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

effspm/htminer/src/freq_miner.hpp

@@ -0,0 +1,60 @@
+#pragma once
+
+#include "load_inst.hpp"
+#include "build_mdd.hpp"
+
+namespace htminer {
+void Freq_miner();
+
+class Pattern {
+public:
+
+	vector<int> seq;
+	vector<unsigned int> str_pnt;
+	vector<int> list;
+
+	unsigned long long int freq;
+
+	Pattern(int item) {
+		seq.push_back(item);
+		freq = 0;
+	}
+
+	Pattern(size_t _pnt, bool _res) {
+		str_pnt.reserve(_pnt);
+		freq = 0;
+	}
+
+	Pattern() {
+		freq = 0;
+	}
+
+
+};
+
+class VPattern {
+public:
+
+	unsigned long long int ass_patt;
+	
+	vector<int> str_pnt;
+	vector<unsigned int> seq_ID;
+
+	VPattern(unsigned long long int _patt) {
+		ass_patt = _patt;
+	}
+
+	VPattern(size_t _pnt, bool a) {
+		str_pnt.reserve(_pnt);
+	}
+
+	VPattern() {}
+};
+
+
+extern unsigned long long int num_patt;
+extern vector<Pattern> DFS;
+extern vector<VPattern> VDFS;
+
+}
+

effspm/htminer/src/load_inst.cpp

@@ -0,0 +1,381 @@
+#include <iostream>
+#include <fstream>
+#include <sstream>
+#include <cmath>
+#include <ctime>
+#include "load_inst.hpp"
+#include "freq_miner.hpp"
+#include "utility.hpp"
+#include "build_mdd.hpp"
+
+namespace htminer {
+using namespace std;
+
+// unsigned int M = 0, mlim;
+// unsigned long long E = 0;
+
+// unsigned long long int N = 0, L = 0, theta;
+
+bool itmset_exists = 0;
+
+vector<int> item_dic;
+// vector<Pattern> DFS;
+// vector<VPattern> VDFS;
+
+void Load_items_pre(string &inst_name);
+bool Load_items(string &inst_name);
+bool Preprocess(string& inst, double thresh);
+
+bool Load_instance(std::string& items_file, double thresh) {
+    // Debug: entry
+    // std::cerr << "[HTMiner::Load_instance] called with file=\"" << items_file
+    //           << "\"  minsup=" << thresh << std::endl;
+
+    std::clock_t kk = std::clock();
+
+    // Initialize root of Tree
+    Tree.emplace_back(0, 0, 0);
+	Tree[0].itmset = 1;
+	{
+	std::vector<unsigned int> dummyAnc(L, 0);
+         std::vector<int>          dummyItems;
+         CTree.emplace_back(dummyAnc, dummyItems);
+         VTree.emplace_back();  // calls VArc() default ctor
+     }
+
+    // ─── Initialize DFS and VDFS to length L so Add_vec can index them ─
+     DFS.clear();
+     DFS.reserve(L);
+     for (unsigned int i = 0; i < L; ++i)
+        DFS.emplace_back(-static_cast<int>(i) - 1);
+        VDFS.clear();
+        VDFS.resize(L);
+
+    if (pre_pro) {
+        // std::cerr << "[HTMiner::Load_instance] pre_pro flag is true; calling Preprocess" << std::endl;
+        if (!Preprocess(items_file, thresh)) {
+           // std::cerr << "[HTMiner::Load_instance] Preprocess failed; returning false" << std::endl;
+            return false;
+        }
+       // std::cout << "\nPreprocess done in " << give_time(std::clock() - kk) << " seconds\n\n";
+
+        // Reserve DFS stack
+        DFS.reserve(L);
+        for (int i = 0; i < static_cast<int>(L); ++i)
+            DFS.emplace_back(-i - 1);
+
+        kk = std::clock();
+        //std::cerr << "[HTMiner::Load_instance] calling Load_items_pre" << std::endl;
+        Load_items_pre(items_file);
+
+        // If Tree is very large, shrink vectors
+        if (Tree.size() > 100000000) {
+            Tree.shrink_to_fit();
+            CTree.shrink_to_fit();
+            VTree.shrink_to_fit();
+           // std::cerr << "[HTMiner::Load_instance] Shrunk Tree, CTree, VTree to fit" << std::endl;
+        }
+        // std::cerr << "[HTMiner::Load_instance] Load_items_pre completed; N=" << N
+        //           << "  M=" << M << "  L=" << L << "  E=" << E << std::endl;
+    }
+    else if (!Load_items(items_file)) {
+        // std::cerr << "[HTMiner::Load_instance] Load_items failed; returning false" << std::endl;
+        return false;
+    }
+    else {
+        // Calculate absolute theta if user passed a fraction
+        if (thresh < 1.0) {
+            theta = static_cast<unsigned long long>(std::ceil(thresh * N));
+            //std::cerr << "[HTMiner::Load_instance] Computed theta = ceil(" << thresh << " * " << N << ") = " << theta << std::endl;
+        } else {
+            theta = static_cast<unsigned long long>(thresh);
+           // std::cerr << "[HTMiner::Load_instance] Using theta = " << theta << " (absolute)" << std::endl;
+        }
+        // std::cerr << "[HTMiner::Load_instance] No preprocessing; Load_items succeeded; N=" << N
+        //           << "  M=" << M << "  L=" << L << "  E=" << E << std::endl;
+    }
+
+    std::cout << "\nMDD Database built in " << give_time(std::clock() - kk) << " seconds\n\n";
+    std::cout << "Found " << N << " sequence, with max line len " << M
+              << ", and " << L << " items, and " << E << " enteries\n";
+    // std::cout << "Total Trie nodes: " << Tree.size()
+    //           << " Total CTree nodes: " << CTree.size()
+    //           << " Total VTree nodes: " << VTree.size() << std::endl;
+
+    // std::cerr << "[HTMiner::Load_instance] Exiting normally; returning true" << std::endl;
+    return true;
+}
+
+
+bool Preprocess(string &inst, double thresh) {
+
+         //std::cerr << "[HTMiner::Preprocess] called with \"" << inst << "\"  minsup=" << thresh << std::endl; //DEBUG
+
+
+	vector<unsigned long long int> MN(100, 0);
+	vector<vector<bool>> ML(100, vector<bool>(1000000, 0));
+	
+
+	ifstream file(inst);
+
+	vector<unsigned int> freq(1000000, 0);
+	vector<unsigned long long int> counted(1000000, 0);
+
+	if (file.good()) {
+		string line;
+		int ditem;
+		while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
+			++N;
+			
+//if (N % 10000000 == 0)
+//cout << "N: " << N << endl;
+			istringstream word(line);
+			string itm;
+			int size_m = 0;
+			while (word >> itm) {
+				++size_m;
+				ditem = stoi(itm);
+
+				if (ditem > 0) 
+					itmset_exists = 1;
+				else 
+					ditem *= -1;
+
+				if (size_m < MN.size()) {
+					++MN[size_m - 1];
+					if (ML[size_m - 1].size() < ditem) {
+						ML[size_m - 1].reserve(ditem);
+						while (ML[size_m - 1].size() < ditem)
+							ML[size_m - 1].push_back(0);
+					}
+					ML[size_m - 1][ditem - 1] = 1;
+				}
+	
+				if (L < ditem)
+					L = ditem;
+				
+				if (freq.size() < L) {
+					freq.reserve(L);
+					counted.reserve(L);
+					while (freq.size() < L) { 
+						freq.push_back(0);
+						counted.push_back(0);
+					}
+				}
+
+				if (counted[ditem - 1] != N) {
+					++freq[ditem - 1];
+					counted[ditem - 1] = N;
+				}
+
+			}
+			if (size_m > M)
+				M = size_m;
+		}
+	}
+	else {
+		//cout << "!!!!!! No such file exists: " << inst << " !!!!!!\n";
+		return 0;
+	}
+
+	if (thresh < 1)
+		theta = ceil(thresh * N);
+	else
+		theta = thresh;
+
+	int real_L = 0;
+	item_dic = vector<int>(L, -1);
+	vector<bool> item_in(L, 0);
+	for (int i = 0; i < L; ++i) {
+		if (freq[i] >= theta) {
+			item_dic[i] = ++real_L;
+			item_in[i] = 1;
+		}
+	}
+	
+	//cout << "Original number of items: " << L << " Reduced to: " << real_L << endl;
+
+	unsigned long long int LpM = 1;
+	mlim = M;
+	int orgmlim;
+	int ulim = min(3 + real_L / 5, 10);
+	unsigned long long int ml;
+	int coef = 1 + 1 * itmset_exists;
+	for (int i = 0; i + ulim < MN.size() && i + ulim < M; ++i) {
+		ml = 0;
+		for (int j = 0; j < L; ++j) {
+			if (ML[i][j] && item_in[j]) 
+				++ml;
+		}
+		LpM *= ml * coef;
+		// cout << ml << " " << LpM << " " << MN[i] << endl;
+		if (LpM * ulim > MN[i]) {
+		if (6 * (MN[i] - LpM) >= 5 * MN[i])
+			orgmlim = i;
+			while (i + ulim - 1 < MN.size() && i + ulim - 1 < M) {
+				// cout << MN[i - 1] - MN[i + ulim - 1] << " " << MN[i + ulim - 1] << endl;
+				if (MN[i - 1] - MN[i + ulim - 1] < MN[i + ulim - 1] && MN[i + ulim - 1] < 600000000 ) {
+					mlim = i - 1;
+					break;
+				}					
+				i += 1;
+			}
+			break;
+		}
+	}
+
+	// cout << "M is: " << M << " Mlim is: " << mlim << " ulim is: " << ulim << " original mlim is: " << orgmlim << " guess is: " << round((log(N) - log(6)) / log(real_L)) << endl;
+
+
+	if (mlim < M) {
+		for (int i = 0; i < real_L; ++i) 
+			VDFS.emplace_back(i);
+		if (MN[mlim + ulim] > 100000000) {
+			CTree.reserve(MN[mlim + ulim] / 2);
+			VTree.reserve(MN[mlim + ulim] / 2);
+			Tree.reserve((N - MN[mlim + ulim]) * 2);
+		}
+	}
+	else if (N > 100000000) 
+		Tree.reserve(500000000);
+
+
+	L = real_L;
+	N = 0;
+	M = 0;
+      
+	return 1;
+}
+
+
+void Load_items_pre(string &inst_name) {
+
+	ifstream file(inst_name);
+
+	if (file.good()) {
+		string line;
+		int ditem;
+		while (getline(file, line) && give_time(clock() - start_time) < time_limit) {
+			istringstream word(line);
+			string itm;
+			vector<int> temp_vec;
+			vector<int> temp_lim;
+			bool sgn = 0;
+			while (word >> itm) {
+				
+				ditem = stoi(itm);
+				
+				if (item_dic[abs(ditem) - 1] == -1) {
+					if (!sgn)
+						sgn = ditem < 0;
+					continue;
+				}
+				else {
+					if (ditem > 0)
+						ditem = item_dic[ditem - 1];
+					else
+						ditem = -item_dic[-ditem - 1];
+				}
+
+				if (sgn) {
+					if (ditem > 0)
+						ditem = -ditem;
+					sgn = 0;
+				}
+				
+				if (temp_vec.size() <= mlim)
+					temp_vec.push_back(ditem);
+				else
+					temp_lim.push_back(ditem);
+			}
+
+			if (temp_vec.empty())
+				continue;
+
+			++N;
+			//if (N % 1000000 == 0)
+			//	cout << N << " " << Tree.size() << " " << CTree.size() << " " << VTree.size() << endl;
+
+			if (temp_vec.size() + temp_lim.size() > M)
+				M = temp_vec.size() + temp_lim.size();
+
+			Build_MDD(temp_vec, temp_lim);
+		}
+	}
+
+}
+
+bool Load_items(std::string &inst_name) {
+    //std::cerr << "[HTMiner::Load_items] called with filename=\"" << inst_name << "\"" << std::endl;  // DEBUG
+    unsigned int mlim = 0;
+// reset the global M and E here instead of redeclaring:
+     M = 0;
+     E = 0;
+
+    std::ifstream file(inst_name);
+    if (!file.good()) {
+        //std::cerr << "[HTMiner::Load_items] ERROR: No such file: " << inst_name << std::endl;
+        return false;
+    }
+
+    std::string line;
+    int ditem;
+    while (std::getline(file, line) && give_time(std::clock() - start_time) < time_limit) {
+        ++N;
+        if (N % 1000000 == 0) {
+            // std::cerr << "[HTMiner::Load_items] Read " << N << " sequences so far; "
+            //           << "current M=" << M << ", L=" << L << ", E=" << E << std::endl;
+        }
+
+        std::istringstream word(line);
+        std::vector<int> temp_vec;
+        std::vector<int> temp_lim;
+        while (word >> ditem) {
+			E+=1;
+            if (ditem > 0)
+                itmset_exists = 1;
+
+            if (std::abs(ditem) > static_cast<int>(L)) {
+                L = std::abs(ditem);
+                //std::cerr << "[HTMiner::Load_items] Updated L to " << L << " (expanding DFS and VDFS)" << std::endl;
+               // Expand DFS
+                 while (DFS.size() < L) {
+                    DFS.reserve(L);
+                    DFS.emplace_back(-static_cast<int>(DFS.size()) - 1);
+               }
+                // ALSO expand VDFS so that VDFS[cur_itm-1] is valid later
+               VDFS.resize(L);
+            }
+
+            if (temp_vec.size() < mlim)
+                temp_vec.push_back(ditem);
+            else
+                temp_lim.push_back(ditem);
+        }
+
+        if (temp_vec.size() + temp_lim.size() > M) {
+            M = temp_vec.size() + temp_lim.size();
+            //std::cerr << "[HTMiner::Load_items] Updated M to " << M << std::endl;
+        }
+
+        // std::cerr << "[HTMiner::Load_items] Calling Build_MDD with temp_vec size=" << temp_vec.size()
+        //           << ", temp_lim size=" << temp_lim.size() << std::endl;
+        Build_MDD(temp_vec, temp_lim);
+        //std::cerr << "[HTMiner::Load_items] Build_MDD returned; Tree size now=" << Tree.size() << std::endl;
+    }
+
+    // std::cerr << "[HTMiner::Load_items] Finished loading. Final counts: N=" << N
+    //           << ", M=" << M << ", L=" << L << ", E=" << E << std::endl;
+    return true;
+}
+
+}
+
+
+
+
+
+
+
+
+

effspm/htminer/src/load_inst.hpp

@@ -0,0 +1,23 @@
+#pragma once
+
+#include <vector>
+#include <string>
+#include <fstream>
+#include <map>
+#include <unordered_set>
+#include <unordered_map>
+namespace htminer {
+using namespace std;
+
+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, itmset_exists;
+
+extern unsigned int M, mlim, time_limit;
+
+extern unsigned long long int N, L, theta;
+
+extern clock_t start_time; 
+}

effspm/htminer/src/main.cpp

@@ -0,0 +1,96 @@
+#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"
+
+using namespace std;
+
+string out_file;
+
+bool  b_disp = 0, b_write = 0, use_dic = 0, just_build = 0, pre_pro = 1;
+
+unsigned int time_limit = 10 * 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], "-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
+			cout << "Command " << argv[i] << " not recognized and skipped.\n";
+	}
+
+
+
+    cout << "\n********************** " << VV << "**********************\n";
+
+	string item_file = folder + VV + ".txt";
+
+	cout << "loading instances...\n";
+
+	clock_t start_time_all = clock();
+	start_time = clock();
+
+	if (!Load_instance(item_file, thresh)) {
+		cout << "Files invalid, exiting.\n";
+		cin.get();
+		return 0;
+	}
+
+	start_time = 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_all) << " seconds\n\n";
+	}
+
+
+	return 0;
+}

effspm/htminer/src/utility.cpp

@@ -0,0 +1,72 @@
+#include "utility.hpp"
+#include "load_inst.hpp"
+#include "freq_miner.hpp"
+#include <vector>
+namespace htminer {
+
+// ─── Flag‐like globals ──────────────────────────────────────────────────────
+bool use_list   = false;
+bool just_build = false;
+bool b_disp     = false;
+bool b_write    = false;
+bool use_dic    = false;
+bool pre_pro    = false;
+
+unsigned int time_limit = 0;
+std::string out_file    = "";
+std::clock_t start_time = 0;
+
+// ─── Dataset‐level globals ─────────────────────────────────────────────────
+std::vector<std::vector<int>> items;
+unsigned long long N     = 0;
+unsigned long long L     = 0;
+unsigned long long theta = 0;
+unsigned int M           = 0;
+unsigned long long E     = 0;
+unsigned int mlim        = 0; 
+// ─── DFS stacks ─────────────────────────────────────────────────────────────
+std::vector<Pattern> DFS;
+std::vector<VPattern> VDFS;
+
+// ─── Collected patterns storage ────────────────────────────────────────────
+std::vector<std::vector<int>> collectedPatterns;
+const std::vector<std::vector<int>>& GetCollected() {
+    return collectedPatterns;
+}
+
+// ─── give_time and check_parent get their definitions here (as provided) ───
+float give_time(std::clock_t kk) {
+    return static_cast<float>(kk) / static_cast<float>(CLOCKS_PER_SEC);
+}
+bool check_parent(unsigned int cur_anct, unsigned int str_pnt, unsigned int start, vector<unsigned int>& strpnt_vec) {
+
+	vector<unsigned int> ancestors;
+	
+	while (abs(Tree[cur_anct].itmset) > abs(Tree[str_pnt].itmset)) {
+		if (Tree[cur_anct].item > 0)
+			ancestors.push_back(cur_anct);
+		cur_anct = Tree[cur_anct].anct;
+	}
+
+	if (abs(Tree[cur_anct].itmset) == abs(Tree[str_pnt].itmset))
+		return 1;
+	else {
+		for (vector<unsigned int>::reverse_iterator it = ancestors.rbegin(); it != ancestors.rend(); ++it) {
+			for (unsigned int i = start; i < strpnt_vec.size(); ++i) {
+				if (strpnt_vec[i] == *it)
+					return 1;
+			}
+		}
+	}
+
+
+	return 0;
+
+}
+
+
+
+
+
+
+}