* Licensed under GPLv2 or later, see file LICENSE in this source tree.
*/
-/* BB_AUDIT SUSv3 _NOT_ compliant -- option -m is not currently supported. */
+/* BB_AUDIT SUSv3 compliant. */
/* http://www.opengroup.org/onlinepubs/007904975/utilities/wc.html */
/* Mar 16, 2003 Manuel Novoa III (mjn3@codepoet.org)
* 3) no checking of ferror on EOF returns
* 4) isprint() wasn't considered when word counting.
*
- * TODO:
- *
- * When locale support is enabled, count multibyte chars in the '-m' case.
- *
* NOTES:
*
* The previous busybox wc attempted an optimization using stat for the
*
* for which 'wc -c' should output '0'.
*/
-
#include "libbb.h"
+#include "unicode.h"
#if !ENABLE_LOCALE_SUPPORT
# undef isprint
# define COUNT_FMT "u"
#endif
+/* We support -m even when UNICODE_SUPPORT is off,
+ * we just don't advertise it in help text,
+ * since it is the same as -c in this case.
+ */
+
+//usage:#define wc_trivial_usage
+//usage: "[-c"IF_UNICODE_SUPPORT("m")"lwL] [FILE]..."
+//usage:
+//usage:#define wc_full_usage "\n\n"
+//usage: "Count lines, words, and bytes for each FILE (or stdin)\n"
+//usage: "\n -c Count bytes"
+//usage: IF_UNICODE_SUPPORT(
+//usage: "\n -m Count characters"
+//usage: )
+//usage: "\n -l Count newlines"
+//usage: "\n -w Count words"
+//usage: "\n -L Print longest line length"
+//usage:
+//usage:#define wc_example_usage
+//usage: "$ wc /etc/passwd\n"
+//usage: " 31 46 1365 /etc/passwd\n"
+
+/* Order is important if we want to be compatible with
+ * column order in "wc -cmlwL" output:
+ */
enum {
- WC_LINES = 0,
- WC_WORDS = 1,
- WC_CHARS = 2,
- WC_LENGTH = 3
+ WC_LINES = 0, /* -l */
+ WC_WORDS = 1, /* -w */
+ WC_UNICHARS = 2, /* -m */
+ WC_BYTES = 3, /* -c */
+ WC_LENGTH = 4, /* -L */
+ NUM_WCS = 5,
};
int wc_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
const char *start_fmt = " %9"COUNT_FMT + 1;
const char *fname_fmt = " %s\n";
COUNT_T *pcounts;
- COUNT_T counts[4];
- COUNT_T totals[4];
+ COUNT_T counts[NUM_WCS];
+ COUNT_T totals[NUM_WCS];
int num_files;
smallint status = EXIT_SUCCESS;
unsigned print_type;
- print_type = getopt32(argv, "lwcL");
+ init_unicode();
+
+ print_type = getopt32(argv, "lwmcL");
if (print_type == 0) {
- print_type = (1 << WC_LINES) | (1 << WC_WORDS) | (1 << WC_CHARS);
+ print_type = (1 << WC_LINES) | (1 << WC_WORDS) | (1 << WC_BYTES);
}
argv += optind;
pcounts = counts;
num_files = 0;
- while ((arg = *argv++) != 0) {
+ while ((arg = *argv++) != NULL) {
FILE *fp;
const char *s;
unsigned u;
linepos = 0;
in_word = 0;
- do {
+ while (1) {
int c;
/* Our -w doesn't match GNU wc exactly... oh well */
- ++counts[WC_CHARS];
c = getc(fp);
if (c == EOF) {
if (ferror(fp)) {
bb_simple_perror_msg(arg);
status = EXIT_FAILURE;
}
- --counts[WC_CHARS];
- goto DO_EOF; /* Treat an EOF as '\r'. */
+ goto DO_EOF; /* Treat an EOF as '\r'. */
}
- if (isprint_asciionly(c)) {
+
+ /* Cater for -c and -m */
+ ++counts[WC_BYTES];
+ if (unicode_status != UNICODE_ON /* every byte is a new char */
+ || (c & 0xc0) != 0x80 /* it isn't a 2nd+ byte of a Unicode char */
+ ) {
+ ++counts[WC_UNICHARS];
+ }
+
+ if (isprint_asciionly(c)) { /* FIXME: not unicode-aware */
++linepos;
if (!isspace(c)) {
in_word = 1;
*/
if (c == '\t') {
linepos = (linepos | 7) + 1;
- } else { /* '\n', '\r', '\f', or '\v' */
+ } else { /* '\n', '\r', '\f', or '\v' */
DO_EOF:
if (linepos > counts[WC_LENGTH]) {
counts[WC_LENGTH] = linepos;
if (c == EOF) {
break;
}
- } while (1);
+ }
+
+ fclose_if_not_stdin(fp);
if (totals[WC_LENGTH] < counts[WC_LENGTH]) {
totals[WC_LENGTH] = counts[WC_LENGTH];
}
totals[WC_LENGTH] -= counts[WC_LENGTH];
- fclose_if_not_stdin(fp);
-
OUTPUT:
/* coreutils wc tries hard to print pretty columns
- * (saves results for all files, find max col len etc...)
+ * (saves results for all files, finds max col len etc...)
* we won't try that hard, it will bloat us too much */
s = start_fmt;
u = 0;
s = " %9"COUNT_FMT; /* Ok... restore the leading space. */
}
totals[u] += pcounts[u];
- } while (++u < 4);
+ } while (++u < NUM_WCS);
printf(fname_fmt, arg);
}
* effect of trashing the totals array after outputting it, but that's
* irrelavent since we no longer need it. */
if (num_files > 1) {
- num_files = 0; /* Make sure we don't get here again. */
+ num_files = 0; /* Make sure we don't get here again. */
arg = "total";
pcounts = totals;
--argv;