1 /* join - join lines of two files on a common field
2 Copyright (C) 91, 95, 1996 Free Software Foundation, Inc.
4 This program is free software; you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 2, or (at your option)
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program; if not, write to the Free Software Foundation,
16 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
18 Written by Mike Haertel, mike@gnu.ai.mit.edu. */
23 #define alloca __builtin_alloca
24 #else /* not __GNUC__ */
27 #else /* not HAVE_ALLOCA_H */
33 #endif /* not HAVE_ALLOCA_H */
34 #endif /* not __GNUC__ */
36 /* Get isblank from GNU libc. */
42 #include <sys/types.h>
50 # define UINT_MAX ((unsigned int) ~(unsigned int) 0)
54 # define INT_MAX ((int) (UINT_MAX >> 1))
57 #if _LIBC || STDC_HEADERS
58 # define TOLOWER(c) tolower (c)
60 # define TOLOWER(c) (ISUPPER (c) ? tolower (c) : (c))
64 #include "long-options.h"
67 #include "memcasecmp.h"
69 #define join system_join
74 /* Undefine, to avoid warning about redefinition on some systems. */
77 #define min(A, B) ((A) < (B) ? (A) : (B))
78 #define max(A, B) ((A) > (B) ? (A) : (B))
80 /* An element of the list identifying which fields to print for each
84 /* File number: 0, 1, or 2. 0 means use the join field.
85 1 means use the first file argument, 2 the second. */
88 /* Field index (zero-based), specified only when FILE is 1 or 2. */
94 /* A field of a line. */
97 const char *beg; /* First character in field. */
98 size_t len; /* The length of the field. */
101 /* A line read from an input file. Newlines are not stored. */
104 char *beg; /* First character in line. */
105 char *lim; /* Character after last character in line. */
106 int nfields; /* Number of elements in `fields'. */
107 int nfields_allocated; /* Number of elements in `fields'. */
108 struct field *fields;
111 /* One or more consecutive lines read from a file that all have the
112 same join field value. */
115 int count; /* Elements used in `lines'. */
116 int alloc; /* Elements allocated in `lines'. */
120 /* The name this program was run with. */
123 /* If nonzero, print unpairable lines in file 1 or 2. */
124 static int print_unpairables_1, print_unpairables_2;
126 /* If nonzero, print pairable lines. */
127 static int print_pairables;
129 /* Empty output field filler. */
130 static char *empty_filler;
132 /* Field to join on. */
133 static int join_field_1, join_field_2;
135 /* List of fields to print. */
136 static struct outlist outlist_head;
138 /* Last element in `outlist', where a new element can be added. */
139 static struct outlist *outlist_end = &outlist_head;
141 /* Tab character separating fields; if this is NUL fields are separated
142 by any nonempty string of white space, otherwise by exactly one
146 /* When using getopt_long_only, no long option can start with
147 a character that is a short option. */
148 static struct option const longopts[] =
150 {"ignore-case", no_argument, NULL, 'i'},
151 {"j", required_argument, NULL, 'j'},
152 {"j1", required_argument, NULL, '1'},
153 {"j2", required_argument, NULL, '2'},
157 /* Used to print non-joining lines */
158 static struct line uni_blank;
160 /* If nonzero, ignore case when comparing join fields. */
161 static int ignore_case;
167 fprintf (stderr, _("Try `%s --help' for more information.\n"),
172 Usage: %s [OPTION]... FILE1 FILE2\n\
176 For each pair of input lines with identical join fields, write a line to\n\
177 standard output. The default join field is the first, delimited\n\
178 by whitespace. When FILE1 or FILE2 (not both) is -, read standard input.\n\
180 -a SIDE print unpairable lines coming from file SIDE\n\
181 -e EMPTY replace missing input fields with EMPTY\n\
182 -i, --ignore-case ignore differences in case when comparing fields\n\
183 -j FIELD (obsolescent) equivalent to `-1 FIELD -2 FIELD'\n\
184 -j1 FIELD (obsolescent) equivalent to `-1 FIELD'\n\
185 -j2 FIELD (obsolescent) equivalent to `-2 FIELD'\n\
186 -o FORMAT obey FORMAT while constructing output line\n\
187 -t CHAR use CHAR as input and output field separator\n\
188 -v SIDE like -a SIDE, but suppress joined output lines\n\
189 -1 FIELD join on this FIELD of file 1\n\
190 -2 FIELD join on this FIELD of file 2\n\
191 --help display this help and exit\n\
192 --version output version information and exit\n\
194 Unless -t CHAR is given, leading blanks separate fields and are ignored,\n\
195 else fields are separated by CHAR. Any FIELD is a field number counted\n\
196 from 1. FORMAT is one or more comma or blank separated specifications,\n\
197 each being `SIDE.FIELD' or `0'. Default FORMAT outputs the join field,\n\
198 the remaining fields from FILE1, the remaining fields from FILE2, all\n\
199 separated by CHAR.\n\
201 puts (_("\nReport bugs to textutils-bugs@gnu.ai.mit.edu"));
203 exit (status == 0 ? EXIT_SUCCESS : EXIT_FAILURE);
207 ADD_FIELD (struct line *line, const char *field, size_t len)
209 if (line->nfields >= line->nfields_allocated)
211 line->nfields_allocated = (3 * line->nfields_allocated) / 2 + 1;
212 line->fields = (struct field *) xrealloc ((char *) line->fields,
213 (line->nfields_allocated
214 * sizeof (struct field)));
216 line->fields[line->nfields].beg = field;
217 line->fields[line->nfields].len = len;
221 /* Fill in the `fields' structure in LINE. */
224 xfields (struct line *line)
227 register char *ptr, *lim;
234 /* Skip leading blanks before the first field. */
235 while (ptr < lim && ISSPACE (*ptr))
239 for (i = 0; ptr < lim; ++i)
246 while (ptr < lim && *ptr != tab)
248 ADD_FIELD (line, beg, ptr - beg);
257 while (ptr < lim && !ISSPACE (*ptr))
259 ADD_FIELD (line, beg, ptr - beg);
260 while (ptr < lim && ISSPACE (*ptr))
265 if (ptr > line->beg && ((tab && ISSPACE (ptr[-1])) || ptr[-1] == tab))
267 /* Add one more (empty) field because the last character of the
268 line was a delimiter. */
269 ADD_FIELD (line, NULL, 0);
273 /* Read a line from FP into LINE and split it into fields.
274 Return 0 if EOF, 1 otherwise. */
277 get_line (FILE *fp, struct line *line)
279 static int linesize = 80;
286 ptr = xmalloc (linesize);
288 for (i = 0; (c = getc (fp)) != EOF && c != '\n'; ++i)
293 ptr = xrealloc (ptr, linesize);
298 if (c == EOF && i == 0)
305 line->lim = line->beg + i;
306 line->nfields_allocated = 0;
314 freeline (struct line *line)
316 free ((char *) line->fields);
322 initseq (struct seq *seq)
326 seq->lines = (struct line *) xmalloc (seq->alloc * sizeof (struct line));
329 /* Read a line from FP and add it to SEQ. Return 0 if EOF, 1 otherwise. */
332 getseq (FILE *fp, struct seq *seq)
334 if (seq->count == seq->alloc)
337 seq->lines = (struct line *)
338 xrealloc ((char *) seq->lines, seq->alloc * sizeof (struct line));
341 if (get_line (fp, &seq->lines[seq->count]))
350 delseq (struct seq *seq)
353 for (i = 0; i < seq->count; i++)
354 if (seq->lines[i].beg)
355 freeline (&seq->lines[i]);
356 free ((char *) seq->lines);
359 /* Return <0 if the join field in LINE1 compares less than the one in LINE2;
360 >0 if it compares greater; 0 if it compares equal. */
363 keycmp (struct line *line1, struct line *line2)
365 const char *beg1, *beg2; /* Start of field to compare in each file. */
366 int len1, len2; /* Length of fields to compare. */
369 if (join_field_1 < line1->nfields)
371 beg1 = line1->fields[join_field_1].beg;
372 len1 = line1->fields[join_field_1].len;
380 if (join_field_2 < line2->nfields)
382 beg2 = line2->fields[join_field_2].beg;
383 len2 = line2->fields[join_field_2].len;
392 return len2 == 0 ? 0 : -1;
396 /* Use an if-statement here rather than a function variable to
397 avoid portability hassles of getting a non-conflicting declaration
400 diff = memcasecmp (beg1, beg2, min (len1, len2));
402 diff = memcmp (beg1, beg2, min (len1, len2));
409 /* Print field N of LINE if it exists and is nonempty, otherwise
410 `empty_filler' if it is nonempty. */
413 prfield (int n, struct line *line)
417 if (n < line->nfields)
419 len = line->fields[n].len;
421 fwrite (line->fields[n].beg, 1, len, stdout);
422 else if (empty_filler)
423 fputs (empty_filler, stdout);
425 else if (empty_filler)
426 fputs (empty_filler, stdout);
429 /* Print the join of LINE1 and LINE2. */
432 prjoin (struct line *line1, struct line *line2)
434 const struct outlist *outlist;
436 outlist = outlist_head.next;
439 const struct outlist *o;
449 if (line1 == &uni_blank)
452 field = join_field_2;
457 field = join_field_1;
462 line = (o->file == 1 ? line1 : line2);
463 assert (o->field >= 0);
466 prfield (field, line);
470 putchar (tab ? tab : ' ');
478 if (line1 == &uni_blank)
485 prfield (join_field_1, line1);
486 for (i = 0; i < join_field_1 && i < line1->nfields; ++i)
488 putchar (tab ? tab : ' ');
491 for (i = join_field_1 + 1; i < line1->nfields; ++i)
493 putchar (tab ? tab : ' ');
497 for (i = 0; i < join_field_2 && i < line2->nfields; ++i)
499 putchar (tab ? tab : ' ');
502 for (i = join_field_2 + 1; i < line2->nfields; ++i)
504 putchar (tab ? tab : ' ');
511 /* Print the join of the files in FP1 and FP2. */
514 join (FILE *fp1, FILE *fp2)
516 struct seq seq1, seq2;
518 int diff, i, j, eof1, eof2;
520 /* Read the first line of each file. */
526 while (seq1.count && seq2.count)
528 diff = keycmp (&seq1.lines[0], &seq2.lines[0]);
531 if (print_unpairables_1)
532 prjoin (&seq1.lines[0], &uni_blank);
533 freeline (&seq1.lines[0]);
540 if (print_unpairables_2)
541 prjoin (&uni_blank, &seq2.lines[0]);
542 freeline (&seq2.lines[0]);
548 /* Keep reading lines from file1 as long as they continue to
549 match the current line from file2. */
552 if (!getseq (fp1, &seq1))
558 while (!keycmp (&seq1.lines[seq1.count - 1], &seq2.lines[0]));
560 /* Keep reading lines from file2 as long as they continue to
561 match the current line from file1. */
564 if (!getseq (fp2, &seq2))
570 while (!keycmp (&seq1.lines[0], &seq2.lines[seq2.count - 1]));
574 for (i = 0; i < seq1.count - 1; ++i)
575 for (j = 0; j < seq2.count - 1; ++j)
576 prjoin (&seq1.lines[i], &seq2.lines[j]);
579 for (i = 0; i < seq1.count - 1; ++i)
580 freeline (&seq1.lines[i]);
583 seq1.lines[0] = seq1.lines[seq1.count - 1];
589 for (i = 0; i < seq2.count - 1; ++i)
590 freeline (&seq2.lines[i]);
593 seq2.lines[0] = seq2.lines[seq2.count - 1];
600 if (print_unpairables_1 && seq1.count)
602 prjoin (&seq1.lines[0], &uni_blank);
603 freeline (&seq1.lines[0]);
604 while (get_line (fp1, &line))
606 prjoin (&line, &uni_blank);
611 if (print_unpairables_2 && seq2.count)
613 prjoin (&uni_blank, &seq2.lines[0]);
614 freeline (&seq2.lines[0]);
615 while (get_line (fp2, &line))
617 prjoin (&uni_blank, &line);
626 /* Add a field spec for field FIELD of file FILE to `outlist'. */
629 add_field (int file, int field)
633 assert (file == 0 || file == 1 || file == 2);
634 assert (file == 0 ? field < 0 : field >= 0);
636 o = (struct outlist *) xmalloc (sizeof (struct outlist));
641 /* Add to the end of the list so the fields are in the right order. */
642 outlist_end->next = o;
646 /* Convert a single field specifier string, S, to a *FILE_INDEX, *FIELD_INDEX
647 pair. In S, the field index string is 1-based; *FIELD_INDEX is zero-based.
648 If S is valid, return zero. Otherwise, give a diagnostic, don't update
649 *FILE_INDEX or *FIELD_INDEX, and return nonzero. */
652 decode_field_spec (const char *s, int *file_index, int *field_index)
656 /* The first character must be 0, 1, or 2. */
663 /* Give *field_index an invalid value. */
669 /* `0' must be all alone -- no `.FIELD'. */
670 error (0, 0, _("invalid field specifier: `%s'"), s);
676 if (s[1] == '.' && s[2] != '\0')
681 s_err = xstrtol (s + 2, NULL, 10, &tmp_long, "");
682 if (s_err != LONGINT_OK || tmp_long <= 0 || tmp_long > INT_MAX)
684 error (0, 0, _("invalid field number: `%s'"), s + 2);
688 *file_index = s[0] - '0';
689 /* Convert to a zero-based index. */
690 *field_index = (int) tmp_long - 1;
697 error (0, 0, _("invalid file number in field spec: `%s'"), s);
703 /* Add the comma or blank separated field spec(s) in STR to `outlist'.
704 Return nonzero to indicate failure. */
707 add_field_list (const char *c_str)
711 /* Make a writable copy of c_str. */
712 str = (char *) alloca (strlen (c_str) + 1);
719 int file_index, field_index;
722 p = strpbrk (p, ", \t");
725 invalid = decode_field_spec (spec_item, &file_index, &field_index);
728 add_field (file_index, field_index);
729 uni_blank.nfields = max (uni_blank.nfields, field_index);
735 /* Create a blank line with COUNT fields separated by tabs. */
738 make_blank (struct line *blank, int count)
741 blank->nfields = count;
742 blank->beg = xmalloc (blank->nfields + 1);
743 blank->fields = (struct field *) xmalloc (sizeof (struct field) * count);
744 for (i = 0; i < blank->nfields; i++)
746 blank->beg[i] = '\t';
747 blank->fields[i].beg = &blank->beg[i];
748 blank->fields[i].len = 0;
750 blank->beg[i] = '\0';
751 blank->lim = &blank->beg[i];
755 main (int argc, char **argv)
759 int optc, prev_optc = 0, nfiles;
761 program_name = argv[0];
762 setlocale (LC_ALL, "");
763 bindtextdomain (PACKAGE, LOCALEDIR);
764 textdomain (PACKAGE);
766 /* Initialize this before parsing options. In parsing options,
767 it may be increased. */
768 uni_blank.nfields = 1;
770 parse_long_options (argc, argv, "join", GNU_PACKAGE, VERSION, usage);
775 while ((optc = getopt_long_only (argc, argv, "-a:e:i1:2:o:t:v:", longopts,
790 if (xstrtol (optarg, NULL, 10, &val, "") != LONGINT_OK
791 || (val != 1 && val != 2))
792 error (EXIT_FAILURE, 0, _("invalid field number: `%s'"), optarg);
794 print_unpairables_1 = 1;
796 print_unpairables_2 = 1;
800 empty_filler = optarg;
808 if (xstrtol (optarg, NULL, 10, &val, "") != LONGINT_OK
809 || val <= 0 || val > INT_MAX)
811 error (EXIT_FAILURE, 0,
812 _("invalid field number for file 1: `%s'"), optarg);
814 join_field_1 = (int) val - 1;
818 if (xstrtol (optarg, NULL, 10, &val, "") != LONGINT_OK
819 || val <= 0 || val > INT_MAX)
820 error (EXIT_FAILURE, 0,
821 _("invalid field number for file 2: `%s'"), optarg);
822 join_field_2 = (int) val - 1;
826 if (xstrtol (optarg, NULL, 10, &val, "") != LONGINT_OK
827 || val <= 0 || val > INT_MAX)
828 error (EXIT_FAILURE, 0, _("invalid field number: `%s'"), optarg);
829 join_field_1 = join_field_2 = (int) val - 1;
833 if (add_field_list (optarg))
841 case 1: /* Non-option argument. */
842 if (prev_optc == 'o' && optind <= argc - 2)
844 if (add_field_list (optarg))
847 /* Might be continuation of args to -o. */
848 continue; /* Don't change `prev_optc'. */
853 error (0, 0, _("too many non-option arguments"));
856 names[nfiles++] = optarg;
865 /* Now that we've seen the options, we can construct the blank line
867 make_blank (&uni_blank, uni_blank.nfields);
871 error (0, 0, _("too few non-option arguments"));
875 fp1 = strcmp (names[0], "-") ? fopen (names[0], "r") : stdin;
877 error (EXIT_FAILURE, errno, "%s", names[0]);
878 fp2 = strcmp (names[1], "-") ? fopen (names[1], "r") : stdin;
880 error (EXIT_FAILURE, errno, "%s", names[1]);
882 error (EXIT_FAILURE, errno, _("both files cannot be standard input"));
885 if (fp1 != stdin && fclose (fp1) == EOF)
886 error (EXIT_FAILURE, errno, "%s", names[0]);
887 if (fp2 != stdin && fclose (fp2) == EOF)
888 error (EXIT_FAILURE, errno, "%s", names[1]);
889 if ((fp1 == stdin || fp2 == stdin) && fclose (stdin) == EOF)
890 error (EXIT_FAILURE, errno, "-");
891 if (ferror (stdout) || fclose (stdout) == EOF)
892 error (EXIT_FAILURE, errno, _("write error"));