1/* tr.c - translate or delete characters
2 *
3 * Copyright 2014 Sandeep Sharma <sandeep.jack2756@gmail.com>
4 *
5 * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/tr.html
6
7USE_TR(NEWTOY(tr, "^>2<1Ccsd[+cC]", TOYFLAG_USR|TOYFLAG_BIN))
8
9config TR
10  bool "tr"
11  default n
12  help
13    usage: tr [-cds] SET1 [SET2]
14
15    Translate, squeeze, or delete characters from stdin, writing to stdout
16
17    -c/-C  Take complement of SET1
18    -d     Delete input characters coded SET1
19    -s     Squeeze multiple output characters of SET2 into one character
20*/
21
22#define FOR_tr
23#include "toys.h"
24
25GLOBALS(
26  short map[256]; //map of chars
27  int len1, len2;
28)
29
30enum {
31  class_alpha, class_alnum, class_digit,
32  class_lower,class_upper,class_space,class_blank,
33  class_punct,class_cntrl,class_xdigit,class_invalid
34};
35
36static void map_translation(char *set1 , char *set2)
37{
38  int i = TT.len1, k = 0;
39
40  if (toys.optflags & FLAG_d)
41    for (; i; i--, k++) TT.map[set1[k]] = set1[k]|0x100; //set delete bit
42
43  if (toys.optflags & FLAG_s) {
44    for (i = TT.len1, k = 0; i; i--, k++)
45      TT.map[set1[k]] = TT.map[set1[k]]|0x200;
46    for (i = TT.len2, k = 0; i; i--, k++)
47      TT.map[set2[k]] = TT.map[set2[k]]|0x200;
48  }
49  i = k = 0;
50  while (!(toys.optflags & FLAG_d) && set2 && TT.len1--) { //ignore set2 if -d present
51    TT.map[set1[i]] = ((TT.map[set1[i]] & 0xFF00) | set2[k]);
52    if (set2[k + 1]) k++;
53    i++;
54  }
55}
56
57static int handle_escape_char(char **esc_val) //taken from printf
58{
59  char *ptr = *esc_val;
60  int esc_length = 0;
61  unsigned  base = 0, num = 0, result = 0, count = 0;
62
63  if (*ptr == 'x') {
64    ptr++;
65    esc_length++;
66    base = 16;
67  } else if (isdigit(*ptr)) base = 8;
68
69  while (esc_length < 3 && base) {
70    num = tolower(*ptr) - '0';
71    if (num > 10) num += ('0' - 'a' + 10);
72    if (num >= base) {
73      if (base == 16) {
74        esc_length--;
75        if (!esc_length) {// Invalid hex value eg. /xvd, print as it is /xvd
76          result = '\\';
77          ptr--;
78        }
79      }
80      break;
81    }
82    esc_length++;
83    count = result = (count * base) + num;
84    ptr++;
85  }
86  if (base) {
87    ptr--;
88    *esc_val = ptr;
89    return (char)result;
90  } else {
91    switch (*ptr) {
92      case 'n':  result = '\n'; break;
93      case 't':  result = '\t'; break;
94      case 'e':  result = (char)27; break;
95      case 'b':  result = '\b'; break;
96      case 'a':  result = '\a'; break;
97      case 'f':  result = '\f'; break;
98      case 'v':  result = '\v'; break;
99      case 'r':  result = '\r'; break;
100      case '\\': result = '\\'; break;
101      default :
102        result = '\\';
103        ptr--; // Let pointer pointing to / we will increment after returning.
104        break;
105    }
106  }
107  *esc_val = ptr;
108  return (char)result;
109}
110
111static int find_class(char *class_name)
112{
113  int i;
114  static char *class[] = {
115    "[:alpha:]","[:alnum:]","[:digit:]",
116    "[:lower:]","[:upper:]","[:space:]",
117    "[:blank:]","[:punct:]","[:cntrl:]",
118    "[:xdigit:]","NULL"
119  };
120
121  for (i = 0; i != class_invalid; i++) {
122    if (!memcmp(class_name, class[i], (class_name[0] == 'x')?10:9)) break;
123  }
124  return i;
125}
126
127static char *expand_set(char *arg, int *len)
128{
129  int i = 0, j, k, size = 256;
130  char *set = xzalloc(size*sizeof(char));
131
132  while (*arg) {
133
134    if (i >= size) {
135      size += 256;
136      set = xrealloc(set, size);
137    }
138    if (*arg == '\\') {
139      arg++;
140      set[i++] = (int)handle_escape_char(&arg);
141      arg++;
142      continue;
143    }
144    if (arg[1] == '-') {
145      if (arg[2] == '\0') goto save;
146      j = arg[0];
147      k = arg[2];
148      if (j > k) perror_exit("reverse colating order");
149      while (j <= k) set[i++] = j++;
150      arg += 3;
151      continue;
152    }
153    if (arg[0] == '[' && arg[1] == ':') {
154
155      if ((j = find_class(arg)) == class_invalid) goto save;
156
157      if ((j == class_alpha) || (j == class_upper) || (j == class_alnum)) {
158      for (k = 'A'; k <= 'Z'; k++) set[i++] = k;
159      }
160      if ((j == class_alpha) || (j == class_lower) || (j == class_alnum)) {
161        for (k = 'a'; k <= 'z'; k++) set[i++] = k;
162      }
163      if ((j == class_alnum) || (j == class_digit) || (j == class_xdigit)) {
164        for (k = '0'; k <= '9'; k++) set[i++] = k;
165      }
166      if (j == class_space || j == class_blank) {
167        set[i++] = '\t';
168        if (j == class_space) {
169          set[i++] = '\n';
170          set[i++] = '\f';
171          set[i++] = '\r';
172          set[i++] = '\v';
173        }
174        set[i++] = ' ';
175      }
176      if (j == class_punct) {
177        for (k = 0; k <= 255; k++)
178          if (ispunct(k)) set[i++] = k;
179      }
180      if (j == class_cntrl) {
181        for (k = 0; k <= 255; k++)
182          if (iscntrl(k)) set[i++] = k;
183      }
184      if (j == class_xdigit) {
185        for (k = 'A'; k <= 'F'; k++) {
186          set[i + 6] = k | 0x20;
187          set[i++] = k;
188        }
189        i += 6;
190        arg += 10;
191        continue;
192      }
193
194      arg += 9; //never here for class_xdigit.
195      continue;
196    }
197    if (arg[0] == '[' && arg[1] == '=') { //[=char=] only
198      arg += 2;
199      if (*arg) set[i++] = *arg;
200      if (!arg[1] || arg[1] != '=' || arg[2] != ']')
201        error_exit("bad equiv class");
202      continue;
203    }
204save:
205    set[i++] = *arg++;
206  }
207  *len = i;
208  return set;
209}
210
211static void print_map(char *set1, char *set2)
212{
213  int r = 0, i, prev_char = -1;
214
215  while (1)
216  {
217    i = 0;
218    r = read(STDIN_FILENO, (toybuf), sizeof(toybuf));
219    if (!r) break;
220    for (;r > i;i++) {
221
222      if ((toys.optflags & FLAG_d) && (TT.map[(int)toybuf[i]] & 0x100)) continue;
223      if (toys.optflags & FLAG_s) {
224        if ((TT.map[(int)toybuf[i]] & 0x200) &&
225            (prev_char == TT.map[(int)toybuf[i]])) {
226          continue;
227        }
228      }
229      xputc(TT.map[(int)toybuf[i]] & 0xFF);
230      prev_char = TT.map[(int)toybuf[i]];
231      fflush(stdout);
232    }
233  }
234}
235
236static void do_complement(char **set)
237{
238  int i, j;
239  char *comp = xmalloc(256);
240
241  for (i = 0, j = 0;i < 256; i++) {
242    if (memchr(*set, i, TT.len1)) continue;
243    else comp[j++] = (char)i;
244  }
245  free(*set);
246  TT.len1 = j;
247  *set = comp;
248}
249
250void tr_main(void)
251{
252  char *set1, *set2 = NULL;
253  int i;
254
255  for (i = 0; i < 256; i++) TT.map[i] = i; //init map
256
257  set1 = expand_set(toys.optargs[0], &TT.len1);
258  if (toys.optflags & FLAG_c) do_complement(&set1);
259  if (toys.optargs[1]) {
260    if (toys.optargs[1][0] == '\0') error_exit("set2 can't be empty string");
261    set2 = expand_set(toys.optargs[1], &TT.len2);
262  }
263  map_translation(set1, set2);
264
265  print_map(set1, set2);
266  free(set1);
267  free(set2);
268}
269