From: "Igor B. Poretsky" Date: Wed, 15 Feb 2023 14:25:38 +0300 Subject: New lexical database holding utility --- lib/Makefile | 50 ++++++---- lib/lexholder.c | 271 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 301 insertions(+), 20 deletions(-) create mode 100644 lib/lexholder.c --- a/lib/Makefile +++ b/lib/Makefile @@ -10,41 +10,51 @@ # # ############################################################################ + +# Installation paths +prefix = /usr/local +bindir = ${prefix}/bin +datadir = ${prefix}/share/freespeech +docdir = ${prefix}/share/doc/enlex-data + +# Installed database filename +DBF = enlex.db + +# Installed utility name +UTILITY = lexholder-en + # Select the appropriate Makefile options for your system. then type make. ############################################################################ -# Sun users -# CC = gcc -# CFLAGS = -O2 -s -# LIBS = -lm -############################################################################ # FreeBSD users # CC = gcc -# CFLAGS = -O2 -DFBSD_DATABASE -s +# CFLAGS = -O2 -DFBSD_DATABASE # LIBS = -lm ############################################################################ # Linux users CC = gcc -CFLAGS = -O2 -DBERKELEYDB -s -LIBS = -lm -ldb-3 +CFLAGS = -g -Wall -O2 -DBERKELEYDB +LIBS = -ldb ############################################################################ -all: lexicon +all: lexicon -lexicon: text710.edin dbm_prog - ./dbm_prog -o lexicon -i text710.edin +lexicon: text710.edin lexholder + ./lexholder -qf $< $@ #text710.edin: text710.dat read_ox_dict # read_ox_dict text710.dat > text710.edin -dbm_prog: dbm_prog.c - $(CC) $(CFLAGS) -o $@ $< $(LIBS) - -clean: - rm -f dbm_prog *.o core gmon.out lexicon +lexholder: lexholder.c + $(CC) $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) -o $@ $< $(LIBS) -distribution: - echo "This needs written." - -# dependencies +.PHONY: install +install: lexholder lexicon + install -d ${bindir} ${datadir} ${docdir} + install -m 0755 -p lexholder ${bindir}/${UTILITY} + install -m 0644 -p lexicon ${datadir}/${DBF} + install -m 0644 -p text710.doc ${docdir} +.PHONY: clean +clean: + rm -f lexholder *.o lexicon --- /dev/null +++ b/lib/lexholder.c @@ -0,0 +1,271 @@ +/* lexholder.c + * + * Lexicon database holding utility. + */ + +/* + * Copyright (C) 2010 Igor B. Poretsky + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + + +#include +#include +#include +#include +#include +#include + +#ifdef FBSD_DATABASE +#include +#elif defined(BERKELEYDB) +#include +#else +#include +#endif + +#include + + +/* Command line error codes */ +#define CL_OK 0 +#define CS_CONFLICT 1 +#define NO_DB_FILE 2 + + +static const char *usage = + "Lexical database holding utility.\n\n" + + "This utility is designed for constructing, managing, testing and querying\n" + "lexical database providing pronunciation info for English words.\n\n" + + "Usage:\t%s [options] \n\n" + + "When filling and updating the database, new records are read\n" + "from the standard input. When extracting data from the database\n" + "the result is printed to the standard output.\n" + "This behaviour can be changed by the \"-f\" switch.\n\n" + + "If no options are specified, the program reads it's standard input\n" + "and stores it's content in the database.\n\n" + + "The recognized options are as follows:\n\n" + + "-h -- Print this help (the only option not requiring the database path)\n" + "-l -- List database content\n" + "-s -- Search specified word\n" + "-d -- Delete record for specified word\n" + "-f -- Use specified file instead of standard input or output\n" + "-r -- Replace mode\n" + "-q -- Don't print messages about ignoring duplicates\n\n"; + + +int main(int argc, char *argv[]) +{ + FILE *fp = NULL; + void *db; +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + DBT key, value; +#else + datum key, value; +#endif + char line[256]; + char *f = NULL, *d = NULL, *s = NULL; + int i, n, q = 0, ret = NO_DB_FILE; +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + int mode = R_NOOVERWRITE; +#else + int mode = GDBM_INSERT; +#endif + + /* Parse command line */ + if(argc==1) + { + (void)fprintf(stderr, usage, argv[0]); + return EXIT_FAILURE; + } + while((n = getopt(argc,argv,"f:s:d:lqrh")) != -1) + switch(n) + { + case 'f': + if (strcmp(optarg, "-")) + f = optarg; + break; + case 'd': + if (d || s) ret = CS_CONFLICT; + else d = optarg; + break; + case 's': + if (d || s) ret = CS_CONFLICT; + else s = optarg; + break; + case 'l': + if (d || s) ret = CS_CONFLICT; + else s = line; + break; + case 'r': +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + mode = 0; +#else + mode = GDBM_REPLACE; +#endif + break; + case 'q': + q = 1; + break; + case 'h': + (void)fprintf(stderr, usage, argv[0]); + return EXIT_SUCCESS; + default: + (void)fprintf(stderr, usage, argv[0]); + return EXIT_FAILURE; + } + if (optind && argv[optind]) + ret = CL_OK; + switch (ret) + { + case CS_CONFLICT: + (void)fprintf(stderr, + "Ambiguous options in command line\n"); + return EXIT_FAILURE; + case NO_DB_FILE: + (void)fprintf(stderr, "DB file must be specified\n"); + return EXIT_FAILURE; + default: + break; + } + + /* Open file for input or output if necessary */ + if (!d) + { + if (f) + fp = fopen(f, s ? "w" : "r"); + else fp = s ? stdout : stdin; + if (!fp) + { + (void)fprintf(stderr, "Can't open file %s\n", f); + return EXIT_FAILURE; + } + } + + /* Open the database in appropriate mode */ +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + db = dbopen(argv[optind], s ? O_RDONLY : (O_RDWR | O_CREAT), 0644, DB_HASH, NULL); +#else + db = dbm_open(argv[optind], s ? O_RDONLY : (O_RDWR | O_CREAT), 0644); +#endif + if (!db) + { + (void)fprintf(stderr, "Cannot open the database %s\n", argv[optind]); + return EXIT_FAILURE; + } + + ret = EXIT_SUCCESS; + if (s == line) /* List database content */ + { +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + for (n = 0; !((DB*)db)->seq(db, &key, &value, R_NEXT); n++) +#else + for (n = 0, key = dbm_firstkey(db); key.dptr; key = dbm_nextkey(db), n++) +#endif + { +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + (void)strncpy(line, key.data, key.size); + (void)fprintf(fp, "%s %s\n", line, (char *)value.data); +#else + value = dbm_fetch(db, key); + (void)strncpy(line, key.dptr, key.dsize); + (void)fprintf(fp, "%s %s\n", line, (char *)value.dptr); +#endif + } + (void)fprintf(stderr, "%i records extracted.\n", n); + } + else if (s) /* Lookup database for specified word */ + { +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + key.data = s; + key.size = strlen(s); + if (((DB*)db)->get(db, &key, &value, 0)) + ret = EXIT_FAILURE; + else (void)fprintf(fp, "%s\n", (char *)value.data); +#else + key.dptr = s; + key.dsize = strlen(s)+1; + value = dbm_fetch(db, key); + if (!value.dptr) + ret = EXIT_FAILURE; + else (void)fprintf(fp, "%s\n", (char *)value.dptr); +#endif + } + else if (d) /* Delete record for specified word */ + { +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + key.data = d; + key.size = strlen(d); + ret = ((DB*)db)->del(db, &key, 0) ? EXIT_FAILURE : EXIT_SUCCESS; +#else + key.dptr = d; + key.dsize = strlen(d)+1; + dbm_delete(db, key); +#endif + } + else /* Fill the database */ + { + for (i = 0, n = 1; fgets(line, 256, fp); n++) + { +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + key.data = strtok(line," "); + key.size = strlen(key.data); + value.data = strtok(NULL,"\n"); + value.size = strlen(value.data) + 1; + ret = ((DB*)db)->put(db, &key, &value, mode); +#else + key.dptr = strtok(line," "); + key.dsize = strlen(key.dptr)+1; + value.dptr = strtok(NULL,"\n"); + value.dsize = strlen(value.dptr) + 1; + ret = dbm_store(db, key, value, mode); +#endif + if (ret < 0) + break; + else if (ret) + { + if (!q) + (void)fprintf(stderr, "%s:%i: warning: Duplicate entry. Ignored.\n", + f ? f : "stdin", n); + i++; + } + } + if (ret < 0) + { + (void)fprintf(stderr, "%s:%i: error: Database insertion failure. Exiting now.\n", + f ? f : "stdin", n); + ret = EXIT_FAILURE; + } + else + { + (void)fprintf(stderr, "%i words processed.\n", --n); + (void)fprintf(stderr, "%i duplicates ignored.\n", i); + (void)fprintf(stderr, "%i records put into the database.\n", n - i); + ret = EXIT_SUCCESS; + } + } + + /* All done */ +#if defined(FBSD_DATABASE) || defined(BERKELEYDB) + (void)((DB*)db)->close(db); +#else + dbm_close(db); +#endif + (void)fclose(fp); + return ret; +}