一个简单的文本查询程序------bash sample(2)

时间：2010-08-25 来源：小妮子8308

功能：

1. 检查参数个数；

2. 检查input file是否存在；

3. 列出input file中出现频率最高的10个单词；

4. 列出这10个单词所在的行数和句子内容；

5. 手动输入input string，在input file中查询string并记录。

#! /bin/bash
#
#run:./textsh.sh inputfilename
ARGNO=1 #expecting number of parameter
E_WRONGARGS=65
temp="tempfile"
output="freout.txt"
# clean
rm $output
# check the parameter
#if [ $# != $ARGNO ]
if [ $# -ne $ARGNO ]
then
echo "Parameter error. Expecting only $ARGNO parameter(s)."
exit $E_WRONGARGS
fi
# check file exists
filename=$1 #get input file name
if [ -e $filename ]; then
echo "Input file is $filename."
else
echo "File $filename could not be found. Please check!"
exit $E_WRONGARGS
fi
# list 10 most frequent words
#echo "(1) Here are 10 most frequent words in file $filename:" > $output
printf "(1) Here are 10 most frequent words in file $filename:\n" > $output
cat $filename | sed -e "s/[^a-zA-Z]/\n/g" | grep -v ^$ | sort | uniq -c | sort -n -k 1 -r | head -10 > $temp
cat $temp >> $output
printf "\n(2) Lists of words:\n" >> $output
for x in $(awk '{print $NF}' $temp)
do
# echo -e >> $output
printf "\n[$x]:\n" >> $output
grep -En "\b$x\b" $1 >> $output
done
# manual search
printf "\n(3) To search what you want?\n" >> $output
echo "To search what you want? Please input your word!"
read input
while [ "$input" != "exit" ]
do
if [ "$input" != "exit" ]
then
printf "\nsearch for [$input]:\n" >> $output
grep -En "\b$input\b" $1 | tee -a $output
echo -e | tee -a $output
echo "Continue? Please input your word. (To stop by using 'exit'.)"
read input
fi
done
echo -e
echo "Exit of searching."
#clean
rm $temp