Date: Fri, 5 Feb 2010 04:51:30 GMT From: Gabor Kovesdan <gabor@FreeBSD.org> To: Perforce Change Reviews <perforce@freebsd.org> Subject: PERFORCE change 174343 for review Message-ID: <201002050451.o154pU3l079670@repoman.freebsd.org>
next in thread | raw e-mail | index | archive | help
http://p4web.freebsd.org/chv.cgi?CH=174343 Change 174343 by gabor@gabor_aspire on 2010/02/05 04:51:22 - Remove custom merge sort; use qsort() instead of reinventing the wheel - Remove swapping to disk; we have to choose between performance or memory consumption and we prefer the former - Make it WARNS=6 clean Affected files ... .. //depot/projects/soc2008/gabor_textproc/newsort/Makefile#2 edit .. //depot/projects/soc2008/gabor_textproc/newsort/coll.c#11 edit .. //depot/projects/soc2008/gabor_textproc/newsort/file.c#4 edit .. //depot/projects/soc2008/gabor_textproc/newsort/msort.c#3 delete .. //depot/projects/soc2008/gabor_textproc/newsort/sort.c#5 edit .. //depot/projects/soc2008/gabor_textproc/newsort/sort.h#3 edit Differences ... ==== //depot/projects/soc2008/gabor_textproc/newsort/Makefile#2 (text+ko) ==== @@ -1,7 +1,9 @@ PROG= sort -SRCS= coll.c file.c mem.c msort.c sort.c +SRCS= coll.c file.c mem.c sort.c + +WARNS= 6 -CFLAGS+= -std=c99 -Wall -pedantic +CFLAGS+= --param max-inline-insns-single=64 .if !defined(WITHOUT_NLS) NLS+= hu_HU.ISO8859-2 ==== //depot/projects/soc2008/gabor_textproc/newsort/coll.c#11 (text+ko) ==== @@ -107,7 +107,7 @@ *preproc(const wchar_t *s) { wchar_t *ret, *sp, *ep; size_t len; - int i; + unsigned int i; if (kflag && (sfield > 1)) { sp = wcschr(s, field_sep); @@ -150,11 +150,13 @@ * the underlying collate functions, which done the actual comparison. */ int -coll(const wchar_t *s1, const wchar_t *s2) { - wchar_t *ps1, *ps2; +coll(const void *s1, const void *s2) { + const wchar_t *const *ss1 = (const wchar_t * const *)s1; + const wchar_t *const *ss2 = (const wchar_t * const *)s2; + wchar_t *ps1, *ps2; - ps1 = preproc(s1); - ps2 = preproc(s2); + ps1 = preproc(*ss1); + ps2 = preproc(*ss2); if (rflag) { wchar_t *tmp = ps1; @@ -324,7 +326,7 @@ len = wcslen(s); ls = sort_malloc(sizeof(wint_t) * (len + 1)); - for (int i = 0; i < len; i++) + for (unsigned int i = 0; i < len; i++) ls[i] = towlower(s[i]); if (months == NULL) { @@ -340,7 +342,7 @@ continue; } m[len] = L'\0'; - for (int j = 0; j < len; j++) + for (unsigned int j = 0; j < len; j++) m[j] = towlower(m[j]); months[i] = m; } ==== //depot/projects/soc2008/gabor_textproc/newsort/file.c#4 (text+ko) ==== @@ -37,27 +37,11 @@ #include "sort.h" /* - * Prints the internal buffer to a file or to stdout if - * "-" is given as output filename. - */ -void -print(char *fn) { - FILE *file = NULL; - - file = openfile(fn, "w"); - - for (int i = 0; i < count; i++) { - fprintf(file, "%ls%s", list[i], zflag ? "\0" : "\n"); - } - fclose(file); -} - -/* * Checks if the given file is sorted. Stops at the first disorder, * prints the disordered line and returns 1. */ int -check(char *fn) { +check(const char *fn) { int pos = 1; wchar_t *s1, *s2, *tmp; FILE *file; @@ -94,19 +78,10 @@ * opened. */ FILE -*openfile(char *fn, char *mode) { +*openfile(const char *fn, const char *mode) { FILE *file; - int len, fd; - if (fn == NULL) { - tempfiles = sort_realloc(tempfiles, sizeof(char *) * ++tempno); - len = strlen(tmpdir) + strlen(TMPPAT); - tempfiles[tempno - 1] = sort_malloc(sizeof(char) * (len + 1)); - sprintf(tempfiles[tempno - 1], "%s" TMPPAT, tmpdir); - fd = mkstemp(tempfiles[tempno - 1]); - if ((file = fdopen(fd, "w")) == NULL) - err(2, NULL); - } else if (strcmp(fn, "-") == 0) { + if (strcmp(fn, "-") == 0) { return (stdout); } else { if((file = fopen(fn, mode)) == NULL) @@ -132,35 +107,12 @@ list = sort_realloc(list, sizeof(struct qentry*) * ++listlen); } list[count++] = item; - if ((bufcnt -= sizeof(item)) < 0) - return (1); } fclose (file); return (0); } /* - * Calculates the optimal buffer size. - */ -int -optbufsize(char *fn) { - struct stat st; - - if (bufsiz > 0) - return (bufsiz); - - if (stat(fn, &st) != 0) - err(2, NULL); - - if (st.st_size/OPT_SLICES < OPT_BUFSIZE) - return (OPT_BUFSIZE); - else if (st.st_size/OPT_SLICES > MAX_BUFSIZE) - return (MAX_BUFSIZE); - else - return (st.st_size/OPT_SLICES); -} - -/* * Prints the content of f1 to f2, which can also be * stdout. */ @@ -183,7 +135,7 @@ * stdout. */ void -merge_files(char *fn1, char *fn2, char *fn3) { +merge_files(const char *fn1, const char *fn2, const char *fn3) { FILE *f1, *f2, *f3; wchar_t *s1, *s2, *tmp; size_t len; ==== //depot/projects/soc2008/gabor_textproc/newsort/sort.c#5 (text+ko) ==== @@ -51,7 +51,7 @@ * Default messages to use when NLS is disabled or no catalogue * is found. */ -char *nlsstr[] = { +const char *nlsstr[] = { "", /* 1*/ "you cannot use -%c and -%c together", /* 2*/ "extra argument not allowed with -c" @@ -60,11 +60,9 @@ extern char *__progname; -int count, bufcnt, listlen = 0, tempno = 0; +int count, listlen = 0; wchar_t **list; wchar_t field_sep = L' '; -char **tempfiles; -char *tmpdir = "/var/tmp"; unsigned long sfield, lfield; unsigned long long bufsiz = 0; bool cflag; @@ -100,7 +98,7 @@ {"reverse", no_argument, NULL, 'r'}, {"buffer-size", required_argument, NULL, 'S'}, {"field-separator", required_argument, NULL, 't'}, - {"temporary-directory", required_argument, NULL, 'T'}, +// {"temporary-directory", required_argument, NULL, 'T'}, {"unique", no_argument, NULL, 'u'}, {"version", no_argument, NULL, 'V'}, {"zero-terminated", no_argument, NULL, 'z'}, @@ -118,22 +116,12 @@ main(int argc, char *argv[]) { char c; char *sptr; - char *outfile = "-"; - struct stat st; + char *outfile; + + outfile = strdup("-"); setlocale(LC_ALL, ""); - if ((sptr = getenv("TMPDIR")) != NULL) { - stat(sptr, &st); - - if (!S_ISDIR(st.st_mode)) { - errno = ENOTDIR; - err(2, NULL); - } - - tmpdir = sptr; - } - while (((c = getopt_long(argc, argv, OPTIONS, long_options, NULL)) != -1)) { switch (c) { case 'b': @@ -226,17 +214,6 @@ printf("field_sep = %lc\n", field_sep); } break; - case 'T': - stat(optarg, &st); - - if (!S_ISDIR(st.st_mode)) { - errno = ENOTDIR; - err(2, NULL); - } - - tmpdir = sort_malloc(sizeof(char) * (strlen(optarg) + 2)); - strlcpy(tmpdir, optarg, strlen(optarg) + 1); - break; case 'u': uflag = true; break; @@ -287,44 +264,15 @@ FILE *file; file = openfile(*argv, "r"); - bufcnt = optbufsize(*argv); - while (procfile(file)) { - list = merge_sort(list, count); - print(NULL); - file = openfile(*argv, "r"); - for (int i = 0; i < count; i++) - free(list[i]); - free(list); - count = 0; - } - list = merge_sort(list, count); - print(NULL); + procfile(file); --argc; ++argv; } - if (tempno == 0) { - list = merge_sort(list, count); - print(outfile); - } else { - for (int i = 0; i < tempno;) { - if (i == (tempno - 2)) { - merge_files(tempfiles[i], tempfiles[i + 1], "-"); - unlink(tempfiles[i]); - unlink(tempfiles[i + 1]); - break; - } else if ( i == (tempno - 1)) { - printfile(openfile(tempfiles[i], "r"), stdout); - unlink(tempfiles[i]); - break; - } else { - merge_files(tempfiles[i], tempfiles[i + 1], NULL); - unlink(tempfiles[i]); - unlink(tempfiles[i + 1]); - i += 2; - } - } - } + qsort(list, count, sizeof(wchar_t *), coll); + + for (int i = 0; i < count; i++) + printf("%ls\n", list[i]); return (0); } else if (cflag) { ==== //depot/projects/soc2008/gabor_textproc/newsort/sort.h#3 (text+ko) ==== @@ -43,7 +43,7 @@ #define getstr(n) catgets(catalog, 1, n, nlsstr[n]) #endif -extern char *nlsstr[]; +extern const char *nlsstr[]; /* * sort.c @@ -59,16 +59,14 @@ /* * coll.c */ -int coll(const wchar_t *, const wchar_t *); +int coll(const void *, const void *); /* * file.c */ -int check(char *); -void merge_files(char *, char *, char *); -FILE *openfile(char *, char *); -int optbufsize(char *); -void print(char *); +int check(const char *); +void merge_files(const char *, const char *, const char *); +FILE *openfile(const char *, const char *); void printfile(FILE *, FILE *); int procfile(FILE *);
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201002050451.o154pU3l079670>