1

我有一个带有以下几行的 bash 脚本

for file in $diff_file_list
    do

        # replace any ? with $current_date and replace any % with $file
        formatted_output_filename=$(echo $output_filename | sed "s|?|$current_date|g" | sed "s|%|_$file|g")
        $pig_bin_dir/pig -param preceding=$hdfs_hadoop_pre_dir/$file -param current=$hdfs_hadoop_cur_dir/$file -param output_added=$hdfs_hadoop_delta_dir/${file}_added -param output_removed=$hdfs_hadoop_delta_dir/${file}_removed -param delimiter=$delimiter diff.pig
        [ $? -ne 0 ] && die "diff of data between $previous_date and $current_date using pig failed. exiting `basename $0` script"
        $hadoop_bin_dir/hadoop dfs -cat $hdfs_hadoop_delta_dir/${file}_added/* | gzip > $file_output_dir/${formatted_output_filename}_added.gz
        $hadoop_bin_dir/hadoop dfs -cat $hdfs_hadoop_delta_dir/${file}_removed/* | gzip > $file_output_dir/${formatted_output_filename}_removed.gz
        [ $? -ne 0 ] && die "there was a problem gzipping ${formatted_output_filename}. exiting `basename $0` script"
        [ $post_diff_script ] && ./$post_diff_script $source $previous_date $current_date

    done

我只希望它在文件不为空时创建 _removed.gz 和 _added.gz 文件。我尝试在下面执行此操作，但我的脚本有问题吗？

    for file in $diff_file_list
    do

        # replace any ? with $current_date and replace any % with $file
        formatted_output_filename=$(echo $output_filename | sed "s|?|$current_date|g" | sed "s|%|_$file|g")
        $pig_bin_dir/pig -param preceding=$hdfs_hadoop_pre_dir/$file -param current=$hdfs_hadoop_cur_dir/$file -param output_added=$hdfs_hadoop_delta_dir/${file}_added -param output_removed=$hdfs_hadoop_delta_dir/${file}_removed -param delimiter=$delimiter diff.pig
        [ $? -ne 0 ] && die "diff of data between $previous_date and $current_date using pig failed. exiting `basename $0` script"
        if [[ -s $hdfs_hadoop_delta_dir/${file}_added/* ]] ; then
        echo "$hdfs_hadoop_delta_dir/${file}_added/* has data."
        $hadoop_bin_dir/hadoop dfs -cat $hdfs_hadoop_delta_dir/${file}_added/* | gzip > $file_output_dir/${formatted_output_filename}_added.gz
        $hadoop_bin_dir/hadoop dfs -cat $hdfs_hadoop_delta_dir/${file}_removed/* | gzip > $file_output_dir/${formatted_output_filename}_removed.gz
        else
        echo "$hdfs_hadoop_delta_dir/${file}_added/*is empty."
        fi ;
        [ $? -ne 0 ] && die "there was a problem gzipping ${formatted_output_filename}. exiting `basename $0` script"
        [ $post_diff_script ] && ./$post_diff_script $source $previous_date $current_date

    done
4

2 回答 2

2 
if [[ -s file ]]
then
    do_file_creation
fi

或者

for f in dir/*
do
    if [[ -s $f ]]
    then
        do_file_creation
    fi
done

使用小写或混合大小写的变量名称。

使用if代替[[ ]] &&

使用缩进。

于 2012-05-29T15:14:20.247 回答
0

我试图找到一个不添加循环的解决方案。但是，它并不那么干净：

if [[ -s `ls -S $hdfs_hadoop_delta_dir/${file}_added/* 2>/dev/null | head -1` ]] ; then

ls -S按文件大小对文件进行排序，head -1取最大的一个，然后通过if [[ -s.

不幸的是，必须处理没有文件的情况。我用过2>/dev/null。有人有更好的主意吗？

于 2012-05-29T16:10:52.927 回答