From 7347305730fb9cc87bee02744c06fbc2089b1680 Mon Sep 17 00:00:00 2001 From: "yuta.256" Date: Sat, 12 Jul 2008 23:38:41 +0000 Subject: Rewrote bwt.c. --- examples/bwt.c | 208 ++++++++++++++++++++++++++++++++++++++------------------- 1 file changed, 138 insertions(+), 70 deletions(-) diff --git a/examples/bwt.c b/examples/bwt.c index 842712d..5a362d0 100644 --- a/examples/bwt.c +++ b/examples/bwt.c @@ -24,129 +24,197 @@ * OTHER DEALINGS IN THE SOFTWARE. */ -#ifdef HAVE_CONFIG_H +#if HAVE_CONFIG_H # include "config.h" #endif -#include #include +#if HAVE_STRING_H +# include +#endif #if HAVE_STDLIB_H # include #endif -#if HAVE_STRING_H -# if !STDC_HEADERS && HAVE_MEMORY_H -# include -# endif -# include +#if HAVE_MEMORY_H +# include +#endif +#if HAVE_STDDEF_H +# include #endif #if HAVE_STRINGS_H # include #endif +#if HAVE_SYS_TYPES_H +# include +#endif +#if HAVE_IO_H && HAVE_FCNTL_H +# include +# include +#endif #include +#include +#include "lfs.h" static -saidx_t -_str2size(const char *str) { - saidx_t s[3]; - saidx_t t; - int i, c; - for(i = 0, t = s[0] = s[1] = s[2] = 0; (c = str[i]) != '\0'; ++i) { - if(('0' <= c) && (c <= '9')) { - t = (t * 10) + (c - '0'); - } else { - switch(c) { - case 'm': - case 'M': - s[0] += t << 20; - break; - case 'k': - case 'K': - s[1] += t << 10; - break; - case 'b': - case 'B': - s[2] += t; - break; - } - t = 0; - } - } - return s[0] + s[1] + s[2]; +size_t +write_int(FILE *fp, saidx_t n) { + unsigned char c[4]; + c[0] = (unsigned char)((n >> 0) & 0xff), c[1] = (unsigned char)((n >> 8) & 0xff), + c[2] = (unsigned char)((n >> 16) & 0xff), c[3] = (unsigned char)((n >> 24) & 0xff); + return fwrite(c, sizeof(unsigned char), 4, fp); +} + +static +void +print_help(const char *progname, int status) { + fprintf(stderr, + "bwt, a burrows-wheeler transform program, version %s.\n", + divsufsort_version()); + fprintf(stderr, "usage: %s [-b num] INFILE OUTFILE\n", progname); + fprintf(stderr, " -b num set block size to num MiB [1..512] (default: 32)\n\n"); + exit(status); } int main(int argc, const char *argv[]) { + FILE *fp, *ofp; + const char *fname, *ofname; sauchar_t *T; saidx_t *SA; - saidx_t m, n, blocksize, idx; + LFS_OFF_T n; + size_t m; + saidx_t pidx; clock_t start,finish; + saint_t i, blocksize = 32, needclose = 3; - /* Check argument. */ - if(((argc != 1) && (argc != 2)) || - ((argc == 2) && (strcmp(argv[1], "-h") == 0)) || - ((argc == 2) && (strcmp(argv[1], "--help") == 0))) { - fprintf(stderr, - "bwt, a burrows-wheeler transform program, version %s.\n" - , divsufsort_version()); - fprintf(stderr, - "usage: %s [BLOCKSIZE] < STDIN > STDOUT\n\n" - , argv[0]); - exit(EXIT_FAILURE); + /* Check arguments. */ + if((argc == 1) || + (strcmp(argv[1], "-h") == 0) || + (strcmp(argv[1], "--help") == 0)) { print_help(argv[0], EXIT_SUCCESS); } + if((argc != 3) && (argc != 5)) { print_help(argv[0], EXIT_FAILURE); } + i = 1; + if(argc == 5) { + if(strcmp(argv[i], "-b") != 0) { print_help(argv[0], EXIT_FAILURE); } + blocksize = atoi(argv[i + 1]); + if(blocksize < 0) { blocksize = 1; } + else if(512 < blocksize) { blocksize = 512; } + i += 2; + } + blocksize <<= 20; + + /* Open a file for reading. */ + if(strcmp(argv[i], "-") != 0) { +#if HAVE_FOPEN_S + if(fopen_s(&fp, fname = argv[i], "rb") != 0) { +#else + if((fp = LFS_FOPEN(fname = argv[i], "rb")) == NULL) { +#endif + fprintf(stderr, "%s: Cannot open file `%s': ", argv[0], fname); + perror(NULL); + exit(EXIT_FAILURE); + } + } else { +#if HAVE__SETMODE && HAVE__FILENO + if(_setmode(_fileno(stdin), _O_BINARY) == -1) { + fprintf(stderr, "%s: Cannot set mode: ", argv[0]); + perror(NULL); + exit(EXIT_FAILURE); + } +#endif + fp = stdin; + fname = "stdin"; + needclose ^= 1; } - blocksize = (argc == 2) ? _str2size(argv[1]) : 0; - if(blocksize <= 0) { - fseek(stdin, 0, SEEK_END); - blocksize = ftell(stdin); - if(blocksize < 0) { blocksize = BUFSIZ; } - rewind(stdin); + i += 1; + + /* Open a file for writing. */ + if(strcmp(argv[i], "-") != 0) { +#if HAVE_FOPEN_S + if(fopen_s(&ofp, ofname = argv[i], "wb") != 0) { +#else + if((ofp = LFS_FOPEN(ofname = argv[i], "wb")) == NULL) { +#endif + fprintf(stderr, "%s: Cannot open file `%s': ", argv[0], ofname); + perror(NULL); + exit(EXIT_FAILURE); + } + } else { +#if HAVE__SETMODE && HAVE__FILENO + if(_setmode(_fileno(stdout), _O_BINARY) == -1) { + fprintf(stderr, "%s: Cannot set mode: ", argv[0]); + perror(NULL); + exit(EXIT_FAILURE); + } +#endif + ofp = stdout; + ofname = "stdout"; + needclose ^= 2; } + /* Get the file size. */ + if(LFS_FSEEK(fp, 0, SEEK_END) == 0) { + n = LFS_FTELL(fp); + rewind(fp); + if(n < 0) { + fprintf(stderr, "%s: Cannot ftell `%s': ", argv[0], fname); + perror(NULL); + exit(EXIT_FAILURE); + } + if(0x20000000L < n) { n = 0x20000000L; } + if((blocksize == 0) || (n < blocksize)) { blocksize = (saidx_t)n; } + } else if(blocksize == 0) { blocksize = 32 << 20; } + /* Allocate 5blocksize bytes of memory. */ - if(((T = malloc(blocksize * sizeof(sauchar_t))) == NULL) || - ((SA = malloc(blocksize * sizeof(saidx_t))) == NULL)) { + T = (sauchar_t *)malloc(blocksize * sizeof(sauchar_t)); + SA = (saidx_t *)malloc(blocksize * sizeof(saidx_t)); + if((T == NULL) || (SA == NULL)) { fprintf(stderr, "%s: Cannot allocate memory.\n", argv[0]); exit(EXIT_FAILURE); } /* Write the blocksize. */ - if(fwrite(&blocksize, sizeof(saidx_t), 1, stdout) != 1) { - fprintf(stderr, "%s: Cannot write to `stdout': ", argv[0]); + if(write_int(ofp, blocksize) != 4) { + fprintf(stderr, "%s: Cannot write to `%s': ", argv[0], ofname); perror(NULL); exit(EXIT_FAILURE); } - fprintf(stderr, " BWT (blocksize %d) ... ", (int)blocksize); - start=clock(); - for(n = 0; 0 < (m = fread(T, sizeof(sauchar_t), blocksize, stdin)); n += m) { + fprintf(stderr, " BWT (blocksize %" PRIdSAINT_T ") ... ", blocksize); + start = clock(); + for(n = 0; 0 < (m = fread(T, sizeof(sauchar_t), blocksize, fp)); n += m) { /* Burrows-Wheeler Transform. */ - idx = divbwt(T, T, SA, m); - if(idx < 0) { + pidx = divbwt(T, T, SA, m); + if(pidx < 0) { fprintf(stderr, "%s (bw_transform): %s.\n", argv[0], - (idx == -1) ? "Invalid arguments" : "Cannot allocate memory"); + (pidx == -1) ? "Invalid arguments" : "Cannot allocate memory"); exit(EXIT_FAILURE); } /* Write the bwted data. */ - if((fwrite(&idx, sizeof(saidx_t), 1, stdout) != 1) || - (fwrite(T, sizeof(sauchar_t), m, stdout) != m)) { - fprintf(stderr, "%s: Cannot write to `stdout': ", argv[0]); + if((write_int(ofp, pidx) != 4) || + (fwrite(T, sizeof(sauchar_t), m, ofp) != m)) { + fprintf(stderr, "%s: Cannot write to `%s': ", argv[0], ofname); perror(NULL); exit(EXIT_FAILURE); } } - if(ferror(stdin)) { - fprintf(stderr, "%s: Cannot read from `stdin': ", argv[0]); + if(ferror(fp)) { + fprintf(stderr, "%s: Cannot read from `%s': ", argv[0], fname); perror(NULL); exit(EXIT_FAILURE); } finish = clock(); - fprintf(stderr, "%d bytes: %.4f sec\n", - (int)n, (double)(finish - start) / (double)CLOCKS_PER_SEC); + fprintf(stderr, "%" PRIdOFF_T " bytes: %.4f sec\n", + n, (double)(finish - start) / (double)CLOCKS_PER_SEC); + + /* Close files */ + if(needclose & 1) { fclose(fp); } + if(needclose & 2) { fclose(ofp); } /* Deallocate memory. */ - free(T); free(SA); + free(T); return 0; } -- cgit v1.2.3