重写一个for循环与“并行”一起使用

Question

我想重写一个for循环（这很好用）并将其作为并行运行，但我有各种各样的问题。这里是我的循环

function no_sam { 
    filename=$(basename "$file") 
    extension="${file##*.}" 

    if [ $extension = "sam" ]; 
     then 
     filename="${filename%.*}" 
     feat_out=$filename.out 
     htseq-count -f $types -r "$pos" -m "$mode" -i "$attribute" -s "$strand" -t "$feature" -a "$qual" "$file" "$input_gff" > "$feat_out" 
     grep -v "_" "$feat_out" > temp && mv temp "$feat_out" 
     mv "$feat_out" "$counts_folder" 
    elif [ $extension = "bam" ]; 
     then 
     filename="${filename%.*}" 
     feat_out=$filename.out 
     htseq-count -f $types -r "$pos" -m "$mode" -i "$attribute" -s "$strand" -t "$feature" -a "$qual" "$file" "$input_gff" > "$feat_out" 
     grep -v "_" "$feat_out" > temp && mv temp "$feat_out" 
     mv "$feat_out" "$counts_folder" 
    fi  
} 

for file in "${multi[@]}"; do 
    no_sam 
done 

当我更换了与GNU并行循环，我得到错误

"${multi[@]}" no_sam | parallel 

testfile.sam: command not found 

Answer 1

试试这个（基于https://www.gnu.org/software/parallel/man.html#EXAMPLE:-Using-shell-variables和https://www.gnu.org/software/parallel/man.html#EXAMPLE:-Calling-Bash-functions）：

function no_sam { 
    file="$1" 
    filename=$(basename "$file") 
    extension="${file##*.}" 

    if [ $extension = "sam" ]; 
     then 
     filename="${filename%.*}" 
     feat_out=$filename.out 
     htseq-count -f $types -r "$pos" -m "$mode" -i "$attribute" -s "$strand" -t "$feature" -a "$qual" "$file" "$input_gff" > "$feat_out" 
     grep -v "_" "$feat_out" > temp && mv temp "$feat_out" 
     mv "$feat_out" "$counts_folder" 
    elif [ $extension = "bam" ]; 
     then 
     filename="${filename%.*}" 
     feat_out=$filename.out 
     htseq-count -f $types -r "$pos" -m "$mode" -i "$attribute" -s "$strand" -t "$feature" -a "$qual" "$file" "$input_gff" > "$feat_out" 
     grep -v "_" "$feat_out" > temp && mv temp "$feat_out" 
     mv "$feat_out" "$counts_folder" 
    fi  
} 
export -f no_sam 

parallel no_sam ::: "${multi[@]}" 

从您尝试使用GNU Parallel的方式来看，我认为您将从步行穿过man parallel_tutorial花费一个小时获益。 你的命令行会爱你。