effspm 0.2.6__cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl

effspm/freq_miner.cpp

@@ -0,0 +1,143 @@
+#include <iostream>
+#include <time.h>
+#include <fstream>
+#include <cmath>
+#include "freq_miner.hpp"
+#include "utility.hpp"
+
+
+
+// Forward declarations from the original code:
+void Extend_patt(Pattern& _patt);
+
+// Globals from original:
+unsigned long long int num_patt = 0;
+Pattern _patt;
+
+// Main miner function from original:
+void Freq_miner() {
+    vector<int> islist;
+    if (use_list) {
+        for (int i = 0; i < L; ++i) {
+            if (DFS[i].freq >= theta)
+                islist.push_back(i);
+        }
+        for (int i = 0; i < DFS.size(); ++i) {
+            DFS[i].ilist = islist;
+            DFS[i].slist = islist;
+        }
+    }
+
+    while (!DFS.empty() && give_time(clock() - start_time) < time_limit) {
+        if (DFS.back().freq >= theta)
+            Extend_patt(DFS.back());
+        else
+            DFS.pop_back();
+    }
+}
+
+// The recursive extension from original:
+void Extend_patt(Pattern& _pattern) {
+    swap(_patt, _pattern);
+    DFS.pop_back();
+
+    vector<bool> slist;
+    vector<bool> ilist;
+
+    if (use_list) {
+        slist = vector<bool>(L, 0);
+        ilist = vector<bool>(L, 0);
+        for (int idx : _patt.slist) slist[idx] = 1;
+        for (int idx : _patt.ilist) ilist[idx] = 1;
+    }
+
+    vector<Pattern> pot_patt(L * 2);
+
+    int last_neg = _patt.seq.size() - 1;
+    while (_patt.seq[last_neg] > 0) --last_neg;
+
+    for (int i = 0; i < _patt.str_pnt.size(); ++i) {
+        vector<bool> found(L * 2, 0);
+        unsigned int seq = _patt.seq_ID[i];
+        unsigned int j = _patt.str_pnt[i] + 1;
+        // positive extensions
+        while (j < items[seq].size() && items[seq][j] > 0) {
+            int cur_itm = items[seq][j];
+            if (!use_list || ilist[cur_itm - 1]) {
+                pot_patt[cur_itm - 1].seq_ID.push_back(seq);
+                pot_patt[cur_itm - 1].str_pnt.push_back(j);
+                ++pot_patt[cur_itm - 1].freq;
+                found[cur_itm - 1] = 1;
+            }
+            ++j;
+        }
+        // negative and cross-itemset extensions...
+        int num_itmfnd = 0;
+        for (int k = j; k < items[seq].size(); ++k) {
+            int cur_itm = abs(items[seq][k]);
+            if (items[seq][k] < 0) num_itmfnd = 0;
+            if ((!use_list || slist[cur_itm - 1]) && !found[L + cur_itm - 1]) {
+                pot_patt[L + cur_itm - 1].seq_ID.push_back(seq);
+                pot_patt[L + cur_itm - 1].str_pnt.push_back(k);
+                ++pot_patt[L + cur_itm - 1].freq;
+                found[L + cur_itm - 1] = 1;
+            }
+            if (num_itmfnd == _patt.seq.size() - last_neg) {
+                if ((!use_list || ilist[cur_itm - 1]) && !found[cur_itm - 1]) {
+                    pot_patt[cur_itm - 1].seq_ID.push_back(seq);
+                    pot_patt[cur_itm - 1].str_pnt.push_back(k);
+                    ++pot_patt[cur_itm - 1].freq;
+                    found[cur_itm - 1] = 1;
+                }
+            } else if (cur_itm == abs(_patt.seq[last_neg + num_itmfnd])) {
+                ++num_itmfnd;
+            }
+        }
+    }
+
+    // Now generate new DFS states
+    if (use_list) {
+        // itemset extensions
+        vector<int> slistp, ilistp;
+        for (int idx : _patt.ilist)
+            if (pot_patt[idx].freq >= theta) ilistp.push_back(idx);
+        for (int idx : _patt.slist)
+            if (pot_patt[idx + L].freq >= theta) slistp.push_back(idx);
+
+        for (int idx : ilistp) {
+            DFS.emplace_back();
+            swap(DFS.back(), pot_patt[idx]);
+            DFS.back().seq = _patt.seq;
+            DFS.back().seq.push_back(idx + 1);
+            DFS.back().slist = slistp;
+            DFS.back().ilist = ilistp;
+            Out_patt(DFS.back().seq, DFS.back().freq);
+            ++num_patt;
+        }
+        for (int idx : slistp) {
+            DFS.emplace_back();
+            swap(DFS.back(), pot_patt[idx + L]);
+            DFS.back().seq = _patt.seq;
+            DFS.back().seq.push_back(-idx - 1);
+            DFS.back().slist = slistp;
+            DFS.back().ilist = slistp;
+            Out_patt(DFS.back().seq, DFS.back().freq);
+            ++num_patt;
+        }
+    } else {
+        // no list optimization
+        for (int i = 0; i < 2 * L; ++i) {
+            if (pot_patt[i].freq >= theta) {
+                DFS.emplace_back();
+                swap(DFS.back(), pot_patt[i]);
+                DFS.back().seq = _patt.seq;
+                if (i >= L)
+                    DFS.back().seq.push_back(-(i - L + 1));
+                else
+                    DFS.back().seq.push_back(i + 1);
+                Out_patt(DFS.back().seq, DFS.back().freq);
+                ++num_patt;
+            }
+        }
+    }
+}

effspm/freq_miner.hpp

@@ -0,0 +1,48 @@
+#pragma once
+#include <vector>
+#include "load_inst.hpp"
+#include <cstdlib> 
+#include <cmath> 
+#include <cstddef>  // for std::size_t
+
+using namespace std;
+void Freq_miner();
+void Out_patt(std::vector<int>& seq, unsigned int freq);
+
+
+class Pattern {
+public:
+
+	vector<int> seq;
+	vector<unsigned int> str_pnt;
+	vector<unsigned int> seq_ID;
+
+	vector<int> slist;
+	vector<int> ilist;
+
+	unsigned int freq;
+
+	Pattern(vector<int>& _seq, int item) {
+		seq.reserve(_seq.size());
+		for (std::size_t i = 0; i < _seq.size(); ++i)
+
+			seq.push_back(_seq[i]);
+		seq.push_back(item);
+		freq = 0;
+	}
+
+
+	Pattern(int item) {
+		seq.push_back(item);
+		freq = 0;
+	}
+
+	Pattern() {
+		freq = 0;
+	}
+
+};
+
+extern vector<Pattern> DFS;				//DFS queue of potential patterns to extend
+
+extern unsigned long long int num_patt;

effspm/htminer/src/build_mdd.cpp

@@ -0,0 +1,192 @@
+#include <vector>
+#include <iostream>
+#include "load_inst.hpp"
+#include "build_mdd.hpp"
+#include "freq_miner.hpp"
+#include "utility.hpp"
+
+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);
+
+// Global trees (unchanged)
+std::vector<Arc> Tree;
+std::vector<VArc> VTree;
+std::vector<CArc> CTree;
+
+void Build_MDD(std::vector<int>& items, std::vector<int>& items_lim) {
+    // DEBUG: entry into Build_MDD
+    // std::cerr << "[HTMiner::Build_MDD] called with items.size()=" << items.size()
+    //           << "  items_lim.size()=" << items_lim.size() << std::endl;
+
+    // // Prepare ancestor map of size L
+    std::vector<unsigned int> ancest_map(L, 0);    
+
+    unsigned int last_arc = 0;
+    int itmset = 0;
+
+    // Iterate over items
+    for (size_t idx = 0; idx < items.size(); ++idx) {
+        int curr_item = items[idx];
+        // std::cerr << "[HTMiner::Build_MDD] processing items[" << idx 
+        //           << "]=" << curr_item << "  last_arc=" << last_arc 
+        //           << "  itmset=" << itmset << std::endl;
+
+        last_arc = Add_arc(curr_item, last_arc, itmset, ancest_map);
+
+        // std::cerr << "[HTMiner::Build_MDD] returned from Add_arc, new last_arc=" 
+        //           << last_arc << "  itmset=" << itmset << std::endl;
+    }
+
+    // If there are limited items, handle them
+    if (!items_lim.empty()) {
+        // std::cerr << "[HTMiner::Build_MDD] items_lim is not empty; size=" 
+        //           << items_lim.size() << std::endl;
+        Add_vec(items_lim, ancest_map, last_arc, itmset);
+        // std::cerr << "[HTMiner::Build_MDD] returned from Add_vec" << std::endl;
+    } else {
+        // std::cerr << "[HTMiner::Build_MDD] items_lim is empty; skipping Add_vec" << std::endl;
+    }
+
+    // DEBUG: exit Build_MDD
+//     std::cerr << "[HTMiner::Build_MDD] exiting; Tree.size()=" << Tree.size() 
+//               << "  CTree.size()=" << CTree.size() 
+//               << "  VTree.size()=" << VTree.size() << std::endl;
+// 
+}
+
+int Add_arc(int item, unsigned int last_arc, int& itmset, std::vector<unsigned int>& ancest_map) {
+    unsigned int anct = ancest_map[std::abs(item) - 1];
+    if (item < 0) {
+        ++itmset;
+        // std::cerr << "[HTMiner::Add_arc] negative item detected; itmset incremented to " 
+        //           << itmset << std::endl;
+    }
+
+    unsigned int last_sibl = Tree[last_arc].chld;
+    // std::cerr << "[HTMiner::Add_arc] starting with last_sibl=" << last_sibl 
+    //           << "  anct=" << anct << std::endl;
+
+    if (last_sibl == 0) {
+        Tree.emplace_back(item, itmset, anct);
+        last_sibl = static_cast<unsigned int>(Tree.size() - 1);
+        Tree[last_arc].chld = last_sibl;
+        // std::cerr << "[HTMiner::Add_arc] created new arc at index=" << last_sibl 
+        //           << "  setting Tree[" << last_arc << "].chld=" << last_sibl << std::endl;
+        if (anct == 0) {
+            DFS[std::abs(item) - 1].str_pnt.push_back(last_sibl);
+            // std::cerr << "[HTMiner::Add_arc] appended to DFS[" << (std::abs(item) - 1) 
+            //           << "].str_pnt -> " << last_sibl << std::endl;
+        }
+    }
+    else {
+        // std::cerr << "[HTMiner::Add_arc] traversing siblings starting at " << last_sibl << std::endl;
+        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);
+                // std::cerr << "[HTMiner::Add_arc] created sibling arc at index=" << last_sibl 
+                //           << "  setting Tree[" << (last_sibl - 1) << "].sibl=" << last_sibl << std::endl;
+                if (anct == 0) {
+                    DFS[std::abs(item) - 1].str_pnt.push_back(last_sibl);
+                    // std::cerr << "[HTMiner::Add_arc] appended to DFS[" << (std::abs(item) - 1) 
+                    //           << "].str_pnt -> " << last_sibl << std::endl;
+                }
+                break;
+            }
+            last_sibl = Tree[last_sibl].sibl;
+            // std::cerr << "[HTMiner::Add_arc] moving to next sibling: " << last_sibl << std::endl;
+        }
+    }
+
+    if (anct == 0) {
+        ++DFS[std::abs(item) - 1].freq;
+        // std::cerr << "[HTMiner::Add_arc] incremented DFS[" << (std::abs(item) - 1) 
+        //           << "].freq -> " << DFS[std::abs(item) - 1].freq << std::endl;
+    }
+
+    ++Tree[last_sibl].freq;
+    // std::cerr << "[HTMiner::Add_arc] incremented Tree[" << last_sibl << "].freq -> " 
+            //  << Tree[last_sibl].freq << std::endl;
+
+    ancest_map[std::abs(item) - 1] = last_sibl;
+    // std::cerr << "[HTMiner::Add_arc] updated ancest_map[" << (std::abs(item) - 1) 
+    //           << "] -> " << last_sibl << std::endl;
+
+    return static_cast<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();
+    // std::cerr << "[HTMiner::Add_vec] called with items_lim.size()=" << items_lim.size()
+    //           << "  last_arc=" << last_arc << "  itmset=" << itmset << std::endl;
+
+    std::vector<bool> counted(L, false);
+
+    if (Tree[last_arc].itmset > 0) {
+        ancest.push_back(0);
+        ancest.shrink_to_fit();
+        // std::cerr << "[HTMiner::Add_vec] Tree[" << last_arc << "].itmset > 0; pushing 0 to ancest" << std::endl;
+
+        for (size_t i = 0; i < items_lim.size(); ++i) {
+            int cur_itm = std::abs(items_lim[i]);
+            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()));
+                    // std::cerr << "[HTMiner::Add_vec] appended negative str_pnt to VDFS[" 
+                    //           << (cur_itm - 1) << "] -> " << (-static_cast<int>(i) - 1) << std::endl;
+                }
+                ++DFS[cur_itm - 1].freq;
+                counted[cur_itm - 1] = true;
+                // std::cerr << "[HTMiner::Add_vec] incremented DFS[" << (cur_itm - 1) 
+                //           << "].freq -> " << DFS[cur_itm - 1].freq << std::endl;
+            }
+        }
+
+        CTree.emplace_back(ancest, items_lim);
+        //std::cerr << "[HTMiner::Add_vec] added new CTree node; CTree.size()=" << CTree.size() << std::endl;
+
+        Tree[last_arc].chld = static_cast<unsigned int>(CTree.size() - 1);
+        Tree[last_arc].itmset = -itmset;
+    //     std::cerr << "[HTMiner::Add_vec] updated Tree[" << last_arc 
+    //               << "].chld=" << Tree[last_arc].chld 
+    //               << "  Tree[" << last_arc << "].itmset=" << Tree[last_arc].itmset << std::endl;
+    //
+     }
+    else {
+        std::vector<unsigned int>& ancest_ct = CTree[Tree[last_arc].chld].ancest;
+        // std::cerr << "[HTMiner::Add_vec] Tree[" << last_arc << "].itmset <= 0; using existing CTree node " 
+        //           << Tree[last_arc].chld << std::endl;
+
+        for (size_t i = 0; i < items_lim.size(); ++i) {
+            int cur_itm = std::abs(items_lim[i]);
+            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()));
+                    // std::cerr << "[HTMiner::Add_vec] appended positive str_pnt to VDFS[" 
+                    //           << (cur_itm - 1) << "] -> " << (static_cast<unsigned int>(i) + 1) << std::endl;
+                }
+                ++DFS[cur_itm - 1].freq;
+                counted[cur_itm - 1] = true;
+                // std::cerr << "[HTMiner::Add_vec] incremented DFS[" << (cur_itm - 1) 
+                //           << "].freq -> " << DFS[cur_itm - 1].freq << std::endl;
+            }
+        }
+
+        VTree.emplace_back(items_lim, ancest_ct.back());
+        // std::cerr << "[HTMiner::Add_vec] added new VTree node; VTree.size()=" << VTree.size() << std::endl;
+
+        CTree[Tree[last_arc].chld].ancest.back() = static_cast<unsigned int>(VTree.size());
+        // std::cerr << "[HTMiner::Add_vec] updated CTree[" << Tree[last_arc].chld 
+        //           << "].ancest.back()=" << CTree[Tree[last_arc].chld].ancest.back() << std::endl;
+    }
+
+    //std::cerr << "[HTMiner::Add_vec] exiting" << std::endl;
+}
+
+} // namespace htminer 

effspm/htminer/src/build_mdd.hpp

@@ -0,0 +1,64 @@
+#pragma once
+
+#include <vector>
+#include <cmath>
+#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;
+
+        Arc(unsigned int _itm, int _itmset, unsigned int _anc) {
+            chld = 0;    
+            sibl = 0;
+            freq = 0;
+            itmset = _itmset;
+            anct = _anc;
+            item = _itm;
+        }
+
+        Arc() {
+            chld = 0;    
+            sibl = 0;
+            freq = 0;
+        }
+    };
+
+    class VArc {
+    public:
+        unsigned int sibl;    
+        std::vector<int> seq;
+
+        VArc(std::vector<int>& items, unsigned int _sib) {
+            sibl = _sib;
+            seq.swap(items);
+        }
+        
+        VArc() {
+            sibl = 0;
+        }
+    };
+
+    class CArc {
+    public:
+        std::vector<int> seq;
+        std::vector<unsigned int> ancest;
+
+        CArc(std::vector<unsigned int>& _anc, std::vector<int>& items) {
+            ancest.swap(_anc);
+            seq.swap(items);
+        }
+    };
+
+    extern std::vector<Arc> Tree;
+    extern std::vector<VArc> VTree;
+    extern std::vector<CArc> CTree;
+}