meillo@14: /* cut - remove parts of lines of files
meillo@14:    Copyright (C) 1984 by David M. Ihnat
meillo@14:  
meillo@14:    This program is a total rewrite of the Bell Laboratories Unix(Tm)
meillo@14:    command of the same name, as of System V.  It contains no proprietary
meillo@14:    code, and therefore may be used without violation of any proprietary
meillo@14:    agreements whatsoever.  However, you will notice that the program is
meillo@14:    copyrighted by me.  This is to assure the program does *not* fall
meillo@14:    into the public domain.  Thus, I may specify just what I am now:
meillo@14:    This program may be freely copied and distributed, provided this notice
meillo@14:    remains; it may not be sold for profit without express written consent of
meillo@14:    the author.
meillo@14:    Please note that I recreated the behavior of the Unix(Tm) 'cut' command
meillo@14:    as faithfully as possible; however, I haven't run a full set of regression
meillo@14:    tests.  Thus, the user of this program accepts full responsibility for any
meillo@14:    effects or loss; in particular, the author is not responsible for any losses,
meillo@14:    explicit or incidental, that may be incurred through use of this program.
meillo@14: 
meillo@14:    I ask that any bugs (and, if possible, fixes) be reported to me when
meillo@14:    possible.  -David Ihnat (312) 784-4544 ignatz@homebru.chi.il.us
meillo@14: 
meillo@14:    POSIX changes, bug fixes, long-named options, and cleanup
meillo@14:    by David MacKenzie <djm@ai.mit.edu>.
meillo@14: 
meillo@14:    Options:
meillo@14:    --bytes=byte-list
meillo@14:    -b byte-list			Print only the bytes in positions listed
meillo@14: 				in BYTE-LIST.
meillo@14: 				Tabs and backspaces are treated like any
meillo@14: 				other character; they take up 1 byte.
meillo@14: 
meillo@14:    --characters=character-list
meillo@14:    -c character-list		Print only characters in positions listed
meillo@14: 				in CHARACTER-LIST.
meillo@14: 				The same as -b for now, but
meillo@14: 				internationalization will change that.
meillo@14: 				Tabs and backspaces are treated like any
meillo@14: 				other character; they take up 1 character.
meillo@14: 
meillo@14:    --fields=field-list
meillo@14:    -f field-list		Print only the fields listed in FIELD-LIST.
meillo@14: 				Fields are separated by a TAB by default.
meillo@14: 
meillo@14:    --delimiter=delim
meillo@14:    -d delim			For -f, fields are separated by the first
meillo@14: 				character in DELIM instead of TAB.
meillo@14: 
meillo@14:    -n				Do not split multibyte chars (no-op for now).
meillo@14: 
meillo@14:    --only-delimited
meillo@14:    -s				For -f, do not print lines that do not contain
meillo@14: 				the field separator character.
meillo@14: 
meillo@14:    The BYTE-LIST, CHARACTER-LIST, and FIELD-LIST are one or more numbers
meillo@14:    or ranges separated by commas.  The first byte, character, and field
meillo@14:    are numbered 1.
meillo@14: 
meillo@14:    A FILE of `-' means standard input. */
meillo@14: 
meillo@14: #define _GNU_SOURCE
meillo@14: #include <ctype.h>
meillo@14: #ifndef isblank
meillo@14: #define isblank(c) ((c) == ' ' || (c) == '\t')
meillo@14: #endif
meillo@14: #include <stdio.h>
meillo@14: #include <getopt.h>
meillo@14: #include <sys/types.h>
meillo@14: #include "system.h"
meillo@14: 
meillo@14: #ifdef isascii
meillo@14: #define ISDIGIT(c) (isascii ((c)) && isdigit ((c)))
meillo@14: #else
meillo@14: #define ISDIGIT(c) (isdigit ((c)))
meillo@14: #endif
meillo@14: 
meillo@14: char *xmalloc ();
meillo@14: char *xrealloc ();
meillo@14: int set_fields ();
meillo@14: int cut_file ();
meillo@14: void cut_stream ();
meillo@14: void cut_bytes ();
meillo@14: void cut_fields ();
meillo@14: void enlarge_line ();
meillo@14: void error ();
meillo@14: void invalid_list ();
meillo@14: void usage ();
meillo@14: 
meillo@14: /* The number of elements allocated for the input line
meillo@14:    and the byte or field number.
meillo@14:    Enlarged as necessary. */
meillo@14: int line_size;
meillo@14: 
meillo@14: /* Processed output buffer. */
meillo@14: char *outbuf;
meillo@14: 
meillo@14: /* Where to save next char to output. */
meillo@14: char *outbufptr;
meillo@14: 
meillo@14: /* Raw line buffer for field mode. */
meillo@14: char *inbuf;
meillo@14: 
meillo@14: /* Where to save next input char. */
meillo@14: char *inbufptr;
meillo@14: 
meillo@14: /* What can be done about a byte or field. */
meillo@14: enum field_action
meillo@14: {
meillo@14:   field_omit,
meillo@14:   field_output
meillo@14: };
meillo@14: 
meillo@14: /* In byte mode, which bytes to output.
meillo@14:    In field mode, which `delim'-separated fields to output.
meillo@14:    Both bytes and fields are numbered starting with 1,
meillo@14:    so the first element of `fields' is unused. */
meillo@14: enum field_action *fields;
meillo@14: 
meillo@14: enum operating_mode
meillo@14: {
meillo@14:   undefined_mode,
meillo@14: 
meillo@14:   /* Output characters that are in the given bytes. */
meillo@14:   byte_mode,
meillo@14: 
meillo@14:   /* Output the given delimeter-separated fields. */
meillo@14:   field_mode
meillo@14: };
meillo@14: 
meillo@14: enum operating_mode operating_mode;
meillo@14: 
meillo@14: /* If nonzero,
meillo@14:    for field mode, do not output lines containing no delimeter characters. */
meillo@14: int delimited_lines_only;
meillo@14: 
meillo@14: /* The delimeter character for field mode. */
meillo@14: unsigned char delim;
meillo@14: 
meillo@14: /* Nonzero if we have ever read standard input. */
meillo@14: int have_read_stdin;
meillo@14: 
meillo@14: /* The name this program was run with. */
meillo@14: char *program_name;
meillo@14: 
meillo@14: struct option longopts[] =
meillo@14: {
meillo@14:   {"bytes", 1, 0, 'b'},
meillo@14:   {"characters", 1, 0, 'c'},
meillo@14:   {"fields", 1, 0, 'f'},
meillo@14:   {"delimiter", 1, 0, 'd'},
meillo@14:   {"only-delimited", 0, 0, 's'},
meillo@14:   {0, 0, 0, 0}
meillo@14: };
meillo@14: 
meillo@14: void
meillo@14: main (argc, argv)
meillo@14:      int argc;
meillo@14:      char **argv;
meillo@14: {
meillo@14:   int optc, exit_status = 0;
meillo@14: 
meillo@14:   program_name = argv[0];
meillo@14: 
meillo@14:   line_size = 512;
meillo@14:   operating_mode = undefined_mode;
meillo@14:   delimited_lines_only = 0;
meillo@14:   delim = '\0';
meillo@14:   have_read_stdin = 0;
meillo@14: 
meillo@14:   fields = (enum field_action *)
meillo@14:     xmalloc (line_size * sizeof (enum field_action));
meillo@14:   outbuf = (char *) xmalloc (line_size);
meillo@14:   inbuf = (char *) xmalloc (line_size);
meillo@14: 
meillo@14:   for (optc = 0; optc < line_size; optc++)
meillo@14:     fields[optc] = field_omit;
meillo@14: 
meillo@14:   while ((optc = getopt_long (argc, argv, "b:c:d:f:ns", longopts, (int *) 0))
meillo@14: 	 != EOF)
meillo@14:     {
meillo@14:       switch (optc)
meillo@14: 	{
meillo@14: 	case 'b':
meillo@14: 	case 'c':
meillo@14: 	  /* Build the byte list. */
meillo@14: 	  if (operating_mode != undefined_mode)
meillo@14: 	    usage ();
meillo@14: 	  operating_mode = byte_mode;
meillo@14: 	  if (set_fields (optarg) == 0)
meillo@14: 	    error (2, 0, "no fields given");
meillo@14: 	  break;
meillo@14: 
meillo@14: 	case 'f':
meillo@14: 	  /* Build the field list. */
meillo@14: 	  if (operating_mode != undefined_mode)
meillo@14: 	    usage ();
meillo@14: 	  operating_mode = field_mode;
meillo@14: 	  if (set_fields (optarg) == 0)
meillo@14: 	    error (2, 0, "no fields given");
meillo@14: 	  break;
meillo@14: 
meillo@14: 	case 'd':
meillo@14: 	  /* New delimiter. */
meillo@14: 	  if (optarg[0] == '\0')
meillo@14: 	    error (2, 0, "no delimiter given");
meillo@14: 	  if (optarg[1] != '\0')
meillo@14: 	    error (2, 0, "delimiter must be a single character");
meillo@14: 	  delim = optarg[0];
meillo@14: 	  break;
meillo@14: 
meillo@14: 	case 'n':
meillo@14: 	  break;
meillo@14: 
meillo@14: 	case 's':
meillo@14: 	  delimited_lines_only++;
meillo@14: 	  break;
meillo@14: 
meillo@14: 	default:
meillo@14: 	  usage ();
meillo@14: 	}
meillo@14:     }
meillo@14: 
meillo@14:   if (operating_mode == undefined_mode)
meillo@14:     usage ();
meillo@14: 
meillo@14:   if ((delimited_lines_only || delim != '\0') && operating_mode != field_mode)
meillo@14:     usage ();
meillo@14: 
meillo@14:   if (delim == '\0')
meillo@14:     delim = '\t';
meillo@14: 
meillo@14:   if (optind == argc)
meillo@14:     exit_status |= cut_file ("-");
meillo@14:   else
meillo@14:     for (; optind < argc; optind++)
meillo@14:       exit_status |= cut_file (argv[optind]);
meillo@14: 
meillo@14:   if (have_read_stdin && fclose (stdin) == EOF)
meillo@14:     {
meillo@14:       error (0, errno, "-");
meillo@14:       exit_status = 1;
meillo@14:     }
meillo@14:   if (ferror (stdout) || fclose (stdout) == EOF)
meillo@14:     error (1, 0, "write error");
meillo@14: 
meillo@14:   exit (exit_status);
meillo@14: }
meillo@14: 
meillo@14: /* Select for printing the positions in `fields' that are listed in
meillo@14:    byte or field specification FIELDSTR.  FIELDSTR should be
meillo@14:    composed of one or more numbers or ranges of numbers, separated by
meillo@14:    blanks or commas.  Incomplete ranges may be given: `-m' means
meillo@14:    `1-m'; `n-' means `n' through end of line or last field.
meillo@14: 
meillo@14:    Return the number of fields selected. */
meillo@14: 
meillo@14: int
meillo@14: set_fields (fieldstr)
meillo@14:      char *fieldstr;
meillo@14: {
meillo@14:   int initial = 1;		/* Value of first number in a range. */
meillo@14:   int dash_found = 0;		/* Nonzero if a '-' is found in this field. */
meillo@14:   int value = 0;		/* If nonzero, a number being accumulated. */
meillo@14:   int fields_selected = 0;	/* Number of fields selected so far. */
meillo@14:   /* If nonzero, index of first field in a range that goes to end of line. */
meillo@14:   int eol_range_start = 0;
meillo@14: 
meillo@14:   for (;;)
meillo@14:     {
meillo@14:       if (*fieldstr == '-')
meillo@14: 	{
meillo@14: 	  /* Starting a range. */
meillo@14: 	  if (dash_found)
meillo@14: 	    invalid_list ();
meillo@14: 	  dash_found++;
meillo@14: 	  fieldstr++;
meillo@14: 
meillo@14: 	  if (value)
meillo@14: 	    {
meillo@14: 	      if (value >= line_size)
meillo@14: 		enlarge_line (value);
meillo@14: 	      initial = value;
meillo@14: 	      value = 0;
meillo@14: 	    }
meillo@14: 	  else
meillo@14: 	    initial = 1;
meillo@14: 	}
meillo@14:       else if (*fieldstr == ',' || isblank (*fieldstr) || *fieldstr == '\0')
meillo@14: 	{
meillo@14: 	  /* Ending the string, or this field/byte sublist. */
meillo@14: 	  if (dash_found)
meillo@14: 	    {
meillo@14: 	      dash_found = 0;
meillo@14: 
meillo@14: 	      /* A range.  Possibilites: -n, m-n, n-.
meillo@14: 		 In any case, `initial' contains the start of the range. */
meillo@14: 	      if (value == 0)
meillo@14: 		{
meillo@14: 		  /* `n-'.  From `initial' to end of line. */
meillo@14: 		  eol_range_start = initial;
meillo@14: 		  fields_selected++;
meillo@14: 		}
meillo@14: 	      else
meillo@14: 		{
meillo@14: 		  /* `m-n' or `-n' (1-n). */
meillo@14: 		  if (value < initial)
meillo@14: 		    invalid_list ();
meillo@14: 
meillo@14: 		  if (value >= line_size)
meillo@14: 		    enlarge_line (value);
meillo@14: 
meillo@14: 		  /* Is there already a range going to end of line? */
meillo@14: 		  if (eol_range_start != 0)
meillo@14: 		    {
meillo@14: 		      /* Yes.  Is the new sequence already contained
meillo@14: 			 in the old one?  If so, no processing is
meillo@14: 			 necessary. */
meillo@14: 		      if (initial < eol_range_start)
meillo@14: 			{
meillo@14: 			  /* No, the new sequence starts before the
meillo@14: 			     old.  Does the old range going to end of line
meillo@14: 			     extend into the new range?  */
meillo@14: 			  if (eol_range_start < value)
meillo@14: 			    /* Yes.  Simply move the end of line marker. */
meillo@14: 			    eol_range_start = initial;
meillo@14: 			  else
meillo@14: 			    {
meillo@14: 			      /* No.  A simple range, before and disjoint from
meillo@14: 				 the range going to end of line.  Fill it. */
meillo@14: 			      for (; initial <= value; initial++)
meillo@14: 				fields[initial] = field_output;
meillo@14: 			    }
meillo@14: 
meillo@14: 			  /* In any case, some fields were selected. */
meillo@14: 			  fields_selected++;
meillo@14: 			}
meillo@14: 		    }
meillo@14: 		  else
meillo@14: 		    {
meillo@14: 		      /* There is no range going to end of line. */
meillo@14: 		      for (; initial <= value; initial++)
meillo@14: 			fields[initial] = field_output;
meillo@14: 		      fields_selected++;
meillo@14: 		    }
meillo@14: 		  value = 0;
meillo@14: 		}
meillo@14: 	    }
meillo@14: 	  else if (value != 0)
meillo@14: 	    {
meillo@14: 	      /* A simple field number, not a range. */
meillo@14: 	      if (value >= line_size)
meillo@14: 		enlarge_line (value);
meillo@14: 
meillo@14: 	      fields[value] = field_output;
meillo@14: 	      value = 0;
meillo@14: 	      fields_selected++;
meillo@14: 	    }
meillo@14: 
meillo@14: 	  if (*fieldstr == '\0')
meillo@14: 	    {
meillo@14: 	      /* If there was a range going to end of line, fill the
meillo@14: 		 array from the end of line point.  */
meillo@14: 	      if (eol_range_start)
meillo@14: 		for (initial = eol_range_start; initial < line_size; initial++)
meillo@14: 		  fields[initial] = field_output;
meillo@14: 
meillo@14: 	      return fields_selected;
meillo@14: 	    }
meillo@14: 
meillo@14: 	  fieldstr++;
meillo@14: 	}
meillo@14:       else if (ISDIGIT (*fieldstr))
meillo@14: 	{
meillo@14: 	  value = 10 * value + *fieldstr - '0';
meillo@14: 	  fieldstr++;
meillo@14: 	}
meillo@14:       else
meillo@14: 	invalid_list ();
meillo@14:     }
meillo@14: }
meillo@14: 
meillo@14: /* Process file FILE to standard output.
meillo@14:    Return 0 if successful, 1 if not. */
meillo@14: 
meillo@14: int
meillo@14: cut_file (file)
meillo@14:      char *file;
meillo@14: {
meillo@14:   FILE *stream;
meillo@14: 
meillo@14:   if (!strcmp (file, "-"))
meillo@14:     {
meillo@14:       have_read_stdin = 1;
meillo@14:       stream = stdin;
meillo@14:     }
meillo@14:   else
meillo@14:     {
meillo@14:       stream = fopen (file, "r");
meillo@14:       if (stream == NULL)
meillo@14: 	{
meillo@14: 	  error (0, errno, "%s", file);
meillo@14: 	  return 1;
meillo@14: 	}
meillo@14:     }
meillo@14: 
meillo@14:   cut_stream (stream);
meillo@14: 
meillo@14:   if (ferror (stream))
meillo@14:     {
meillo@14:       error (0, errno, "%s", file);
meillo@14:       return 1;
meillo@14:     }
meillo@14:   if (!strcmp (file, "-"))
meillo@14:     clearerr (stream);		/* Also clear EOF. */
meillo@14:   else if (fclose (stream) == EOF)
meillo@14:     {
meillo@14:       error (0, errno, "%s", file);
meillo@14:       return 1;
meillo@14:     }
meillo@14:   return 0;
meillo@14: }
meillo@14: 
meillo@14: void
meillo@14: cut_stream (stream)
meillo@14:      FILE *stream;
meillo@14: {
meillo@14:   if (operating_mode == byte_mode)
meillo@14:     cut_bytes (stream);
meillo@14:   else
meillo@14:     cut_fields (stream);
meillo@14: }
meillo@14: 
meillo@14: /* Print the file open for reading on stream STREAM
meillo@14:    with the bytes marked `field_omit' in `fields' removed from each line. */
meillo@14: 
meillo@14: void
meillo@14: cut_bytes (stream)
meillo@14:      FILE *stream;
meillo@14: {
meillo@14:   register int c;		/* Each character from the file. */
meillo@14:   int doneflag = 0;		/* Nonzero if EOF reached. */
meillo@14:   int char_count;		/* Number of chars in the line so far. */
meillo@14: 
meillo@14:   while (doneflag == 0)
meillo@14:     {
meillo@14:       /* Start processing a line. */
meillo@14:       outbufptr = outbuf;
meillo@14:       char_count = 0;
meillo@14: 
meillo@14:       do
meillo@14: 	{
meillo@14: 	  c = getc (stream);
meillo@14: 	  if (c == EOF)
meillo@14: 	    {
meillo@14: 	      doneflag++;
meillo@14: 	      break;
meillo@14: 	    }
meillo@14: 
meillo@14: 	  /* If this character is to be sent, stow it in the outbuffer. */
meillo@14: 
meillo@14: 	  if (++char_count == line_size - 1)
meillo@14: 	    enlarge_line (char_count);
meillo@14: 
meillo@14: 	  if (fields[char_count] == field_output || c == '\n')
meillo@14: 	    *outbufptr++ = c;
meillo@14: 	}
meillo@14:       while (c != '\n');
meillo@14: 
meillo@14:       if (char_count)
meillo@14: 	fwrite (outbuf, sizeof (char), outbufptr - outbuf, stdout);
meillo@14:     }
meillo@14: }
meillo@14: 
meillo@14: /* Print the file open for reading on stream STREAM
meillo@14:    with the fields marked `field_omit' in `fields' removed from each line.
meillo@14:    All characters are initially stowed in the raw input buffer, until
meillo@14:    at least one field has been found. */
meillo@14: 
meillo@14: void
meillo@14: cut_fields (stream)
meillo@14:      FILE *stream;
meillo@14: {
meillo@14:   register int c;		/* Each character from the file. */
meillo@14:   int doneflag = 0;		/* Nonzero if EOF reached. */
meillo@14:   int char_count;		/* Number of chars in line before any delim. */
meillo@14:   int fieldfound;		/* Nonzero if any fields to print found. */
meillo@14:   int curr_field;		/* Current index in `fields'. */
meillo@14: 
meillo@14:   while (doneflag == 0)
meillo@14:     {
meillo@14:       char_count = 0;
meillo@14:       fieldfound = 0;
meillo@14:       curr_field = 1;
meillo@14:       outbufptr = outbuf;
meillo@14:       inbufptr = inbuf;
meillo@14: 
meillo@14:       do
meillo@14: 	{
meillo@14: 	  c = getc (stream);
meillo@14: 	  if (c == EOF)
meillo@14: 	    {
meillo@14: 	      doneflag++;
meillo@14: 	      break;
meillo@14: 	    }
meillo@14: 
meillo@14: 	  if (fields[curr_field] == field_output && c != '\n')
meillo@14: 	    {
meillo@14: 	      /* Working on a field.  It, and its terminating
meillo@14: 		 delimiter, go only into the processed buffer. */
meillo@14: 	      fieldfound = 1;
meillo@14: 	      if (outbufptr - outbuf == line_size - 2)
meillo@14: 		enlarge_line (outbufptr - outbuf);
meillo@14: 	      *outbufptr++ = c;
meillo@14: 	    }
meillo@14: 	  else if (fieldfound == 0)
meillo@14: 	    {
meillo@14: 	      if (++char_count == line_size - 1)
meillo@14: 		enlarge_line (char_count);
meillo@14: 	      *inbufptr++ = c;
meillo@14: 	    }
meillo@14: 
meillo@14: 	  if (c == delim && ++curr_field == line_size - 1)
meillo@14: 	    enlarge_line (curr_field);
meillo@14: 	}
meillo@14:       while (c != '\n');
meillo@14: 
meillo@14:       if (fieldfound)
meillo@14: 	{
meillo@14: 	  /* Something was found. Print it. */
meillo@14: 	  if (outbufptr[-1] == delim)
meillo@14: 	    --outbufptr;	/* Suppress trailing delimiter. */
meillo@14: 
meillo@14: 	  fwrite (outbuf, sizeof (char), outbufptr - outbuf, stdout);
meillo@14: 	  if (c == '\n')
meillo@14: 	    putc (c, stdout);
meillo@14: 	}
meillo@14:       else if (!delimited_lines_only && char_count)
meillo@14: 	/* A line with some characters, no delimiters, and no
meillo@14: 	   suppression.  Print it. */
meillo@14: 	fwrite (inbuf, sizeof (char), inbufptr - inbuf, stdout);
meillo@14:     }
meillo@14: }
meillo@14: 
meillo@14: /* Extend the buffers to accomodate at least NEW_SIZE characters. */
meillo@14: 
meillo@14: void
meillo@14: enlarge_line (new_size)
meillo@14:      int new_size;
meillo@14: {
meillo@14:   char *newp;
meillo@14:   int i;
meillo@14: 
meillo@14:   new_size += 256;		/* Leave some room to grow. */
meillo@14: 
meillo@14:   fields = (enum field_action *)
meillo@14:     xrealloc (fields, new_size * sizeof (enum field_action));
meillo@14: 
meillo@14:   newp = (char *) xrealloc (outbuf, new_size);
meillo@14:   outbufptr += newp - outbuf;
meillo@14:   outbuf = newp;
meillo@14: 
meillo@14:   newp = (char *) xrealloc (inbuf, new_size);
meillo@14:   inbufptr += newp - inbuf;
meillo@14:   inbuf = newp;
meillo@14: 
meillo@14:   for (i = line_size; i < new_size; i++)
meillo@14:     fields[i] = field_omit;
meillo@14:   line_size = new_size;
meillo@14: }
meillo@14: 
meillo@14: void
meillo@14: invalid_list ()
meillo@14: {
meillo@14:   error (2, 0, "invalid byte or field list");
meillo@14: }
meillo@14: 
meillo@14: void
meillo@14: usage ()
meillo@14: {
meillo@14:   fprintf (stderr, "\
meillo@14: Usage: %s {-b byte-list,--bytes=byte-list} [-n] [file...]\n\
meillo@14:        %s {-c character-list,--characters=character-list} [file...]\n\
meillo@14:        %s {-f field-list,--fields=field-list} [-d delim] [-s]\n\
meillo@14:        [--delimiter=delim] [--only-delimited] [file...]\n",
meillo@14: 	   program_name, program_name, program_name);
meillo@14:   exit (2);
meillo@14: }