reading from a word list and picking a random word

[Mia_tech]

this I'm doing it in shell, so I appreciate if the code is en same script... I trying to generate a random word. I'm not aware of any command that can do this in ubuntu; however, I know that with `cat file | wc -l` you can get the number of lines in a given file, if I just could pass that number into a var and generate a random number then extracting that word from the file and put it into a variable... that's the idea... if there is an easier way of generating random words with build in linux commands, I would like to know


thanks

[Vaphell]

#!/bin/bash

dict='dict.txt'
n=$( cat "$dict" | wc -l )
while true
do
rnd=$( date '+%N' | sed 's/^[0]*//' )
lnum=$(( rnd % n + 1 ))
word=$( awk -v lnum=$lnum 'NR==lnum { print }' "$dict" )
echo $lnum $word
done

date formatted to nanoseconds should provide enough randomness
row_number = rnd modulo line_count + 1

[Mia_tech]

#!/bin/bash

dict='dict.txt'
n=$( cat "$dict" | wc -l )
while true
do
rnd=$( date '+%N' | sed 's/^[0]*//' )
lnum=$(( rnd % n + 1 ))
word=$( awk -v lnum=$lnum 'NR==lnum { print }' "$dict" )
echo $lnum $word
done

date formatted to nanoseconds should provide enough randomness
row_number = rnd modulo line_count + 1

are you sure you're not missing a condition somewhere?.... and I don't see much diff b/t "date +%N" and "date '+%N' | sed 's/^[0]*//'"... it generate random number either way

[Vaphell]

depends on what that condition is supposed to do

[Mia_tech]

much simpler... just using command substitution


FILE=file.txt
LINES=`wc -l $FILE`
RANDOM=`date +%N`
NUMBER=$[ ( $RANDOM % $LINES ) + 1 ]
WORD=`sed -n $NUMBER\p $FILE`

[PunkLV]

#!/bin/bash
dictionary="/usr/share/dict/american-english"
word=""
wordlist="./wordlist"


function choose_words(){
characters=$1
for word in $(cat $dictionary); do
word_length=$(expr length $word)
if [ $word_length == $characters ]; then
echo $word >> ./wordlist$characters
fi
done

}

function get_word(){
characters=$1
if [ -f "$wordlist$characters" ]; then
random=$RANDOM
range=$( wc -l $wordlist$characters | grep -o [0-9]*[0-9])
random=$(($random%$range))
random_word=$(awk NR==$random $wordlist)
else
choose_words $characters
fi
}

echo "Word length?: "
read word_length

# makes a list of all words. Takes up some time
# because dictionary has 100k entries.
# However, you could make 10-12 files once and use them.

# choose_words $word_length
get_word $word_length
echo $random_word

Coding is not as nice as the previous, yet this was fun.
Should work fine.

[Vaphell]

much simpler... just using command substitution

FILE=file.txt
LINES=`wc -l $FILE`
RANDOM=`date +%N`
NUMBER=$[ ( $RANDOM % $LINES ) + 1 ]
WORD=`sed -n $NUMBER\p $FILE`


i merely showcased randomness with an infinite loop
besides your script does pretty much what my did, except you used backticks instead of better $()
http://mywiki.wooledge.org/BashFAQ/082

also on my box wc -l $file returns '<int> <filename>' and that doesn't look like a valid division parameter.

./rnd.sh: line 8: ( 2327 % 196 dict.txt ) + 1 : syntax error

Also i had to remove leading zeroes in my version, because it threw an error (leading zeroes indicate base-8 )

./rnd.sh: line 19: 006382406: value too great for base (error token is "006382406")
For some reason your modulo line doesn't trigger it. It's interesting because my tests say that in your version backticked date +%N seems to degrade to 16bit integer while my $( ) preserves all 9digits

[PunkLV]

There is grep for wc. And had used a modulo division to limit $RANDOM.
Anyways. Seems to work for me, if this is what OP was asking for.
And yeah, needs proper testing anyways.

[Vaphell]

i was replying to OP's post :)

[kaibob]

The Greg's Wiki site has an informative FAQ on this topic:

http://mywiki.wooledge.org/BashFAQ/026

[Some Penguin]

sort -R <infile> | head -1


would also work; sort -R randomizes.

[Mia_tech]

i merely showcased randomness with an infinite loop
besides your script does pretty much what my did, except you used backticks instead of better $()
http://mywiki.wooledge.org/BashFAQ/082

also on my box wc -l $file returns '<int> <filename>' and that doesn't look like a valid division parameter.

./rnd.sh: line 8: ( 2327 % 196 dict.txt ) + 1 : syntax error

Also i had to remove leading zeroes in my version, because it threw an error (leading zeroes indicate base-8 )

./rnd.sh: line 19: 006382406: value too great for base (error token is "006382406")
For some reason your modulo line doesn't trigger it. It's interesting because my tests say that in your version backticked date +%N seems to degrade to 16bit integer while my $( ) preserves all 9digits

you're right!.. I forgot to fix it. here's what it should've been


LINES=`wc -l $FILE | cut -d " " -f1`

[Mia_tech]

sort -R <infile> | head -1


would also work; sort -R randomizes.

you know what? simplicity is king when it comes to programming!... nice!!!

[cgroza]

In python this would be child play.

[cgroza]

People should switch to python:



#!/usr/bin/python
import os, random, sys

def readfile(file):
if os.path.exists(file):
txt_file = open(file, "r")
return txt_file.read()
else:
print "Invalid path"
return

def pick_random_word(text):
words = text.split()
return words[random.randrange(0, len(words))]

def main(path):
print pick_random_word(readfile( path))

if __name__ == "__main__":
main(sys.argv[1])

[hakermania]

Common!

LINES=$(wc -l < test)
is the right way!
Then you simply get a random number using the variable $RANDOM which is at the $LINES's range (let's call this number $random_line) and finally, the random word comes:

random_word=$(sed -n "${random_line}p;${random_line}q" dict.txt)

I agree that simplicity is good but you have to do some things at first alone (e.g. to generate a random number or to read a specific line from a file), automation isn't always good.

[bashologist]

I know it's solved but for future google searches or something...

Does it have to be indexed? Did you need the index for something? If not then here's another solution.

Not sure if this's what you wanted but

sort -R file | head -n 1

A faster possible option would be to use seek to a random byte position then read until the next newline then read that line and exit.

Edit: Someone beat me to it lol. Great minds think alike.