在带引号的字符串 csv 文件中有多个双引号

Question

我有一个 csv 文件，每个字段都有引号。

有一些字段，里面可以有多个双引号。我想用额外的双引号转义它们中的每一个。

","ABC "XYZ" PQRS","
","ABC "XYZ"","
","ABC "A" "B" TEST","
","ABC 2.5" "C" Test","

我从 link 那里得到了帮助，并且能够使用正则表达式 [regex]$r='(","[^"]+"[^"]+?",")' 涵盖内容中包含单双引号的场景。但是，遇到了内容中有多个双引号的情况。

[regex]$r='(","[^"]+"[^"]+"",")' # Not working
get-content C:\Projects\MyProject\testRegexFordoublequotes.csv | foreach {

  #save each line to a variable to make it easier to track

  $line=$_

  #look for a regex match

  $find=$r.matches($line)
  
  if ($find[0].Success) { 

      foreach ($match in $find) {

        #the original string we matched on

        $found=$match.value

        #replace the substring

        $replace= '","'+  $found.Trim('","').Replace('""','"').Replace('"','""')+ '","'

        #replace the full string and write to the pipeline

        $line -replace $found,$replace

      } #foreach
       

  } #if

  else {

        #no match so write the line to pipeline

        $line

    }

 } | Set-Content C:\Projects\MyProject\modified.csv -Force

你能帮我定义正则表达式吗，这将有助于字段内的多个双引号。

Answer 1

搜索 valid 分隔符（例如 "\s*,\s*"）并将行拆分为字段可能比简单地更正每个（无效的）单双更容易在每个字段中引用 2 个引号。
通过用双引号将字段括起来并将它们与 csv（逗号）分隔符

连接起来，将字段重建为记录

输入

$Csv = @'
"Field","ABC "XYZ" PQRS","Field"
"Field","ABC "XYZ"","Field"
"Field","ABC "A" "B" TEST","Field"
"Field","ABC 2.5" "C" Test","Field"
'@ -Split '[\r\n]+'

脚本

$Csv | # replace with: get-content .\testRegexFordoublequotes.csv |
Foreach-Object {
    $Line = $_ -Replace '^\s*"' -Replace '"\s*$' # Strip outer double quotes
    $Fields = $Line -Split '"\s*,\s*"'           # Split line into fields
    $Fields = $Fields -Replace '"', '""'         # Escape each " in each field
    '"' + ($Fields -Join '","') + '"'            # Rejoin the fields to line
} # append: | Set-Content .\modified.csv -Force

输出

"Field","ABC ""XYZ"" PQRS","Field"
"Field","ABC ""XYZ""","Field"
"Field","ABC ""A"" ""B"" TEST","Field"
"Field","ABC 2.5"" ""C"" Test","Field"

Answer 2

根据我们在 post 评论中的对话，这些文件是不一致的 CSV 文件，因此 CSV 解析器没有帮助。

请注意，如果单个单元格恰好有一个 some textext","more text，则您有一个未定义的案例。由于未转义引号，该单元格将被视为两个单元格。

现在是正则表达式。你可以找到一个前瞻和后视的正则表达式，但我认为盲目地将所有引号加倍更容易，然后清理不需要的引号，例如在行首和行尾，以及单元格之间。

我不熟悉 powershell，但这里有一个 JavaScript/pseudo 代码，您可以轻松将其转换为 powershell 语法。我正在使用包含您所有陈述的测试用例的单行；您将遍历文件中的行：

/* assume $line is:
"Start","ABC "XYZ" PQRS","ABC "XYZ"","ABC "A" "B" TEST","ABC 2.5" "C" Test","End"
*/

$fixed = $line.replace(/"/g, '""')
              .replace(/"",""/g, '","')
              .replace(/^""/, '"')
              .replace(/""$/, '"')

/* $fixed is:
"Start","ABC ""XYZ"" PQRS","ABC ""XYZ""","ABC ""A"" ""B"" TEST","ABC 2.5"" ""C"" Test","End"
*/

解释：

.replace(/"/g, '""') - 盲目地将所有引号加倍
.replace(/"",""/g, '","') - 将 "","" 恢复为 ","
.replace(/^""/, '"') - 将行首的 "" 恢复为 "
.replace(/""$/, '"') - 将行尾的 "" 恢复为 "

Answer 3

您可以执行以下操作以查看更改内容：

(Get-Content file.csv) -replace '(?<!^|",)"(?!,"|$)','""'

您可以简单地通过管道传输到 Set-Content 以保存新内容：

(Get-Content file.csv) -replace '(?<!^|",)"(?!,"|$)','""' |
    Set-Content file.csv

解释：

(?<!^|",) 是对任何不是行首 (^) 或 ", 的先前位置的负向后视。 (?!,"|$) 是对不是行尾 ($) 或 ," 的任何下一个位置的负前瞻。如果满足这些环视条件，" 将替换为 ""。

在带引号的字符串 csv 文件中有多个双引号

Having multiple double quotes inside quoted string csv file

regex

csv

powershell

str-replace

double-quotes