From: "Igor B. Poretsky" Date: Wed, 15 Feb 2023 14:03:25 +0300 Subject: Berkeley DB instead of GDBM --- lib/Makefile | 20 +++++--------- lib/dbm_prog.c | 78 +++++++++++++++++++++++++++++++++++-------------------- src/Makefile | 8 ++--- src/conv1.c | 12 ++++---- src/nrl_edin.c | 12 ++++---- src/space.c | 11 ++++--- src/transcribe.c | 11 ++++--- 7 files changed, 88 insertions(+), 64 deletions(-) --- a/lib/Makefile +++ b/lib/Makefile @@ -14,35 +14,31 @@ ############################################################################ # Sun users # CC = gcc -# CFLAGS = -g -Wall -pedantic +# CFLAGS = -O2 -s # LIBS = -lm ############################################################################ # FreeBSD users # CC = gcc -# CFLAGS = -O2 -Wall -pedantic -DFBSD_DATABASE +# CFLAGS = -O2 -DFBSD_DATABASE -s # LIBS = -lm ############################################################################ # Linux users CC = gcc -CFLAGS = -g -Wall -pedantic -LIBS = -lm -lgdbm +CFLAGS = -O2 -DBERKELEYDB -s +LIBS = -lm -ldb-3 ############################################################################ -OBJ = dbm_prog.c - all: lexicon lexicon: text710.edin dbm_prog - /bin/rm -i lexicon* ./dbm_prog -o lexicon -i text710.edin - touch lexicon -text710.edin: text710.dat read_ox_dict - read_ox_dict text710.dat > text710.edin +#text710.edin: text710.dat read_ox_dict +# read_ox_dict text710.dat > text710.edin -dbm_prog: $(OBJ) - $(CC) $(CFLAGS) -o $@ $(OBJ) $(LIBS) +dbm_prog: dbm_prog.c + $(CC) $(CFLAGS) -o $@ $< $(LIBS) clean: rm -f dbm_prog *.o core gmon.out lexicon --- a/lib/dbm_prog.c +++ b/lib/dbm_prog.c @@ -1,20 +1,26 @@ #include #include +#include #include #include #include #include -#ifdef FBSD_DATABASE + +#if defined(FBSD_DATABASE) #include +#elif defined(BERKELEYDB) +#include #else #include +#define RET_SUCCESS 0 +#define RET_SPECIAL 1 #endif #ifdef FBSD_DATABASE /* this needs to be explicit */ HASHINFO openinfo = { 128, /* bsize */ 8, /* ffactor */ - 75000, /* nelem */ + 155000, /* nelem */ 2048 * 1024, /* cachesize */ NULL, /* hash */ 0 /* lorder */ @@ -23,13 +29,12 @@ HASHINFO openinfo = { int main(int argc, char *argv[]) { -#ifdef FBSD_DATABASE +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) DBT inKey, inVal; DB * db; #else DBM * db; - datum Key; - datum Val; + datum inKey, inVal; #endif FILE *ifd; @@ -40,7 +45,7 @@ int main(int argc, char *argv[]) extern int optind; char c; - int i; + int ret, n = 0, i = 0; /* read in file name */ if(argc==1) { @@ -70,38 +75,55 @@ int main(int argc, char *argv[]) exit(1); } - -#ifdef FBSD_DATABASE +#if defined(BERKELEYDB) + db = dbopen(optr,O_RDWR|O_CREAT, 0000644, DB_HASH, NULL); +#elif defined(FBSD_DATABASE) db = dbopen(optr,O_RDWR|O_CREAT, 0000644, DB_HASH, &openinfo); +#else + db = dbm_open(optr,O_RDWR|O_CREAT, 0000644); +#endif while(fgets(line,256,ifd)) { +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) inKey.data = strtok(line," "); - inKey.size = strlen(inKey.data)+1; + inKey.size = strlen(inKey.data); inVal.data = strtok(NULL,"\n"); inVal.size = strlen(inVal.data)+1; - (db->put)(db,&inKey,&inVal,R_NOOVERWRITE); - - i++; - fprintf(stderr,"%d\r",i); - } - - (void)(db->close)(db); + ret = (db->put)(db,&inKey,&inVal,R_NOOVERWRITE); #else - db = dbm_open(optr,O_RDWR|O_CREAT, 0000644); - - while(fgets(line,256,ifd)) { - Key.dptr = strtok(line," "); - Key.dsize = strlen(Key.dptr)+1; - Val.dptr = strtok(NULL,"\n"); - Val.dsize = strlen(Val.dptr)+1; - dbm_store(db,Key,Val,DBM_INSERT); - - i++; - fprintf(stderr,"%d\r",i); + inKey.dptr = strtok(line," "); + inKey.dsize = strlen(Key.dptr); + inVal.dptr = strtok(NULL,"\n"); + inVal.dsize = strlen(Val.dptr)+1; + ret = dbm_store(db,Key,Val,DBM_INSERT); +#endif + n++; + switch(ret) { + case RET_SUCCESS: + i++; + break; + case RET_SPECIAL: + fprintf(stderr,"%s:%i: warning: Duplicate entry. Ignored.\n",iptr,n); + break; + default: +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + (void)(db->close)(db); +#else + dbm_close(db); +#endif + fprintf(stderr,"%s:%i: error: storing error\n",iptr,n); + fprintf(stderr,"%d words processed.\n",i); + exit(1); + } } - dbm_close(db); +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + (void)(db->close)(db); +#else + dbm_close(db); #endif + fprintf(stderr,"%d words processed.\n",i); + return(0); } --- a/src/Makefile +++ b/src/Makefile @@ -14,18 +14,18 @@ ############################################################################ # Sun users # CC = gcc -# CFLAGS = -g -Wall -pedantic +# CFLAGS = -O2 -s # LIBS = -lm ############################################################################ # FreeBSD users # CC = gcc -# CFLAGS = -O2 -Wall -pedantic +# CFLAGS = -O2 -DFBSD_DATABASE -s # LIBS = -lm ############################################################################ # Linux users CC = gcc -CFLAGS = -g -Wall -pedantic -LIBS = -lm -lgdbm +CFLAGS = -O2 -DBERKELEYDB -pedantic -Wall -s +LIBS = -lm -ldb ############################################################################ --- a/src/conv1.c +++ b/src/conv1.c @@ -40,19 +40,19 @@ export void conv1(CONFIG *config, LING_L sil_fill(j++,"L-H]",0.7,50,sent); buffer_add_str(slist," ##"); } else if(ling_list->text[i]->word[0] == ':') { - sil_fill(j++,"L-H]",0.7,750,sent); + sil_fill(j++,"L-H]",0.7,50,sent); buffer_add_str(slist," ##"); } else if(ling_list->text[i]->word[0] == ';') { - sil_fill(j++,"L-L]",0.7,750,sent); + sil_fill(j++,"L-L]",0.7,50,sent); buffer_add_str(slist," ##"); } else if(ling_list->text[i]->word[0] == '.') { - sil_fill(j++,"L-L]",1.0,1200,sent); + sil_fill(j++,"L-L]",1.0,50,sent); buffer_add_str(slist," ##"); } else if(ling_list->text[i]->word[0] == '!') { - sil_fill(j++,"L-L]",1.0,1200,sent); + sil_fill(j++,"L-L]",1.0,50,sent); buffer_add_str(slist," ##"); } else if(ling_list->text[i]->word[0] == '?') { - sil_fill(j++,"L-H]",1.0,1200,sent); + sil_fill(j++,"L-H]",1.0,50,sent); buffer_add_str(slist," ##"); } else { sil_fill(j++,"L-H]",0.7,200,sent); @@ -88,7 +88,7 @@ export void conv1(CONFIG *config, LING_L } } if(buffer_last_char(slist) != '#') { - sil_fill(j++,"",0.0,1200,sent); + sil_fill(j++,"",0.0,50,sent); buffer_add_str(slist," ##"); } sent->sil_sz = j; --- a/src/nrl_edin.c +++ b/src/nrl_edin.c @@ -15,7 +15,7 @@ */ #include "t2s.h" -static char *nrl_edin[][2] = { +static PKEY nrl_edin[] = { { " ", "| ",}, { "IY", "ee ",}, { "EY", "ai ",}, @@ -62,7 +62,7 @@ static char *nrl_edin[][2] = { { "w", "w ",}, { "r", "r ",}, { "JH", "j ",}, /* problem */ - { "","",}, + { "",""} }; @@ -74,12 +74,12 @@ export void nrl_edin_conv(char *str, cha while(*str) { i = 0; - while(nrl_edin[i][0] != '\0') { - if(!strncmp(str,nrl_edin[i][0],strlen(nrl_edin[i][0]))) + while(nrl_edin[i].keyword != "") { + if(!strncmp(str,nrl_edin[i].keyword,strlen(nrl_edin[i].keyword))) break; i++; } - strcat(str2,nrl_edin[i][1]); - str += strlen(nrl_edin[i][0]); + strcat(str2,nrl_edin[i].keyvalue); + str += strlen(nrl_edin[i].keyword); } } --- a/src/space.c +++ b/src/space.c @@ -11,12 +11,15 @@ #include #include -/* FreeBSD, and Linux? */ + #ifdef FBSD_DATABASE #include +#elif defined(BERKELEYDB) +#include #else #include #endif + #include int ft_endian_loc = 1; /* for deciding if we need to byte-swap */ @@ -50,7 +53,7 @@ export void init(CONFIG *config, BUFFER /* set up database if present */ if(strcmp("-",config->hash_file)) { -#ifdef FBSD_DATABASE +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) config->db = (void *)dbopen(config->hash_file,O_RDONLY, 0000644, DB_HASH, NULL); #else config->db = (void *)dbm_open(config->hash_file,O_RDONLY, 0000644); @@ -125,8 +128,8 @@ void terminate(CONFIG *config, BUFFER *b { if(config->db != NULL) -#ifdef FBSD_DATABASE - (void)(config->db->close)(config->db); +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + (void)(((DB *)(config->db))->close)((DB *)(config->db)); #else dbm_close(config->db); #endif --- a/src/transcribe.c +++ b/src/transcribe.c @@ -9,12 +9,15 @@ #include #include -/* FreeBSD, and Linux? */ + #ifdef FBSD_DATABASE #include +#elif defined(BERKELEYDB) +#include #else #include #endif + #include static char *lookup_db(char *word, CONFIG *config); @@ -54,21 +57,21 @@ export void transcribe(CONFIG *config, L } } -#ifdef FBSD_DATABASE +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) static char *lookup_db(char *word, CONFIG *config) { DBT inKey, inVal; inKey.data = word; - inKey.size = strlen(word)+1; + inKey.size = strlen(word); inVal.data = NULL; inVal.size = 0; if(config->db != NULL) { - (config->db->get)((DB *)config->db,&inKey,&inVal,0); + (((DB *)(config->db))->get)((DB *)(config->db),&inKey,&inVal,0); return(inVal.data); } else return(NULL);