effspm 0.2.8__cp39-cp39-macosx_11_0_arm64.whl → 0.3.2__cp39-cp39-macosx_11_0_arm64.whl

effspm/btminer/src/load_inst.cpp

@@ -1,12 +1,11 @@
-
+// effspm/btminer/src/load_inst.cpp
 #include <iostream>
-#include <sstream>
 #include <fstream>
-#include <cmath>
-#include <ctime>
-#include <map>
-#include <vector>
+#include <sstream>
 #include <algorithm>
+#include <math.h>
+#include <time.h>
+
 #include "load_inst.hpp"
 #include "utility.hpp"
 #include "build_mdd.hpp"
@@ -16,184 +15,261 @@ namespace btminer {
 
 using namespace std;
 
-extern int num_nodes, cur_node;
-
-
-map<string, int> item_map;
-map<int, string> item_map_rev;
-vector<int> freq;
-vector<int> item_dic;
-
-void Load_items_pre(string& inst_name);
-bool Load_items(string& inst_name);
-bool Preprocess(string& inst, double thresh);
-
-
-
-
-bool Load_instance(string& items_file, double thresh) {
+// ---------------------------------------------------------------------
+// global definitions (must match load_inst.hpp)
+// ---------------------------------------------------------------------
+int  M = 0;
+int  N = 0;
+int  L = 0;
+unsigned long long E = 0ULL;    // matches header: extern unsigned long long E;
+int  num_nodes = 0;
+int  theta = 0;
+int  cur_node = 0;
+
+map<string, int>  item_map;
+map<int, string>  item_map_rev;
+
+std::vector<int>  freq;
+std::vector<int>  item_dic;
+
+// ✅ REAL DEFINITION lives here:
+std::vector<Pattern> DFS;
+
+string out_file, folder;
+bool   b_disp    = 0;
+bool   b_write   = 0;
+bool   use_dic   = 0;
+bool   just_build= 0;
+bool   pre_pro   = 1;
+
+int    N_mult    = 1;
+int    M_mult    = 1;
+int    time_limit= 30 * 3600;    // 30 hours, same as professor
+
+clock_t start_time;
+
+// ---------------------------------------------------------------------
+// forward decls
+// ---------------------------------------------------------------------
+void Load_items_pre(string &inst_name);
+bool Load_items(string &inst_name);
+bool Preprocess(string &inst, double thresh);
+
+// ---------------------------------------------------------------------
+// main loader
+// ---------------------------------------------------------------------
+bool Load_instance(string &items_file, double thresh) {
     clock_t kk = clock();
+
+    // root node for MDD
     Tree.emplace_back(0, 0, 0);
 
     if (pre_pro) {
         if (!Preprocess(items_file, thresh))
             return false;
+        if (b_disp)
+           cout << "\nPreprocess done in " << give_time(clock() - kk) << " seconds\n\n";
 
-        cout << "\nPreprocess done in " << give_time(clock() - kk) << " seconds\n\n";
-
+        // build empty DFS of size L
+        DFS.clear();
         DFS.reserve(L);
         for (int i = 0; i < L; ++i)
             DFS.emplace_back(-i - 1);
 
         kk = clock();
         Load_items_pre(items_file);
-    } else if (!Load_items(items_file))
+    }
+    else if (!Load_items(items_file)) {
         return false;
+    }
     else {
-        theta = (thresh < 1) ? ceil(thresh * N * N_mult) : thresh;
+        if (thresh < 1)
+            theta = static_cast<int>(ceil(thresh * N * N_mult));
+        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;
 }
 
-bool Preprocess(string& inst, double thresh) {
+// ---------------------------------------------------------------------
+// preprocessing pass
+// ---------------------------------------------------------------------
+bool Preprocess(string &inst, double thresh) {
     ifstream file(inst);
-    if (!file.good()) {
-        cout << "!!!!!! No such file exists: " << inst << " !!!!!!\n";
-        return false;
-    }
 
-    string line;
-    int size_m, ditem;
-    while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
-        ++N;
-        vector<bool> counted(L, 0);
-        istringstream word(line);
-        string itm;
-        while (word >> itm) {
-            ditem = stoi(itm);
-            if (L < abs(ditem)) L = abs(ditem);
-            while (freq.size() < L) {
-                freq.push_back(0);
-                counted.push_back(0);
-            }
-            if (!counted[abs(ditem) - 1]) {
-                ++freq[abs(ditem) - 1];
-                counted[abs(ditem) - 1] = 1;
+    if (file.good()) {
+        string line;
+        while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
+            ++N;
+            vector<bool> counted(L, false);
+
+            istringstream word(line);
+            string itm;
+            while (word >> itm) {
+                int ditem = stoi(itm);
+                if (L < abs(ditem))
+                    L = abs(ditem);
+
+                // extend freq / counted if L grew
+                while (static_cast<int>(freq.size()) < L) {
+                    freq.push_back(0);
+                    counted.push_back(false);
+                }
+
+                int idx = abs(ditem) - 1;
+                if (!counted[idx]) {
+                    ++freq[idx];
+                    counted[idx] = true;
+                }
             }
         }
+    } else {
+        cout << "!!!!!! No such file exists: " << inst << " !!!!!!\n";
+        return false;
     }
 
-    theta = (thresh < 1) ? ceil(thresh * N * N_mult) : thresh;
+    if (thresh < 1)
+        theta = static_cast<int>(ceil(thresh * N * N_mult));
+    else
+        theta = static_cast<int>(thresh);
 
+    // build item_dic with only frequent items
     int real_L = 0;
     item_dic = vector<int>(L, -1);
     for (int i = 0; i < L; ++i) {
         if (freq[i] >= theta)
             item_dic[i] = ++real_L;
     }
+    if (b_disp)
+      cout << "Original number of items: " << L
+         << " Reduced to: " << real_L << endl;
 
-    cout << "Original number of items: " << L << " Reduced to: " << real_L << endl;
     L = real_L;
-    N = 0;
+    N = 0;   // will be recounted in Load_items_pre
+
     return true;
 }
 
-void Load_items_pre(string& inst_name) {
+// ---------------------------------------------------------------------
+// load after preprocessing
+// ---------------------------------------------------------------------
+void Load_items_pre(string &inst_name) {
     ifstream file(inst_name);
-    if (!file.good()) return;
-
-    string line;
-    int ditem;
-    while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
-        istringstream word(line);
-        string itm;
-        vector<int> temp_vec;
-        bool sgn = 0;
-        while (word >> itm) {
-            if (use_dic) {
-                auto it = item_map.find(itm);
-                if (it == item_map.end()) {
-                    item_map[itm] = ++L;
-                    item_map_rev[L] = itm;
-                    ditem = L;
+
+    if (file.good()) {
+        string line;
+        while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
+            istringstream word(line);
+            string itm;
+            vector<int> temp_vec;
+            bool sgn = false;
+            while (word >> itm) {
+                int ditem;
+                if (use_dic) {
+                    auto it = item_map.find(itm);
+                    if (it == item_map.end()) {
+                        item_map[itm] = ++L;
+                        item_map_rev[L] = itm;
+                        ditem = L;
+                    } else {
+                        ditem = it->second;
+                    }
                 } else {
-                    ditem = it->second;
+                    ditem = stoi(itm);
                 }
-            } else {
-                ditem = stoi(itm);
-            }
 
-            if (pre_pro && freq.size() > abs(ditem) - 1 && freq[abs(ditem) - 1] < theta) {
-                if (!sgn)
-                    sgn = ditem < 0;
-                continue;
-            } else if (pre_pro) {
-                ditem = (ditem > 0) ? item_dic[ditem - 1] : -item_dic[-ditem - 1];
-            }
+                // drop infrequent items
+                if (freq[abs(ditem) - 1] < theta) {
+                    if (!sgn)
+                        sgn = (ditem < 0);
+                    continue;
+                } else {
+                    if (ditem > 0)
+                        ditem = item_dic[ditem - 1];
+                    else
+                        ditem = -item_dic[-ditem - 1];
+                }
 
-            if (sgn && ditem > 0)
-                ditem = -ditem;
-            sgn = 0;
+                if (sgn) {
+                    if (ditem > 0)
+                        ditem = -ditem;
+                    sgn = false;
+                }
 
-            temp_vec.push_back(ditem);
-        }
+                temp_vec.push_back(ditem);
+            }
 
-        if (temp_vec.empty()) continue;
+            if (temp_vec.empty())
+                continue;
+
+            ++N;
 
-        ++N;
-        if (temp_vec.size() > M) M = temp_vec.size();
+            if (static_cast<int>(temp_vec.size()) > M)
+                M = static_cast<int>(temp_vec.size());
 
-        E += temp_vec.size();  // <-- make sure E gets incremented
-        Build_MDD(temp_vec);
+            // this increments E inside Build_MDD
+            Build_MDD(temp_vec);
+        }
     }
 }
 
-bool Load_items(string& inst_name) {
+// ---------------------------------------------------------------------
+// load without preprocessing
+// ---------------------------------------------------------------------
+bool Load_items(string &inst_name) {
     ifstream file(inst_name);
-    if (!file.good()) {
-        cout << "!!!!!! No such file exists: " << inst_name << " !!!!!!\n";
-        return false;
-    }
 
-    string line;
-    int ditem;
-    while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
-        ++N;
-        istringstream word(line);
-        string itm;
-        vector<int> temp_vec;
-        while (word >> itm) {
-            if (use_dic) {
-                auto it = item_map.find(itm);
-                if (it == item_map.end()) {
-                    item_map[itm] = ++L;
-                    item_map_rev[L] = itm;
-                    ditem = L;
+    if (file.good()) {
+        string line;
+        while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
+            ++N;
+            istringstream word(line);
+            string itm;
+            vector<int> temp_vec;
+            while (word >> itm) {
+                int ditem;
+                if (use_dic) {
+                    auto it = item_map.find(itm);
+                    if (it == item_map.end()) {
+                        item_map[itm] = ++L;
+                        item_map_rev[L] = itm;
+                        ditem = L;
+                    } else {
+                        ditem = it->second;
+                    }
                 } else {
-                    ditem = it->second;
-                }
-            } else {
-                ditem = stoi(itm);
-                if (L < abs(ditem)) {
-                    L = abs(ditem);
-                    while (DFS.size() < L && !just_build) {
-                        DFS.reserve(L);
-                        DFS.emplace_back(-DFS.size() - 1);
+                    ditem = stoi(itm);
+                    if (L < abs(ditem)) {
+                        L = abs(ditem);
+                        // make sure DFS is large enough (unless just_build)
+                        while (static_cast<int>(DFS.size()) < L && !just_build) {
+                            DFS.reserve(L);
+                            DFS.emplace_back(-((int)DFS.size()) - 1);
+                        }
                     }
                 }
+
+                temp_vec.push_back(ditem);
             }
-            temp_vec.push_back(ditem);
-        }
 
-        if (temp_vec.size() > M) M = temp_vec.size();
-        E += temp_vec.size();  // <-- make sure E gets incremented
-        Build_MDD(temp_vec);
+            if (static_cast<int>(temp_vec.size()) > M)
+                M = static_cast<int>(temp_vec.size());
+
+            Build_MDD(temp_vec);
+        }
+    } else {
+        cout << "!!!!!! No such file exists: " << inst_name << " !!!!!!\n";
+        return false;
     }
+
     return true;
 }
 

effspm/btminer/src/load_inst.hpp

@@ -10,16 +10,34 @@
 
 namespace btminer {
 
-bool Load_instance(std::string& items_file, double thresh);
+using std::string;
+using std::vector;
+using std::map;
+using std::unordered_map;
+using std::unordered_set;
 
-extern std::string out_file, folder;
+bool Load_instance(string& items_file, double thresh);
+
+extern string out_file, folder;
 
 extern bool b_disp, b_write, use_dic, just_build, pre_pro;
 
-extern int N, M, L, theta, num_nodes, M_mult, N_mult, time_limit, cur_node;
+extern int  N, M, L, theta, num_nodes, M_mult, N_mult, time_limit, cur_node;
+extern unsigned long long E;              // total number of entries (we need this for _effspm.cpp)
+
+extern std::clock_t start_time;
+
+// these 2 are for dictionary mode
+extern map<string,int> item_map;
+extern map<int,string> item_map_rev;
 
-extern clock_t start_time;
+extern vector<int> freq;
+extern vector<int> item_dic;
 
+// expose items so _effspm.cpp can fall back to seeding (it expects btminer::items)
+extern vector<vector<int>> items;
 
+class Pattern;
+extern vector<Pattern> DFS;
 
 } // namespace btminer

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/btminer/src/utility.cpp

@@ -5,61 +5,46 @@
 
 namespace btminer {
 
-// ─── Global definitions ──────────────────────────────────────────
-bool use_dic   = false;
-std::vector<std::vector<int>> items;
-bool use_list  = false;
-bool just_build = false;
-int  E = 0, M = 0, N = 0, L = 0, theta = 0;
-std::vector<Pattern> DFS;
-clock_t start_time = 0;
-bool b_disp = false, b_write = false;
-std::string out_file;
-
-bool pre_pro   = true;
-int  N_mult    = 1, M_mult   = 1;
-int  time_limit = 30 * 3600;
-
-// buffer of mined patterns returned to Python
-std::vector<std::vector<int>> collected;
-
-void ClearCollected()                       { collected.clear(); }
-const std::vector<std::vector<int>>& GetCollected() { return collected; }
-
-// ─── Utility functions ───────────────────────────────────────────
-int find_ID(std::vector<int>& vec, int itm)
-{
+int find_ID(vector<int>& vec, int itm) {
     int plc = 0;
-    while (plc < static_cast<int>(vec.size()) && vec[plc] != itm) ++plc;
-    return (plc == static_cast<int>(vec.size())) ? -1 : plc;
+    while (plc < static_cast<int>(vec.size()) && vec[plc] != itm)
+        ++plc;
+
+    if (plc == static_cast<int>(vec.size()))
+        return -1;
+    else
+        return plc;
 }
 
-bool check_parent(int cur_arc, int str_pnt, int start,
-                  std::vector<int>& strpnt_vec)
-{
+bool check_parent(int cur_arc, int str_pnt, int start, std::vector<int>& strpnt_vec) {
+
     std::vector<int> ancestors;
+
     int 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;
 
-    for (auto it = ancestors.rbegin(); it != ancestors.rend(); ++it)
-        for (int i = start; i < static_cast<int>(strpnt_vec.size()); ++i)
-            if (strpnt_vec[i] == *it) return true;
+    if (Tree[cur_anct].itmset == Tree[str_pnt].itmset)
+        return true;
+    else {
+        for (auto it = ancestors.rbegin(); it != ancestors.rend(); ++it) {
+            for (int i = start; i < static_cast<int>(strpnt_vec.size()); ++i) {
+                if (strpnt_vec[i] == *it)
+                    return true;
+            }
+        }
+    }
 
     return false;
 }
 
-bool find_pnt(Arc* pnt, std::vector<Arc*>& vec, int pos)
-{
-    for (size_t i = pos; i < vec.size(); ++i)
-        if (vec[i] == pnt) return true;
-    return false;
+float give_time(clock_t kk) {
+    float ll = static_cast<float>(kk) / CLOCKS_PER_SEC;
+    return ll;
 }
 
-double give_time(clock_t kk) { return double(kk) / CLOCKS_PER_SEC; }
-
 } // namespace btminer

effspm/btminer/src/utility.hpp

@@ -1,40 +1,16 @@
 #pragma once
 
 #include <vector>
-#include <ctime>
+#include <time.h>
 #include <string>
 #include "build_mdd.hpp"
-#include "freq_miner.hpp"
-#include "load_inst.hpp"
 
 namespace btminer {
 
-// === Utility function declarations ===
-bool find_pnt(Arc* pnt, std::vector<Arc*>& vec, int pos);
-int find_ID(std::vector<int>& vec, int itm);
-double give_time(clock_t kk);
-bool check_parent(int cur_arc, int str_pnt, int start, std::vector<int>& strpnt_vec);
+using std::vector;
 
-// === Global variables (DECLARATIONS ONLY) ===
+int   find_ID(vector<int>& vec, int itm);
+float give_time(clock_t kk);
+bool  check_parent(int cur_arc, int str_pnt, int start, vector<int>& strpnt_vec);
 
-extern bool use_list;
-extern bool just_build;
-extern int E, M, N, L, theta;
-extern std::vector<Pattern> DFS;
-extern clock_t start_time;
-extern bool b_disp, b_write;
-extern std::string out_file;
-extern bool pre_pro;
-extern int N_mult, M_mult;
-extern int time_limit;
-extern std::vector<std::vector<int>> items;
-
- extern std::vector<std::vector<int>> collected;
-
-void ClearCollected();
-const std::vector<std::vector<int>>& GetCollected();
-
-}
-
-
-// namespace btminer
+} // namespace btminer

effspm/freq_miner.hpp

@@ -24,7 +24,8 @@ public:
 
 	Pattern(vector<int>& _seq, int item) {
 		seq.reserve(_seq.size());
-		for (std::size_t i = 0; i < _seq.size(); ++i)
+		for (int i = 0; i < _seq.size(); ++i)
+
 
 			seq.push_back(_seq[i]);
 		seq.push_back(item);