I am reading a source code, and find these lines :
if [ -n "${INIT_NAMENODE+1}" ]
then
echo "Initializing namenode"
else
echo "Starting namenode"
fi
how should I interpred the 'if' condition : if [ -n "${INIT_NAMENODE+1}" ] ? ?
The nice thing about this code is that it is not written for a "Linux shell". It is written for the more general category of "UNIX shell". It will work in everything since V7 UNIX (1979) at least. People with lesser portability goals might write it without the -n.
The first item of interest is the ${foo+bar} syntax. This is a test for existence of the foo parameter. If $foo exists, then ${foo+bar} equals bar. If $foo doesn't exist, then ${foo+bar} equals the empty string.
If you look for this in your shell man page, it's usually documented as ${foo:+bar}, along with some other related forms like ${foo:-bar}, and somewhere nearby there's a note explaining that the colon can be omitted from all of them, resulting in slightly different behavior (with the colon, variables whose value is the empty string are treated the same as nonexistent variables).
Next we have the [ -n ... ] test. -n tests the following string for emptiness. It succeeds if the string is non-empty. From the previous paragraph we know that ${INIT_NAMENODE+1} is empty if and only if $INIT_NAMENODE doesn't exist. So the -n test succeeds if $INIT_NAMENODE exists. The value 1 doesn't really matter here - it would do the same thing if you changed the 1 to 2 or 0 or teapot. All that matters is that it's not an empty string, since -n doesn't care about the rest.
Try some examples from your shell prompt: echo ${PATH+hello} should say hello because you do have a $PATH variable. echo ${asdfghjkl+hello} should print a blank line.
So, in the context of the if statement, the purpose of the test is to do the first echo if the variable $INIT_NAMENODE exists, and the second echo if the variable $INIT_NAMENODE doesn't exist.
Related
This question already has answers here:
Read user input inside a loop
(6 answers)
Closed 5 years ago.
First post here! I really need help on this one, I looked the issue on google, but can't manage to find an useful answer for me. So here's the problem.
I'm having fun coding some like of a framework in bash. Everyone can create their own module and add it to the framework. BUT. To know what arguments the script require, I created an "args.conf" file that must be in every module, that kinda looks like this:
LHOST;true;The IP the remote payload will connect to.
LPORT;true;The port the remote payload will connect to.
The first column is the argument name, the second defines if it's required or not, the third is the description. Anyway, long story short, the framework is supposed to read the args.conf file line by line to ask the user a value for every argument. Here's the piece of code:
info "Reading module $name argument list..."
while read line; do
echo $line > line.tmp
arg=`cut -d ";" -f 1 line.tmp`
requ=`cut -d ";" -f 2 line.tmp`
if [ $requ = "true" ]; then
echo "[This argument is required]"
else
echo "[This argument isn't required, leave a blank space if you don't wan't to use it]"
fi
read -p " $arg=" answer
echo $answer >> arglist.tmp
done < modules/$name/args.conf
tr '\n' ' ' < arglist.tmp > argline.tmp
argline=`cat argline.tmp`
info "Launching module $name..."
cd modules/$name
$interpreter $file $argline
cd ../..
rm arglist.tmp
rm argline.tmp
rm line.tmp
succes "Module $name execution completed."
As you can see, it's supposed to ask the user a value for every argument... But:
1) The read command seems to not be executing. It just skips it, and the argument has no value
2) Despite the fact that the args.conf file contains 3 lines, the loops seems to be executing just a single time. All I see on the screen is "[This argument is required]" just one time, and the module justs launch (and crashes because it has not the required arguments...).
Really don't know what to do, here... I hope someone here have an answer ^^'.
Thanks in advance!
(and sorry for eventual mistakes, I'm french)
Alpha.
As #that other guy pointed out in a comment, the problem is that all of the read commands in the loop are reading from the args.conf file, not the user. The way I'd handle this is by redirecting the conf file over a different file descriptor than stdin (fd #0); I like to use fd #3 for this:
while read -u3 line; do
...
done 3< modules/$name/args.conf
(Note: if your shell's read command doesn't understand the -u option, use read line <&3 instead.)
There are a number of other things in this script I'd recommend against:
Variable references without double-quotes around them, e.g. echo $line instead of echo "$line", and < modules/$name/args.conf instead of < "modules/$name/args.conf". Unquoted variable references get split into words (if they contain whitespace) and any wildcards that happen to match filenames will get replaced by a list of matching files. This can cause really weird and intermittent bugs. Unfortunately, your use of $argline depends on word splitting to separate multiple arguments; if you're using bash (not a generic POSIX shell) you can use arrays instead; I'll get to that.
You're using relative file paths everywhere, and cding in the script. This tends to be fragile and confusing, since file paths are different at different places in the script, and any relative paths passed in by the user will become invalid the first time the script cds somewhere else. Worse, you aren't checking for errors when you cd, so if any cd fails for any reason, then entire rest of the script will run in the wrong place and fail bizarrely. You'd be far better off figuring out where your system's root directory is (as an absolute path), then referencing everything from it (e.g. < "$module_root/modules/$name/args.conf").
Actually, you're not checking for errors anywhere. It's generally a good idea, when writing any sort of program, to try to think of what can go wrong and how your program should respond (and also to expect that things you didn't think of will also go wrong). Some people like to use set -e to make their scripts exit if any simple command fails, but this doesn't always do what you'd expect. I prefer to explicitly test the exit status of the commands in my script, with something like:
command1 || {
echo 'command1 failed!' >&2
exit 1
}
if command2; then
echo 'command2 succeeded!' >&2
else
echo 'command2 failed!' >&2
exit 1
fi
You're creating temp files in the current directory, which risks random conflicts (with other runs of the script at the same time, any files that happen to have names you're using, etc). It's better to create a temp directory at the beginning, then store everything in it (again, by absolute path):
module_tmp="$(mktemp -dt module-system)" || {
echo "Error creating temp directory" >&2
exit 1
}
...
echo "$answer" >> "$module_tmp/arglist.tmp"
(BTW, note that I'm using $() instead of backticks. They're easier to read, and don't have some subtle syntactic oddities that backticks have. I recommend switching.)
Speaking of which, you're overusing temp files; a lot of what you're doing with can be done just fine with shell variables and built-in shell features. For example, rather than reading line from the config file, then storing them in a temp file and using cut to split them into fields, you can simply echo to cut:
arg="$(echo "$line" | cut -d ";" -f 1)"
...or better yet, use read's built-in ability to split fields based on whatever IFS is set to:
while IFS=";" read -u3 arg requ description; do
(Note that since the assignment to IFS is a prefix to the read command, it only affects that one command; changing IFS globally can have weird effects, and should be avoided whenever possible.)
Similarly, storing the argument list in a file, converting newlines to spaces into another file, then reading that file... you can skip any or all of these steps. If you're using bash, store the arg list in an array:
arglist=()
while ...
arglist+=("$answer") # or ("#arg=$answer")? Not sure of your syntax.
done ...
"$module_root/modules/$name/$interpreter" "$file" "${arglist[#]}"
(That messy syntax, with the double-quotes, curly braces, square brackets, and at-sign, is the generally correct way to expand an array in bash).
If you can't count on bash extensions like arrays, you can at least do it the old messy way with a plain variable:
arglist=""
while ...
arglist="$arglist $answer" # or "$arglist $arg=$answer"? Not sure of your syntax.
done ...
"$module_root/modules/$name/$interpreter" "$file" $arglist
... but this runs the risk of arguments being word-split and/or expanded to lists of files.
I am using a bash script to execute a program. The program must take the following argument. (The program is gnuplot.)
gnuplot -e "filename='output_0.csv'" 'plot.p'
I need to be able to assemble the following string: "filename='output_0.csv'"
My plan is to assemble the string STRING=filename='output_0.csv' and then do the following: gnuplot -r "$STRING" 'plot.p'. Note I left the words STRING without stackoverflow syntax style highlighting to emphasise the string I want to produce.
I'm not particularly proficient at bash, and so I have no idea how to do this.
I think that strings can be concatenated by using STRING="$STRING"stuff to append to string? I think that may be required?
As an extra layer of complication the value 0 is actually an integer which should increment by 1 each time the program is run. (Done by a for loop.) If I have n=1 in my program, how can I replace the 0 in the string by the "string value" or text version of the integer n?
A safest way to append something to an existing string would be to include squiggly brackets and quotes:
STRING="something"
STRING="${STRING}else"
You can create the "dynamic" portion of your command line with something like this:
somevalue=0
STRING="filename='output_${somevalue}.csv'"
There are other tools like printf which can handle more complex formatting.
somevalue=1
fmt="filename='output_%s.csv'"
STRING="$(printf "$fmt" "$somevalue")"
Regarding your "extra layer of complication", I gather that this increment has to happen in such a way as to store the value somewhere outside the program, or you'd be able to use a for loop to handle things. You can use temporary files for this:
#!/usr/bin/env bash
# Specify our counter file
counter=/tmp/my_counter
# If it doesn't exist, "prime" it with zero
if [ ! -f "$counter" ]; then
echo "0" > $counter
fi
# And if it STILL doesn't exist, fail.
if [ ! -f "$counter" ]; then
echo "ERROR: can't create counter." >&2
fi
# Read the last value...
read value < "$counter"
# and set up our string, per your question.
STRING="$(printf "filename='output_%d.csv'" "${value}")"
# Last, run your command, and if it succeeds, update the stored counter.
gnuplot -e "$STRING" 'plot.p' && echo "$((value + 1))" > $counter
As always, there's more than one way to solve this problem. With luck, this will give you a head start on your reading of the bash man page and other StackOverflow questions which will help you learn what you need!
An answer was posted, which I thought I had accepted already, but for some reason it has been deleted, possibly because it didn't quite answer the question.
I posted another similar question, and the answer to that helped me also answer this question. You can find said question and answer here: bash: Execute a string as a command
I am trying to read the performance counters and get the IPC. I need to use IPC to control few machine specific parameters. I am using shell script to do the same. Please see the code below:
while true
do
retval=./perf periodic -e instructions -e cycles -s 50 -d -m td &
some_pid=$!
kill some_pid
if ["$retval" -gt "0.5"]
then
***something***
fi
sleep 1
done
I am getting following error:
Algorithm.sh[27]: kill: some_pid: arguments must be jobs or process IDs
Algorithm.sh[27]: periodic: not found
Algorithm.sh[27]: [: missing ]
Algorithm.sh[27]: kill: some_pid: arguments must be jobs or process IDs
Algorithm.sh[27]: [: missing ]
Algorithm.sh[27]: periodic: not found
Algorithm.sh[27]: kill: some_pid: arguments must be jobs or process IDs
Algorithm.sh[27]: [: missing ]
Can someone give me some pointers on how to get/return the value from perf instruction. I tried using function and returning the value, but it also failed.
---------UPDATED----------
Now I am running following, and one of the problem got solved and one is remaining.
./perf periodic -e instructions -e cycles -s 50 -d -m td > result.txt &
And other one is
while true
do
retval=$(tail -n 1 result.txt)
echo $retval
if ["$retval" -gt "0.5"]
then
echo "Hello mate"
fi
sleep 1
done
The echo is giving value, but then the if statement is not getting executed. It is giving following:
Algorithm.sh[30]: [: missing ]
0.302430
Algorithm.sh[30]: [0.302430: not found
0.472716
Algorithm.sh[30]: [0.472716: not found
0.475687
Algorithm.sh[30]: [0.475687: not found
I looked up the if condition syntax and couldn't spot the mistake. Please help.
Couple of shell syntax issues here.
First, retval=... is going to set the retval variable equal to the first part of the string on the right side of the '='. The ampersand will then background the whole thing, essentially throwing that value away. You probably meant to do:
retval=`./perf periodic -e instructions -e cycles -s 50 -d -m td`
which would store the output of the perf command into retval. However, that won't work if you put it into the background with '&'. You'll need to either (a) run it synchronously without the '&' as I've shown above, (b) redirect its output into a file and recover it after it's finished (you'll need to use wait to determine when that's happened), or (c) use a "coprocess" (too complicated to explain here: see the bash man page).
Also, you probably meant kill $some_pid? Without the '$', the string "some_pid" is passed as a literal argument to kill, which is probably not what you intended.
Edit
Following your revisions... The shell operates by splitting the command line up into individual tokens. So spaces are often important. In this case, the initial token being identified by the shell will be the combined value of ["$retval" (after variable substitution and quote removal). The last token will be 0.5] after removal of quotes. In the first invocation line then, the first token was simply '[' (presumably retval was empty the first time through). So there it's complaining about the last token not being the matching ']'. In the other iterations, the first token is '[' plus additional numeric text from $retval which is not providing a valid command name.
Once you fix that, you'll discover that the -gt operator only evaluates integer comparisons. You could use the bc(1) command. For example, this command will produce output of 1 if $retval is greater than 0.5; otherwise 0.
echo "$retval > 0.5" | bc
But note you'll need to ensure retval has a valid numeric expression or you'll cause a syntax error in bc. You would then need to capture the output and put it into a conditional. Something like this should work:
if [ "$retval" ]
then
x=$(echo "$retval > 0.5" | bc)
if [ $x -eq 1 ]
then
echo "hello mate"
fi
fi
(Note that with $(...) you don't need additional spaces next to the parentheses. And in the assignment statement x=foo, you must not have a space on either side of the =.)
I've just discovered a strange behaviour in bash that I don't understand. The expression
${variable:=default}
sets variable to the value default if it isn't already set. Consider the following examples:
#!/bin/bash
file ${foo:=$1}
echo "foo >$foo<"
file ${bar:=$1} | cat
echo "bar >$bar<"
The output is:
$ ./test myfile.txt
myfile.txt: ASCII text
foo >myfile.txt<
myfile.txt: ASCII text
bar ><
You will notice that the variable foo is assigned the value of $1 but the variable bar is not, even though the result of its defaulting is presented to the file command.
If you remove the innocuous pipe into cat from line 4 and re-run it, then it both foo and bar get set to the value of $1
Am I missing somehting here, or is this potentially a bash bug?
(GNU bash, version 4.3.30)
In second case file is a pipe member and runs as every pipe member in its own shell. When file with its subshell ends, $b with its new value from $1 no longer exists.
Workaround:
#!/bin/bash
file ${foo:=$1}
echo "foo >$foo<"
: "${bar:=$1}" # Parameter Expansion before subshell
file $bar | cat
echo "bar >$bar<"
It's not a bug. Parameter expansion happens when the command is evaluated, not parsed, but a command that is part of a pipeline is not evaluated until the new process has been started. Changing this, aside from likely breaking some existing code, would require extra level of expansion before evaluation occurs.
A hypothetical bash session:
> foo=5
> bar='$foo'
> echo "$bar"
$foo
# $bar expands to '$foo' before the subshell is created, but then `$foo` expands to 5
# during the "normal" round of parameter expansion.
> echo "$bar" | cat
5
To avoid that, bash would need some way of marking pieces of text that result from the new first round of pre-evaluation parameter expansion, so that they do not undergo a second
round of evaluation. This type of bookkeeping would quickly lead to unmaintainable code as more corner cases are found to be handled. Far simpler is to just accept that parameter expansions will be deferred until after the subshell starts.
The other alternative is to allow each component to run in the current shell, something that is allowed by the POSIX standard, but is not required, either. bash made the choice long ago to execute each component in a subshell, and reversing that would break too much existing code that relies on the current behavior. (bash 4.2 did introduce the lastpipe option, allowing the last component of a pipeline to execute in the current shell if explicitly enabled.)
I wanted to run a simple bash timer and found this online (user brent7890)
#!/usr/bin/bash
timer=60
until [ "$timer" = 0 ]
do
clear
echo "$timer"
timer=`expr $timer - 1`
sleep 1
done
echo "-------Time to go home--------"
I couldn't copy and paste this code because the server is on another network. I typed it like this (below) and got an error on the line that starts with "until".
#!/usr/bin/bash
timer=60
#Note I forgot the space between [ and "
until ["$timer" = 0 ]
do
clear
echo "$timer"
timer=`expr $timer - 1`
sleep 1
done
echo "-------Time to go home--------"
Where is spacing like this documented? It seems strange that it matters. Bash scripts can be confusing, I want to understand why the space is important.
There are several rules, two basic of that are these:
You must separate all arguments of a command with spaces.
You must separate a command and the argument, that follows after, with a space.
[ here is a command (test).
If you write ["$timer" that means that you start command [60,
and that is, of course, incorrect. The name of the command is [.
The name of the command is always separated from the rest of the command line with a space. (you can have a command with a space in it, but in this case you must write the name of the command in "" or '').