effspm 0.1.11__cp313-cp313-macosx_10_13_universal2.whl → 0.2.1__cp313-cp313-macosx_10_13_universal2.whl

effspm/htminer/src/utility.hpp

@@ -0,0 +1,77 @@
+#pragma once
+
+#include <vector>
+#include <ctime>
+#include <string>
+#include "build_mdd.hpp"
+#include "freq_miner.hpp"
+#include "load_inst.hpp"
+
+namespace htminer {
+
+// ─── Global flags and counters ─────────────────────────────────────────────
+/// Controls whether to mine in “list” mode (unused for HTMiner, but declared)
+extern bool use_list;
+/// If true, only build MDD and exit (don’t actually mine)
+extern bool just_build;
+/// If true, print each pattern to stdout as it’s found
+extern bool b_disp;
+/// If true, write each pattern to file (see out_file)
+extern bool b_write;
+/// If true, use a dictionary‐file mapping items → new IDs
+extern bool use_dic;
+/// If true, preprocess input (create dictionary) instead of mining
+extern bool pre_pro;
+
+/// Time limit (in seconds) for mining before forced exit
+extern unsigned int time_limit;
+/// Output filename (if b_write is true)
+extern std::string out_file;
+/// Clock tick when mining started
+extern std::clock_t start_time;
+
+// ─── Data‐set‐level globals ─────────────────────────────────────────────────
+/// The input sequences (each sequence is a vector of integers)
+extern std::vector<std::vector<int>> items;
+/// Number of sequences (items.size())
+extern unsigned long long N;
+/// Number of distinct items (max absolute item ID)
+extern unsigned long long L;
+/// Minimum support threshold (absolute count, not fraction)
+extern unsigned long long theta;
+/// Maximum sequence length across all items
+extern unsigned int M;
+/// Total number of “entries” (sum of all sequence lengths)
+extern unsigned long long E;
+
+// ─── Per‐pattern DFS stacks ─────────────────────────────────────────────────
+/// DFS stack of “in‐memory” patterns (each Pattern holds its own ilist/slist, freq, str_pnt, etc.)
+extern std::vector<Pattern> DFS;
+
+extern std::vector<std::vector<int>> collectedPatterns;
+// ─── Collected output ───────────────────────────────────────────────────────
+/// Clears any patterns left in DFS (called at the start of each run)
+inline void ClearCollected() {
+    DFS.clear();
+    collectedPatterns.clear();     
+}
+/// Returns a reference to the entire “collected patterns” vector
+/// (each Pattern knows how to output itself as a vector<int>)
+const std::vector<std::vector<int>>& GetCollected();
+
+// ─── Helper functions ───────────────────────────────────────────────────────
+/// Given a clock‐tick difference, return elapsed seconds as a float
+float give_time(std::clock_t kk);
+
+/// Check whether a candidate can extend its parent pattern:
+///   cur_arc  = current Arc node ID in MDD
+///   str_pnt  = string‐pointer of the existing pattern
+///   start    = starting index within the MDD arc‐list
+///   strpnt_vec = parent’s “string pointers” vector
+/// Returns true if `cur_arc` is a valid child of `str_pnt` from position `start`.
+bool check_parent(unsigned int cur_arc,
+                  unsigned int str_pnt,
+                  unsigned int start,
+                  std::vector<unsigned int>& strpnt_vec);
+
+} // namespace htminer

effspm/largebm/src/build_mdd.cpp

@@ -0,0 +1,137 @@
+// File: effspm/largebm/src/load_inst.cpp
+
+#include <vector>
+#include <iostream>
+#include <unordered_map>
+#include "load_inst.hpp"
+#include "build_mdd.hpp"
+#include "freq_miner.hpp"
+#include "utility.hpp"
+
+namespace largebm {
+
+    // Forward declaration for Add_arc
+    int Add_arc(int item, unsigned long long int last_arc, int& itmset,
+                std::unordered_map<int, unsigned long long int>& ancest_map);
+
+    // Global MDD tree and other globals (declared in headers)
+    std::vector<Arc> Tree;
+
+    void Build_MDD(std::vector<int>& items) {
+        std::unordered_map<int, unsigned long long int> ancest_map;
+        unsigned long long int last_arc = 0;
+        int itmset = 0;
+
+        for (auto it = items.begin(); it != items.end(); ++it) {
+            last_arc = Add_arc(*it, last_arc, itmset, ancest_map);
+        }
+    }
+
+
+    int Add_arc(int item, unsigned long long int last_arc, int& itmset,
+                std::unordered_map<int, unsigned long long int>& ancest_map) {
+
+        unsigned idx = std::abs(item) - 1;
+
+        // ─── DEBUG ────────────────────────────────────────────────
+        // std::cout << "[Add_arc] item=" << item
+        //           << "  idx=" << idx
+        //           << "  last_arc=" << last_arc
+        //           << "  Tree.size=" << Tree.size()
+        //           << "  DFS.size="  << DFS.size()
+        //           << std::endl;
+
+        // Ensure DFS can hold this index
+        if (idx >= DFS.size()) {
+           // std::cout << "[Add_arc]  • resizing DFS to " << (idx + 1) << std::endl;
+            DFS.reserve(idx + 1);
+            while (DFS.size() <= idx) {
+                DFS.emplace_back(-static_cast<int>(DFS.size()) - 1); // Pattern(-id)
+            }
+        }
+
+        unsigned long long int anct;
+        auto p = ancest_map.find(std::abs(item));
+        if (p == ancest_map.end()) {
+            anct = 0;
+        } else {
+            anct = p->second;
+        }
+
+        if (item < 0) {
+            ++itmset;
+        }
+
+        // Before accessing Tree[last_arc].chld, check bounds
+        if (last_arc >= Tree.size()) {
+            // std::cout << "[Add_arc]  !!! last_arc OOB  last_arc="
+            //           << last_arc << "  Tree.size=" << Tree.size()
+            //           << std::endl;
+            // We still proceed so we can see crash context:
+        }
+
+        unsigned long long int last_sibl = 0;
+        if (last_arc < Tree.size()) {
+            last_sibl = Tree[last_arc].chld;
+        }
+
+        if (last_sibl == 0) {
+            // Insert new node as first child
+            Tree.emplace_back(item, itmset, anct);
+            last_sibl = Tree.size() - 1;
+
+            if (last_arc < Tree.size()) {
+                Tree[last_arc].chld = last_sibl;
+            }
+            if (anct == 0) {
+                // Debug before DFS access
+                // std::cout << "[Add_arc]  • DFS access at index=" << (std::abs(item) - 1)
+                //           << "  DFS.size=" << DFS.size() << std::endl;
+                DFS[std::abs(item) - 1].str_pnt.push_back(last_sibl);
+            }
+
+        } else {
+
+            // Walk siblings until find matching item or end
+            while (true) {
+                if (last_sibl >= Tree.size()) {
+                    // std::cout << "[Add_arc]  !!! last_sibl OOB  last_sibl="
+                    //           << last_sibl << "  Tree.size=" << Tree.size()
+                    //           << std::endl;
+                    break;
+                }
+                if (Tree[last_sibl].item == item) {
+                    break;
+                }
+                if (Tree[last_sibl].sibl == 0) {
+                    Tree.emplace_back(item, itmset, anct);
+                    Tree[last_sibl].sibl = Tree.size() - 1;
+                    last_sibl = Tree.size() - 1;
+                    if (anct == 0) {
+                        // std::cout << "[Add_arc]  • DFS access at index=" << (std::abs(item) - 1)
+                        //           << "  DFS.size=" << DFS.size() << std::endl;
+                        DFS[std::abs(item) - 1].str_pnt.push_back(last_sibl);
+                    }
+                    break;
+                }
+                last_sibl = Tree[last_sibl].sibl;
+            }
+        }
+
+        if (anct == 0) {
+            // std::cout << "[Add_arc]  • increment DFS.freq at index=" << (std::abs(item) - 1)
+            //           << "  DFS.size=" << DFS.size() << std::endl;
+            DFS[std::abs(item) - 1].freq++;
+        }
+
+        if (last_sibl < Tree.size()) {
+            // std::cout << "[Add_arc]  • increment Tree.freq at node=" << last_sibl
+            //           << "  Tree.size=" << Tree.size() << std::endl;
+            Tree[last_sibl].freq++;
+        }
+
+        ancest_map[std::abs(item)] = last_sibl;
+        return last_sibl;
+    }
+
+}  // namespace largebm

effspm/largebm/src/build_mdd.hpp

@@ -0,0 +1,47 @@
+#pragma once
+
+#include<vector>
+#include <cmath>
+#include "load_inst.hpp"
+
+namespace largebm {
+void Build_MDD(std::vector<int>& items);
+
+class Arc {
+public:
+
+	unsigned long long int chld;	
+	unsigned long long int sibl;
+	unsigned long long int freq;		
+	unsigned long long int anct;
+	int itmset;
+	int item;
+
+	Arc(int _itm, int _itmset, unsigned long long int _anc) {
+		itmset = _itmset;
+		anct = _anc;
+		item = _itm;
+		freq = 0;
+		chld = 0;
+		sibl = 0;
+	}
+
+	Arc(int _itm, int _anc) {
+		item = _itm;
+		anct = _anc;
+		freq = 0;
+		chld = 0;
+		sibl = 0;
+	}
+
+	Arc() {
+		freq = 0;
+		chld = 0;
+		sibl = 0;
+	}
+
+
+};
+
+extern std::vector<Arc> Tree;
+}

effspm/largebm/src/freq_miner.cpp

@@ -0,0 +1,342 @@
+// File: effspm/largebm/src/freq_miner.cpp
+
+#include <vector>
+#include <algorithm>
+#include <iostream>
+#include <fstream>
+#include <ctime>
+#include <unordered_map>
+#include <unordered_set>
+
+#include "freq_miner.hpp"   // must come before load_inst.hpp
+#include "load_inst.hpp"
+#include "utility.hpp"
+#include "build_mdd.hpp"
+
+namespace largebm {
+
+    // Helper declarations (must match headers exactly)
+    static void Out_patt(const std::vector<int>& seq, unsigned long long freq);
+    static void Extend_patt(Pattern& patt);
+
+    // Globals (declared once; types must match freq_miner.hpp)
+    unsigned long long int num_patt = 0;
+    std::vector<bool> ilist;
+    std::vector<bool> slist;
+    std::vector<int> DFS_numfound;
+    Pattern _patt;
+
+    void Freq_miner() {
+        std::vector<int> list;
+
+        if (use_list) {
+            // List-based routine
+            std::vector<int> empty_pref;
+            Freq_miner_list(items, empty_pref, theta, collected);
+            return;
+        }
+
+        // MDD-based initialization
+        for (int i = 0; i < static_cast<int>(L); ++i) {
+            if (DFS[i].freq >= theta) {
+                list.push_back(-i - 1);
+                if (itmset_exists) {
+                    list.push_back(i + 1);
+                }
+            }
+        }
+        for (size_t i = 0; i < DFS.size(); ++i) {
+            DFS[i].list = list;
+        }
+
+        while (!DFS.empty() && give_time(clock() - start_time) < time_limit) {
+            if (DFS.back().freq >= theta) {
+                Extend_patt(DFS.back());
+            } else {
+                DFS.pop_back();
+            }
+        }
+    }
+
+    void Extend_patt(Pattern& _pattern) {
+        swap(_patt, _pattern);
+        DFS.pop_back();
+
+        slist = std::vector<bool>(L, false);
+        bool ilist_nempty = false;
+
+        if (itmset_exists) {
+            ilist = std::vector<bool>(L, false);
+            for (auto it = _patt.list.begin(); it != _patt.list.end(); ++it) {
+                if (*it < 0) {
+                    slist[-(*it) - 1] = true;
+                } else {
+                    ilist[(*it) - 1] = true;
+                    ilist_nempty = true;
+                }
+            }
+        } else {
+            for (auto it = _patt.list.begin(); it != _patt.list.end(); ++it) {
+                slist[-(*it) - 1] = true;
+            }
+        }
+
+        int itmset_size = 1;
+        int last_neg = static_cast<int>(_patt.seq.size()) - 1;
+        while (_patt.seq[last_neg] > 0) {
+            --last_neg;
+            ++itmset_size;
+        }
+
+        std::vector<Pattern> pot_patt(L + (ilist_nempty ? L : 0));
+        std::vector<unsigned long long int> DFS_patt_init;
+        std::vector<unsigned long long int> DFS_patt;
+        if (ilist_nempty) {
+            DFS_numfound.clear();
+        }
+        std::vector<unsigned long long int> last_strpnt(L, 0);
+
+        for (unsigned long long int pnt = 0; pnt < _patt.str_pnt.size(); ++pnt) {
+            DFS_patt_init.push_back(_patt.str_pnt[pnt]);
+            while (!DFS_patt_init.empty()) {
+                unsigned long long int cur_sibl = Tree[DFS_patt_init.back()].chld;
+                DFS_patt_init.pop_back();
+                while (cur_sibl != 0) {
+                    int cur_itm = Tree[cur_sibl].item;
+                    if (cur_itm < 0) {
+                        cur_itm = -cur_itm;
+                        if (slist[cur_itm - 1]) {
+                            pot_patt[cur_itm - 1].freq += Tree[cur_sibl].freq;
+                            if (Tree[cur_sibl].chld != 0) {
+                                pot_patt[cur_itm - 1].str_pnt.push_back(cur_sibl);
+                            }
+                        }
+                        if (Tree[cur_sibl].chld != static_cast<unsigned long long>(-1)) {
+                            DFS_patt.push_back(cur_sibl);
+                            if (ilist_nempty) {
+                                if (cur_itm == -_patt.seq[last_neg]) {
+                                    DFS_numfound.push_back(1);
+                                } else {
+                                    DFS_numfound.push_back(0);
+                                }
+                            }
+                        }
+                    } else {
+                        if (ilist[cur_itm - 1]) {
+                            pot_patt[cur_itm + L - 1].freq += Tree[cur_sibl].freq;
+                            if (Tree[cur_sibl].chld != 0) {
+                                pot_patt[cur_itm + L - 1].str_pnt.push_back(cur_sibl);
+                            }
+                        }
+                        if (Tree[cur_sibl].chld != static_cast<unsigned long long>(-1)) {
+                            DFS_patt_init.push_back(cur_sibl);
+                        }
+                    }
+                    cur_sibl = Tree[cur_sibl].sibl;
+                }
+            }
+            if (ilist_nempty) {
+                for (int i = 0; i < static_cast<int>(L); ++i) {
+                    if (ilist[i]) {
+                        last_strpnt[i] = pot_patt[i + L].str_pnt.size();
+                    }
+                }
+            }
+            while (!DFS_patt.empty()) {
+                unsigned long long int cur_sibl = Tree[DFS_patt.back()].chld;
+                DFS_patt.pop_back();
+                int num_found = 0;
+                if (ilist_nempty) {
+                    num_found = DFS_numfound.back();
+                    DFS_numfound.pop_back();
+                }
+                while (cur_sibl != 0) {
+                    int cur_itm = Tree[cur_sibl].item;
+                    if (cur_itm > 0) {
+                        if (num_found == itmset_size &&
+                            ilist[cur_itm - 1] &&
+                            (Tree[Tree[cur_sibl].anct].itmset < Tree[_patt.str_pnt[pnt]].itmset ||
+                             !check_parent(cur_sibl, _patt.str_pnt[pnt],
+                                           last_strpnt[cur_itm - 1],
+                                           pot_patt[cur_itm + L - 1].str_pnt))) {
+                            pot_patt[cur_itm + L - 1].freq += Tree[cur_sibl].freq;
+                            if (Tree[cur_sibl].chld != 0) {
+                                pot_patt[cur_itm + L - 1].str_pnt.push_back(cur_sibl);
+                            }
+                        }
+                        if (slist[cur_itm - 1] &&
+                            Tree[Tree[cur_sibl].anct].itmset <= Tree[_patt.str_pnt[pnt]].itmset) {
+                            pot_patt[cur_itm - 1].freq += Tree[cur_sibl].freq;
+                            if (Tree[cur_sibl].chld != 0) {
+                                pot_patt[cur_itm - 1].str_pnt.push_back(cur_sibl);
+                            }
+                        }
+                        if (Tree[cur_sibl].chld != static_cast<unsigned long long>(-1)) {
+                            DFS_patt.push_back(cur_sibl);
+                            if (ilist_nempty) {
+                                if (num_found < itmset_size &&
+                                    cur_itm == std::abs(_patt.seq[last_neg + num_found])) {
+                                    DFS_numfound.push_back(num_found + 1);
+                                } else {
+                                    DFS_numfound.push_back(num_found);
+                                }
+                            }
+                        }
+                    } else {
+                        cur_itm = -cur_itm;
+                        if (slist[cur_itm - 1] &&
+                            Tree[Tree[cur_sibl].anct].itmset <= Tree[_patt.str_pnt[pnt]].itmset) {
+                            pot_patt[cur_itm - 1].freq += Tree[cur_sibl].freq;
+                            if (Tree[cur_sibl].chld != 0) {
+                                pot_patt[cur_itm - 1].str_pnt.push_back(cur_sibl);
+                            }
+                        }
+                        if (Tree[cur_sibl].chld != static_cast<unsigned long long>(-1)) {
+                            DFS_patt.push_back(cur_sibl);
+                            if (ilist_nempty) {
+                                if (cur_itm == -_patt.seq[last_neg]) {
+                                    DFS_numfound.push_back(1);
+                                } else {
+                                    DFS_numfound.push_back(0);
+                                }
+                            }
+                        }
+                    }
+                    cur_sibl = Tree[cur_sibl].sibl;
+                }
+            }
+        }
+
+        std::vector<int> ilistp;
+        std::vector<int> slistp;
+        for (auto it = _patt.list.begin(); it != _patt.list.end(); ++it) {
+            int idx = (*it < 0) ? (-(*it) - 1) : ((*it) - 1 + static_cast<int>(L));
+            if (*it > 0 && pot_patt[idx].freq >= theta) {
+                ilistp.push_back(*it);
+            } else if (*it < 0 && pot_patt[-(*it) - 1].freq >= theta) {
+                if (itmset_exists) {
+                    slistp.push_back(-(*it));
+                }
+                ilistp.push_back(*it);
+                slistp.push_back(*it);
+            }
+        }
+
+        for (auto it = ilistp.begin(); it != ilistp.end(); ++it) {
+            int p;
+            if (*it < 0) {
+                p = -(*it) - 1;
+            } else {
+                p = (*it) - 1 + static_cast<int>(L);
+            }
+
+            DFS.emplace_back();
+            swap(DFS.back(), pot_patt[p]);
+            DFS.back().seq = _patt.seq;
+            DFS.back().seq.push_back(*it);
+            if (*it < 0) {
+                DFS.back().list = slistp;
+            } else {
+                DFS.back().list = ilistp;
+            }
+            if (b_disp || b_write) {
+                Out_patt(DFS.back().seq, DFS.back().freq);
+            }
+            ++num_patt;
+        }
+    }
+
+    void Out_patt(const std::vector<int>& seq, unsigned long long freq) {
+        if (b_disp || b_write) {
+            std::ofstream file_o;
+            if (b_write) {
+                file_o.open(out_file, std::ios::app);
+            }
+            for (int v : seq) {
+                if (b_disp) std::cout << v << ' ';
+                if (b_write) file_o << v << ' ';
+            }
+            if (b_disp) std::cout << '\n';
+            if (b_write) file_o << '\n';
+
+            if (b_disp) {
+                std::cout << "************** Freq: " << freq << '\n';
+            }
+            if (b_write) {
+                file_o << "************** Freq: " << freq << '\n';
+                file_o.close();
+            }
+        }
+        collected.push_back(seq);
+    }
+
+    void Freq_miner_list(const std::vector<std::vector<int>>& db,
+                         std::vector<int>& prefix,
+                         unsigned long long minsup,
+                         std::vector<std::vector<int>>& out) {
+        // 1) count single‐item support (one count per sequence)
+        std::unordered_map<int, unsigned long long> freq;
+        for (auto const& seq : db) {
+            std::unordered_set<int> seen;
+            for (int x : seq) {
+                if (seen.insert(x).second) {
+                    ++freq[x];
+                }
+            }
+        }
+
+        // 2) collect the frequent candidates
+        std::vector<std::pair<int, unsigned long long>> cand;
+        cand.reserve(freq.size());
+        for (auto& p : freq) {
+            if (p.second >= minsup) {
+                cand.emplace_back(p.first, p.second);
+            }
+        }
+
+        // 3) sort by absolute item ID
+        std::sort(cand.begin(), cand.end(),
+                  [](const std::pair<int, unsigned long long>& a,
+                     const std::pair<int, unsigned long long>& b) {
+                      return std::abs(a.first) < std::abs(b.first);
+                  });
+
+        // 4) depth-first enumerate them
+        for (auto const& pr : cand) {
+            int item = pr.first;
+            prefix.push_back(item);
+
+            if (use_dic) {
+                // “un-compress” each pattern back to original IDs
+                std::vector<int> unmapped;
+                unmapped.reserve(prefix.size());
+                for (int cid : prefix) {
+                    int abs_id = std::abs(cid);
+                    int o = inv_item_dic[abs_id];
+                    unmapped.push_back(cid < 0 ? -o : o);
+                }
+                out.push_back(std::move(unmapped));
+            } else {
+                // just store the raw prefix
+                out.push_back(prefix);
+            }
+
+            // 5) project on the *first* occurrence of `item`
+            std::vector<std::vector<int>> proj;
+            proj.reserve(db.size());
+            for (auto const& seq : db) {
+                auto it = std::find(seq.begin(), seq.end(), item);
+                if (it != seq.end() && ++it != seq.end()) {
+                    proj.emplace_back(it, seq.end());
+                }
+            }
+
+            if (!proj.empty()) {
+                Freq_miner_list(proj, prefix, minsup, out);
+            }
+
+            prefix.pop_back();
+        }
+    }
+
+}  // namespace largebm

effspm/largebm/src/freq_miner.hpp

@@ -0,0 +1,48 @@
+#pragma once
+
+#include "load_inst.hpp"
+#include "build_mdd.hpp"
+
+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;
+	}
+
+
+};
+
+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;
+
+
+}