In version 2.38, exit-on-eof has been disabled by default. This change
is annoying for users and forces many users to use 'alias more="more
-e"'. It seems better to force POSIX lovers to use POSIXLY_CORRECT
env. variable and stay backwardly compatible by default.
Addresses: https://github.com/util-linux/util-linux/issues/1703
Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=2088493
Signed-off-by: Karel Zak <kzak@redhat.com>
The command 'echo | more -s' ends in infinite loop if exit-on-EOF is
disabled (default now). We need to call more_key_command() to ask for
user's reaction.
Fixes: https://github.com/util-linux/util-linux/issues/1669
Signed-off-by: Karel Zak <kzak@redhat.com>
Sami Kerola pointed out that /proc contains files with zero size but
content in ec4153cc28.
Eventually fixes for better type handling have been committed with
ba105bb5eb and error handling when "more"
is called with directories in 32a4efc567.
This commit fixes a regression and wrong file size handling:
checkf ends execution prematurely if st.st_size is 0. The idea was to
skip the check_magic call which would otherwise claim that a text file
would not be one, e.g. /proc/self/wchan would fail otherwise. This has
two consequences:
1. The last check in checkf is never true, because if st_size is 0 we
already left the function (file_size and st_size are both off_t).
Proof of Concept (use a window with less lines than /proc/cpuinfo):
$ echo "hello world" > /tmp/more-poc
$ more /tmp/more-poc /proc/cpuinfo
- Press space to switch to next file
- You see >100% in status line
2. Leaving early does not set the close on exit flag for empty files.
3. Since the last line is never reached, ~(off_t)0 is never set, which
implies that the check in display_file is never true.
The st.st_size = 0 case is supposed to be special. Simplify this
regression fix by considering ctl->file_size = 0 to be special. This
eleminates the ~(off_t)0 magic, which would be -1.
Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org>
When a quit character is seen, immediately return that command, as it is the correct operation to perform in this case.
A bigger fix is needed here, in that we need to have a *queue* of operations to perform, rather than a single one
The command allows executing arbitrary shell commands while viewing a file by
entering '!' followed by the command. Entering a command that contains a '%',
'!', or '\' causes a segmentation violation.
The same more(1) function has a problem when not file is specified (cat
/etc/passwd | more) on command line.
Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1975153
Signed-off-by: Karel Zak <kzak@redhat.com>
The rule is pretty simple, always use setgid() before setuid().
Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>
Make the code avoid divided by zero. This can happen when file has content
but is zero in size. Such files can be found from procfs, possibly some
other pseudo-filesystems. To reproduce the issue run the following.
$ more /proc/crypto
...
Floating point exception (core dumped)
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Currently ARROW_DOWN and PAGE_DOWN is bound to command 's' (skip lines).
But this behaviour is not what the user expects, since at least one line
is always missing! Furthermore at the end of the file there is typically
this skipping lines message, if the remainder does not fit in a
complete screen. This conflicts also with e.g. less and man PAGE_DOWN
behaviour. The 'natural' behaviour is binding to
more_kc_jump_lines_per_screen.
If more is extended to also jump only a specific number of lines forward,
this would be even a better binding for ARROW_DOWN.
Before this change the error message about directory was not informative,
and a little bit out of place.
$ touch empty ; ./more ./empty /
::::::::::::::
./empty
::::::::::::::
magic failed: cannot read fd 4 (Is a directory)
::::::::::::::
/
::::::::::::::
After the change messaging is reverted back to what it used to be.
$ ./more ./empty /
::::::::::::::
./empty
::::::::::::::
*** /: directory ***
And while on it lets remove repeated fstat() call.
Fixes: 09070e1a65
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
My earlier change that took libmagic in use to identify mime-type of an input
file caused empty files to be marked binary. Before the change empty files
were simply displayed as empty. This change will restore that behavior.
Addresses: 09070e1a65
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Aim was to introduce page and arrow up/down keys to more(1), but that
also required merging colon_command() and more_key_command() functions.
The more_key_commands enum is pointless from computers point of view.
The command identification performed in read_command() inline with
more_key_command() execution -- but that would be hard for humans, and
source code ought to serve both parties.
Reference: https://github.com/karelzak/util-linux/pull/1003
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
As the old commend said: 'This code would best be shared with the file(1)
program or, perhaps, more should not try to be so smart'. Now at configure
time one can choose whether more(1) is sharing code with file(1), or not.
Addresses: http://bugs.debian.org/139205
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Moving backwards has worked fine until reaching start of file. At that
point more printout had one line too much in output causing more seemingly
never to be able to reach the start. That is now fixed, with clean ups to
skip_backwards() making it less confusing.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
This commit also includes fix to how initial skip lines and search are
instructed in the code. Earlier version was pretty close impossible to make
work with getopt_long() and had minor flaw - if both initial skip lines and
search were defined at the same time the skipping did not happen. That is
now corrected.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
When suspending only the more process. Sending signal to process group
makes signal destination unnecessarily vague. After the suspend is over
SIGCONT is expected, and it needs to ensure output terminal settings are
what more needs.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
The wait() is now a little more robust by being more tolerate rogue SIGCHLDs
and unblocked signals. The repeated fork() and sleep() is removed, if first
try does not success give up without delay to provide user timely feedback.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
If streams need to be flushed do not try to be clever, just flush all of
them. That lead to finding unnecessary print out in run_shell() that is
removed in this commit.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
The sys/ttydefaults.h can be used for CTRL() and rubout that is CERASE. For
alarm it's best to keep things simple and call printf() alarm modifier.
QUIT was not in use, so it is just deleted.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
As a minor robustness improvement use ftello() to find out file position
rather than add and substract from old value.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Use the backspace provided dy terminfo, and if that is not possible use
fallback.
Simplify entering and exiting standard mode. Now when these std_enter and
std_exit variables are initialized correctly they do not need to be checked
quite as complex way as earlier.
Do not call prepare_line_buffer() unnecessarily. The line buffer needs only
to be prepared only at times of more(1) initialization and when terminal
size changes.
Add a function to print various separators, that is better than static
strings.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Earlier main() inline code implemented input files and stdin displaying
separately. With small restructuring this code can be shared.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Return rather than use long 'else' statement. Also unwind some overly
compact variable set and test code, that can be best observed how
underlining settings are now handled.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Currently open file is needed a bit here and there. Move it to control
structure to avoid passing it around. This also makes fixing memory leak at
more_exit() very easy.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
This change also removes a ghost prompt that was left in place when printing
help. When looking old commits the output fluke has been around at least
since util-linux v2.2.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Replacing long 'else' with 'continue' allows dropping one level of
indentation. Main aim is to improve code readability by reducing
complexity one needs to keep track while looking the code.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Pagers are not expected to have setuid or setgid bits, but assuming such
mistake has taken place try to avoid privilege escalation.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
This change has three possitive effects: Avoid expansion code when it is
not needed. Stop wasting memory from stack for something rarely needed.
When command buffer is needed stop limiting it's size artificially.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
From man siglongjmp(3) 'longjmp() and siglongjmp() make programs hard to
understand and maintain. If possible, an alternative should be used.' That
manual page remark summarizes quite well why more(1) needs to move away from
jumping around.
Signed-off-by: Sami Kerola <kerolasa@iki.fi>
