Bash - Another thing for the lazy: umount by name

Hi, just another script. You won't believe how many errors you can get before this simple thing works. Sometimes I wonder if my brain does it just for kicks, to see what funny message we can get this time. Or maybe it is too lazy to bother with heuristics and just goes over all the imaginable options. Surprisingly, the second thing feels like an easier one.

Or maybe it does both: 

#!/bin/bash

if [ -z $1 ]
  then printf "\nPlease provide drive name (label)\n\n"
    exit
fi

drivelabel = $1

cat /etc/mtab | grep ${drivelabel} | cut -d " " -f1 | xargs umount

Bash - Bulk file update with find and sed

Doing bulk file update (mind escapes when using complicated patterns):

#!/bin/bash

if [ -z $1 ] || [ -z $2 ]  || [ -z $3 ] || [ -z $4 ]; then
        printf "\nPlease enter:\n
        1. file command-compatible filename pattern
        2. backup suffix for affected files
        3. pattern to replace
        4. replacement\n\n
        For instance, $0 *.txt .xml-backup blue red\n
        This should replace blue with red in all txt files\n\n"
        exit
fi


filename_pattern=$1
backup_suffix=$2
pattern_to_replace=$3
replace_with=$4

find . -type f -name ${filename_pattern} -exec sed -i${backup_suffix} "s|${pattern_to_replace}|$replace_with|g" {} \+

Bash - Cut a fragment of a log by some criteria (like timestamp)

Here is how to cut out a piece of a log between two unique strings (like timestamps):

#!/bin/bash

printf "\nResults are returned *inclusive*\n\n"

if [ -z $1 ] || [ -z $2 ] || [ -z $3 ]; then
        printf "\nPlease enter SUBSTRSTART SUBSTREND and LOGPATH\nFor instance, $0 11:59:17,206 12:02:14,606 test.log\n\n"
        exit
fi

substrstart=$1
substrend=$2
logpath=$3

sed -n "/${substrstart}/,/${substrend}/p" ${logpath} | tee log_excerpt-$(date +%F_%H-%M-%S) | cat

Bash - Getting network data for the port (localhost only)

Getting network data:

#!/bin/bash

printf "\nCurrently for localhost only\n\n"

if [ -z $1 ]; then
    printf "Please enter port and [host]\n\n"
    exit
fi

port=$1
host=localhost

printf '========= lsof -i ============\n\n'

lsof -i :${port}

printf '========= netstat -lnt | grep $port ============\n\n'

netstat -lnt | grep ${port}

printf '========= nc -z $host $port  ============\n\n'

nc -z ${host} ${port}

Bash - kill everything with a matching name

Just a small script for killing matching processes by name:

#!/bin/bash

if [ -z $1 ]
then
    printf "keyword is empty\n\n"
    exit
fi

let k=0
let min=1

# mind the double brakets in do part
for i in $(ps -e | grep "$1"); do ((k++)); done

if [ "${k}" -lt "${min}" ]
then
    printf "no such process to kill\n\n"
    exit
fi


for i in $(ps -e | grep $1 | cut -s -f1 -d\t); do kill -9 ${i}; done

Password generator with bash

Looks like the best way to handle your passwords strongly depends on which password breaking approach is the most effective at the moment, and apparently those ways are changing as quickly as technology develops. Which is why it's unlikely that there will ever be a perfect way to avoid all the risks, but we can design an optimal one.

It occurred to me that first we need to select between two strategies: storing generated password or memorizing them with mnemonics. I don't think it would be a good idea to store memorizable data cause it's too easy to break the pattern, or to try to remember a generated one. As I already can do mnemonics, so I decided to write my own password generator (I like having my own collection of tools).

Solution is based on LINK. Stuff in use: cat, tr, fold, head, /dev/urandom, echo, printf, for each and parameters testing