替换括号中的单词以放入引号中

Question

我想替换括号中的单词以放入引号中。 我的previous question 在括号中有一个值。这个问题和我之前的问题一样，但是想到了一个场景，我将在括号中包含多个值。

以下是我在其中一个可变数据中的数据-

SELECT          * 
FROM            ( 
                       SELECT table1 file2.txt file.txt queues qdefinitions parameters trap-deposit-dstran.dat.2016-08-07 
                       FROM   cs_case 
                       WHERE  ant_cd='FI_BASE_TENANT') t1 
LEFT OUTER JOIN table2 t2 
ON              t2.case_id=t1.case_id 
LEFT OUTER JOIN table3 t3 
ON              t3.service_xid=t1.service_xid 
LEFT OUTER JOIN table4 t4 
ON              t4.service_id=t1.service_id 
WHERE           ( 
                                t1.casestatus_cd = (new,retired,pending,OPEN,closed) 
                OR              t1.case_status_num = (1,2,3,4) ) 
GROUP BY        t1.case_reference, 
                t2.last_scrfp, 
                t1.service_id 
ORDER BY        t2.last_scrfp DESC

这就是我想要的。

SELECT          * 
FROM            ( 
                       SELECT table1 file2.txt file.txt queues qdefinitions parameters trap-deposit-dstran.dat.2016-08-07 
                       FROM   cs_case 
                       WHERE  ant_cd='FI_BASE_TENANT') t1 
LEFT OUTER JOIN table2 t2 
ON              t2.case_id=t1.case_id 
LEFT OUTER JOIN table3 t3 
ON              t3.service_xid=t1.service_xid 
LEFT OUTER JOIN table4 t4 
ON              t4.service_id=t1.service_id 
WHERE           ( 
                                t1.casestatus_cd = ('NEW','RETIRED','PENDING','OPEN','CLOSED') 
                or              t1.case_status_num = (1,2,3,4) ) 
GROUP BY        t1.case_reference, 
                t2.last_scrfp, 
                t1.service_id 
ORDER BY        t2.last_scrfp DESC

之前我使用过如下的 sed 命令

sed -E 's/\(([^(,$1)'\'']+)\)/('\''\1'\'')/g' Filename.txt

Answer 1

这不再是对单个行的简单替换，因此它不再是 sed 的工作，而是 awk 的工作：

$ cat tst.awk
BEGIN { RS="^$"; ORS="" }
{
    while ( match($0,/(=\s*\()([^()]+)(\))/,a) ) {
        if ( a[2] ~ /[^0-9,]/) {
            gsub(/,/,"\047,\047",a[2])
            a[2] = "\047" toupper(a[2]) "\047"
        }
        print substr($0,1,RSTART-1) a[1] a[2] a[3]
        $0 = substr($0,RSTART+RLENGTH)
    }
    print
}

$ awk -f tst.awk file
SELECT          *
FROM            (
                       SELECT table1 file2.txt file.txt queues qdefinitions parameters
trap-deposit-dstran.dat.2016-08-07
                       FROM   cs_case
                       WHERE  ant_cd='FI_BASE_TENANT') t1
LEFT OUTER JOIN table2 t2
ON              t2.case_id=t1.case_id
LEFT OUTER JOIN table3 t3
ON              t3.service_xid=t1.service_xid
LEFT OUTER JOIN table4 t4
ON              t4.service_id=t1.service_id
WHERE           (
                                t1.casestatus_cd = ('NEW','RETIRED','PENDING','OPEN','CLOSED')
                OR              t1.case_status_num = (1,2,3,4) )
GROUP BY        t1.case_reference,
                t2.last_scrfp,
                t1.service_id
ORDER BY        t2.last_scrfp DESC

上面使用 GNU awk 作为第三个参数来 match() 和多字符 RS。

Answer 2

为什么会出现如此疯狂的复杂情况？

怎么样：

$line =~ s/(([^(]+))/"\1"/g;

注意 () 组合了一个子表达式； ( 匹配文字 '(' （无需省略 ')'）； [^(]+ 匹配断言括号内的一组非括号； "\1" 引用匹配的内容，即。 , 括号内发生了什么，括号本身除外。

Answer 3

使用sln 提供的solution 使用Regex Quote-Like operators，以提供可读性/可维护性。

加上mob 提供的另一个solution 用于优雅的引用机制。

加上一些编码，为您的特定场景将所有内容组合在一起：

use strict;
use warnings 'all';

# create a regex quote-like string for our match
my $rx = qr{\(([^)]+)\)};

# step through STDIN
while(<>) {

    # Replace matching (...) sections by passing the matched part
    # to QuoteText() and subsituting the result it returns.
    #
    # Note the 'e' and 'g' flags to s///, e enables "extended"
    # operations which allows us to call QuoteText() and g 
    # replaces _all_ occurances on a line
    s/$rx/QuoteText($1)/eg;

    # print out the line (post any possible substitutions)
    print;
}

sub QuoteText {
    my $text = shift;

    # the variable we will use to return our result
    my $result;

    # Does the text contain commas and also contains alphabetical characters?   
    if($text =~ /,/ && $text =~ /[a-z]/i) {

        # split up the text around commas, then rejoin them with the 
        # string ',', and then prepend and append a single quote 
        # at the begining and end of line. So one,two becomes 'one','two'
        $result = q/'/ . join(q/','/, split (',', $text)) . q/'/;

    # The word _only_ contains numbers, spaces and commas, leave as is
    } elsif($text =~ /^[\s\d,]+$/i) {
        $result = $text;

    # We possibly have just a single word? default to quoting the word
    } else {
        $result = qq/'$text'/;
    }

    # return our result wrapped in ()'s
    return "($result)";
}

注意：我没有将引用的单词大写。如果你想实现这一点，你需要map 和uc split 的部分。同样在要引用的单词是数字和单词的混合的情况下，它最终会引用所有的数字和单词。如果这是一个问题，那么最好通过拆分单词、循环遍历每个单词、评估其内容以查看它是否需要引用，然后将结果连接到返回的字符串来更好地解决此问题。