===== Links =====

Terminals:
  * [[habrahabr>164687|Альтернативный терминал для Windows]]
  * [[habrahabr>176077|Интересная замена стандартного терминала: Terminology]]
  * [[habrahabr>185944|Final Term: новый взгляд на терминал]]
  * [[habrahabr>248881|Обзор Friendly interactive shell (fish) и почему она лучше bash]]
  * [[habrahabr>204368|Linux-like терминал в Windows]]
  * [[http://software.jessies.org/terminator/|Terminator]] -- a cross-platform (Java) GPL terminal emulator with advanced features

Shells:
  * [[https://packages.debian.org/jessie/shells/|Shells in Debian]]
  * [[http://fishshell.com/|Fish]] -- smart and user-friendly command line shell
  * [[habrahabr>267797|Консоль 21 века: mosh, tmux, fish]]

Other tips:
  * [[http://www.spinics.net/lists/util-linux-ng/msg13866.html|New ''columns'' syntax]]


===== Questions answered =====

=== How to use variable in while loop? ===

<note question>
The following scripts do not work properly:
  * <code bash>
VAR=0
ls -1 | while read file
do
	[ "$file" = "tmp" ] && VAR=1
done
echo $VAR # It turns out that $VAR is always zero here
</code>
  * <code bash>
echo "tmp.txt" | read file
</code>
</note>
<note tip>
If ''while'' loop consumes the input from the pipe, it is launched in a subshell, thus all changes to the variables inside the loop do not work as expected (are not propagated to calling shell). Solution: loop should be executed by primary shell:
  - From [[stackoverflowa>16855194/267197|Modifying variable inside while loop is not remembered]]:
    <code bash>
VAR=0
while read file
do
	[ "$file" = "tmp" ] && VAR=1
done < <(ls -1)
</code>
    Process substitution which is used in above example, is described [[http://wiki.bash-hackers.org/syntax/expansion/proc_subst|here]]. Another good application of process substitution is [[stackoverflowa>43972501/267197|capturing the exit code of first command in a pipe]]:
    <code bash>
if mount /dev/sda /mnt 2> >(head -1)
then
	...
fi
</code>
    The same way ''read'' and ''arrayread'' when applied after the pipe initialize the variable in a forked shell (hence ''echo file.txt | read line'' does not work) so the solution looks this:
    <code bash>
read file < <(echo file.txt)
</code>
  - From [[http://unix.stackexchange.com/questions/197121/|Create variables and assign values via loop]]:
    <code bash>
VAR=0
for file in $(ls -1)
do
	[ "$file" = "tmp" ] && VAR=1
done
</code>
</note>


=== [[stackoverflowa>22439016/267197|How to evaluate the intersection of two lists]]? ===

<note tip>
[[wp>Intersection (set theory)|Intersection]] of two lists:
  * ''cat 1.txt 2.txt | sort | uniq -d''
  * ''comm -12 1.txt 2.txt''

[[wp>Symmetric difference]] of two lists:
  * ''cat 1.txt 2.txt | sort | uniq -u''
</note>


=== How to redirect output to STDERR? ===

<note tip>
  * ''%%( >&2 echo "bla-bla" );%%''
  * ''%%echo "bla-bla" > /dev/stderr%%''
  * ''%%... | tee /dev/stderr | ...%%''
</note>


=== How to initialize variable with //here//-document? ===

<note tip>
From [[stackoverflow>1167746|How to assign a heredoc value to a variable in Bash]]:

<code bash>
read -r -d '' SCRIPT <<'EOS'
use strict;

my $lines = 0;

while (<>) {
	chomp;
	
	$files++;
}

print "Number of lines: $lines\n";

EOS

perl -e "$SCRIPT"
</code>
</note>


=== [[stackoverflow>604864|How to skip 1st line of output?]] ===

<note tip>
''%%... | tail -n +2%%''
</note>


=== How to read lines from file preserving leading spaces and without splitting? ===

<note tip>
From [[stackoverflowa>7314111/267197|Use bash to read line by line and keep space]] and [[stackoverflowa>29689199/267197|Bash read line does not read leading spaces]]:

<code bash>
while IFS= read -r line
do
	echo "$line"
done < some_file.txt
</code>

Similarly to read the whole file into variable:

<code bash>
IFS= FILE=`cat some_file`
</code>
</note>


=== How to list files with their sizes and MD5 sum? ===

<note tip>
<code bash>
(
echo -e "File\tMD5\tSize\tCTime"
find . -type f -printf '%p\t%s\t%T+\n' | sort -k1 -t$'\t' | while IFS= read -r line
do
     echo -e "$(md5sum "`echo "$line" | cut -f1`" | cut -f1 -d' ')\t$line"; \
done
) | column -t -s $'\t'
</code>

See also:
  * [[https://unix.stackexchange.com/questions/57222|How can I use column to delimit on tabs and not spaces?]], [[https://unix.stackexchange.com/questions/35369|How to define 'tab' delimiter for 'cut'?]]
  * [[https://unix.stackexchange.com/questions/118433|Quoting within $(command substitution) in Bash]]
</note>


=== How to feed file names containing space to ''xargs''? ===

<note tip>
[[stackoverflowa>17325825/267197|How can I use xargs to copy files that have spaces and quotes in their names]] actually says that ''xargs'' should be tweaked, but if the list is produced by ''find'' then solutions are:
  * Use null separator: ''%%find ... -print0 | xargs -0 wc -l%%''
  * Use ''find'' ability to append arguments: ''%%find ... -exec wc -l {} +%%''
</note>


=== [[stackoverflowa>451204/267197|How to sum sequence of numbers?]] ===

<note tip>
<code bash>
seq 1 10 | paste -s -d+ | bc
</code>
</note>


=== [[https://unix.stackexchange.com/a/18742/36095|How to count number of specific characters in the file?]] ===

<note tip>
<code bash>
$ cat file | tr -d -c '#' | wc -c
</code>
</note>


=== [[stackoverflow>2172352|How to check if a string begins or ends with some value?]] ===

<note tip>
See [[http://tldp.org/LDP/abs/html/testconstructs.html#DBLBRACKETS|Advanced Bash-Scripting Guide, Chapter 7]]:

<code bash>
if [[ $fullpath == *.txt ]]
then
	echo "This is text file"
fi
</code>
</note>


=== How to extract filename and extension? ===

<note tip>
From [[stackoverflow>965053|Extract filename and extension in bash]]:

<code bash>
file=`basename "$fullpath"`
extension=${file##*.}
filename=${file%.*}
</code>
</note>


=== [[stackoverflow>6426142|How to append/prepend the string to each array element?]] ===

<note tip>
<code bash>
array=( "${array[@]/#/prefix_}" )
array=( "${array[@]/%/_suffix}" )
</code>
To print an array as new-line separated string:
<code bash>
IFS=$'\n' echo "${array[*]}"
IFS=$'\n' echo "$*" | xargs ...
</code>
</note>


=== How to generate random strings? ===

<note tip>
To generate e.g. 5 random passwords each having 16 characters in alphabet "A-Za-z0-9" use:
<code bash>
tr -dc A-Za-z0-9 < /dev/urandom | fold -w 16 | head -5
</code>
</note>


=== [[http://unix.stackexchange.com/a/4035/36095|How to re-attach the process to another terminal?]] ===

<note tip>
<code bash>
$ myprog
# press Ctrl+Z to suspend the program
$ bg
$ disown
# switch to another terminal / screen / tmux session
$ reptyr $(pgrep myprog)
</code>
</note>


=== How to turn off bell? ===

<note tip>
To disable bash from sending bell to terminal add/uncomment the following in ''~/.bashrc'':

<code>
set bell-style none
</code>

However bell can be useful to notify long running task:

  * [[stackoverflowa>7895288/267197|Configure PuTTY to flash Windows taskbar on bell]]
  * Latest ConEmu (v170807) supports this out-of-box unless //Disable all flashing// is not enabled in //[[https://conemu.github.io/en/SettingsFeatures.html|Settings → Features]]//.
</note>


=== [[http://unix.stackexchange.com/questions/25372/turn-off-buffering-in-pipe|Turn off buffering for pipe]] ===

<note question>
When data is output to stdout, one can see the result immediately (it flushes after each write). However if output is piped to another command, it is buffered, thus causing difficulties e.g. with ''tee'' utility.
</note>
<note tip>
The problem is solved using ''stdbuf'' utility:
<code bash>
stdbuf -oL tail -f /var/log/messages | tee logs.txt
</code>
</note>


=== [[habrahabr>148988|Можно ли в UNIX обнулять (truncate) файл, в который пишет некоторый процесс?]] ===

<blockquote>
Таким образом, оболочки sh и csh, а также tcsh при открытии нового файла, не используют флаг ''O_APPEND'', а описанный выше способ обнуления используемых файлов в этих оболочках использовать не получится. В более современных оболочках (ksh, bash, zsh) этой проблемы нет.
</blockquote>


{{tag>terminal shell}}