nysol-take 3.0.0

data/ext/lcmseqrun/src/vec.h

@@ -0,0 +1,172 @@
+/* library for sparse vector */
+/* Takeaki Uno    27/Dec/2008 */
+
+#ifndef _vec_h_
+#define _vec_h_
+
+//#define USE_MATH
+
+#include"math.h"
+#include"queue.h"
+
+#ifndef SVEC_VAL
+ #ifdef SVEC_VAL_INT
+  #define SVEC_VAL int
+  #define SVEC_VAL2 LONG
+  #define SVEC_VAL_END INTHUGE
+  #define SVEC_VAL2_END LONGHUGE
+  #define SVEC_VALF "%d"
+ #else
+  #define SVEC_VAL double
+  #define SVEC_VAL2 double
+  #define SVEC_VAL_END DOUBLEHUGE
+  #define SVEC_VAL2_END DOUBLEHUGE
+  #define SVEC_VALF "%f"
+ #endif
+#endif
+
+#define VEC_LOAD_BIN 16777216   // read binary file 
+#define VEC_LOAD_BIN2 33554432   // read binary file with 2byte for each number 
+#define VEC_LOAD_BIN4 67108864   // read binary file with 4byte for each number
+#define VEC_LOAD_CENTERIZE 134217728  // read binary file, and minus the half(128) from each number
+#define VEC_NORMALIZE 268435456  // read binary file, and minus the half(128) from each number
+
+/* matrix */
+typedef struct {
+  unsigned char type;  // mark to identify type of the structure
+  char *fname;      // input file name
+  int flag;         // flag
+
+  VEC *v;
+  VEC_ID end;
+  VEC_ID t;
+  VEC_VAL *buf, *buf2;
+  VEC_ID clms;
+  size_t eles;
+  VEC_VAL *buf_org, *buf2_org;
+} MAT;
+
+/* sparse vector, element */
+typedef struct {
+  QUEUE_ID i;
+  SVEC_VAL a;
+} SVEC_ELE;
+
+/* sparse vector, vector */
+typedef struct {
+  unsigned char type;  // mark to identify type of the structure
+  SVEC_ELE *v;
+  VEC_ID end;
+  VEC_ID t;
+} SVEC;
+
+/* sparse vector, matrix */
+typedef struct {
+  unsigned char type;  // mark to identify type of the structure
+  char *fname;      // input file name
+  int flag;         // flag
+
+  SVEC *v;
+  VEC_ID end;
+  VEC_ID t;
+  SVEC_ELE *buf, *buf2;
+  VEC_ID clms;
+  size_t eles, ele_end;
+} SMAT;
+
+/* set family */
+typedef struct {
+  unsigned char type;  // mark to identify type of the structure
+  char *fname;      // input file name
+  int flag;         // flag
+
+  QUEUE *v;
+  VEC_ID end;
+  VEC_ID t;
+  QUEUE_INT *buf, *buf2;
+  VEC_ID clms;
+  size_t eles, ele_end;
+  WEIGHT *cw, *rw, **w, *wbuf;
+  int unit;
+  char *wfname, *cwfname, *rwfname;     // weight file name
+  PERM *rperm, *cperm;  // row permutation
+} SETFAMILY;
+
+#define INIT_SETFAMILY_ {TYPE_SETFAMILY,NULL,0,NULL,0,0,NULL,NULL,0,0,0,NULL,NULL,NULL,NULL,sizeof(QUEUE_INT),NULL,NULL,NULL,NULL,NULL}
+
+extern MAT INIT_MAT;
+extern SVEC INIT_SVEC;
+extern SMAT INIT_SMAT;
+extern SETFAMILY INIT_SETFAMILY;
+
+QSORT_TYPE_HEADER (SVEC_VAL, SVEC_VAL)
+QSORT_TYPE_HEADER (SVEC_VAL2, SVEC_VAL2)
+
+#define   ARY_QUEUE_INPRO(f,U,V)  do{(f)=0;FLOOP(common_QUEUE_ID, 0, (QUEUE_ID)(U).t)(f)+=(V)[(U).v[common_QUEUE_ID]];}while(0)
+#define   ARY_SVEC_INPRO(f,U,V)  do{(f)=0;FLOOP(common_VEC_ID, 0, (VEC_ID)(U).t)(f)+=((double)(U).v[common_VEC_ID].a)*(V)[(U).v[common_VEC_ID].i];}while(0)
+
+/* terminate routine for VEC */
+void VEC_end (VEC *V);
+void MAT_end (MAT *M);
+void SVEC_end (SVEC *V);
+void SMAT_end (SMAT *M);
+void SETFAMILY_end (SETFAMILY *M);
+
+/* allocate memory according to rows and rowt */
+void VEC_alloc (VEC *V, VEC_ID clms);
+void MAT_alloc (MAT *M, VEC_ID rows, VEC_ID clms);
+void SVEC_alloc (SVEC *V, VEC_ID end);
+void SMAT_alloc (SMAT *M, VEC_ID rows, VEC_ID *rowt, VEC_ID clms, size_t eles);
+void SETFAMILY_alloc (SETFAMILY *M, VEC_ID rows, VEC_ID *rowt, VEC_ID clms, size_t eles);
+void SETFAMILY_alloc_weight (SETFAMILY *M, QUEUE_ID *t);
+
+/* count/read the number in file for MAT */
+/* if *rows>0, only read count the numbers in a row, for the first scan. */
+void MAT_load_bin (MAT *M, FILE2 *fp, int unit);
+void MAT_file_load (MAT *M, FILE2 *fp);
+void MAT_load (MAT *M);
+void SMAT_load (SMAT *M);
+void SETFAMILY_load (SETFAMILY *M);
+void SETFAMILY_load_weight (SETFAMILY *M);
+void SETFAMILY_load_row_weight (SETFAMILY *M);
+void SETFAMILY_load_column_weight (SETFAMILY *M);
+
+void MAT_print (FILE *fp, MAT *M);
+void SVEC_print (FILE *fp, SVEC *M);
+void SMAT_print (FILE *fp, SMAT *M);
+void SETFAMILY_print (FILE *fp, SETFAMILY *M);
+void SETFAMILY_print_weight (FILE *fp, SETFAMILY *M);
+  
+
+/* norm, normalization **************************/
+double SVEC_norm (SVEC *V);
+void SVEC_normalize (SVEC *V);
+
+/* inner product **************************/
+SVEC_VAL2 SVEC_inpro (SVEC *V1, SVEC *V2);
+
+/** Euclidean distance routines *********************************/
+double VEC_eucdist (VEC *V1, VEC *V2);
+double SVEC_eucdist (SVEC *V1, SVEC *V2);
+double VEC_SVEC_eucdist (VEC *V1, SVEC *V2);
+double QUEUE_eucdist (QUEUE *Q1, QUEUE *Q2);
+double VEC_QUEUE_eucdist (VEC *V, QUEUE *Q);
+
+void VEC_rand_gaussian (VEC *V);
+
+double VEC_linfdist (VEC *V1, VEC *V2);
+
+/* compute the inner product, Euclidean distance for multi vector */
+double MVEC_norm (void *V);
+double MVEC_inpro (void *V, void *U);
+double MVEC_double_inpro (void *V, double *p);
+double MVEC_eucdist (void *V, void *U);
+
+/* compute the inner product, euclidean distance for i,jth vector */
+double MMAT_inpro_ij (void *M, int i, int j);
+double MMAT_double_inpro_i (void *M, int i, double *p);
+double MMAT_eucdist_ij (void *M, int i, int j);
+double MMAT_norm_i (void *M, int i);
+
+
+#endif

data/ext/lcmtransrun/extconf.rb

@@ -0,0 +1,18 @@
+require "rubygems"
+require "mkmf"
+
+
+$CPPFLAGS += " -Wall"
+$LOCAL_LIBS += " -lstdc++ "
+
+cp = "$(srcdir)"
+
+$CFLAGS = " -O3 -Wall -I. -I#{cp}/src -DB_STATIC -D_NO_MAIN_ -DLINE -fPIC -Wno-error=format-security"
+$CPPFLAGS = " -O3 -Wall -I. -I#{cp}/src  -DB_STATIC -D_NO_MAIN_ -DLINE -fPIC -Wno-error=format-security"
+$CXXFLAGS = " -O3 -Wall -I. -I#{cp}/src  -DB_STATIC -D_NO_MAIN_ -DLINE -fPIC -Wno-error=format-security"
+
+$LOCAL_LIBS += " -lstdc++ -lkgmod3"
+
+
+create_makefile("nysol/lcmtransrun")
+

data/ext/lcmtransrun/lcmtransrun.cpp

@@ -0,0 +1,264 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <ruby.h>
+#include <kgMethod.h>
+
+#define MAXC 1024000
+
+extern "C" {
+	void Init_lcmtransrun(void);
+}
+
+int lcmtrans(int argc, char **argv){
+	char mode;
+	char *hd;
+	if(argc==2 && 0==strcmp(*(argv+1),"-v" )){
+		fprintf(stderr,"lcm_trans 1.0\n");
+		return 1;
+	}
+
+	if(argc!=4){
+hd="\
+usage) lcm_trans filename p|t\n\
+filename: output file by lcm(lcm_seq)\n\
+p: output patterns and their frequency\n\
+t: output transaction number with pattern ID\n\
+\n\
+ex. of input file(output file by lcm)\n\
+-----------\n\
+ 3\n\
+ 0 1 2\n\
+4 2\n\
+ 0 2\n\
+4 8 1\n\
+ 0\n\
+4 8 6 1\n\
+ 0\n\
+-----------\n\
+odd line: pattern with frequency\n\
+even line: transaction number \n\
+\n\
+standard output (2nd parameter is 'p')\n\
+pattern,count,size,pid\n\
+,3,0,0\n\
+4,2,1,1\n\
+4 8,1,2,2\n\
+4 8 6,1,3,3\n\
+\n\
+standard output (2nd parameter is 't')\n\
+__tid,pid\n\
+0,0\n\
+1,0\n\
+2,0\n\
+0,1\n\
+2,1\n\
+0,2\n\
+0,3\n\
+";
+		fprintf(stderr,"%s",hd);
+		return EXIT_FAILURE;
+	}
+
+	if(**(argv+2)=='p'){
+		mode='p';
+	}else if(**(argv+2)=='e'){
+		mode='e';
+	}else if(**(argv+2)=='t'){
+		mode='t';
+	}else{
+		fprintf(stderr,"argument error2: %s\n",argv+2);
+	}
+
+	FILE *fp;
+	if((fp=fopen(*(argv+1),"rb"))==NULL){
+		fprintf(stderr,"file open error\n");
+		return EXIT_FAILURE;
+	}
+	FILE *ofp;
+	if((ofp=fopen(*(argv+3),"w"))==NULL){
+		fprintf(stderr,"file open error\n");
+		return EXIT_FAILURE;
+	}
+
+
+
+	int odd=1; // 奇数行フラグ
+
+	// enumLcmSeq.rb,enumLcmIs.rbでのパターン
+	// 4 5 (2)
+	if(mode=='p'){
+		fprintf(ofp,"pattern,count,size,pid\n");
+		size_t recNo=0;
+		char buf[MAXC]; // 出力バッファ
+		char buf2[MAXC]; // 出力バッファ
+		int spcCount=0; // スペースのカウント
+		int opos=-1;
+
+		while(1){
+			int rsize = fread(buf, sizeof(char), MAXC, fp);
+			if( rsize < 0 ){ 
+				fprintf(stderr,"file read error\n");
+				return EXIT_FAILURE;
+			}
+			if( rsize == 0 ) { break;}
+			int i;
+			for(i=0 ; i<rsize ;i++){
+				if(buf[i]=='\n'){
+					if(odd==1){
+						fprintf(ofp,"%s,%d,%ld\n",buf2,spcCount,recNo);
+						recNo++;
+					}
+					spcCount=0;
+					opos=-1;
+					odd*=(-1);
+					continue;
+				}
+				if(odd!=1) { continue;}
+				if(buf[i]==' '){
+					buf2[++opos]=buf[i];
+					if(opos!=0){ spcCount++; }
+				}
+				else if(buf[i]=='('){
+					buf2[opos]=',';
+				}
+				else if(buf[i]==')'){
+					buf2[++opos]='\0';
+				}
+				else { buf2[++opos]=buf[i];}
+			}
+		}
+	}else if(mode=='e'){
+		fprintf(ofp,"pattern,countP,countN,size,pid\n");
+		size_t recNo=0;
+		char buf[MAXC]; // 出力バッファ
+		char buf2[MAXC]; // 出力バッファ
+		int spcCount=0; // スペースのカウント
+		int comCount=0; // スペースのカウント
+		int opos=-1;
+
+		while(1){
+			int rsize = fread(buf, sizeof(char), MAXC, fp);
+			if( rsize < 0 ){ 
+				fprintf(stderr,"file read error\n");
+				return EXIT_FAILURE;
+			}
+			if( rsize == 0 ) { break;}
+			int i;
+			for(i=0 ; i<rsize ;i++){
+				if(buf[i]=='\n'){
+					if(odd==1){
+						fprintf(ofp,"%s,%d,%ld\n",buf2,spcCount,recNo);
+						recNo++;
+					}
+					spcCount=0;
+					comCount=0;
+					opos=-1;
+					odd*=(-1);
+					continue;
+				}
+				if(odd!=1) { continue;}
+				if(buf[i]==' '){
+					buf2[++opos]=buf[i];
+					if(opos!=0){ spcCount++; }
+				}
+				else if(buf[i]=='('){
+					buf2[opos]=',';
+				}
+				else if(buf[i]==','){
+					comCount++;
+					if(comCount==2){ buf2[++opos]='\0';}
+					else{ buf2[++opos]=buf[i];}
+				}
+				else { buf2[++opos]=buf[i];}
+			}
+		}
+
+	}else{
+		fprintf(ofp,"__tid,pid\n");
+		size_t recNo=0;
+		int opos=-1;
+		char buf[MAXC]; // 出力バッファ
+		char buf2[MAXC]; // 出力バッファ
+
+		while(1){
+			int rsize = fread(buf, sizeof(char), MAXC, fp);
+			if( rsize < 0 ){ 
+				fprintf(stderr,"file read error\n");
+				return EXIT_FAILURE;
+			}
+			if( rsize == 0 ) { break;}
+			int i;
+			for(i=0 ; i<rsize ;i++){
+				if(odd==1){//奇数行は¥nまでなにもしない
+					if(buf[i]=='\n'){
+						opos=-1;
+						odd*=(-1);
+						buf2[0]='\0';
+					}
+				}
+				else{
+					if(buf[i]==' '||buf[i]=='\n'){
+						buf2[++opos]='\0';
+						if(buf2[0] != '\0'){
+							fprintf(ofp,"%s,%ld\n",buf2,recNo);
+						}
+						opos=-1;
+						if(buf[i]=='\n'){ 
+							odd*=(-1);
+							recNo++;
+						}
+					}
+					else{ buf2[++opos]=buf[i]; }
+				}
+			}
+		}
+	}
+
+	if(0!=fclose(fp)){
+		fprintf(stderr,"file close error\n");
+		return EXIT_FAILURE;
+	}
+	if(0!=fclose(ofp)){
+		fprintf(stderr,"file close error\n");
+		return EXIT_FAILURE;
+	}
+	return 0;
+}
+
+VALUE lcm_trans(int argc, VALUE *argvV, VALUE self){
+	VALUE inf;
+	VALUE para;
+	VALUE outf;
+	rb_scan_args(argc, argvV, "30",&inf,&para,&outf);
+
+	// 引数文字列へのポインタの領域はここでauto変数に確保する
+	size_t vvSize=4;
+	
+	kglib::kgAutoPtr2<char*> argv;
+	char** vv;
+	try{
+		argv.set(new char*[vvSize]);
+		vv = argv.get();
+	}catch(...){
+		rb_raise(rb_eRuntimeError,"memory allocation error");
+	}
+	// vv配列0番目はコマンド名
+	vv[0]=const_cast<char*>("lcm_trans");
+	vv[1]=RSTRING_PTR(inf);
+	vv[2]=RSTRING_PTR(para);
+	vv[3]=RSTRING_PTR(outf);
+	lcmtrans(vvSize,vv);
+	return Qtrue;
+
+}
+
+
+
+void Init_lcmtransrun(void) 
+{
+	// モジュール定義:MCMD::xxxxの部分
+	VALUE mtake=rb_define_module("TAKE");
+	rb_define_module_function(mtake,"run_lcmtrans" , (VALUE (*)(...))lcm_trans,-1);
+
+}

data/ext/macerun/extconf.rb

@@ -0,0 +1,20 @@
+require "rubygems"
+require "mkmf"
+
+unless have_library("kgmod3")
+  puts("need libkgmod.")
+  puts("refer https://github.com/nysol/mcmd")
+  exit 1
+end
+
+
+cp = "$(srcdir)"
+
+$CFLAGS = " -O3 -Wall -I. -I#{cp}/src -DB_STATIC -D_NO_MAIN_ -DLINE -fPIC -Wno-error=format-security"
+$CPPFLAGS = " -O3 -Wall -I. -I#{cp}/src  -DB_STATIC -D_NO_MAIN_ -DLINE -fPIC -Wno-error=format-security"
+$CXXFLAGS = " -O3 -Wall -I. -I#{cp}/src  -DB_STATIC -D_NO_MAIN_ -DLINE -fPIC -Wno-error=format-security"
+
+$LOCAL_LIBS += " -lstdc++ -lm -lkgmod3"
+
+create_makefile("nysol/macerun")
+

data/ext/macerun/macerun.cpp

@@ -0,0 +1,57 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <sys/stat.h>
+#include <string>
+#include <ruby.h>
+
+#include "src/mace.c"
+
+
+#include <kgMethod.h>
+
+extern "C" {
+	void Init_macerun(void);
+}
+
+
+VALUE macerun(VALUE self,VALUE argvV){
+
+	string argstr=RSTRING_PTR(argvV);
+	vector<char *> opts = kglib::splitToken(const_cast<char*>(argstr.c_str()), ' ',true);
+
+	// 引数文字列へのポインタの領域はここでauto変数に確保する
+	kglib::kgAutoPtr2<char*> argv;
+	char** vv;
+	try{
+		argv.set(new char*[opts.size()+1]);
+		vv = argv.get();
+	}catch(...){
+		rb_raise(rb_eRuntimeError,"memory allocation error");
+	}
+
+	// vv配列0番目はコマンド名
+	vv[0]=const_cast<char*>("lcm");
+
+	size_t vvSize;
+	for(vvSize=0; vvSize<opts.size(); vvSize++){
+		vv[vvSize+1] = opts.at(vvSize);
+	}
+	vvSize+=1;
+
+	MACE_main(vvSize,vv);
+	return Qtrue;
+}
+
+// -----------------------------------------------------------------------------
+// ruby Mcsvin クラス init
+// -----------------------------------------------------------------------------
+void Init_macerun(void) 
+{
+	// モジュール定義:MCMD::xxxxの部分
+	VALUE mtake=rb_define_module("TAKE");
+	rb_define_module_function(mtake,"run_mace"       , (VALUE (*)(...))macerun,1);
+}
+
+