[PATCH] add paste implementation
Tito
farmatito at tiscali.it
Sat Mar 11 17:59:52 UTC 2017
Hi,
a few more hints. Hope this helps.
Ciao,
Tito
On 03/11/2017 12:18 PM, Maxime Coste wrote:
> Applied the suggested changes, except for dynamically allocating the
> files array. I can do it if thats preferred.
>
> I traced the getopt32 call to make sure, and it does not touch the
> argument pointer when the switch is not present, so giving it a default
> values works.
>
> ---
> AUTHORS | 3 +
> coreutils/paste.c | 160 +++++++++++++++++++++++++++++++++
> docs/posix_conformance.txt | 8 +-
> testsuite/paste/paste | 20 +++++
> testsuite/paste/paste-back-cuted-lines | 9 ++
> testsuite/paste/paste-multi-stdin | 16 ++++
> testsuite/paste/paste-pairs | 16 ++++
> testsuite/paste/paste-separate | 19 ++++
> 8 files changed, 250 insertions(+), 1 deletion(-)
> create mode 100644 coreutils/paste.c
> create mode 100644 testsuite/paste/paste
> create mode 100644 testsuite/paste/paste-back-cuted-lines
> create mode 100644 testsuite/paste/paste-multi-stdin
> create mode 100644 testsuite/paste/paste-pairs
> create mode 100644 testsuite/paste/paste-separate
>
> diff --git a/AUTHORS b/AUTHORS
> index fa58697f7..5c9a634c9 100644
> --- a/AUTHORS
> +++ b/AUTHORS
> @@ -178,3 +178,6 @@ Mike Frysinger <vapier at gentoo.org>
>
> Jie Zhang <jie.zhang at analog.com>
> fixed two bugs in msh and hush (exitcode of killed processes)
> +
> +Maxime Coste <mawww at kakoune.org>
> + paste implementation
> diff --git a/coreutils/paste.c b/coreutils/paste.c
> new file mode 100644
> index 000000000..34426fea2
> --- /dev/null
> +++ b/coreutils/paste.c
> @@ -0,0 +1,160 @@
> +/* vi: set sw=4 ts=4: */
> +/*
> + * paste.c - implementation of the posix paste command
> + *
> + * Written by Maxime Coste <mawww at kakoune.org>
> + *
> + * Licensed under GPLv2 or later, see file LICENSE in this source tree.
> + */
> +//config:config PASTE
> +//config: bool "paste"
> +//config: default y
> +//config: help
> +//config: paste is used to paste lines of different files together
> +//config: and write the result to stdout
> +
> +//applet:IF_PASTE(APPLET_NOEXEC(paste, paste, BB_DIR_USR_BIN, BB_SUID_DROP, paste))
> +
> +//kbuild:lib-$(CONFIG_PASTE) += paste.o
> +
> +//usage:#define paste_trivial_usage
> +//usage: "[OPTIONS] [FILE]..."
> +//usage:#define paste_full_usage "\n\n"
> +//usage: "Paste lines from each input files together using a tabulation character\n"
> +//usage: "\n -d LIST use delimiters from LIST instead of tabulations"
> +//usage: "\n -s paste lines of each input files separately"
> +//usage:
> +//usage:#define paste_example_usage
> +//usage: "# write out directory in four columns\n"
> +//usage: "$ ls | paste - - - -\n"
> +//usage: "# combine pairs of lines from a file into single lines\n"
> +//usage: "$ paste -s -d '\t\n' file\n"
> +
> +#include "libbb.h"
> +
> +static const char optstring[] = "d:s";
Remove optstring and add it directly to getopt32 for better readability
> +#define PASTE_OPT_DELIMITERS (1 << 0)
> +#define PASTE_OPT_SEPARATE (1 << 1)
> +
> +#define PASTE_MAX_FILES 32
> +
Function used only once maybe it could be inlined
> +static char parse_escaped_delimiter(char c)
> +{
> + switch (c) {
> + case 'n': return '\n';
> + case 't': return '\t';
> + case '\\': return '\\';
> + case '0': return 0;
> + }
> + bb_error_msg_and_die("invalid escaped delimiter %c", c);
> + return 0;
> +}
> +
> +static char get_next_delimiter(char* delimiters, char** current)
> +{
> + char res = **current;
> + if (res == '\\')
> + res = parse_escaped_delimiter(*(++(*current)));
> + if (*(++(*current)) == 0)
> + *current = delimiters;
> + return res;
> +}
> +
> +static void paste_files(FILE** files, int file_count, char* delimiters)
> +{
> + char *lines[PASTE_MAX_FILES];
> + char *current_delimiter;
> + char delim;
> + int active_files = file_count;
> + int i;
> +
> + while (active_files > 0) {
> + current_delimiter = delimiters;
> + for (i = 0; i < file_count; ++i) {
> + if (files[i] == NULL)
> + continue;
> +
> + lines[i] = xmalloc_fgetline(files[i]);
> + if (lines[i] == NULL) {
> + fclose_if_not_stdin(files[i]);
> + files[i] = NULL;
> + --active_files;
> + }
> + }
> +
> + if (active_files == 0)
> + break;
> +
> + for (i = 0; i < file_count; ++i) {
> + if (lines[i] != NULL) {
> + fputs(lines[i], stdout);
> + free(lines[i]);
> + }
> +
> + if (i == file_count-1)
> + fputs("\n", stdout);
> + else if ((delim = get_next_delimiter(delimiters, ¤t_delimiter)) != 0)
> + fputc(delim, stdout);
> + }
> + }
> +}
> +
> +static void paste_files_separate(FILE** files, int file_count, char* delimiters)
> +{
> + char *line, *next_line;
> + char *current_delimiter;
> + int end;
> + int i;
> +
> + for (i = 0; i < file_count; ++i) {
> + line = NULL;
> + current_delimiter = delimiters;
> + while ((next_line = xmalloc_fgets(files[i])) != NULL) {
> + if (line != NULL) {
> + end = strlen(line)-1;
> + line[end] = get_next_delimiter(delimiters, ¤t_delimiter);
> + fputs(line, stdout);
> + free(line);
> + }
> + line = next_line;
> + }
> + if (line) {
> + fputs(line, stdout);
> + free(line);
> + }
> + fclose_if_not_stdin(files[i]);
> + }
> +}
> +
> +int paste_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
> +int paste_main(int argc, char **argv)
> +{
> + char *delimiters = (char*)"\t";
> + unsigned opt;
> + int i;
> + FILE *files[PASTE_MAX_FILES]; // Posix mandates at least 12 file operands
> +
> + opt = getopt32(argv, optstring, &delimiters);
opt = getopt32(argv, "d:s", &delimiters);
> + argc -= optind;
use only argv?
> + argv += optind;
> +
not needed getopt32 will error out if argument to -d is missing
with error message "option requires an argument"
> + if ((opt & PASTE_OPT_DELIMITERS) && delimiters[0] == 0)
> + bb_error_msg_and_die("empty delimiters is not supported");
> +
Use argv pointer and remove PASTE_MAX_FILES limit?
> + if (argc > PASTE_MAX_FILES)
if possible rephrase error message to be shorter e.g.:
"only %d file operands supported"
> + bb_error_msg_and_die("this implementation of paste only supports up to %d file operands", PASTE_MAX_FILES);
> +
busybox has a linked list api see libbb/llist.c
maybe it could be used here to dynamically create
a files list (without PASTE_MAX_FILES limit)
that could be passed later on to paste_files_separate
and paste_files (could eventually save a few function
parameters and help reduce size)
> + files[0] = stdin;
Use argv for the loop?
> + for (i = 0; i < argc; ++i) {
files[1] = xfopen_stdin(argv[i]);
> + //files[i] = fopen_or_warn_stdin(argv[i]);
> + //if (files[i] == NULL)
> + // xfunc_die();
> + }
> +
> + if (opt & PASTE_OPT_SEPARATE)
> + paste_files_separate(files, argc == 0 ? 1 : argc, delimiters);
> + else
> + paste_files(files, argc == 0 ? 1 : argc, delimiters);
> +
> + fflush_stdout_and_exit(0);
> +}
> diff --git a/docs/posix_conformance.txt b/docs/posix_conformance.txt
> index c0582dc23..8b9112020 100644
> --- a/docs/posix_conformance.txt
> +++ b/docs/posix_conformance.txt
> @@ -22,7 +22,7 @@ POSIX Tools supported only as shell built-ins (ash shell):
> POSIX Tools not supported:
> asa, at, batch, bc, c99, command, compress, csplit, ex, fc, file,
> gencat, getconf, iconv, join, link, locale, localedef, lp, m4,
> - mailx, newgrp, nl, paste, pathchk, pax, pr, qalter, qdel, qhold, qmove,
> + mailx, newgrp, nl, pathchk, pax, pr, qalter, qdel, qhold, qmove,
> qmsg, qrerun, qrls, qselect, qsig, qstat, qsub, tabs, talk, tput,
> tsort, unlink, uucp, uustat, uux
>
> @@ -469,6 +469,12 @@ od POSIX options
> -x | no | no |
> od Busybox specific options: None
>
> +paste POSIX options
> + option | exists | compliant | remarks
> + -d list | yes | yes |
> + -s | yes | yes |
> +paste Busybox specific options: None
> +
> patch POSIX options
> option | exists | compliant | remarks
> -D define | no | no |
> diff --git a/testsuite/paste/paste b/testsuite/paste/paste
> new file mode 100644
> index 000000000..349b49d49
> --- /dev/null
> +++ b/testsuite/paste/paste
> @@ -0,0 +1,20 @@
> +cat > foo <<EOF
> +foo1
> +foo2
> +foo3
> +EOF
> +
> +cat > bar <<EOF
> +bar1
> +bar2
> +bar3
> +EOF
> +
> +cat > baz <<EOF
> +foo1 bar1
> +foo2 bar2
> +foo3 bar3
> +EOF
> +
> +busybox paste foo bar > qux
> +diff -u baz qux
> diff --git a/testsuite/paste/paste-back-cuted-lines b/testsuite/paste/paste-back-cuted-lines
> new file mode 100644
> index 000000000..4059b3ec5
> --- /dev/null
> +++ b/testsuite/paste/paste-back-cuted-lines
> @@ -0,0 +1,9 @@
> +cat > foo <<EOF
> +this is the first line
> +this is the second line
> +this is the third line
> +EOF
> +cut -b 1-13 -n foo > foo1
> +cut -b 14- -n foo > foo2
> +paste -d '\0' foo1 foo2 > bar
> +cmp foo bar
> diff --git a/testsuite/paste/paste-multi-stdin b/testsuite/paste/paste-multi-stdin
> new file mode 100644
> index 000000000..834c0c52f
> --- /dev/null
> +++ b/testsuite/paste/paste-multi-stdin
> @@ -0,0 +1,16 @@
> +cat > foo <<EOF
> +line1
> +line2
> +line3
> +line4
> +line5
> +line6
> +EOF
> +
> +cat > bar <<EOF
> +line1 line2 line3
> +line4 line5 line6
> +EOF
> +
> +paste - - - < foo > baz
> +cmp bar baz
> diff --git a/testsuite/paste/paste-pairs b/testsuite/paste/paste-pairs
> new file mode 100644
> index 000000000..90725fa87
> --- /dev/null
> +++ b/testsuite/paste/paste-pairs
> @@ -0,0 +1,16 @@
> +cat > foo <<EOF
> +foo1
> +bar1
> +foo2
> +bar2
> +foo3
> +EOF
> +
> +cat > bar <<EOF
> +foo1 bar1
> +foo2 bar2
> +foo3
> +EOF
> +
> +busybox paste -s -d "\t\n" foo > baz
> +cmp bar baz
> diff --git a/testsuite/paste/paste-separate b/testsuite/paste/paste-separate
> new file mode 100644
> index 000000000..40793fb31
> --- /dev/null
> +++ b/testsuite/paste/paste-separate
> @@ -0,0 +1,19 @@
> +cat > foo <<EOF
> +foo1
> +foo2
> +foo3
> +EOF
> +
> +cat > bar <<EOF
> +bar1
> +bar2
> +bar3
> +EOF
> +
> +cat > baz <<EOF
> +foo1 foo2 foo3
> +bar1 bar2 bar3
> +EOF
> +
> +busybox paste -s foo bar > qux
> +cmp baz qux
>
More information about the busybox
mailing list