xref: /third_party/toybox/toys/posix/wc.c (revision 0f66f451) 
1/* wc.c - Word count
2 *
3 * Copyright 2011 Rob Landley <rob@landley.net>
4 *
5 * See http://opengroup.org/onlinepubs/9699919799/utilities/wc.html
6
7USE_WC(NEWTOY(wc, "mcwl", TOYFLAG_USR|TOYFLAG_BIN|TOYFLAG_LOCALE))
8
9config WC
10  bool "wc"
11  default y
12  help
13    usage: wc -lwcm [FILE...]
14
15    Count lines, words, and characters in input.
16
17    -l	Show lines
18    -w	Show words
19    -c	Show bytes
20    -m	Show characters
21
22    By default outputs lines, words, bytes, and filename for each
23    argument (or from stdin if none). Displays only either bytes
24    or characters.
25*/
26
27#define FOR_wc
28#include "toys.h"
29
30GLOBALS(
31  unsigned long totals[4];
32)
33
34static void show_lengths(unsigned long *lengths, char *name)
35{
36  int i, space = 0, first = 1;
37
38  // POSIX says there should never be leading spaces, but accepts that
39  // traditional implementations use 7 spaces, unless only one file (or
40  // just stdin) is being counted, when there should be no leading spaces,
41  // *except* for the case where we're going to output multiple numbers.
42  // And, yes, folks have test scripts that rely on all this nonsense :-(
43  // Note: sufficiently modern versions of coreutils wc will use the smallest
44  // column width necessary to have all columns be equal width rather than 0.
45  if (!(!toys.optc && !(toys.optflags & (toys.optflags-1))) && toys.optc!=1)
46    space = 7;
47
48  for (i = 0; i<4; i++) {
49    if (toys.optflags&(1<<i)) {
50      printf(" %*ld"+first, space, lengths[i]);
51      first = 0;
52    }
53    TT.totals[i] += lengths[i];
54  }
55  if (*toys.optargs) printf(" %s", name);
56  xputc('\n');
57}
58
59static void do_wc(int fd, char *name)
60{
61  int len = 0, clen = 1, space = 0;
62  unsigned long word = 0, lengths[] = {0,0,0,0};
63
64  // Speed up common case: wc -c normalfile is file length.
65  if (toys.optflags == FLAG_c) {
66    struct stat st;
67
68    // On Linux, files in /proc often report their size as 0.
69    if (!fstat(fd, &st) && S_ISREG(st.st_mode) && st.st_size) {
70      lengths[2] = st.st_size;
71      goto show;
72    }
73  }
74
75  for (;;) {
76    int pos, done = 0, len2 = read(fd, toybuf+len, sizeof(toybuf)-len);
77    unsigned wchar;
78
79    if (len2<0) perror_msg_raw(name);
80    else len += len2;
81    if (len2<1) done++;
82
83    for (pos = 0; pos<len; pos++) {
84      if (toybuf[pos]=='\n') lengths[0]++;
85      lengths[2]++;
86      if (FLAG(m)) {
87        // If we've consumed next wide char
88        if (--clen<1) {
89          // next wide size, don't count invalid, fetch more data if necessary
90          clen = utf8towc(&wchar, toybuf+pos, len-pos);
91          if (clen == -1) continue;
92          if (clen == -2 && !done) break;
93
94          lengths[3]++;
95          space = iswspace(wchar);
96        }
97      } else space = isspace(toybuf[pos]);
98
99      if (space) word=0;
100      else {
101        if (!word) lengths[1]++;
102        word=1;
103      }
104    }
105    if (done) break;
106    if (pos != len) memmove(toybuf, toybuf+pos, len-pos);
107    len -= pos;
108  }
109
110show:
111  show_lengths(lengths, name);
112}
113
114void wc_main(void)
115{
116  if (!toys.optflags) toys.optflags = FLAG_l|FLAG_w|FLAG_c;
117  loopfiles(toys.optargs, do_wc);
118  if (toys.optc>1) show_lengths(TT.totals, "total");
119}
120