extern std::vector<int>              inv_item_dic;
+extern std::vector<Pattern>          DFS;
+extern std::vector<std::vector<int>> collected;
 
-    // ─── Config globals (must match btminer types exactly) ────────────────────
-    extern string        out_file, folder;
-    extern bool          use_list;
-    extern bool          b_disp, b_write, use_dic, just_build, pre_pro, itmset_exists;
-    extern unsigned int  M, L, time_limit;
-    extern unsigned long long N, num_nodes, theta, E;
-    extern clock_t       start_time;
-   // extern std::vector<Arc>                 Tree;
-    // ─── Data for list-based (LargeBTMiner) mode ───────────────────────────────
-    extern std::vector<std::vector<int>> items;
-    extern std::vector<int>              item_dic;
-    extern std::vector<Pattern>          DFS;        // Pattern is now declared
-    extern std::vector<std::vector<int>> collected;
-        void ClearCollected();
-    const std::vector<std::vector<int>>& GetCollected();
+// Loader API
+bool Load_instance(const std::string& items_file, double thresh);
+bool Preprocess(const std::string& fname, double thresh);
+void Load_items_pre(const std::string& fname);
+bool Load_items(const std::string& fname);
 
+void ClearCollected();
+const std::vector<std::vector<int>>& GetCollected();
 
 } // namespace largebm

effspm/largebm/src/main.cpp

@@ -0,0 +1,95 @@
+#include <iostream>
+#include <time.h>
+#include <string.h>
+#include <string>
+#include "load_inst.hpp"
+#include "build_mdd.hpp"
+#include "utility.hpp"
+#include "freq_miner.hpp"
+
+namespace largebm{
+using namespace std;
+
+string out_file;
+
+bool  b_disp = 0, b_write = 0, just_build = 0, pre_pro = 1;
+
+int time_limit = 30 * 3600;
+
+clock_t start_time;
+
+string folder;
+
+int main(int argc, char* argv[]) {
+		
+	string VV, attr;
+
+	double thresh = 0;
+	for (int i = 1; i<argc; i++) {
+		if (argv[i][0] != '-' || isdigit(argv[i][1]))
+			continue;
+		else if (strcmp(argv[i], "-thr") == 0)
+			thresh = stod(argv[i + 1]);
+		else if (strcmp(argv[i], "-file") == 0)
+			VV = argv[i + 1];
+		else if (strcmp(argv[i], "-time") == 0)
+			time_limit = stoi(argv[i + 1]);
+		else if (strcmp(argv[i], "-jbuild") == 0)
+			just_build = 1;
+		else if (strcmp(argv[i], "-folder") == 0)
+			folder = argv[i + 1];
+		else if (strcmp(argv[i], "-npre") == 0)
+			pre_pro = 0;
+		else if (strcmp(argv[i], "-out") == 0) {
+			if (i + 1 == argc || argv[i + 1][0] == '-')
+				b_disp = 1;
+			else if (argv[i + 1][0] == '+') {
+				b_disp = 1;
+				b_write = 1;
+				if (strlen(argv[i + 1]) > 1) {
+					out_file = argv[i + 1];
+					out_file = out_file.substr(1, out_file.size() - 1);
+				}
+				else
+					out_file = VV;
+			}
+			else {
+				b_write = 1;
+				out_file = argv[i + 1];
+			}
+		}
+
+		else
+			cout << "Command " << argv[i] << " not recognized and skipped.\n";
+	}
+
+
+
+    cout << "\n********************** " << VV << "**********************\n";
+
+	string item_file = folder + VV + ".txt";
+
+	cout << "loading instances...\n";
+
+	start_time = clock();
+
+	if (!Load_instance(item_file, thresh)) {
+		cout << "Files invalid, exiting.\n";
+		cin.get();
+		return 0;
+	}
+
+	//kk = clock();
+
+	if (!just_build && give_time(clock() - start_time) < time_limit) {
+		Freq_miner();
+		if (give_time(clock() - start_time) >= time_limit)
+			cout << "TIME LIMIT REACHED\n";
+		cout << "Mining Complete\n\nFound a total of " << num_patt << " patterns\n";	
+		cout << "\nTotal CPU time " << give_time(clock() - start_time) << " seconds\n\n";
+	}
+
+
+	return 0;
+}
+}

effspm/largebm/src/utility.cpp

@@ -1,45 +1,35 @@
 #include "utility.hpp"
 #include "build_mdd.hpp"
-#include "load_inst.hpp"
-
 #include <vector>
-#include <ctime>
-#include <algorithm>
 
 namespace largebm {
 
+double give_time(std::clock_t ticks) {
+    return static_cast<double>(ticks) / CLOCKS_PER_SEC;
+}
+
 bool check_parent(unsigned long long cur_arc,
                   unsigned long long str_pnt,
                   unsigned long long start,
-                  std::vector<unsigned long long>& strpnt_vec)
-{
+                  std::vector<unsigned long long>& strpnt_vec) {
     std::vector<unsigned long long> ancestors;
-    
     unsigned long long cur_anct = Tree[cur_arc].anct;
 
     while (Tree[cur_anct].itmset > Tree[str_pnt].itmset) {
-        if (Tree[cur_anct].item > 0)
-            ancestors.push_back(cur_anct);
+        if (Tree[cur_anct].item > 0) ancestors.push_back(cur_anct);
         cur_anct = Tree[cur_anct].anct;
     }
 
     if (Tree[cur_anct].itmset == Tree[str_pnt].itmset)
         return true;
-    else {
-        for (auto it = ancestors.rbegin(); it != ancestors.rend(); ++it) {
-            for (unsigned long long i = start; i < strpnt_vec.size(); ++i) {
-                if (strpnt_vec[i] == *it)
-                    return true;
-            }
+
+    for (std::vector<unsigned long long>::reverse_iterator it = ancestors.rbegin();
+         it != ancestors.rend(); ++it) {
+        for (unsigned long long i = start; i < strpnt_vec.size(); ++i) {
+            if (strpnt_vec[i] == *it) return true;
         }
     }
-
     return false;
 }
 
-// return elapsed time in seconds
-double give_time(std::clock_t ticks) {
-    return static_cast<double>(ticks) / CLOCKS_PER_SEC;
-}
-
 } // namespace largebm

effspm/largebm/src/utility.hpp

@@ -1,18 +1,15 @@
 #pragma once
-
 #include <vector>
-#include <time.h>
-#include <string>
-#include "build_mdd.hpp"
 #include <ctime>
 
 namespace largebm {
-using namespace std;
-
-double give_time(std::clock_t kk);
-
 
+double give_time(std::clock_t ticks);
 
-bool check_parent(unsigned long long int cur_arc, unsigned long long int str_pnt, unsigned long long int start, vector<unsigned long long int>& strpnt_vec);
+// Check if `str_pnt` is an ancestor of `cur_arc` respecting itemset boundaries
+bool check_parent(unsigned long long cur_arc,
+                  unsigned long long str_pnt,
+                  unsigned long long start,
+                  std::vector<unsigned long long>& strpnt_vec);
 
-}
+} // namespace largebm