Fix bug (wrong value computed) when reading file from stdin, implement
[oweals/busybox.git] / coreutils / tr.c
index 48cdd47bddfe5503c48e18ea5a6870cf3c694c3b..a00e3613427318e10eabffa07c867c250cad5610 100644 (file)
  * Original copyright notice is retained at the end of this file.
  */
 
-#include "internal.h"
 #include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <sys/types.h>
-#define BB_DECLARE_EXTERN
-#define bb_need_write_error
-#include "messages.c"
+#include "busybox.h"
 
-const char *tr_usage="tr [-cds] STRING1 [STRING2]\n"
-#ifndef BB_FEATURE_TRIVIAL_HELP
-       "\nTranslate, squeeze, and/or delete characters from\n"
-       "standard input, writing to standard output.\n\n"
-       "Options:\n"
-       "\t-c\ttake complement of STRING1\n"
-       "\t-d\tdelete input characters coded STRING1\n"
-       "\t-s\tsqueeze multiple output characters of STRING2 into one character\n"
-#endif
-;
+/* This must be a #define, since when DODEBUG and BUFFERS_GO_IN_BSS are
+ * enabled, we otherwise get a "storage size isn't constant error. */
+#define ASCII 0377
 
-
-
-#ifdef TRUE
-#undef TRUE
-#undef FALSE
-#define TRUE   1
-#define FALSE  0
-#endif
-
-#define ASCII          0377
-
-/* some glabals shared across this file */
+/* some "globals" shared across this file */
 static char com_fl, del_fl, sq_fl;
-static unsigned char output[BUFSIZ], input[BUFSIZ];
-static unsigned char vector[ASCII + 1];
-static char invec[ASCII + 1], outvec[ASCII + 1];
 static short in_index, out_index;
+/* these last are pointers to static buffers declared in tr_main */
+static unsigned char *poutput, *pinput;
+static unsigned char *pvector;
+static char *pinvec, *poutvec;
 
 
-static void convert()
+static void convert(void)
 {
        short read_chars = 0;
        short c, coded;
@@ -72,25 +52,23 @@ static void convert()
 
        for (;;) {
                if (in_index == read_chars) {
-                       if ((read_chars = read(0, (char *) input, BUFSIZ)) <= 0) {
-                               if (write(1, (char *) output, out_index) != out_index)
-                                       write(2, write_error, strlen(write_error));
+                       if ((read_chars = read(0, (char *) pinput, BUFSIZ)) <= 0) {
+                               if (write(1, (char *) poutput, out_index) != out_index)
+                                       bb_error_msg(bb_msg_write_error);
                                exit(0);
                        }
                        in_index = 0;
                }
-               c = input[in_index++];
-               coded = vector[c];
-               if (del_fl && invec[c])
+               c = pinput[in_index++];
+               coded = pvector[c];
+               if (del_fl && pinvec[c])
                        continue;
-               if (sq_fl && last == coded && outvec[coded])
+               if (sq_fl && last == coded && (pinvec[c] || poutvec[coded]))
                        continue;
-               output[out_index++] = last = coded;
+               poutput[out_index++] = last = coded;
                if (out_index == BUFSIZ) {
-                       if (write(1, (char *) output, out_index) != out_index) {
-                               write(2, write_error, strlen(write_error));
-                               exit(1);
-                       }
+                       if (write(1, (char *) poutput, out_index) != out_index)
+                               bb_error_msg_and_die(bb_msg_write_error);
                        out_index = 0;
                }
        }
@@ -98,27 +76,43 @@ static void convert()
        /* NOTREACHED */
 }
 
-static void map(register unsigned char *string1, register unsigned char *string2)
+static void map(register unsigned char *string1, unsigned int string1_len,
+               register unsigned char *string2, unsigned int string2_len)
 {
        unsigned char last = '0';
+       unsigned int i, j;
 
-       while (*string1) {
-               if (*string2 == '\0')
-                       vector[*string1] = last;
+       for (j = 0, i = 0; i < string1_len; i++) {
+               if (string2_len <= j)
+                       pvector[string1[i]] = last;
                else
-                       vector[*string1] = last = *string2++;
-               string1++;
+                       pvector[string1[i]] = last = string2[j++];
        }
 }
 
-static void expand(char *arg, register unsigned char *buffer)
+/* supported constructs:
+ *   Ranges,  e.g.,  [0-9]  ==>  0123456789
+ *   Escapes, e.g.,  \a     ==>  Control-G
+ */
+static unsigned int expand(const char *arg, register unsigned char *buffer)
 {
+       unsigned char *buffer_start = buffer;
        int i, ac;
 
        while (*arg) {
                if (*arg == '\\') {
                        arg++;
-                       *buffer++ = process_escape_sequence(&arg);
+                       *buffer++ = bb_process_escape_sequence(&arg);
+               } else if (*(arg+1) == '-') {
+                       ac = *(arg+2);
+                       if(ac == 0) {
+                               *buffer++ = *arg++;
+                               continue;
+                       }
+                       i = *arg;
+                       while (i <= ac)
+                               *buffer++ = i++;
+                       arg += 3; /* Skip the assumed a-z */
                } else if (*arg == '[') {
                        arg++;
                        i = *arg++;
@@ -130,38 +124,52 @@ static void expand(char *arg, register unsigned char *buffer)
                        ac = *arg++;
                        while (i <= ac)
                                *buffer++ = i++;
-                       arg++;                          /* Skip ']' */
+                       arg++;                          /* Skip the assumed ']' */
                } else
                        *buffer++ = *arg++;
        }
+
+       return (buffer - buffer_start);
 }
 
-static void complement(unsigned char *buffer)
+static int complement(unsigned char *buffer, int buffer_len)
 {
-       register unsigned char *ptr;
-       register short i, index;
-       unsigned char conv[ASCII + 2];
+       register short i, j, ix;
+       char conv[ASCII + 2];
 
-       index = 0;
-       for (i = 1; i <= ASCII; i++) {
-               for (ptr = buffer; *ptr; ptr++)
-                       if (*ptr == i)
+       ix = 0;
+       for (i = 0; i <= ASCII; i++) {
+               for (j = 0; j < buffer_len; j++)
+                       if (buffer[j] == i)
                                break;
-               if (*ptr == '\0')
-                       conv[index++] = i & ASCII;
+               if (j == buffer_len)
+                       conv[ix++] = i & ASCII;
        }
-       conv[index] = '\0';
-       strcpy((char *) buffer, (char *) conv);
+       memcpy(buffer, conv, ix);
+       return ix;
 }
 
 extern int tr_main(int argc, char **argv)
 {
        register unsigned char *ptr;
-       int index = 1;
-       short i;
-
-       if (argc > 1 && argv[index][0] == '-') {
-               for (ptr = (unsigned char *) &argv[index][1]; *ptr; ptr++) {
+       int output_length=0, input_length;
+       int idx = 1;
+       int i;
+       RESERVE_CONFIG_BUFFER(output, BUFSIZ);
+       RESERVE_CONFIG_BUFFER(input,  BUFSIZ);
+       RESERVE_CONFIG_UBUFFER(vector, ASCII+1);
+       RESERVE_CONFIG_BUFFER(invec,  ASCII+1);
+       RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
+
+       /* ... but make them available globally */
+       poutput = output;
+       pinput  = input;
+       pvector = vector;
+       pinvec  = invec;
+       poutvec = outvec;
+
+       if (argc > 1 && argv[idx][0] == '-') {
+               for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
                        switch (*ptr) {
                        case 'c':
                                com_fl = TRUE;
@@ -173,28 +181,30 @@ extern int tr_main(int argc, char **argv)
                                sq_fl = TRUE;
                                break;
                        default:
-                               usage(tr_usage);
+                               bb_show_usage();
                        }
                }
-               index++;
+               idx++;
        }
        for (i = 0; i <= ASCII; i++) {
                vector[i] = i;
                invec[i] = outvec[i] = FALSE;
        }
 
-       if (argv[index] != NULL) {
-               expand(argv[index++], input);
+       if (argv[idx] != NULL) {
+               input_length = expand(argv[idx++], input);
                if (com_fl)
-                       complement(input);
-               if (argv[index] != NULL)
-                       expand(argv[index], output);
-               if (argv[index] != NULL)
-                       map(input, output);
-               for (ptr = input; *ptr; ptr++)
-                       invec[*ptr] = TRUE;
-               for (ptr = output; *ptr; ptr++)
-                       outvec[*ptr] = TRUE;
+                       input_length = complement(input, input_length);
+               if (argv[idx] != NULL) {
+                       if (*argv[idx] == '\0')
+                               bb_error_msg_and_die("STRING2 cannot be empty");
+                       output_length = expand(argv[idx], output);
+                       map(input, input_length, output, output_length);
+               }
+               for (i = 0; i < input_length; i++)
+                       invec[(int)input[i]] = TRUE;
+               for (i = 0; i < output_length; i++)
+                       outvec[(int)output[i]] = TRUE;
        }
        convert();
        return (0);