effspm 0.3.0__cp310-cp310-macosx_11_0_arm64.whl → 0.3.3__cp310-cp310-macosx_11_0_arm64.whl

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;
 
@@ -67,8 +67,8 @@ bool Load_instance(string &items_file, double thresh) {
     if (pre_pro) {
         if (!Preprocess(items_file, thresh))
             return false;
-
-        cout << "\nPreprocess done in " << give_time(clock() - kk) << " seconds\n\n";
+        if (b_disp)
+           cout << "\nPreprocess done in " << give_time(clock() - kk) << " seconds\n\n";
 
         // build empty DFS of size L
         DFS.clear();
@@ -88,12 +88,13 @@ bool Load_instance(string &items_file, double thresh) {
         else
             theta = static_cast<int>(thresh);
     }
-
-    cout << "\nMDD Database built in " << give_time(clock() - kk) << " seconds\n\n";
-    cout << "Found " << N * N_mult
-         << " sequence, with max line len " << M
-         << ", and " << L << " items, and " << E << " enteries\n";
-    cout << "Total MDD nodes: " << Tree.size() << endl;
+    if (b_disp)
+         cout << "\nMDD Database built in " << give_time(clock() - kk) << " seconds\n\n";
+    if (b_disp)
+        cout << "Found " << N * N_mult
+             << " sequence, with max line len " << M
+             << ", and " << L << " items, and " << E << " enteries\n";
+    //cout << "Total MDD nodes: " << Tree.size() << endl;
 
     return true;
 }
@@ -102,6 +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()) {
@@ -147,8 +157,8 @@ bool Preprocess(string &inst, double thresh) {
         if (freq[i] >= theta)
             item_dic[i] = ++real_L;
     }
-
-    cout << "Original number of items: " << L
+    if (b_disp)
+      cout << "Original number of items: " << L
          << " Reduced to: " << real_L << endl;
 
     L = real_L;

effspm/btminer/src/main.cpp

@@ -0,0 +1,83 @@
+#include <iostream>
+#include <string.h>
+#include "load_inst.hpp"
+#include "freq_miner.hpp"
+#include "utility.hpp"
+#include "build_mdd.hpp"
+
+namespace btminer {
+    // everything is already declared
+}
+
+int main(int argc, char* argv[]) {
+    using namespace btminer;
+
+    std::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 = std::stod(argv[i + 1]);
+        else if (strcmp(argv[i], "-file") == 0)
+            VV = argv[i + 1];
+        else if (strcmp(argv[i], "-N_mult") == 0)
+            N_mult = std::stoi(argv[i + 1]);
+        else if (strcmp(argv[i], "-M_mult") == 0)
+            M_mult = std::stoi(argv[i + 1]);
+        else if (strcmp(argv[i], "-time") == 0)
+            time_limit = std::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], "-dic") == 0)
+            use_dic = 1;
+        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
+            std::cout << "Command " << argv[i] << " not recognized and skipped.\n";
+    }
+
+    std::cout << "\n********************** " << VV << " N_mult: " << N_mult << " M_mult: " << M_mult << "**********************\n";
+
+    std::string item_file = folder + VV + ".txt";
+
+    std::cout << "loading instances...\n";
+
+    start_time = clock();
+
+    if (!Load_instance(item_file, thresh)) {
+        std::cout << "Files invalid, exiting.\n";
+        return 0;
+    }
+
+    if (!just_build && give_time(clock() - start_time) < time_limit) {
+        Freq_miner();
+        if (give_time(clock() - start_time) >= time_limit)
+            std::cout << "TIME LIMIT REACHED\n";
+        std::cout << "Mining Complete\n\nFound a total of " << num_patt << " patterns\n";
+        std::cout << "\nTotal CPU time " << give_time(clock() - start_time) << " seconds\n\n";
+    }
+
+    return 0;
+}

effspm/htminer/src/build_mdd.cpp

@@ -7,43 +7,32 @@
 
 namespace htminer {
 
-// Forward declarations (unchanged)
-int Add_arc(int item, unsigned int last_arc, int& itmset, std::vector<unsigned int>& ancest_map);
-void Add_vec(std::vector<int>& items_lim, std::vector<unsigned int>& ancest_map, unsigned int last_arc, int itmset);
+using std::vector;
 
-// Global trees (unchanged)
-std::vector<Arc>  Tree;
-std::vector<VArc> VTree;
-std::vector<CArc> CTree;
+int Add_arc(int item, unsigned int last_arc, int& itmset, vector<unsigned int>& ancest_map);
+void Add_vec(vector<int>& items_lim, vector<unsigned int>& ancest_map, unsigned int last_arc, int itmset);
 
-void Build_MDD(std::vector<int>& items, std::vector<int>& items_lim) {
-    // Prepare ancestor map of size L
-    std::vector<unsigned int> ancest_map(L, 0);
+vector<Arc> Tree;
+vector<VArc> VTree;
+vector<CArc> CTree;
 
-    unsigned int last_arc = 0;
-    int          itmset   = 0;
-
-    // 1) normal items
-    for (size_t idx = 0; idx < items.size(); ++idx) {
-        int curr_item = items[idx];
+void Build_MDD(vector<int>& items, vector<int>& items_lim) {
 
-        ++E;   // ✅ count this entry, just like in btminer
+    vector<unsigned int> ancest_map(L, 0);
 
-        last_arc = Add_arc(curr_item, last_arc, itmset, ancest_map);
-    }
+    unsigned int last_arc = 0;
+    int itmset = 0;
+    for (vector<int>::iterator it = items.begin(); it != items.end(); ++it)
+        last_arc = Add_arc(*it, last_arc, itmset, ancest_map);
 
-    // 2) tail / limited items
-    if (!items_lim.empty()) {
+    if (!items_lim.empty())
         Add_vec(items_lim, ancest_map, last_arc, itmset);
-    }
 }
 
-int Add_arc(int item,
-            unsigned int last_arc,
-            int& itmset,
-            std::vector<unsigned int>& ancest_map)
-{
+int Add_arc(int item, unsigned int last_arc, int& itmset, vector<unsigned int>& ancest_map) {
+
     unsigned int anct = ancest_map[std::abs(item) - 1];
+
     if (item < 0)
         ++itmset;
 
@@ -51,9 +40,8 @@ int Add_arc(int item,
 
     if (last_sibl == 0) {
         Tree.emplace_back(item, itmset, anct);
-        last_sibl = static_cast<unsigned int>(Tree.size() - 1);
+        last_sibl = (unsigned int)Tree.size() - 1;
         Tree[last_arc].chld = last_sibl;
-
         if (anct == 0)
             DFS[std::abs(item) - 1].str_pnt.push_back(last_sibl);
     }
@@ -61,8 +49,8 @@ int Add_arc(int item,
         while (Tree[last_sibl].item != item) {
             if (Tree[last_sibl].sibl == 0) {
                 Tree.emplace_back(item, itmset, anct);
-                Tree[last_sibl].sibl = static_cast<unsigned int>(Tree.size() - 1);
-                last_sibl            = static_cast<unsigned int>(Tree.size() - 1);
+                Tree[last_sibl].sibl = (unsigned int)Tree.size() - 1;
+                last_sibl = (unsigned int)Tree.size() - 1;
                 if (anct == 0)
                     DFS[std::abs(item) - 1].str_pnt.push_back(last_sibl);
                 break;
@@ -78,61 +66,48 @@ int Add_arc(int item,
 
     ancest_map[std::abs(item) - 1] = last_sibl;
 
-    return static_cast<int>(last_sibl);
+    return (int)last_sibl;
 }
 
-void Add_vec(std::vector<int>& items_lim,
-             std::vector<unsigned int>& ancest,
-             unsigned int last_arc,
-             int itmset)
-{
-    items_lim.shrink_to_fit();
+void Add_vec(vector<int>& items_lim, vector<unsigned int>& ancest, unsigned int last_arc, int itmset) {
 
-    std::vector<bool> counted(L, false);
+    items_lim.shrink_to_fit();
+    vector<bool> counted(L, 0);
 
     if (Tree[last_arc].itmset > 0) {
-        ancest.push_back(0);
+        ancest.push_back(0);  // last element of ancest is CArc child
         ancest.shrink_to_fit();
-
-        for (size_t i = 0; i < items_lim.size(); ++i) {
+        for (int i = 0; i < (int)items_lim.size(); ++i) {
             int cur_itm = std::abs(items_lim[i]);
-
-            ++E;   // ✅ count this limited-entry too
-
             if (ancest[cur_itm - 1] == 0 && !counted[cur_itm - 1]) {
-                if (i + 1 < static_cast<int>(items_lim.size())) {
-                    VDFS[cur_itm - 1].str_pnt.push_back(-static_cast<int>(i) - 1);
-                    VDFS[cur_itm - 1].seq_ID.push_back(static_cast<unsigned int>(CTree.size()));
+                if (i + 1 < (int)items_lim.size()) {
+                    VDFS[cur_itm - 1].str_pnt.push_back(-i - 1);  // CTree positions: negative pointers
+                    VDFS[cur_itm - 1].seq_ID.push_back((unsigned int)CTree.size());
                 }
                 ++DFS[cur_itm - 1].freq;
-                counted[cur_itm - 1] = true;
+                counted[cur_itm - 1] = 1;
             }
         }
-
         CTree.emplace_back(ancest, items_lim);
-        Tree[last_arc].chld   = static_cast<unsigned int>(CTree.size() - 1);
-        Tree[last_arc].itmset = -itmset;
+        Tree[last_arc].chld = (unsigned int)CTree.size() - 1;
+        Tree[last_arc].itmset = -itmset;  // Tree→CTree edge is marked by negative itmset
     }
     else {
-        std::vector<unsigned int>& ancest_ct = CTree[Tree[last_arc].chld].ancest;
-
-        for (size_t i = 0; i < items_lim.size(); ++i) {
+        vector<unsigned int>& ancest_ref = CTree[Tree[last_arc].chld].ancest;
+        for (int i = 0; i < (int)items_lim.size(); ++i) {
             int cur_itm = std::abs(items_lim[i]);
-
-            ++E;   // ✅ also count in this branch
-
-            if (!counted[cur_itm - 1] && ancest_ct[cur_itm - 1] == 0) {
-                if (i + 1 < static_cast<int>(items_lim.size())) {
-                    VDFS[cur_itm - 1].str_pnt.push_back(static_cast<unsigned int>(i) + 1);
-                    VDFS[cur_itm - 1].seq_ID.push_back(static_cast<unsigned int>(VTree.size()));
+            if (!counted[cur_itm - 1] && ancest_ref[cur_itm - 1] == 0) {
+                if (i + 1 < (int)items_lim.size()) {
+                    VDFS[cur_itm - 1].str_pnt.push_back(i + 1);
+                    VDFS[cur_itm - 1].seq_ID.push_back((unsigned int)VTree.size());
                 }
                 ++DFS[cur_itm - 1].freq;
-                counted[cur_itm - 1] = true;
+                counted[cur_itm - 1] = 1;
             }
         }
-
-        VTree.emplace_back(items_lim, ancest_ct.back());
-        CTree[Tree[last_arc].chld].ancest.back() = static_cast<unsigned int>(VTree.size());
+        VTree.emplace_back(items_lim, CTree[Tree[last_arc].chld].ancest.back());
+        CTree[Tree[last_arc].chld].ancest.back() = (unsigned int)VTree.size();
+        // VTree siblings and CTree children are +1 of actual index to mark end
     }
 }
 

effspm/htminer/src/build_mdd.hpp

@@ -5,60 +5,67 @@
 #include "load_inst.hpp"
 
 namespace htminer {
-    void Build_MDD(std::vector<int>& items, std::vector<int>& items_lim);
 
-    class Arc {
-    public:
-        unsigned int chld;    
-        unsigned int sibl;
-        unsigned int freq;        
-        unsigned int anct;
-        int itmset;
-        int item;
+using std::vector;
 
-        Arc(unsigned int _itm, int _itmset, unsigned int _anc) {
-            chld = 0;    
-            sibl = 0;
-            freq = 0;
-            itmset = _itmset;
-            anct = _anc;
-            item = _itm;
-        }
+void Build_MDD(vector<int>& items, vector<int>& items_lim);
 
-        Arc() {
-            chld = 0;    
-            sibl = 0;
-            freq = 0;
-        }
-    };
+class Arc {
+public:
+    unsigned int chld;
+    unsigned int sibl;
+    unsigned int freq;
+    unsigned int anct;
+    int itmset;
+    int item;
 
-    class VArc {
-    public:
-        unsigned int sibl;    
-        std::vector<int> seq;
+    Arc(unsigned int _itm, int _itmset, unsigned int _anc) {
+        chld = 0;
+        sibl = 0;
+        freq = 0;
+        itmset = _itmset;
+        anct = _anc;
+        item = _itm;
+    }
 
-        VArc(std::vector<int>& items, unsigned int _sib) {
-            sibl = _sib;
-            seq.swap(items);
-        }
-        
-        VArc() {
-            sibl = 0;
-        }
-    };
+    Arc() {
+        chld = 0;
+        sibl = 0;
+        freq = 0;
+        itmset = 0;
+        anct = 0;
+        item = 0;
+    }
+};
 
-    class CArc {
-    public:
-        std::vector<int> seq;
-        std::vector<unsigned int> ancest;
+class VArc {
+public:
+    unsigned int sibl;
+    vector<int> seq;
 
-        CArc(std::vector<unsigned int>& _anc, std::vector<int>& items) {
-            ancest.swap(_anc);
-            seq.swap(items);
-        }
-    };
+    VArc(vector<int>& items, unsigned int _sib) {
+        sibl = _sib;
+        seq.swap(items);
+    }
 
-    extern std::vector<Arc> Tree;
-    extern std::vector<VArc> VTree;
-    extern std::vector<CArc> CTree;
-}
+    VArc() {
+        sibl = 0;
+    }
+};
+
+class CArc {
+public:
+    vector<int> seq;
+    vector<unsigned int> ancest;
+
+    CArc(vector<unsigned int>& _anc, vector<int>& items) {
+        ancest.swap(_anc);
+        seq.swap(items);
+    }
+};
+
+extern vector<Arc> Tree;
+extern vector<VArc> VTree;
+extern vector<CArc> CTree;
+
+} // namespace htminer