PowerShell: Splitting by comma unless in quotes

I believe you're looking for a solution that involves manually parsing your string with regex but just in case going to post this as a PowerShell alternative that is more object oriented and uses ConvertFrom-Csv to parse your string. By no means this is an efficient solution but seems to do what you're looking for.

$string = 'Data1,Data2,"Data 3, more", Data 4'
$parsed = $string, $string | ConvertFrom-Csv
$parsed.PSObject.Properties.Value | & {
    begin   { $i = 1 }
    process {
        $delim = ''
        if($_.Contains(',')) { $delim = '"' }
        ($i++).ToString().PadRight(5) + [string]::Format("{0}$_{0}", $delim)
    }
}

Outputs:

1    Data1
2    Data2
3    "Data 3, more"
4    Data 4

If there may be duplicate data, meaning, ConvertFrom-Csv may attempt to create objects with same Property Names use this code instead:

$string = 'Data1,Data2,"Data 3, more", Data 4, Data 4'
$parsed = $string | ConvertFrom-Csv -Header (1..$string.Split(',').Count)
$parsed.PSObject.Properties.Value | & {
    begin   { $i = 1 }
    process {
        $delim = ''
        if([string]::IsNullOrWhiteSpace($_)) { return }
        if($_.Contains(',')) { $delim = '"' }
        ($i++).ToString().PadRight(5) + [string]::Format("{0}$_{0}", $delim)
    }
}

Using 'Data1,Data2,"Data 3, more", Data 4, Data 4' as the example, the output would become:

1    Data1
2    Data2
3    "Data 3, more"
4    Data 4
5    Data 4

To complement Santiago Squarzon's helpful answer with a regex-based solution:

# Split the string into fields.
$fields = (
  'Data1,Data2,"Data 3, more", Data 4' -split '([^",]+|"[^"]*")' -notmatch '^,?$'
).Trim()

# Output the resulting fields, preceded by their index.
$i = 0; $fields.ForEach({ [pscustomobject] @{ Ndx=($i++); Value=$_ } })

Output:

Ndx Value
--- -----
  0 Data1
  1 Data2
  2 "Data 3, more"
  3 Data 4

Explanation:

• The regex passed to the -split operator is somewhat repurposed in that it normally is designed to match what separates the fields, but here matches the fields, and by virtue of using a capture group ((...)), includes the "separator" matches in the output.

  • For an explanation of the regex and the ability to experiment with it, see this regex101.com page.

• Thus, the resulting tokens comprise both the "fields" - which in this case are the actual separators - and the "separators" - which in this case are the actual fields.
  -notmatch '^,?$' is then used to filter out the actual separators to leave just the array of actual fields (that is, array elements that are either the empty string or contain just , are eliminated).

As an aside: The -split + -notmatch approach above is really a [regex]::Matches() solution in disguise (see zett42's helpful answer). Use of [regex]::Matches() is a more direct expression of the solution's intent, but comes with its own complexity, not least due to not using PowerShell-native features. Bringing the functionality of [regex]::Matches() to PowerShell in the future in the form of a -matchall operator has been green-lighted, but is still awaiting implementation as of this writing: see GitHub issue #7867

mklement0 was faster with the RegEx-based solution. To complement, I'm adding a different variant. Instead of splitting the string, I search for each comma-separated field, which can be either quoted or unquoted.

$string = 'Data1,Data2,"Data 3, more", Data 4'

$i = 0 
[regex]::Matches( $string, '(?:"[^"]+"|[^,]+)(?=,|$)' ).ForEach{
    ($i++).ToString().PadRight(5) + $_.Value.Trim()
}

Output:

0    Data1
1    Data2
2    "Data 3, more"
3    Data 4

• The RegEx patterns consists of two alternations (pattern1|pattern2). First one is quoted string and if it doesn't match, unquoted string will be matched.
• Demo and full explanation at regex101. (Note that I had to escape " for regex101, although it isn't needed for PowerShell. This is because regex101 assumes use from C#. Please upvote this issue if you'd like a true PowerShell flavor for regex101.)
• This isn't a full-fledged CSV parser. If you need to handle edge cases, such as embedded quotation marks, refer to this answer, which presents a more robust solution.

As you asked for it in comment, here is a way to access the data fields directly:

# Create an array that consists of the matched group values
$data = [regex]::Matches( $string, '(?:"[^"]+"|[^,]+)(?=,|$)' ).Value.Trim()

$x = $data[0]
$y = $data[1]
$z = $data[2]
$w = $data[3]

Another, more concise way:

$x, $y, $z, $w = [regex]::Matches( $string, '(?:"[^"]+"|[^,]+)(?=,|$)' ).Value.Trim()

In these samples we don't have to use .ForEach{}, because of PowerShell's convenient member-access enumeration.