#!/bin/bash
cnt=1
mkdir -p backup
mkdir -p attaches
rm -fr attaches/*
rm -f new.zip
cd res
for f in *.zip
do
#fname=`date -D yyddmm | sed -e 's/ //g' -e 's/://g'`
fname=`date +%Y%m%d`
echo "extracting $f ..."
echo "cp $f "../backup/${fname}-${cnt}.zip" "
unzip -o -O CP936 $f -d ../attaches
cnt=$(( $cnt + 1 ))
done
cd ..
mkdir -p new
rm -fr new/*
cd attaches
useful=0
total=0
rm -f ../jiaoben.sql
for f in *
do
ext=`echo $f |awk -F. '{print $2}'`
rm -f c.txt
if [ "$ext" = "pdf" ] ; then
echo $f | grep -q 51job
r=$?
if [ $? -eq 0 ] ; then
name=`echo $f | awk -F. '{print $1}'`
else
name=`echo $f |sed -e 's/【//g' -e 's/】/ /' -e 's/\./ /g' |awk '{print $4}'`
fi
pdftotext "$f" c.txt 2>/dev/null
cat c.txt 2>/dev/null | tr '\n' 'K' |sed 's/KK/\n/g' | sed 's/K/\n/g' >mc.txt
mv mc.txt c.txt
phonenum=`cat c.txt 2>/dev/null | sed 's/-//g' |grep -Eo '1[3-9][0-9]{9}'`
phonecount=`cat c.txt 2>/dev/null | sed 's/-//g' |grep -Eoc '1[3-9][0-9]{9}'`
content=''
elif [ "$ext" = "html" ] ; then
phonenum=`cat $f |sed 's/-//g' |grep -Eo '1[3-9][0-9]{9}'`
phonecount=`cat $f | sed 's/-//g' |grep -Eoc '1[3-9][0-9]{9}'`
content=""
name=`echo $f | awk -F. '{print $1}'`
else
name=`echo $f |sed -e 's/【//g' -e 's/】/ /' -e 's/\./ /g' |awk '{print $4}'`
phonenum=""
phonecount=0
fi
grep -q ",$name," ../old.txt
retname=$?
if [ $phonecount -gt 1 ] ; then
echo " $f A is $phonenum ,AS $phonecount"
phonenum=`echo $phonenum |awk '{print $1}' `
fi
retphone=1
if [ "ZZ$phonenum" != "ZZ" ] ; then
grep -q $phonenum ../old.txt
retphone=$?
fi
curtime=`date +%s`
if [ "$retphone" = "0" ] ; then
echo " $f 重复 "
elif [ "ZZ$phonenum" = "ZZ" -a $retname = "0" ] ; then
echo " $f 名字重复,重复"
else
if echo $f |grep -q -E '[6-9]年|[1-9][0-9]年' ; then
echo "$f 忽略"
else
echo "开始转换$f文件,$phonenum, : $name"
echo "$phonenum$f" |sed -e 's/【//g' -e 's/】/ /' -e 's/\./ /g' |awk '{print $1,$4,$6}' | sed 's/ /,/g' >>../old.txt
pdfurl="/ok/$phonenum${name}.$ext"
sshpass -p '1111111@' scp "$f" root@192.168.10.110:/www/wwwroot/skilluploads/oks/$phonenum${name}.$ext
sql1="INSERT INTO fatable ( user_id, title, createtime, updatetime, status, phone, feetype, zhuanye, xuexiao, xueli, dhzt, bmzt, zixunshi, pdfurl, content) VALUES ( 1, "
sql2="'"$name"', $curtime, $curtime, '', '"$phonenum"', '', '', '', '', '', '', '', '"$pdfurl"', '"$content"');"
echo $sql1$sql2 >>../jiaoben.sql
useful=$(( $useful + 1 ))
echo "$f 有效:"
mv "$f" "../new/$phonenum$f"
echo "$f文件处理结束"
fi
fi
total=$(( $total + 1 ))
done
cd ..
echo "useful $useful, total $total, 有效率: $(( $useful * 100 / $total))"
echo inporting...
mysql -udatabasen -ppass -h 192.168.10.110 database <jiaoben.sql >sqlerror 2>&1
echo done
贡献批处理文件并上传到文件和数据库服务器
最新推荐文章于 2021-12-02 19:02:27 发布