master xplshn/aruu / cmd / posix / cut.c
  1/* See LICENSE file for copyright and license details. */
  2
  3#include <stdio.h>
  4#include <stdlib.h>
  5#include <string.h>
  6
  7#include "text.h"
  8#include "utf.h"
  9#include "util.h"
 10
 11typedef struct Range {
 12  size_t        min, max;
 13  struct Range *next;
 14} Range;
 15
 16static Range *list     = NULL;
 17static char   mode     = 0;
 18static char  *delim    = "\t";
 19static size_t delimlen = 1;
 20static int    nflag    = 0;
 21static int    sflag    = 0;
 22
 23static void
 24insert(Range *r)
 25{
 26  Range *l, *p, *t;
 27
 28  for (p = NULL, l = list; l; p = l, l = l->next) {
 29    if (r->max && r->max + 1 < l->min) {
 30      r->next = l;
 31      break;
 32    } else if (!l->max || r->min < l->max + 2) {
 33      l->min = MIN(r->min, l->min);
 34      for (p = l, t = l->next; t; p = t, t = t->next)
 35        if (r->max && r->max + 1 < t->min)
 36          break;
 37      l->max  = (p->max && r->max) ? MAX(p->max, r->max) : 0;
 38      l->next = t;
 39      return;
 40    }
 41  }
 42  if (p)
 43    p->next = r;
 44  else
 45    list = r;
 46}
 47
 48static void
 49parselist(char *str)
 50{
 51  char  *s;
 52  size_t n = 1;
 53  Range *r;
 54
 55  if (!*str)
 56    eprintf("empty list\n");
 57  for (s = str; *s; s++) {
 58    if (*s == ' ')
 59      *s = ',';
 60    if (*s == ',')
 61      n++;
 62  }
 63  r = ereallocarray(NULL, n, sizeof(*r));
 64  for (s = str; n; n--, s++) {
 65    r->min  = (*s == '-') ? 1 : strtoul(s, &s, 10);
 66    r->max  = (*s == '-') ? strtoul(s + 1, &s, 10) : r->min;
 67    r->next = NULL;
 68    if (!r->min || (r->max && r->max < r->min) || (*s && *s != ','))
 69      eprintf("bad list value\n");
 70    insert(r++);
 71  }
 72}
 73
 74static size_t
 75seek(struct line *s, size_t pos, size_t *prev, size_t count)
 76{
 77  size_t n = pos - *prev, i, j;
 78
 79  if (mode == 'b') {
 80    if (n >= s->len)
 81      return s->len;
 82    if (nflag)
 83      while (n && !UTF8_POINT(s->data[n]))
 84        n--;
 85    *prev += n;
 86    return n;
 87  } else if (mode == 'c') {
 88    for (n++, i = 0; i < s->len; i++)
 89      if (UTF8_POINT(s->data[i]) && !--n)
 90        break;
 91  } else {
 92    for (i = (count < delimlen + 1) ? 0 : delimlen; n && i < s->len;) {
 93      if ((s->len - i) >= delimlen && !memcmp(s->data + i, delim, delimlen)) {
 94        if (!--n && count)
 95          break;
 96        i += delimlen;
 97        continue;
 98      }
 99      for (j = 1; j + i <= s->len && !fullrune(s->data + i, j); j++)
100        ;
101      i += j;
102    }
103  }
104  *prev = pos;
105
106  return i;
107}
108
109static void
110cut(FILE *fp, const char *fname)
111{
112  Range             *r;
113  struct line        s;
114  static struct line line;
115  static size_t      size;
116  size_t             i, n, p;
117  ssize_t            len;
118
119  while ((len = getline(&line.data, &size, fp)) > 0) {
120    line.len = len;
121    if (line.data[line.len - 1] == '\n')
122      line.data[--line.len] = '\0';
123    if (mode == 'f' && !memmem(line.data, line.len, delim, delimlen)) {
124      if (!sflag) {
125        fwrite(line.data, 1, line.len, stdout);
126        fputc('\n', stdout);
127      }
128      continue;
129    }
130    for (i = 0, p = 1, s = line, r = list; r; r = r->next) {
131      n = seek(&s, r->min, &p, i);
132      s.data += n;
133      s.len -= n;
134      i += (mode == 'f') ? delimlen : 1;
135      if (!s.len)
136        break;
137      if (!r->max) {
138        fwrite(s.data, 1, s.len, stdout);
139        break;
140      }
141      n = seek(&s, r->max + 1, &p, i);
142      i += (mode == 'f') ? delimlen : 1;
143      if (fwrite(s.data, 1, n, stdout) != n)
144        eprintf("fwrite <stdout>:");
145      s.data += n;
146      s.len -= n;
147    }
148    putchar('\n');
149  }
150  if (ferror(fp))
151    eprintf("getline %s:", fname);
152}
153
154static void
155usage(void)
156{
157  eprintf(
158      "usage: %s -b list [-n] [file ...]\n"
159      "       %s -c list [file ...]\n"
160      "       %s -f list [-d delim] [-s] [file ...]\n",
161      argv0,
162      argv0,
163      argv0
164  );
165}
166
167// ?man cut: cut out fields from lines
168// ?man arguments: -b list [file ...]
169// ?man print selected parts of lines from files
170int
171main(int argc, char *argv[])
172{
173  FILE *fp;
174  int   ret = 0;
175
176  ARGBEGIN
177  {
178    // ?man -b: specify block size or base directory
179    case 'b':
180    // ?man -c: print count or perform stdout action
181    case 'c':
182    // ?man -f:mode: force the operation
183    case 'f':
184      mode = ARGC();
185      parselist(EARGF(usage()));
186      break;
187    // ?man -d:str: specify directory
188    case 'd':
189      delim = EARGF(usage());
190      if (!*delim)
191        eprintf("empty delimiter\n");
192      delimlen = unescape(delim);
193      break;
194    // ?man -n: print line numbers or counts
195    case 'n':
196      nflag = 1;
197      break;
198    // ?man -s: silent mode or print summary
199    case 's':
200      sflag = 1;
201      break;
202    default:
203      usage();
204  }
205  ARGEND
206
207  if (!mode)
208    usage();
209
210  if (!argc)
211    cut(stdin, "<stdin>");
212  else {
213    for (; *argv; argc--, argv++) {
214      if (!strcmp(*argv, "-")) {
215        *argv = "<stdin>";
216        fp    = stdin;
217      } else if (!(fp = fopen(*argv, "r"))) {
218        weprintf("fopen %s:", *argv);
219        ret = 1;
220        continue;
221      }
222      cut(fp, *argv);
223      if (fp != stdin && fshut(fp, *argv))
224        ret = 1;
225    }
226  }
227
228  ret |= fshut(stdin, "<stdin>") | fshut(stdout, "<stdout>");
229
230  return ret;
231}