1 /* pr -- convert text files for printing.
2 Copyright (C) 1988-2024 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 3 of the License, or
7 (at your option) any later version.
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, see <https://www.gnu.org/licenses/>. */
17 /* By Pete TerMaat, with considerable refinement by Roland Huebner. */
19 /* Things to watch: Sys V screws up on ...
20 pr -n -3 -s: /usr/dict/words
21 pr -m -o10 -n /usr/dict/words{,,,}
22 pr -6 -a -n -o5 /usr/dict/words
26 Keep a things_to_do list of functions to call when we know we have
27 something to print. Cleaner than current series of checks.
29 Improve the printing of control prefixes.
31 Expand the file name in the centered header line to a full file name.
36 If the input_tab_char differs from the default value TAB
37 ('-e[CHAR[...]]' is used), any input text tab is expanded to the
38 default width of 8 spaces (compare char_to_clump). - Same as SunOS
41 The treatment of the number_separator (compare add_line_number):
42 The default value TAB of the number_separator ('-n[SEP[...]]') doesn't
43 be thought to be an input character. An optional '-e'-input has no
45 - With single column output
46 only one POSIX requirement has to be met:
47 The default n-separator should be a TAB. The consequence is a
48 different width between the number and the text if the output position
49 of the separator changes, i.e., it depends upon the left margin used.
50 That's not nice but easy-to-use together with the defaults of other
51 utilities, e.g. sort or cut. - Same as SunOS does.
52 - With multicolumn output
53 two conflicting POSIX requirements exist:
54 First "default n-separator is TAB", second "output text columns shall
55 be of equal width". Moreover POSIX specifies the number+separator a
56 part of the column, together with '-COLUMN' and '-a -COLUMN'.
57 (With -m output the number shall occupy each line only once. Exactly
58 the same situation as single column output exists.)
59 GNU pr gives priority to the 2nd requirement and observes POSIX
60 column definition. The n-separator TAB is expanded to the same number
61 of spaces in each column using the default value 8. Tabification is
62 only performed if it is compatible with the output position.
63 Consequence: The output text columns are of equal width. The layout
64 of a page does not change if the left margin varies. - Looks better
65 than the SunOS approach.
66 SunOS pr gives priority to the 1st requirement. n-separator TAB
67 width varies with each column. Only the width of text part of the
69 Consequence: The output text columns don't have equal width. The
70 widths and the layout of the whole page varies with the left margin.
71 An overflow of the line length (without margin) over the input value
74 The interference of the POSIX-compliant small letter options -w and -s:
75 ("interference" means "setting a _separator_ with -s switches off the
76 column structure and the default - not generally - page_width,
78 options: text form / separator: equivalent new options:
80 --------------------------------------------------------------------
81 1. -- -- columns / space --
82 trunc. to page_width = 72
83 2. -- -s[:] full lines / TAB[:] -J --sep-string[="<TAB>"|:]
85 3. -w l -- columns / space -W l
86 trunc. to page_width = l
87 4. -w l -s[:] columns / no sep.[:] -W l --sep-string[=:]
88 trunc. to page_width = l
89 --------------------------------------------------------------------
94 Including version 1.22i:
95 Some SMALL LETTER options have been redefined with the object of a
96 better POSIX compliance. The output of some further cases has been
97 adapted to other UNIXes. A violation of downward compatibility has to
99 Some NEW CAPITAL LETTER options ( -J, -S, -W) has been introduced to
100 turn off unexpected interference of small letter options (-s and -w
101 together with the three column options).
102 -N option and the second argument LAST_PAGE of +FIRST_PAGE offer more
103 flexibility; The detailed handling of form feeds set in the input
104 files requires -T option.
106 Capital letter options dominate small letter ones.
108 Some of the option-arguments cannot be specified as separate arguments
109 from the preceding option letter (already stated in POSIX specification).
111 Form feeds in the input cause page breaks in the output. Multiple
112 form feeds produce empty pages.
114 +FIRST_PAGE[:LAST_PAGE], --pages=FIRST_PAGE[:LAST_PAGE]
115 begin [stop] printing with page FIRST_[LAST_]PAGE
117 -COLUMN, --columns=COLUMN
118 Produce output that is COLUMN columns wide and
119 print columns down, unless -a is used. Balance number of
120 lines in the columns on each page.
122 -a, --across Print columns across rather than down, used
123 together with -COLUMN. The input
128 will be printed with '-a -3' as
132 -b Balance columns on the last page.
133 -b is no longer an independent option. It's always used
134 together with -COLUMN (unless -a is used) to get a
135 consistent formulation with "FF set by hand" in input
136 files. Each formfeed found terminates the number of lines
137 to be read with the actual page. The situation for
138 printing columns down is equivalent to that on the last
139 page. So we need a balancing.
141 Keeping -b as an underground option guarantees some
142 downward compatibility. Utilities using pr with -b
143 (a most frequently used form) still work as usual.
145 -c, --show-control-chars
146 Print nonprintable characters as control prefixes.
147 Control-g is printed as ^G (use hat notation) and
148 octal backslash notation.
150 -d, --double-space Double space the output.
152 -D FORMAT, --date-format=FORMAT Use FORMAT for the header date.
154 -e[CHAR[WIDTH]], --expand-tabs[=CHAR[WIDTH]]
155 Expand tabs to spaces on input. Optional argument CHAR
156 is the input TAB character. (Default is TAB). Optional
157 argument WIDTH is the input TAB character's width.
160 -F, -f, --form-feed Use formfeeds instead of newlines to separate
161 pages. A three line HEADER is used, no TRAILER with -F,
162 without -F both HEADER and TRAILER are made of five lines.
164 -h HEADER, --header=HEADER
165 Replace the filename in the header with the string HEADER.
166 A centered header is used.
168 -i[CHAR[WIDTH]], --output-tabs[=CHAR[WIDTH]]
169 Replace spaces with tabs on output. Optional argument
170 CHAR is the output TAB character. (Default is TAB).
171 Optional argument WIDTH is the output TAB character's
172 width. (Default is 8)
174 -J, --join-lines Merge lines of full length, turns off -W/-w
175 line truncation, no column alignment, --sep-string[=STRING]
176 sets separators, works with all column options
177 (-COLUMN | -a -COLUMN | -m).
178 -J has been introduced (together with -W and --sep-string) to
179 disentangle the old (POSIX compliant) options -w, -s
180 along with the 3 column options.
182 -l PAGE_LENGTH, --length=PAGE_LENGTH
183 Set the page length to PAGE_LENGTH lines. Default is 66,
184 including 5 lines of HEADER and 5 lines of TRAILER
185 without -F, but only 3 lines of HEADER and no TRAILER
186 with -F (i.e the number of text lines defaults to 56 or
189 -m, --merge Print files in parallel; pad_across_to align
190 columns; truncate lines and print separator strings;
191 Do it also with empty columns to get a continuous line
192 numbering and column marking by separators throughout
193 the whole merged file.
195 Empty pages in some input files produce empty columns
196 [marked by separators] in the merged pages. Completely
197 empty merged pages show no column separators at all.
199 The layout of a merged page is ruled by the largest form
200 feed distance of the single pages at that page. Shorter
201 columns will be filled up with empty lines.
203 Together with -J option join lines of full length and
204 set separators when -S option is used.
206 -n[SEP[DIGITS]], --number-lines[=SEP[DIGITS]]
207 Provide DIGITS digit line numbering (default for DIGITS
208 is 5). With multicolumn output the number occupies the
209 first DIGITS column positions of each text column or only
210 each line of -m output.
211 With single column output the number precedes each line
213 Optional argument SEP is the character appended to the
214 line number to separate it from the text followed.
215 The default separator is a TAB. In a strict sense a TAB
216 is always printed with single column output only. The
217 TAB-width varies with the TAB-position, e.g. with the
218 left margin specified by -o option.
219 With multicolumn output priority is given to "equal width
220 of output columns" (a POSIX specification). The TAB-width
221 is fixed to the value of the 1st column and does not
222 change with different values of left margin. That means a
223 fixed number of spaces is always printed in the place of
224 a TAB. The tabification depends upon the output
227 Default counting of the line numbers starts with 1st
228 line of the input file (not the 1st line printed,
229 compare the --page option and -N option).
231 -N NUMBER, --first-line-number=NUMBER
232 Start line counting with the number NUMBER at the 1st
233 line of first page printed (mostly not the 1st line of
236 -o MARGIN, --indent=MARGIN
237 Offset each line with a margin MARGIN spaces wide.
238 Total page width is the size of the margin plus the
239 PAGE_WIDTH set with -W/-w option.
241 -r, --no-file-warnings
242 Omit warning when a file cannot be opened.
244 -s[CHAR], --separator[=CHAR]
245 Separate columns by a single character CHAR, default for
246 CHAR is the TAB character without -w and 'no char' with -w.
247 Without '-s' default separator 'space' is set.
248 -s[CHAR] turns off line truncation of all 3 column options
249 (-COLUMN|-a -COLUMN|-m) except -w is set. That is a POSIX
250 compliant formulation. The source code translates -s into
251 the new options -S and -J, also -W if required.
253 -S[STRING], --sep-string[=STRING]
254 Separate columns by any string STRING. The -S option
255 doesn't react upon the -W/-w option (unlike -s option
256 does). It defines a separator nothing else.
257 Without -S: Default separator TAB is used with -J and
258 'space' otherwise (same as -S" ").
259 With -S "": No separator is used.
260 Quotes should be used with blanks and some shell active
262 -S is problematic because in its obsolete form you
263 cannot use -S "STRING", but in its standard form you
264 must use -S "STRING" if STRING is empty. Use
265 --sep-string to avoid the ambiguity.
267 -t, --omit-header Do not print headers or footers but retain form
268 feeds set in the input files.
270 -T, --omit-pagination
271 Do not print headers or footers, eliminate any pagination
272 by form feeds set in the input files.
274 -v, --show-nonprinting
275 Print nonprintable characters as escape sequences. Use
276 octal backslash notation. Control-G becomes \007.
278 -w PAGE_WIDTH, --width=PAGE_WIDTH
279 Set page width to PAGE_WIDTH characters for multiple
280 text-column output only (default for PAGE_WIDTH is 72).
281 -s[CHAR] turns off the default page width and any line
282 truncation. Lines of full length will be merged,
283 regardless of the column options set. A POSIX compliant
286 -W PAGE_WIDTH, --page-width=PAGE_WIDTH
287 Set the page width to PAGE_WIDTH characters. That's valid
288 with and without a column option. Text lines will be
289 truncated, unless -J is used. Together with one of the
290 column options (-COLUMN| -a -COLUMN| -m) column alignment
292 Default is 72 characters.
293 Without -W PAGE_WIDTH
294 - but with one of the column options default truncation of
295 72 characters is used (to keep downward compatibility
296 and to simplify most frequently met column tasks).
297 Column alignment and column separators are used.
298 - and without any of the column options NO line truncation
299 is used (to keep downward compatibility and to meet most
300 frequent tasks). That's equivalent to -W 72 -J .
302 With/without -W PAGE_WIDTH the header line is always
303 truncated to avoid line overflow.
305 (In pr versions newer than 1.14 -S option does no longer
314 #include <sys/types.h>
317 #include "hard-locale.h"
318 #include "mbswidth.h"
320 #include "stat-time.h"
322 #include "strftime.h"
324 #include "xstrtol-error.h"
325 #include "xdectoint.h"
327 /* The official name of this program (e.g., no 'g' prefix). */
328 #define PROGRAM_NAME "pr"
331 proper_name ("Pete TerMaat"), \
332 proper_name ("Roland Huebner")
334 /* Used with start_position in the struct COLUMN described below.
335 If start_position == ANYWHERE, we aren't truncating columns and
336 can begin printing a column anywhere. Otherwise we must pad to
337 the horizontal position start_position. */
340 /* Each column has one of these structures allocated for it.
341 If we're only dealing with one file, fp is the same for all
344 The general strategy is to spend time setting up these column
345 structures (storing columns if necessary), after which printing
346 is a matter of flitting from column to column and calling
349 Parallel files, single files printing across in multiple
350 columns, and single files printing down in multiple columns all
351 fit the same printing loop.
353 print_func Function used to print lines in this column.
354 If we're storing this column it will be
355 print_stored(), Otherwise it will be read_line().
357 char_func Function used to process characters in this column.
358 If we're storing this column it will be store_char(),
359 otherwise it will be print_char().
361 current_line Index of the current entry in line_vector, which
362 contains the index of the first character of the
363 current line in buff[].
365 lines_stored Number of lines in this column which are stored in
368 lines_to_print If we're storing this column, lines_to_print is
369 the number of stored_lines which remain to be
370 printed. Otherwise it is the number of lines
371 we can print without exceeding lines_per_body.
373 start_position The horizontal position we want to be in before we
374 print the first character in this column.
376 numbered True means precede this column with a line number. */
378 /* FIXME: There are many unchecked integer overflows in this file,
379 that will cause this command to misbehave given large inputs or
380 options. Many of the "int" values below should be "size_t" or
381 something else like that. */
386 FILE *fp
; /* Input stream for this column. */
387 char const *name
; /* File name. */
391 FF_FOUND
, /* used with -b option, set with \f, changed
392 to ON_HOLD after print_header */
393 ON_HOLD
, /* Hit a form feed. */
396 status
; /* Status of the file pointer. */
398 /* Func to print lines in this col. */
399 bool (*print_func
) (struct COLUMN
*);
401 /* Func to print/store chars in this col. */
402 void (*char_func
) (char);
404 int current_line
; /* Index of current place in line_vector. */
405 int lines_stored
; /* Number of lines stored in buff. */
406 int lines_to_print
; /* No. lines stored or space left on page. */
407 int start_position
; /* Horizontal position of first char. */
409 bool full_page_printed
; /* True means printed without a FF found. */
411 /* p->full_page_printed controls a special case of "FF set by hand":
412 True means a full page has been printed without FF found. To avoid an
413 additional empty page we have to ignore a FF immediately following in
417 typedef struct COLUMN COLUMN
;
419 static int char_to_clump (char c
);
420 static bool read_line (COLUMN
*p
);
421 static bool print_page (void);
422 static bool print_stored (COLUMN
*p
);
423 static bool open_file (char *name
, COLUMN
*p
);
424 static bool skip_to_page (uintmax_t page
);
425 static void print_header (void);
426 static void pad_across_to (int position
);
427 static void add_line_number (COLUMN
*p
);
428 static void getoptnum (char const *n_str
, int min
, int *num
,
430 static void getoptarg (char *arg
, char switch_char
, char *character
,
432 static void print_files (int number_of_files
, char **av
);
433 static void init_parameters (int number_of_files
);
434 static void init_header (char const *filename
, int desc
);
435 static bool init_fps (int number_of_files
, char **av
);
436 static void init_funcs (void);
437 static void init_store_cols (void);
438 static void store_columns (void);
439 static void balance (int total_stored
);
440 static void store_char (char c
);
441 static void pad_down (unsigned int lines
);
442 static void read_rest_of_line (COLUMN
*p
);
443 static void skip_read (COLUMN
*p
, int column_number
);
444 static void print_char (char c
);
445 static void cleanup (void);
446 static void print_sep_string (void);
447 static void separator_string (char const *optarg_S
);
449 /* All of the columns to print. */
450 static COLUMN
*column_vector
;
452 /* When printing a single file in multiple downward columns,
453 we store the leftmost columns contiguously in buff.
454 To print a line from buff, get the index of the first character
455 from line_vector[i], and print up to line_vector[i + 1]. */
458 /* Index of the position in buff where the next character
460 static unsigned int buff_current
;
462 /* The number of characters in buff.
463 Used for allocation of buff and to detect overflow of buff. */
464 static size_t buff_allocated
;
466 /* Array of indices into buff.
467 Each entry is an index of the first character of a line.
468 This is used when storing lines to facilitate shuffling when
469 we do column balancing on the last page. */
470 static int *line_vector
;
472 /* Array of horizontal positions.
473 For each line in line_vector, end_vector[line] is the horizontal
474 position we are in after printing that line. We keep track of this
475 so that we know how much we need to pad to prepare for the next
477 static int *end_vector
;
479 /* (-m) True means we're printing multiple files in parallel. */
480 static bool parallel_files
= false;
482 /* (-m) True means a line starts with some empty columns (some files
483 already CLOSED or ON_HOLD) which we have to align. */
484 static bool align_empty_cols
;
486 /* (-m) True means we have not yet found any printable column in a line.
487 align_empty_cols = true has to be maintained. */
488 static bool empty_line
;
490 /* (-m) False means printable column output precedes a form feed found.
491 Column alignment is done only once. No additional action with that form
493 True means we found only a form feed in a column. Maybe we have to do
494 some column alignment with that form feed. */
497 /* (-[0-9]+) True means we're given an option explicitly specifying
498 number of columns. Used to detect when this option is used with -m
499 and when translating old options to new/long options. */
500 static bool explicit_columns
= false;
502 /* (-t|-T) False means we aren't printing headers and footers. */
503 static bool extremities
= true;
505 /* (-t) True means we retain all FF set by hand in input files.
506 False is set with -T option. */
507 static bool keep_FF
= false;
508 static bool print_a_FF
= false;
510 /* True means we need to print a header as soon as we know we've got input
511 to print after it. */
512 static bool print_a_header
;
514 /* (-f) True means use formfeeds instead of newlines to separate pages. */
515 static bool use_form_feed
= false;
517 /* True means we have read the standard input. */
518 static bool have_read_stdin
= false;
520 /* True means the -a flag has been given. */
521 static bool print_across_flag
= false;
523 /* True means we're printing one file in multiple (>1) downward columns. */
524 static bool storing_columns
= true;
526 /* (-b) True means balance columns on the last page as Sys V does. */
527 /* That's no longer an independent option. With storing_columns = true
528 balance_columns = true is used too (s. function init_parameters).
529 We get a consistent formulation with "FF set by hand" in input files. */
530 static bool balance_columns
= false;
532 /* (-l) Number of lines on a page, including header and footer lines. */
533 static int lines_per_page
= 66;
535 /* Number of lines in the header and footer can be reset to 0 using
537 enum { lines_per_header
= 5 };
538 static int lines_per_body
;
539 enum { lines_per_footer
= 5 };
541 /* (-w|-W) Width in characters of the page. Does not include the width of
543 static int chars_per_line
= 72;
545 /* (-w|W) True means we truncate lines longer than chars_per_column. */
546 static bool truncate_lines
= false;
548 /* (-J) True means we join lines without any line truncation. -J
549 dominates -w option. */
550 static bool join_lines
= false;
552 /* Number of characters in a column. Based on col_sep_length and
554 static int chars_per_column
;
556 /* (-e) True means convert tabs to spaces on input. */
557 static bool untabify_input
= false;
559 /* (-e) The input tab character. */
560 static char input_tab_char
= '\t';
562 /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ...
563 where the leftmost column is 1. */
564 static int chars_per_input_tab
= 8;
566 /* (-i) True means convert spaces to tabs on output. */
567 static bool tabify_output
= false;
569 /* (-i) The output tab character. */
570 static char output_tab_char
= '\t';
572 /* (-i) The width of the output tab. */
573 static int chars_per_output_tab
= 8;
575 /* Keeps track of pending white space. When we hit a nonspace
576 character after some whitespace, we print whitespace, tabbing
577 if necessary to get to output_position + spaces_not_printed. */
578 static int spaces_not_printed
;
580 /* (-o) Number of spaces in the left margin (tabs used when possible). */
581 static int chars_per_margin
= 0;
583 /* Position where the next character will fall.
584 Leftmost position is 0 + chars_per_margin.
585 Rightmost position is chars_per_margin + chars_per_line - 1.
586 This is important for converting spaces to tabs on output. */
587 static int output_position
;
589 /* Horizontal position relative to the current file.
590 (output_position depends on where we are on the page;
591 input_position depends on where we are in the file.)
592 Important for converting tabs to spaces on input. */
593 static int input_position
;
595 /* True if there were any failed opens so we can exit with nonzero
597 static bool failed_opens
= false;
599 /* The number of spaces taken up if we print a tab character with width
600 c_ from position h_. */
601 #define TAB_WIDTH(c_, h_) ((c_) - ((h_) % (c_)))
603 /* The horizontal position we'll be at after printing a tab character
604 of width c_ from the position h_. */
605 #define POS_AFTER_TAB(c_, h_) ((h_) + TAB_WIDTH (c_, h_))
607 /* (-NNN) Number of columns of text to print. */
608 static int columns
= 1;
610 /* (+NNN:MMM) Page numbers on which to begin and stop printing.
611 first_page_number = 0 will be used to check input only. */
612 static uintmax_t first_page_number
= 0;
613 static uintmax_t last_page_number
= UINTMAX_MAX
;
615 /* Number of files open (not closed, not on hold). */
616 static int files_ready_to_read
= 0;
618 /* Current page number. Displayed in header. */
619 static uintmax_t page_number
;
621 /* Current line number. Displayed when -n flag is specified.
623 When printing files in parallel (-m flag), line numbering is as follows:
627 When printing files across (-a flag), ...
631 Otherwise, line numbering is as follows:
634 static int line_number
;
636 /* (-n) True means lines should be preceded by numbers. */
637 static bool numbered_lines
= false;
639 /* (-n) Character which follows each line number. */
640 static char number_separator
= '\t';
642 /* (-n) line counting starts with 1st line of input file (not with 1st
643 line of 1st page printed). */
644 static int line_count
= 1;
646 /* (-n) True means counting of skipped lines starts with 1st line of
647 input file. False means -N option is used in addition, counting of
648 skipped lines not required. */
649 static bool skip_count
= true;
651 /* (-N) Counting starts with start_line_number = NUMBER at 1st line of
652 first page printed, usually not 1st page of input file. */
653 static int start_line_num
= 1;
655 /* (-n) Width in characters of a line number. */
656 static int chars_per_number
= 5;
658 /* Used when widening the first column to accommodate numbers -- only
659 needed when printing files in parallel. Includes width of both the
660 number and the number_separator. */
661 static int number_width
;
663 /* Buffer sprintf uses to format a line number. */
664 static char *number_buff
;
666 /* (-v) True means nonprintable characters are printed as escape sequences.
667 control-g becomes \007. */
668 static bool use_esc_sequence
= false;
670 /* (-c) True means nonprintable characters are printed as control prefixes.
671 control-g becomes ^G. */
672 static bool use_cntrl_prefix
= false;
674 /* (-d) True means output is double spaced. */
675 static bool double_space
= false;
677 /* Number of files opened initially in init_files. Should be 1
678 unless we're printing multiple files in parallel. */
679 static int total_files
= 0;
681 /* (-r) True means don't complain if we can't open a file. */
682 static bool ignore_failed_opens
= false;
684 /* (-S) True means we separate columns with a specified string.
685 -S option does not affect line truncation nor column alignment. */
686 static bool use_col_separator
= false;
688 /* String used to separate columns if the -S option has been specified.
689 Default without -S but together with one of the column options
690 -a|COLUMN|-m is a 'space' and with the -J option a 'tab'. */
691 static char const *col_sep_string
= "";
692 static int col_sep_length
= 0;
693 static char *column_separator
= (char *) " ";
694 static char *line_separator
= (char *) "\t";
696 /* Number of separator characters waiting to be printed as soon as we
697 know that we have any input remaining to be printed. */
698 static int separators_not_printed
;
700 /* Position we need to pad to, as soon as we know that we have input
701 remaining to be printed. */
702 static int padding_not_printed
;
704 /* True means we should pad the end of the page. Remains false until we
705 know we have a page to print. */
706 static bool pad_vertically
;
708 /* (-h) String of characters used in place of the filename in the header. */
709 static char *custom_header
;
711 /* (-D) Date format for the header. */
712 static char const *date_format
;
714 /* The local time zone rules, as per the TZ environment variable. */
715 static timezone_t localtz
;
717 /* Date and file name for the header. */
718 static char *date_text
;
719 static char const *file_text
;
721 /* Output columns available, not counting the date and file name. */
722 static int header_width_available
;
724 static char *clump_buff
;
726 /* True means we read the line no. lines_per_body in skip_read
727 called by skip_to_page. That variable controls the coincidence of a
728 "FF set by hand" and "full_page_printed", see above the definition of
730 static bool last_line
= false;
732 /* For long options that have no equivalent short option, use a
733 non-character as a pseudo short option, starting with CHAR_MAX + 1. */
736 COLUMNS_OPTION
= CHAR_MAX
+ 1,
740 static char const short_options
[] =
741 "-0123456789D:FJN:S::TW:abcde::fh:i::l:mn::o:rs::tvw:";
743 static struct option
const long_options
[] =
745 {"pages", required_argument
, nullptr, PAGES_OPTION
},
746 {"columns", required_argument
, nullptr, COLUMNS_OPTION
},
747 {"across", no_argument
, nullptr, 'a'},
748 {"show-control-chars", no_argument
, nullptr, 'c'},
749 {"double-space", no_argument
, nullptr, 'd'},
750 {"date-format", required_argument
, nullptr, 'D'},
751 {"expand-tabs", optional_argument
, nullptr, 'e'},
752 {"form-feed", no_argument
, nullptr, 'f'},
753 {"header", required_argument
, nullptr, 'h'},
754 {"output-tabs", optional_argument
, nullptr, 'i'},
755 {"join-lines", no_argument
, nullptr, 'J'},
756 {"length", required_argument
, nullptr, 'l'},
757 {"merge", no_argument
, nullptr, 'm'},
758 {"number-lines", optional_argument
, nullptr, 'n'},
759 {"first-line-number", required_argument
, nullptr, 'N'},
760 {"indent", required_argument
, nullptr, 'o'},
761 {"no-file-warnings", no_argument
, nullptr, 'r'},
762 {"separator", optional_argument
, nullptr, 's'},
763 {"sep-string", optional_argument
, nullptr, 'S'},
764 {"omit-header", no_argument
, nullptr, 't'},
765 {"omit-pagination", no_argument
, nullptr, 'T'},
766 {"show-nonprinting", no_argument
, nullptr, 'v'},
767 {"width", required_argument
, nullptr, 'w'},
768 {"page-width", required_argument
, nullptr, 'W'},
769 {GETOPT_HELP_OPTION_DECL
},
770 {GETOPT_VERSION_OPTION_DECL
},
771 {nullptr, 0, nullptr, 0}
774 static _Noreturn
void
775 integer_overflow (void)
777 error (EXIT_FAILURE
, 0, _("integer overflow"));
780 /* Return the number of columns that have either an open file or
785 cols_ready_to_print (void)
792 for (q
= column_vector
, i
= 0; i
< columns
; ++q
, ++i
)
793 if (q
->status
== OPEN
794 || q
->status
== FF_FOUND
/* With -b: To print a header only */
795 || (storing_columns
&& q
->lines_stored
> 0 && q
->lines_to_print
> 0))
800 /* Estimate first_ / last_page_number
801 using option +FIRST_PAGE:LAST_PAGE */
804 first_last_page (int oi
, char c
, char const *pages
)
808 uintmax_t last
= UINTMAX_MAX
;
809 strtol_error err
= xstrtoumax (pages
, &p
, 10, &first
, "");
810 if (err
!= LONGINT_OK
&& err
!= LONGINT_INVALID_SUFFIX_CHAR
)
811 xstrtol_fatal (err
, oi
, c
, long_options
, pages
);
813 if (p
== pages
|| !first
)
818 char const *p1
= p
+ 1;
819 err
= xstrtoumax (p1
, &p
, 10, &last
, "");
820 if (err
!= LONGINT_OK
)
821 xstrtol_fatal (err
, oi
, c
, long_options
, pages
);
822 if (p1
== p
|| last
< first
)
829 first_page_number
= first
;
830 last_page_number
= last
;
834 /* Parse column count string S, and if it's valid (1 or larger and
835 within range of the type of 'columns') set the global variables
836 columns and explicit_columns. Otherwise, exit with a diagnostic. */
839 parse_column_count (char const *s
)
841 getoptnum (s
, 1, &columns
, _("invalid number of columns"));
842 explicit_columns
= true;
845 /* Estimate length of col_sep_string with option -S. */
848 separator_string (char const *optarg_S
)
850 size_t len
= strlen (optarg_S
);
853 col_sep_length
= len
;
854 col_sep_string
= optarg_S
;
858 main (int argc
, char **argv
)
860 unsigned int n_files
;
861 bool old_options
= false;
866 /* Accumulate the digits of old-style options like -99. */
867 char *column_count_string
= nullptr;
871 initialize_main (&argc
, &argv
);
872 set_program_name (argv
[0]);
873 setlocale (LC_ALL
, "");
874 bindtextdomain (PACKAGE
, LOCALEDIR
);
875 textdomain (PACKAGE
);
877 atexit (close_stdout
);
880 file_names
= (argc
> 1
881 ? xnmalloc (argc
- 1, sizeof (char *))
887 int c
= getopt_long (argc
, argv
, short_options
, long_options
, &oi
);
893 /* Accumulate column-count digits specified via old-style options. */
894 if (n_digits
+ 1 >= n_alloc
)
896 = X2REALLOC (column_count_string
, &n_alloc
);
897 column_count_string
[n_digits
++] = c
;
898 column_count_string
[n_digits
] = '\0';
906 case 1: /* Non-option argument. */
907 /* long option --page dominates old '+FIRST_PAGE ...'. */
908 if (! (first_page_number
== 0
909 && *optarg
== '+' && first_last_page (-2, '+', optarg
+ 1)))
910 file_names
[n_files
++] = optarg
;
913 case PAGES_OPTION
: /* --pages=FIRST_PAGE[:LAST_PAGE] */
914 { /* dominates old opt +... */
916 error (EXIT_FAILURE
, 0,
917 _("'--pages=FIRST_PAGE[:LAST_PAGE]' missing argument"));
918 else if (! first_last_page (oi
, 0, optarg
))
919 error (EXIT_FAILURE
, 0, _("invalid page range %s"),
924 case COLUMNS_OPTION
: /* --columns=COLUMN */
926 parse_column_count (optarg
);
928 /* If there was a prior column count specified via the
929 short-named option syntax, e.g., -9, ensure that this
930 long-name-specified value overrides it. */
931 free (column_count_string
);
932 column_count_string
= nullptr;
938 print_across_flag
= true;
939 storing_columns
= false;
942 balance_columns
= true;
945 use_cntrl_prefix
= true;
951 date_format
= optarg
;
955 getoptarg (optarg
, 'e', &input_tab_char
,
956 &chars_per_input_tab
);
957 /* Could check tab width > 0. */
958 untabify_input
= true;
962 use_form_feed
= true;
965 custom_header
= optarg
;
969 getoptarg (optarg
, 'i', &output_tab_char
,
970 &chars_per_output_tab
);
971 /* Could check tab width > 0. */
972 tabify_output
= true;
978 getoptnum (optarg
, 1, &lines_per_page
,
979 _("'-l PAGE_LENGTH' invalid number of lines"));
982 parallel_files
= true;
983 storing_columns
= false;
986 numbered_lines
= true;
988 getoptarg (optarg
, 'n', &number_separator
,
993 getoptnum (optarg
, INT_MIN
, &start_line_num
,
994 _("'-N NUMBER' invalid starting line number"));
997 getoptnum (optarg
, 0, &chars_per_margin
,
998 _("'-o MARGIN' invalid line offset"));
1001 ignore_failed_opens
= true;
1006 if (!use_col_separator
&& optarg
)
1007 separator_string (optarg
);
1011 /* Reset an additional input of -s, -S dominates -s */
1012 col_sep_string
= "";
1014 use_col_separator
= true;
1016 separator_string (optarg
);
1019 extremities
= false;
1023 extremities
= false;
1027 use_esc_sequence
= true;
1034 getoptnum (optarg
, 1, &tmp_cpl
,
1035 _("'-w PAGE_WIDTH' invalid number of characters"));
1036 if (! truncate_lines
)
1037 chars_per_line
= tmp_cpl
;
1041 old_w
= false; /* dominates -w */
1042 truncate_lines
= true;
1043 getoptnum (optarg
, 1, &chars_per_line
,
1044 _("'-W PAGE_WIDTH' invalid number of characters"));
1046 case_GETOPT_HELP_CHAR
;
1047 case_GETOPT_VERSION_CHAR (PROGRAM_NAME
, AUTHORS
);
1049 usage (EXIT_FAILURE
);
1054 if (column_count_string
)
1056 parse_column_count (column_count_string
);
1057 free (column_count_string
);
1061 date_format
= (getenv ("POSIXLY_CORRECT") && !hard_locale (LC_TIME
)
1063 : "%Y-%m-%d %H:%M");
1065 localtz
= tzalloc (getenv ("TZ"));
1067 /* Now we can set a reasonable initial value: */
1068 if (first_page_number
== 0)
1069 first_page_number
= 1;
1071 if (parallel_files
&& explicit_columns
)
1072 error (EXIT_FAILURE
, 0,
1073 _("cannot specify number of columns when printing in parallel"));
1075 if (parallel_files
&& print_across_flag
)
1076 error (EXIT_FAILURE
, 0,
1077 _("cannot specify both printing across and printing in parallel"));
1079 /* Translate some old short options to new/long options.
1080 To meet downward compatibility with other UNIX pr utilities
1081 and some POSIX specifications. */
1087 if (parallel_files
|| explicit_columns
)
1090 truncate_lines
= true;
1092 /* adapt HP-UX and SunOS: -s = no separator;
1094 use_col_separator
= true;
1097 /* old -w sets width with columns only
1101 else if (!use_col_separator
)
1103 /* No -S option read */
1104 if (old_s
&& (parallel_files
|| explicit_columns
))
1106 if (!truncate_lines
)
1108 /* old -s (without -w and -W) annuls column alignment,
1109 uses fields, activate -J */
1111 if (col_sep_length
> 0)
1113 use_col_separator
= true;
1117 /* adapt HP-UX and SunOS: -s = no separator;
1119 use_col_separator
= true;
1124 for (; optind
< argc
; optind
++)
1126 file_names
[n_files
++] = argv
[optind
];
1131 /* No file arguments specified; read from standard input. */
1132 print_files (0, nullptr);
1137 print_files (n_files
, file_names
);
1140 for (unsigned int i
= 0; i
< n_files
; i
++)
1141 print_files (1, &file_names
[i
]);
1147 if (have_read_stdin
&& fclose (stdin
) == EOF
)
1148 error (EXIT_FAILURE
, errno
, _("standard input"));
1149 main_exit (failed_opens
? EXIT_FAILURE
: EXIT_SUCCESS
);
1152 /* Parse numeric arguments, ensuring MIN <= number <= INT_MAX. */
1155 getoptnum (char const *n_str
, int min
, int *num
, char const *err
)
1157 intmax_t tnum
= xdectoimax (n_str
, min
, INT_MAX
, "", err
, 0);
1161 /* Parse options of the form -scNNN.
1163 Example: -nck, where 'n' is the option, c is the optional number
1164 separator, and k is the optional width of the field used when printing
1168 getoptarg (char *arg
, char switch_char
, char *character
, int *number
)
1172 error (0, 0, _("'-%c': Invalid argument: %s"), switch_char
, quote (arg
));
1173 usage (EXIT_FAILURE
);
1176 if (!ISDIGIT (*arg
))
1177 *character
= *arg
++;
1181 strtol_error e
= xstrtol (arg
, nullptr, 10, &tmp_long
, "");
1182 if (e
== LONGINT_OK
)
1185 e
= LONGINT_INVALID
;
1186 else if (INT_MAX
< tmp_long
)
1187 e
= LONGINT_OVERFLOW
;
1189 if (e
!= LONGINT_OK
)
1191 error (0, e
& LONGINT_OVERFLOW
? EOVERFLOW
: 0,
1192 _("'-%c' extra characters or invalid number in the argument: %s"),
1193 switch_char
, quote (arg
));
1194 usage (EXIT_FAILURE
);
1200 /* Set parameters related to formatting. */
1203 init_parameters (int number_of_files
)
1205 int chars_used_by_number
= 0;
1207 lines_per_body
= lines_per_page
- lines_per_header
- lines_per_footer
;
1208 if (lines_per_body
<= 0)
1210 extremities
= false;
1213 if (extremities
== false)
1214 lines_per_body
= lines_per_page
;
1217 lines_per_body
= MAX (1, lines_per_body
/ 2);
1219 /* If input is stdin, cannot print parallel files. BSD dumps core
1221 if (number_of_files
== 0)
1222 parallel_files
= false;
1225 columns
= number_of_files
;
1227 /* One file, multi columns down: -b option is set to get a consistent
1228 formulation with "FF set by hand" in input files. */
1229 if (storing_columns
)
1230 balance_columns
= true;
1232 /* Tabification is assumed for multiple columns. */
1235 if (!use_col_separator
)
1237 /* Use default separator */
1239 col_sep_string
= line_separator
;
1241 col_sep_string
= column_separator
;
1244 use_col_separator
= true;
1246 /* It's rather pointless to define a TAB separator with column
1248 else if (!join_lines
&& col_sep_length
== 1 && *col_sep_string
== '\t')
1249 col_sep_string
= column_separator
;
1251 truncate_lines
= true;
1252 if (! (col_sep_length
== 1 && *col_sep_string
== '\t'))
1253 untabify_input
= true;
1254 tabify_output
= true;
1257 storing_columns
= false;
1259 /* -J dominates -w in any case */
1261 truncate_lines
= false;
1265 int chars_per_default_tab
= 8;
1267 line_count
= start_line_num
;
1269 /* To allow input tab-expansion (-e sensitive) use:
1270 if (number_separator == input_tab_char)
1271 number_width = chars_per_number
1272 + TAB_WIDTH (chars_per_input_tab, chars_per_number); */
1274 /* Estimate chars_per_text without any margin and keep it constant. */
1275 if (number_separator
== '\t')
1276 number_width
= (chars_per_number
1277 + TAB_WIDTH (chars_per_default_tab
, chars_per_number
));
1279 number_width
= chars_per_number
+ 1;
1281 /* The number is part of the column width unless we are
1282 printing files in parallel. */
1284 chars_used_by_number
= number_width
;
1287 int sep_chars
, useful_chars
;
1288 if (ckd_mul (&sep_chars
, columns
- 1, col_sep_length
))
1289 sep_chars
= INT_MAX
;
1290 if (ckd_sub (&useful_chars
, chars_per_line
- chars_used_by_number
,
1293 chars_per_column
= useful_chars
/ columns
;
1295 if (chars_per_column
< 1)
1296 error (EXIT_FAILURE
, 0, _("page width too narrow"));
1301 number_buff
= xmalloc (MAX (chars_per_number
,
1302 INT_STRLEN_BOUND (line_number
)) + 1);
1305 /* Pick the maximum between the tab width and the width of an
1307 The width of an escape sequence (4) isn't the lower limit any longer.
1308 We've to use 8 as the lower limit, if we use chars_per_default_tab = 8
1309 to expand a tab which is not an input_tab-char. */
1311 clump_buff
= xmalloc (MAX (8, chars_per_input_tab
));
1314 /* Open the necessary files,
1315 maintaining a COLUMN structure for each column.
1317 With multiple files, each column p has a different p->fp.
1318 With single files, each column p has the same p->fp.
1319 Return false if (number_of_files > 0) and no files can be opened,
1322 With each column/file p, p->full_page_printed is initialized,
1323 see also open_file. */
1326 init_fps (int number_of_files
, char **av
)
1332 free (column_vector
);
1333 column_vector
= xnmalloc (columns
, sizeof (COLUMN
));
1337 int files_left
= number_of_files
;
1338 for (p
= column_vector
; files_left
--; ++p
, ++av
)
1340 if (! open_file (*av
, p
))
1348 init_header ("", -1);
1353 if (number_of_files
> 0)
1355 if (! open_file (*av
, p
))
1357 init_header (*av
, fileno (p
->fp
));
1358 p
->lines_stored
= 0;
1362 p
->name
= _("standard input");
1364 have_read_stdin
= true;
1366 p
->full_page_printed
= false;
1368 init_header ("", -1);
1369 p
->lines_stored
= 0;
1372 char const *firstname
= p
->name
;
1373 FILE *firstfp
= p
->fp
;
1375 for (i
= columns
- 1, ++p
; i
; --i
, ++p
)
1377 p
->name
= firstname
;
1380 p
->full_page_printed
= false;
1381 p
->lines_stored
= 0;
1384 files_ready_to_read
= total_files
;
1388 /* Determine print_func and char_func, the functions
1389 used by each column for printing and/or storing.
1391 Determine the horizontal position desired when we begin
1392 printing a column (p->start_position). */
1400 h
= chars_per_margin
;
1402 if (!truncate_lines
)
1406 /* When numbering lines of parallel files, we enlarge the
1407 first column to accommodate the number. Looks better than
1408 the Sys V approach. */
1409 if (parallel_files
&& numbered_lines
)
1410 h_next
= h
+ chars_per_column
+ number_width
;
1412 h_next
= h
+ chars_per_column
;
1415 /* Enlarge p->start_position of first column to use the same form of
1416 padding_not_printed with all columns. */
1417 h
= h
+ col_sep_length
;
1419 /* This loop takes care of all but the rightmost column. */
1421 for (p
= column_vector
, i
= 1; i
< columns
; ++p
, ++i
)
1423 if (storing_columns
) /* One file, multi columns down. */
1425 p
->char_func
= store_char
;
1426 p
->print_func
= print_stored
;
1429 /* One file, multi columns across; or parallel files. */
1431 p
->char_func
= print_char
;
1432 p
->print_func
= read_line
;
1435 /* Number only the first column when printing files in
1437 p
->numbered
= numbered_lines
&& (!parallel_files
|| i
== 1);
1438 p
->start_position
= h
;
1440 /* If we don't truncate lines, all start_positions are
1441 ANYWHERE, except the first column's start_position when
1444 if (!truncate_lines
)
1451 h
= h_next
+ col_sep_length
;
1452 h_next
= h
+ chars_per_column
;
1456 /* The rightmost column.
1458 Doesn't need to be stored unless we intend to balance
1459 columns on the last page. */
1460 if (storing_columns
&& balance_columns
)
1462 p
->char_func
= store_char
;
1463 p
->print_func
= print_stored
;
1467 p
->char_func
= print_char
;
1468 p
->print_func
= read_line
;
1471 p
->numbered
= numbered_lines
&& (!parallel_files
|| i
== 1);
1472 p
->start_position
= h
;
1475 /* Open a file. Return true if successful.
1477 With each file p, p->full_page_printed is initialized,
1478 see also init_fps. */
1481 open_file (char *name
, COLUMN
*p
)
1483 if (STREQ (name
, "-"))
1485 p
->name
= _("standard input");
1487 have_read_stdin
= true;
1492 p
->fp
= fopen (name
, "r");
1494 if (p
->fp
== nullptr)
1496 failed_opens
= true;
1497 if (!ignore_failed_opens
)
1498 error (0, errno
, "%s", quotef (name
));
1501 fadvise (p
->fp
, FADVISE_SEQUENTIAL
);
1503 p
->full_page_printed
= false;
1508 /* Close the file in P.
1510 If we aren't dealing with multiple files in parallel, we change
1511 the status of all columns in the column list to reflect the close. */
1514 close_file (COLUMN
*p
)
1519 if (p
->status
== CLOSED
)
1523 if (!ferror (p
->fp
))
1525 if (fileno (p
->fp
) == STDIN_FILENO
)
1527 else if (fclose (p
->fp
) != 0 && !err
)
1530 error (EXIT_FAILURE
, err
, "%s", quotef (p
->name
));
1532 if (!parallel_files
)
1534 for (q
= column_vector
, i
= columns
; i
; ++q
, --i
)
1537 if (q
->lines_stored
== 0)
1539 q
->lines_to_print
= 0;
1546 p
->lines_to_print
= 0;
1549 --files_ready_to_read
;
1552 /* Put a file on hold until we start a new page,
1553 since we've hit a form feed.
1555 If we aren't dealing with parallel files, we must change the
1556 status of all columns in the column list. */
1559 hold_file (COLUMN
*p
)
1564 if (!parallel_files
)
1565 for (q
= column_vector
, i
= columns
; i
; ++q
, --i
)
1567 if (storing_columns
)
1568 q
->status
= FF_FOUND
;
1570 q
->status
= ON_HOLD
;
1573 p
->status
= ON_HOLD
;
1575 p
->lines_to_print
= 0;
1576 --files_ready_to_read
;
1579 /* Undo hold_file -- go through the column list and change any
1580 ON_HOLD columns to OPEN. Used at the end of each page. */
1588 for (p
= column_vector
; i
; --i
, ++p
)
1589 if (p
->status
== ON_HOLD
)
1592 files_ready_to_read
++;
1595 if (storing_columns
)
1597 if (column_vector
->status
== CLOSED
)
1598 /* We use the info to output an error message in skip_to_page. */
1599 files_ready_to_read
= 0;
1601 files_ready_to_read
= 1;
1605 /* Print a single file, or multiple files in parallel.
1607 Set up the list of columns, opening the necessary files.
1608 Allocate space for storing columns, if necessary.
1609 Skip to first_page_number, if user has asked to skip leading pages.
1610 Determine which functions are appropriate to store/print lines
1612 Print the file(s). */
1615 print_files (int number_of_files
, char **av
)
1617 init_parameters (number_of_files
);
1618 if (! init_fps (number_of_files
, av
))
1620 if (storing_columns
)
1623 if (first_page_number
> 1)
1625 if (!skip_to_page (first_page_number
))
1628 page_number
= first_page_number
;
1635 line_number
= line_count
;
1636 while (print_page ())
1640 /* Initialize header information.
1641 If DESC is non-negative, it is a file descriptor open to
1642 FILENAME for reading. */
1645 init_header (char const *filename
, int desc
)
1647 char *buf
= nullptr;
1653 /* If parallel files or standard input, use current date. */
1654 if (STREQ (filename
, "-"))
1656 if (0 <= desc
&& fstat (desc
, &st
) == 0)
1657 t
= get_stat_mtime (&st
);
1660 static struct timespec timespec
;
1661 if (! timespec
.tv_sec
)
1662 gettime (×pec
);
1667 if (localtime_rz (localtz
, &t
.tv_sec
, &tm
))
1670 = nstrftime (nullptr, SIZE_MAX
, date_format
, &tm
, localtz
, ns
) + 1;
1671 buf
= xmalloc (bufsize
);
1672 nstrftime (buf
, bufsize
, date_format
, &tm
, localtz
, ns
);
1676 char secbuf
[INT_BUFSIZE_BOUND (intmax_t)];
1677 buf
= xmalloc (sizeof secbuf
+ MAX (10, INT_BUFSIZE_BOUND (int)));
1678 sprintf (buf
, "%s.%09d", timetostr (t
.tv_sec
, secbuf
), ns
);
1683 file_text
= custom_header
? custom_header
: desc
< 0 ? "" : filename
;
1684 header_width_available
= (chars_per_line
1685 - mbswidth (date_text
, 0)
1686 - mbswidth (file_text
, 0));
1689 /* Set things up for printing a page
1691 Scan through the columns ...
1692 Determine which are ready to print
1693 (i.e., which have lines stored or open files)
1694 Set p->lines_to_print appropriately
1695 (to p->lines_stored if we're storing, or lines_per_body
1696 if we're reading straight from the file)
1697 Keep track of this total so we know when to stop printing */
1705 if (storing_columns
)
1708 for (j
= columns
- 1, p
= column_vector
; j
; --j
, ++p
)
1710 p
->lines_to_print
= p
->lines_stored
;
1714 if (balance_columns
)
1716 p
->lines_to_print
= p
->lines_stored
;
1718 /* Since we're not balancing columns, we don't need to store
1719 the rightmost column. Read it straight from the file. */
1722 if (p
->status
== OPEN
)
1724 p
->lines_to_print
= lines_per_body
;
1727 p
->lines_to_print
= 0;
1731 for (j
= columns
, p
= column_vector
; j
; --j
, ++p
)
1732 if (p
->status
== OPEN
)
1734 p
->lines_to_print
= lines_per_body
;
1737 p
->lines_to_print
= 0;
1740 /* Align empty columns and print separators.
1741 Empty columns will be formed by files with status ON_HOLD or CLOSED
1742 when printing multiple files in parallel. */
1745 align_column (COLUMN
*p
)
1747 padding_not_printed
= p
->start_position
;
1748 if (col_sep_length
< padding_not_printed
)
1750 pad_across_to (padding_not_printed
- col_sep_length
);
1751 padding_not_printed
= ANYWHERE
;
1754 if (use_col_separator
)
1755 print_sep_string ();
1758 add_line_number (p
);
1763 As long as there are lines left on the page and columns ready to print,
1764 Scan across the column list
1765 if the column has stored lines or the file is open
1766 pad to the appropriate spot
1768 pad the remainder of the page with \n or \f as requested
1769 reset the status of all files -- any files which where on hold because
1770 of formfeeds are now put back into the lineup. */
1776 int lines_left_on_page
;
1779 /* Used as an accumulator (with | operator) of successive values of
1780 pad_vertically. The trick is to set pad_vertically
1781 to false before each run through the inner loop, then after that
1782 loop, it tells us whether a line was actually printed (whether a
1783 newline needs to be output -- or two for double spacing). But those
1784 values have to be accumulated (in pv) so we can invoke pad_down
1785 properly after the outer loop completes. */
1790 if (cols_ready_to_print () == 0)
1794 print_a_header
= true;
1796 /* Don't pad unless we know a page was printed. */
1797 pad_vertically
= false;
1800 lines_left_on_page
= lines_per_body
;
1802 lines_left_on_page
*= 2;
1804 while (lines_left_on_page
> 0 && cols_ready_to_print () > 0)
1806 output_position
= 0;
1807 spaces_not_printed
= 0;
1808 separators_not_printed
= 0;
1809 pad_vertically
= false;
1810 align_empty_cols
= false;
1813 for (j
= 1, p
= column_vector
; j
<= columns
; ++j
, ++p
)
1816 if (p
->lines_to_print
> 0 || p
->status
== FF_FOUND
)
1819 padding_not_printed
= p
->start_position
;
1820 if (!(p
->print_func
) (p
))
1821 read_rest_of_line (p
);
1822 pv
|= pad_vertically
;
1824 --p
->lines_to_print
;
1825 if (p
->lines_to_print
<= 0)
1827 if (cols_ready_to_print () == 0)
1831 /* File p changed its status to ON_HOLD or CLOSED */
1832 if (parallel_files
&& p
->status
!= OPEN
)
1835 align_empty_cols
= true;
1836 else if (p
->status
== CLOSED
1837 || (p
->status
== ON_HOLD
&& FF_only
))
1841 else if (parallel_files
)
1843 /* File status ON_HOLD or CLOSED */
1845 align_empty_cols
= true;
1850 /* We need it also with an empty column */
1851 if (use_col_separator
)
1852 ++separators_not_printed
;
1858 --lines_left_on_page
;
1861 if (cols_ready_to_print () == 0 && !extremities
)
1864 if (double_space
&& pv
)
1867 --lines_left_on_page
;
1871 if (lines_left_on_page
== 0)
1872 for (j
= 1, p
= column_vector
; j
<= columns
; ++j
, ++p
)
1873 if (p
->status
== OPEN
)
1874 p
->full_page_printed
= true;
1876 pad_vertically
= pv
;
1878 if (pad_vertically
&& extremities
)
1879 pad_down (lines_left_on_page
+ lines_per_footer
);
1880 else if (keep_FF
&& print_a_FF
)
1886 if (last_page_number
< ++page_number
)
1887 return false; /* Stop printing with LAST_PAGE */
1889 reset_status (); /* Change ON_HOLD to OPEN. */
1891 return true; /* More pages to go. */
1894 /* Allocate space for storing columns.
1896 This is necessary when printing multiple columns from a single file.
1897 Lines are stored consecutively in buff, separated by '\0'.
1899 The following doesn't apply any longer - any tuning possible?
1900 (We can't use a fixed offset since with the '-s' flag lines aren't
1903 We maintain a list (line_vector) of pointers to the beginnings
1904 of lines in buff. We allocate one more than the number of lines
1905 because the last entry tells us the index of the last character,
1906 which we need to know in order to print the last line in buff. */
1909 init_store_cols (void)
1911 int total_lines
, total_lines_1
, chars_per_column_1
, chars_if_truncate
;
1912 if (ckd_mul (&total_lines
, lines_per_body
, columns
)
1913 || ckd_add (&total_lines_1
, total_lines
, 1)
1914 || ckd_add (&chars_per_column_1
, chars_per_column
, 1)
1915 || ckd_mul (&chars_if_truncate
, total_lines
, chars_per_column_1
))
1916 integer_overflow ();
1919 /* FIXME: here's where it was allocated. */
1920 line_vector
= xnmalloc (total_lines_1
, sizeof *line_vector
);
1923 end_vector
= xnmalloc (total_lines
, sizeof *end_vector
);
1926 buff
= xnmalloc (chars_if_truncate
, use_col_separator
+ 1);
1927 buff_allocated
= chars_if_truncate
; /* Tune this. */
1928 buff_allocated
*= use_col_separator
+ 1;
1931 /* Store all but the rightmost column.
1932 (Used when printing a single file in multiple downward columns)
1935 set p->current_line to be the index in line_vector of the
1936 first line in the column
1937 For each line in the column
1938 store the line in buff
1939 add to line_vector the index of the line's first char
1940 buff_start is the index in buff of the first character in the
1944 store_columns (void)
1947 unsigned int line
= 0;
1948 unsigned int buff_start
;
1949 int last_col
; /* The rightmost column which will be saved in buff */
1955 if (balance_columns
)
1958 last_col
= columns
- 1;
1960 for (i
= 1, p
= column_vector
; i
<= last_col
; ++i
, ++p
)
1961 p
->lines_stored
= 0;
1963 for (i
= 1, p
= column_vector
; i
<= last_col
&& files_ready_to_read
;
1966 p
->current_line
= line
;
1967 for (j
= lines_per_body
; j
&& files_ready_to_read
; --j
)
1969 if (p
->status
== OPEN
) /* Redundant. Clean up. */
1974 read_rest_of_line (p
);
1976 if (p
->status
== OPEN
1977 || buff_start
!= buff_current
)
1980 line_vector
[line
] = buff_start
;
1981 end_vector
[line
++] = input_position
;
1982 buff_start
= buff_current
;
1987 /* Keep track of the location of the last char in buff. */
1988 line_vector
[line
] = buff_start
;
1990 if (balance_columns
)
1995 balance (int total_stored
)
2001 for (i
= 1, p
= column_vector
; i
<= columns
; ++i
, ++p
)
2003 lines
= total_stored
/ columns
;
2004 if (i
<= total_stored
% columns
)
2007 p
->lines_stored
= lines
;
2008 p
->current_line
= first_line
;
2010 first_line
+= lines
;
2014 /* Store a character in the buffer. */
2019 if (buff_current
>= buff_allocated
)
2021 /* May be too generous. */
2022 buff
= X2REALLOC (buff
, &buff_allocated
);
2024 buff
[buff_current
++] = c
;
2028 add_line_number (COLUMN
*p
)
2034 /* Cutting off the higher-order digits is more informative than
2035 lower-order cut off. */
2036 num_width
= sprintf (number_buff
, "%*d", chars_per_number
, line_number
);
2038 s
= number_buff
+ (num_width
- chars_per_number
);
2039 for (i
= chars_per_number
; i
> 0; i
--)
2040 (p
->char_func
) (*s
++);
2044 /* Tabification is assumed for multiple columns, also for n-separators,
2045 but 'default n-separator = TAB' hasn't been given priority over
2046 equal column_width also specified by POSIX. */
2047 if (number_separator
== '\t')
2049 i
= number_width
- chars_per_number
;
2051 (p
->char_func
) (' ');
2054 (p
->char_func
) (number_separator
);
2057 /* To comply with POSIX, we avoid any expansion of default TAB
2058 separator with a single column output. No column_width requirement
2059 has to be considered. */
2061 (p
->char_func
) (number_separator
);
2062 if (number_separator
== '\t')
2063 output_position
= POS_AFTER_TAB (chars_per_output_tab
,
2067 if (truncate_lines
&& !parallel_files
)
2068 input_position
+= number_width
;
2071 /* Print (or store) padding until the current horizontal position
2075 pad_across_to (int position
)
2077 int h
= output_position
;
2080 spaces_not_printed
= position
- output_position
;
2083 while (++h
<= position
)
2085 output_position
= position
;
2089 /* Pad to the bottom of the page.
2091 If the user has requested a formfeed, use one.
2092 Otherwise, use newlines. */
2095 pad_down (unsigned int lines
)
2100 for (unsigned int i
= lines
; i
; --i
)
2104 /* Read the rest of the line.
2106 Read from the current column's file until an end of line is
2107 hit. Used when we've truncated a line and we no longer need
2108 to print or store its characters. */
2111 read_rest_of_line (COLUMN
*p
)
2116 while ((c
= getc (f
)) != '\n')
2120 if ((c
= getc (f
)) != '\n')
2135 /* Read a line with skip_to_page.
2137 Read from the current column's file until an end of line is
2138 hit. Used when we read full lines to skip pages.
2139 With skip_to_page we have to check for FF-coincidence which is done
2140 in function read_line otherwise.
2141 Count lines of skipped pages to find the line number of 1st page
2142 printed relative to 1st line of input file (start_line_num). */
2145 skip_read (COLUMN
*p
, int column_number
)
2150 bool single_ff
= false;
2153 /* Read 1st character in a line or any character succeeding a FF */
2154 if ((c
= getc (f
)) == '\f' && p
->full_page_printed
)
2155 /* A FF-coincidence with a previous full_page_printed.
2156 To avoid an additional empty page, eliminate the FF */
2157 if ((c
= getc (f
)) == '\n')
2160 p
->full_page_printed
= false;
2162 /* 1st character a FF means a single FF without any printable
2163 characters. Don't count it as a line with -n option. */
2167 /* Preparing for a FF-coincidence: Maybe we finish that page
2168 without a FF found */
2170 p
->full_page_printed
= true;
2176 /* No FF-coincidence possible,
2177 no catching up of a FF-coincidence with next page */
2180 if (!parallel_files
)
2181 for (q
= column_vector
, i
= columns
; i
; ++q
, --i
)
2182 q
->full_page_printed
= false;
2184 p
->full_page_printed
= false;
2187 if ((c
= getc (f
)) != '\n')
2201 if ((!parallel_files
|| column_number
== 1) && !single_ff
)
2205 /* If we're tabifying output,
2207 When print_char encounters white space it keeps track
2208 of our desired horizontal position and delays printing
2209 until this function is called. */
2212 print_white_space (void)
2215 int h_old
= output_position
;
2216 int goal
= h_old
+ spaces_not_printed
;
2218 while (goal
- h_old
> 1
2219 && (h_new
= POS_AFTER_TAB (chars_per_output_tab
, h_old
)) <= goal
)
2221 putchar (output_tab_char
);
2224 while (++h_old
<= goal
)
2227 output_position
= goal
;
2228 spaces_not_printed
= 0;
2231 /* Print column separators.
2233 We keep a count until we know that we'll be printing a line,
2234 then print_sep_string() is called. */
2237 print_sep_string (void)
2239 char const *s
= col_sep_string
;
2240 int l
= col_sep_length
;
2242 if (separators_not_printed
<= 0)
2244 /* We'll be starting a line with chars_per_margin, anything else? */
2245 if (spaces_not_printed
> 0)
2246 print_white_space ();
2250 for (; separators_not_printed
> 0; --separators_not_printed
)
2254 /* 3 types of sep_strings: spaces only, spaces and chars,
2258 /* We're tabifying output; consecutive spaces in
2259 sep_string may have to be converted to tabs */
2261 ++spaces_not_printed
;
2265 if (spaces_not_printed
> 0)
2266 print_white_space ();
2271 /* sep_string ends with some spaces */
2272 if (spaces_not_printed
> 0)
2273 print_white_space ();
2278 /* Print (or store, depending on p->char_func) a clump of N
2282 print_clump (COLUMN
*p
, int n
, char *clump
)
2285 (p
->char_func
) (*clump
++);
2288 /* Print a character.
2290 Update the following comment: process-char hasn't been used any
2292 If we're tabifying, all tabs have been converted to spaces by
2293 process_char(). Keep a count of consecutive spaces, and when
2294 a nonspace is encountered, call print_white_space() to print the
2295 required number of tabs and spaces. */
2304 ++spaces_not_printed
;
2307 else if (spaces_not_printed
> 0)
2308 print_white_space ();
2310 /* Nonprintables are assumed to have width 0, except '\b'. */
2311 if (! isprint (to_uchar (c
)))
2322 /* Skip to page PAGE before printing.
2323 PAGE may be larger than total number of pages. */
2326 skip_to_page (uintmax_t page
)
2328 for (uintmax_t n
= 1; n
< page
; ++n
)
2333 for (int i
= 1; i
< lines_per_body
; ++i
)
2335 for (j
= 1, p
= column_vector
; j
<= columns
; ++j
, ++p
)
2336 if (p
->status
== OPEN
)
2340 for (j
= 1, p
= column_vector
; j
<= columns
; ++j
, ++p
)
2341 if (p
->status
== OPEN
)
2344 if (storing_columns
) /* change FF_FOUND to ON_HOLD */
2345 for (j
= 1, p
= column_vector
; j
<= columns
; ++j
, ++p
)
2346 if (p
->status
!= CLOSED
)
2347 p
->status
= ON_HOLD
;
2352 if (files_ready_to_read
< 1)
2354 /* It's very helpful, normally the total number of pages is
2355 not known in advance. */
2357 _("starting page number %ju exceeds page count %ju"),
2362 return files_ready_to_read
> 0;
2367 Formfeeds are assumed to use up two lines at the beginning of
2373 char page_text
[256 + INT_STRLEN_BOUND (page_number
)];
2374 int available_width
;
2378 output_position
= 0;
2379 pad_across_to (chars_per_margin
);
2380 print_white_space ();
2382 if (page_number
== 0)
2383 error (EXIT_FAILURE
, 0, _("page number overflow"));
2385 /* The translator must ensure that formatting the translation of
2386 "Page %ju" does not generate more than (sizeof page_text - 1)
2388 sprintf (page_text
, _("Page %ju"), page_number
);
2389 available_width
= header_width_available
- mbswidth (page_text
, 0);
2390 available_width
= MAX (0, available_width
);
2391 lhs_spaces
= available_width
>> 1;
2392 rhs_spaces
= available_width
- lhs_spaces
;
2394 printf ("\n\n%*s%s%*s%s%*s%s\n\n\n",
2395 chars_per_margin
, "",
2396 date_text
, lhs_spaces
, " ",
2397 file_text
, rhs_spaces
, " ", page_text
);
2399 print_a_header
= false;
2400 output_position
= 0;
2403 /* Print (or store, if p->char_func is store_char()) a line.
2405 Read a character to determine whether we have a line or not.
2406 (We may hit EOF, \n, or \f)
2408 Once we know we have a line,
2409 set pad_vertically = true, meaning it's safe
2410 to pad down at the end of the page, since we do have a page.
2411 print a header if needed.
2412 pad across to padding_not_printed if needed.
2413 print any separators which need to be printed.
2414 print a line number if it needs to be printed.
2416 Print the clump which corresponds to the first character.
2418 Enter a loop and keep printing until an end of line condition
2419 exists, or until we exceed chars_per_column.
2421 Return false if we exceed chars_per_column before reading
2422 an end of line character, true otherwise. */
2425 read_line (COLUMN
*p
)
2428 int chars
IF_LINT (= 0);
2429 int last_input_position
;
2433 /* read 1st character in each line or any character succeeding a FF: */
2436 last_input_position
= input_position
;
2438 if (c
== '\f' && p
->full_page_printed
)
2439 if ((c
= getc (p
->fp
)) == '\n')
2441 p
->full_page_printed
= false;
2446 if ((c
= getc (p
->fp
)) != '\n')
2449 if (print_a_header
&& !storing_columns
)
2451 pad_vertically
= true;
2464 chars
= char_to_clump (c
);
2467 if (truncate_lines
&& input_position
> chars_per_column
)
2469 input_position
= last_input_position
;
2473 if (p
->char_func
!= store_char
)
2475 pad_vertically
= true;
2477 if (print_a_header
&& !storing_columns
)
2480 if (parallel_files
&& align_empty_cols
)
2482 /* We have to align empty columns at the beginning of a line. */
2483 k
= separators_not_printed
;
2484 separators_not_printed
= 0;
2485 for (j
= 1, q
= column_vector
; j
<= k
; ++j
, ++q
)
2488 separators_not_printed
+= 1;
2490 padding_not_printed
= p
->start_position
;
2492 spaces_not_printed
= chars_per_column
;
2494 spaces_not_printed
= 0;
2495 align_empty_cols
= false;
2498 if (col_sep_length
< padding_not_printed
)
2500 pad_across_to (padding_not_printed
- col_sep_length
);
2501 padding_not_printed
= ANYWHERE
;
2504 if (use_col_separator
)
2505 print_sep_string ();
2509 add_line_number (p
);
2515 print_clump (p
, chars
, clump_buff
);
2526 if ((c
= getc (p
->fp
)) != '\n')
2537 last_input_position
= input_position
;
2538 chars
= char_to_clump (c
);
2539 if (truncate_lines
&& input_position
> chars_per_column
)
2541 input_position
= last_input_position
;
2545 print_clump (p
, chars
, clump_buff
);
2549 /* Print a line from buff.
2551 If this function has been called, we know we have "something to
2552 print". But it remains to be seen whether we have a real text page
2553 or an empty page (a single form feed) with/without a header only.
2554 Therefore first we set pad_vertically to true and print a header
2556 If FF_FOUND and we are using -t|-T option we omit any newline by
2557 setting pad_vertically to false (see print_page).
2558 Otherwise we pad across if necessary, print separators if necessary
2559 and text of COLUMN *p.
2561 Return true, meaning there is no need to call read_rest_of_line. */
2564 print_stored (COLUMN
*p
)
2568 int line
= p
->current_line
++;
2569 char *first
= &buff
[line_vector
[line
]];
2571 UMR: Uninitialized memory read:
2572 * This is occurring while in:
2573 print_stored [pr.c:2239]
2574 * Reading 4 bytes from 0x5148c in the heap.
2575 * Address 0x5148c is 4 bytes into a malloc'd block at 0x51488 of 676 bytes
2576 * This block was allocated from:
2578 xmalloc [xmalloc.c:94]
2579 init_store_cols [pr.c:1648]
2581 char *last
= &buff
[line_vector
[line
+ 1]];
2583 pad_vertically
= true;
2588 if (p
->status
== FF_FOUND
)
2591 for (i
= 1, q
= column_vector
; i
<= columns
; ++i
, ++q
)
2592 q
->status
= ON_HOLD
;
2593 if (column_vector
->lines_to_print
<= 0)
2596 pad_vertically
= false;
2597 return true; /* print a header only */
2601 if (col_sep_length
< padding_not_printed
)
2603 pad_across_to (padding_not_printed
- col_sep_length
);
2604 padding_not_printed
= ANYWHERE
;
2607 if (use_col_separator
)
2608 print_sep_string ();
2610 while (first
!= last
)
2611 print_char (*first
++);
2613 if (spaces_not_printed
== 0)
2615 output_position
= p
->start_position
+ end_vector
[line
];
2616 if (p
->start_position
- col_sep_length
== chars_per_margin
)
2617 output_position
-= col_sep_length
;
2623 /* Convert a character to the proper format and return the number of
2624 characters in the resulting clump. Increment input_position by
2625 the width of the clump.
2627 Tabs are converted to clumps of spaces.
2628 Nonprintable characters may be converted to clumps of escape
2629 sequences or control prefixes.
2631 Note: the width of a clump is not necessarily equal to the number of
2632 characters in clump_buff. (e.g., the width of '\b' is -1, while the
2633 number of characters is 1.) */
2636 char_to_clump (char c
)
2638 unsigned char uc
= c
;
2639 char *s
= clump_buff
;
2644 int chars_per_c
= 8;
2646 if (c
== input_tab_char
)
2647 chars_per_c
= chars_per_input_tab
;
2649 if (c
== input_tab_char
|| c
== '\t')
2651 width
= TAB_WIDTH (chars_per_c
, input_position
);
2655 for (i
= width
; i
; --i
)
2666 else if (! isprint (uc
))
2668 if (use_esc_sequence
)
2673 sprintf (esc_buff
, "%03o", uc
);
2674 for (i
= 0; i
<= 2; ++i
)
2677 else if (use_cntrl_prefix
)
2691 sprintf (esc_buff
, "%03o", uc
);
2692 for (i
= 0; i
<= 2; ++i
)
2716 /* Too many backspaces must put us in position 0 -- never negative. */
2717 if (width
< 0 && input_position
== 0)
2722 else if (width
< 0 && input_position
<= -width
)
2725 input_position
+= width
;
2730 /* We've just printed some files and need to clean up things before
2731 looking for more options and printing the next batch of files.
2733 Free everything we've xmalloc'ed, except 'header'. */
2740 free (column_vector
);
2746 /* Complain, print a usage message, and die. */
2751 if (status
!= EXIT_SUCCESS
)
2756 Usage: %s [OPTION]... [FILE]...\n\
2761 Paginate or columnate FILE(s) for printing.\n\
2765 emit_mandatory_arg_note ();
2768 +FIRST_PAGE[:LAST_PAGE], --pages=FIRST_PAGE[:LAST_PAGE]\n\
2769 begin [stop] printing with page FIRST_[LAST_]PAGE\n\
2770 -COLUMN, --columns=COLUMN\n\
2771 output COLUMN columns and print columns down,\n\
2772 unless -a is used. Balance number of lines in the\n\
2773 columns on each page\n\
2776 -a, --across print columns across rather than down, used together\n\
2778 -c, --show-control-chars\n\
2779 use hat notation (^G) and octal backslash notation\n\
2780 -d, --double-space\n\
2781 double space the output\n\
2784 -D, --date-format=FORMAT\n\
2785 use FORMAT for the header date\n\
2786 -e[CHAR[WIDTH]], --expand-tabs[=CHAR[WIDTH]]\n\
2787 expand input CHARs (TABs) to tab WIDTH (8)\n\
2788 -F, -f, --form-feed\n\
2789 use form feeds instead of newlines to separate pages\n\
2790 (by a 3-line page header with -F or a 5-line header\n\
2791 and trailer without -F)\n\
2794 -h, --header=HEADER\n\
2795 use a centered HEADER instead of filename in page header,\n\
2796 -h \"\" prints a blank line, don't use -h\"\"\n\
2797 -i[CHAR[WIDTH]], --output-tabs[=CHAR[WIDTH]]\n\
2798 replace spaces with CHARs (TABs) to tab WIDTH (8)\n\
2799 -J, --join-lines merge full lines, turns off -W line truncation, no column\n\
2800 alignment, --sep-string[=STRING] sets separators\n\
2803 -l, --length=PAGE_LENGTH\n\
2804 set the page length to PAGE_LENGTH (66) lines\n\
2805 (default number of lines of text 56, and with -F 63).\n\
2806 implies -t if PAGE_LENGTH <= 10\n\
2809 -m, --merge print all files in parallel, one in each column,\n\
2810 truncate lines, but join lines of full length with -J\n\
2813 -n[SEP[DIGITS]], --number-lines[=SEP[DIGITS]]\n\
2814 number lines, use DIGITS (5) digits, then SEP (TAB),\n\
2815 default counting starts with 1st line of input file\n\
2816 -N, --first-line-number=NUMBER\n\
2817 start counting with NUMBER at 1st line of first\n\
2818 page printed (see +FIRST_PAGE)\n\
2821 -o, --indent=MARGIN\n\
2822 offset each line with MARGIN (zero) spaces, do not\n\
2823 affect -w or -W, MARGIN will be added to PAGE_WIDTH\n\
2824 -r, --no-file-warnings\n\
2825 omit warning when a file cannot be opened\n\
2828 -s[CHAR], --separator[=CHAR]\n\
2829 separate columns by a single character, default for CHAR\n\
2830 is the <TAB> character without -w and \'no char\' with -w.\
2832 -s[CHAR] turns off line truncation of all 3 column\n\
2833 options (-COLUMN|-a -COLUMN|-m) except -w is set\n\
2836 -S[STRING], --sep-string[=STRING]\n\
2837 separate columns by STRING,\n\
2838 without -S: Default separator <TAB> with -J and <space>\n\
2839 otherwise (same as -S\" \"), no effect on column options\n\
2842 -t, --omit-header omit page headers and trailers;\n\
2843 implied if PAGE_LENGTH <= 10\n\
2846 -T, --omit-pagination\n\
2847 omit page headers and trailers, eliminate any pagination\n\
2848 by form feeds set in input files\n\
2849 -v, --show-nonprinting\n\
2850 use octal backslash notation\n\
2851 -w, --width=PAGE_WIDTH\n\
2852 set page width to PAGE_WIDTH (72) characters for\n\
2853 multiple text-column output only, -s[char] turns off (72)\n\
2856 -W, --page-width=PAGE_WIDTH\n\
2857 set page width to PAGE_WIDTH (72) characters always,\n\
2858 truncate lines, except -J option is set, no interference\n\
2861 fputs (HELP_OPTION_DESCRIPTION
, stdout
);
2862 fputs (VERSION_OPTION_DESCRIPTION
, stdout
);
2863 emit_ancillary_info (PROGRAM_NAME
);