awk 中用户定义函数的打印输出给出了意外的令牌错误

Question

我想灵活地将两个小 awk 的输出打印到 bash 管道，它们使用变量（它们最初工作）。我最初认为我可以将整个命令存储为变量本身，但对于一个它不起作用并且显然 (store awk command in a variable of bash script) 这不是一个好主意。所以我写了两个函数，但我在“完成”附近得到了一个“意外令牌”，但它的格式如上面的链接所示。

我的错误在哪里？

for coverage_file in */*.cov
do
    #gene_count=$(awk '{print $5}' $coverage_file |sort | uniq -c | wc -l) #this is apparently not a good idea
    #contig_count=$(awk '{print $1}' $coverage_file |sort | uniq -c | wc -l) #this is apparently not a good idea
    cmd_gene() { awk '{print $5}' $coverage_file |sort | uniq -c | wc -l }
    cmd_contig() { awk '{print $1}' $coverage_file |sort | uniq -c | wc -l }
    cmd_gene $coverage_file
    cmd_contig $coverage_file
    #print "we found", $gene_count, "genes on ",$contig_count" contigs
done

cov 文件如下所示：

k141_85332.3 4119 19 A5 phnM_031
k141_85332.3 4119 19 A5 phnM_031
k141_85332.3 4119 28 A1 phnM_031
k141_85332.3 4119 28 A1 phnM_031
k141_85332.3 4119 8 A2 phnM_031
k141_85332.3 4119 8 A2 phnM_031
k141_88684 267 5 B10 phnM_032
k141_88684 268 5 B10 phnM_032
k141_88684 269 5 B10 phnM_032
k141_88684 270 5 B10 phnM_032
k141_88684 271 5 B10 phnM_032
k141_88684 272 5 B10 phnM_032

编辑：这包括接受的答案 + 一种可能的方式来清楚地打印它：

#!/bin/bash

#define variables
gene="phnM"
threshold="5"

#define functions
cmd_gene() { awk '{print $5}' $1 |sort | uniq -c | wc -l ; } #semicolon is important here!
cmd_contig() { awk '{print $1}' $1 |sort | uniq -c | wc -l ; } #semicolon is important here!

#loop over files and print results (would be prettier with printf)
for coverage_file in */*.cov
do
    echo $gene" was found" $(cmd_gene "$coverage_file") "times on" $(cmd_contig "$coverage_file")" contigs with minimum coverage of" $threshold in $coverage_file
done

输出：

phnM was found 67 times on 65 contigs with minimum coverage of 5 in phnm/test.cov
phnM was found 3 times on 2 contigs with minimum coverage of 5 in test/test.cov

Answer 1

出现意外的标记错误是因为在定义函数时，} 必须在它自己的行上或前面有 ;。

另外，由于您在函数定义中使用了$coverage_file，因此您不必传递它。

for coverage_file in */*.cov
do
    cmd_gene() { awk '{print $5}' $coverage_file |sort | uniq -c | wc -l; }
    cmd_contig() { awk '{print $1}' $coverage_file |sort | uniq -c | wc -l; }
    cmd_gene 
    cmd_contig 
    #print "we found", $gene_count, "genes on ",$contig_count" contigs
done

如果您想在 for 循环之外定义函数，您可以使用 $1（不要与 awk 的 $1 混淆）并像以前一样传递 $coverage_file。

编辑：以上示例

$ cat a.sh
cmd_gene() { awk '{print $5}' $1 |sort | uniq -c | wc -l; }
cmd_contig() { awk '{print $1}' $1 |sort | uniq -c | wc -l; }

for coverage_file in */*.cov
do
    cmd_gene $coverage_file
    cmd_contig $coverage_file
done

$ ls */*.cov
bf/a.cov

$ cat */*.cov
k141_85332.3 4119 19 A5 phnM_031
k141_85332.3 4119 19 A5 phnM_031
k141_85332.3 4119 28 A1 phnM_031
k141_85332.3 4119 28 A1 phnM_031
k141_85332.3 4119 8 A2 phnM_031
k141_85332.3 4119 8 A2 phnM_031
k141_88684 267 5 B10 phnM_032
k141_88684 268 5 B10 phnM_032
k141_88684 269 5 B10 phnM_032
k141_88684 270 5 B10 phnM_032
k141_88684 271 5 B10 phnM_032
k141_88684 272 5 B10 phnM_032

$ sh a.sh
       2
       2

Answer 2

@jas 回答了你的问题，所以坚持下去，以下只是一种更好的方法来做你想做的事情，它太大/格式不适合发表评论：

awk '
BEGIN {
    gene = "phnM"
    threshold = "5"
}
{
    genes[$5]
    contigs[$1]
}
ENDFILE {
    printf "%s was found %d times on %d contigs with minimum coverage of %d in %s\n",
        gene, length(genes), length(contigs), threshold, FILENAME
    delete genes
    delete contigs
}
' */*.cov

上面将 GNU awk 用于 ENDFILE，但如果需要，它可以用于其他 awk，这是一个微不足道的更改：

awk '
BEGIN {
    gene = "phnM"
    threshold = "5"
}
FNR==1 { prt() }
{
    genes[$5]
    contigs[$1]
}
END { prt() }
function prt() {
    if (fname != "") {
        printf "%s was found %d times on %d contigs with minimum coverage of %d in %s\n",
            gene, length(genes), length(contigs), threshold, fname
        delete genes
        delete contigs
    }
    fname = FILENAME
}
' */*.cov

请参阅https://unix.stackexchange.com/questions/169716/why-is-using-a-shell-loop-to-process-text-considered-bad-practice 了解一些在处理文本时避免 shell 循环的原因。