I wrote this script to search a lot of text files (~100,000) for 4 different search criteria and export to 4 separate files, I thought it would be more efficient to perform all 4 searches on each file as it is loaded vs doing 4 full searches like the first iteration below does. I may be missing some other major inefficiencies as I am pretty new to powershell.
I have this script re written from the first version to the second, but can't figure out how to get the path and data to display together like the first version did. I am struggling to reference the object within the loop, and have pieced this second version together, which is working, but not giving me the path to the file which is necessary.
It seems like I am just missing one or two little things to get me going in the right direction. Thanks in advance for your help
1st version:
Get-ChildItem -Filter *.txt -Path "\\file\to\search" -Recurse | Select-String -Pattern "abc123" -Context 0,3 | Out-File -FilePath "\\c:\out.txt"
Get-ChildItem -Filter *.txt -Path "\\file\to\search2" -Recurse | Select-String -Pattern "abc124" -Context 0,3 | Out-File -FilePath "\\c:\out2.txt"
Get-ChildItem -Filter *.txt -Path "\\file\to\search3" -Recurse | Select-String -Pattern "abc125" -Context 0,3 | Out-File -FilePath "\\c:\out3.txt"
Get-ChildItem -Filter *.txt -Path "\\file\to\search4" -Recurse | Select-String -Pattern "abc126" -Context 0,3 | Out-File -FilePath "\\c:\out4.txt"
Output:
\\file\that\was\found\example.txt:84: abc123
\\file\that\was\found\example.txt:90: abc123
\\file\that\was\found\example.txt:91: abc123
2nd version:
##$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ Configuration $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
############################################ Global Parameters #############################################
$SearchPath="\\file\to\search"
$ProgressFile=""\\progress\file\ResultsCount.txt"
$records = 105325
##----------------------------------------- End Global Parameters -----------------------------------------
########################################### Search Parameters ##############################################
##Search Pattern 1
$Pattern1="abc123"
$SaveFile1="\\c:\out.txt"
##Search Pattern 2
$Pattern2="abc124"
$SaveFile2="\\c:\out2.txt"
##Search Pattern 3
$Pattern3= "abc125"
$SaveFile3= "\\c:\out3.txt"
##Search Pattern 4
$Pattern4= "abc126"
$SaveFile4="\\c:\out4.txt"
##Search Pattern 5
$Pattern5= ""
$SaveFile5=""
##----------------------------------------- End Search Parameters ------------------------------------------
##$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ End of Config $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
############################### SCRIPT #####################################################################
## NOTES
## ------
##$files=Get-ChildItem -Filter *.txt -Path $SearchPath -Recurse ## Set all files to variable #### Long running, needs to be a better way #######
##$records=$files.count ## Set record #
Get-ChildItem -Filter *.txt -Path $SearchPath -Recurse | Foreach-Object { ## loop through search folder
$i=$i+1 ## increment record
##
Get-Content $_.FullName | Select-String -Pattern $Pattern1 -Context 0,3 | Out-File -FilePath $SaveFile1 ## pattern1 search
Get-Content $_.FullName | Select-String -Pattern $Pattern2 | Out-File -FilePath $SaveFile2 ## pattern2 search
Get-Content $_.FullName | Select-String -Pattern $Pattern3 -Context 0,1 | Out-File -FilePath $SaveFile3 ## pattern3 search
Get-Content $_.FullName | Select-String -Pattern $Pattern4 -Context 0,1 | Out-File -FilePath $SaveFile4 ## pattern4 search
##Get-Content $_.FullName | Select-String -Pattern $Pattern5 -Context 0,1 | Out-File -FilePath $SaveFile5 ## pattern5 search (Comment out unneeded search lines like this one)
$progress ="Record $($i) of $($records)" ## set progress
Write-Host "Record $($i) of $($records)" ## Writes progress to window
$progress | Out-File -FilePath $ProgressFile ## progress file
} ##
############################################################################################################
Output:
abc123
abc123
abc123
Edit: Also I am trying to figure out a good way to not have to hard code in the number of records for a decent progress readout, I commented out the way I thought would work (1st & 2nd line of the script), but there needs to be a more efficient way than rerunning the same search twice, one for a count and one for the for loop.
I would be very interested in any runtime efficiency information you could provide.
[edit - thanks to mklement0 for pointing out the errors about speed and the -SimpleMatch switch. [grin]]
the Select-String cmdlet will accept a -Path parameter ... and it is FAR [i was thinking of Get-Content, not Get-ChidItem] faster than using Get-ChildItem to feed the files to S-S. [grin]
also, the -Pattern parameter accepts a regex OR pattern like Thing|OtherThing|YetAnotherThing - and it accepts simple string patterns if you use the -SimpleMatch switch parameter.
what the code does ...
defines the source dir
defines the file spec
joins those two into a wildcard file path
builds an array of string patterns to use
calls Select-String with a path and an array of strings to search for
uses Group-Object and a calculated property to group the matches by the last part of .Line property from the S-S call
saves that to a $Var
shows that on screen
at that point, you can use the .Name property of each GroupInfo to select the items to send out to each file AND to build your file names.
the code ...
$SourceDir = 'D:\Temp\zzz - Copy'
$FileSpec = '*.log'
$SD_FileSpec = Join-Path -Path $SourceDir -ChildPath $FileSpec
$TargetPatternList = #(
'Accordion Cajun Zydeco'
'better-not-be-there'
'Piano Rockabilly Rowdy'
)
$GO_Results = Select-String -Path $SD_FileSpec -SimpleMatch $TargetPatternList |
Group-Object -Property {$_.Line.Split(':')[-1]}
$GO_Results
output ...
Count Name Group
----- ---- -----
6 Accordion Cajun Zydeco {D:\Temp\zzz - Copy\Grouping-List_08-02.log:11:Accordion Cajun Zydeco, D:\Temp\zzz - Copy\Grouping-List_08-09.log:11:Accordion Cajun Zy...
6 Bawdy Dupe Piano Rocka... {D:\Temp\zzz - Copy\Grouping-List_08-02.log:108:Bawdy Dupe Piano Rockabilly Rowdy, D:\Temp\zzz - Copy\Grouping-List_08-09.log:108:Bawdy...
6 Bawdy Piano Rockabilly... {D:\Temp\zzz - Copy\Grouping-List_08-02.log:138:Bawdy Piano Rockabilly Rowdy, D:\Temp\zzz - Copy\Grouping-List_08-09.log:138:Bawdy Pian...
6 Dupe Piano Rockabilly ... {D:\Temp\zzz - Copy\Grouping-List_08-02.log:948:Dupe Piano Rockabilly Rowdy, D:\Temp\zzz - Copy\Grouping-List_08-09.log:948:Dupe Piano ...
6 Instrumental Piano Roc... {D:\Temp\zzz - Copy\Grouping-List_08-02.log:1563:Instrumental Piano Rockabilly Rowdy, D:\Temp\zzz - Copy\Grouping-List_08-09.log:1563:I...
6 Piano Rockabilly Rowdy {D:\Temp\zzz - Copy\Grouping-List_08-02.log:1781:Piano Rockabilly Rowdy, D:\Temp\zzz - Copy\Grouping-List_08-09.log:1781:Piano Rockabil...
note that the .Group contains an array of lines from the matches sent out by the S-S call. you can send that to your output file.
Here is my take at solving this problem, very similar to Lee_Dailey's nice answer but with a foreach loop. I would recommend investing some time into researching the multi-threading options available on PowerShell in case you need to increase the performance of the script, you can look specifically at the ThreadJob module by Microsoft which is really easy to use or if you can't install modules due to some work policy, you can use Runspace.
It is worth adding that you can use the -List switch on Select-String, this way the performance of the script would be increased even more:
-List
Only the first instance of matching text is returned from each input file. This is the most efficient way to retrieve a list of files that have contents matching the regular expression.
$map = #{
abc123 = 'C:\out_abc123.txt'
abc124 = 'C:\out_abc124.txt'
abc125 = 'C:\out_abc125.txt'
}
$pattern = $map.Keys -join '|'
$match = foreach($file in Get-ChildItem *.txt)
{
Select-String -LiteralPath $file.FullName -Pattern $pattern
}
$match | Group-Object { $_.Matches.Value } | ForEach-Object {
$_.Group | Select-Object Path, LineNumber, Line | Out-File $map[$_.Name]
}
To compliment the answers #Santiago Squarzon and Lee_Dailey, I think you were actually on the good way yourself knowing that the Group-Object cmdlet is pretty expensive especially in memory usage as it chokes the PowerShell pipeline causing all the search results to be piled up in memory.
Besides, the Select-String cmdlet supports multiple (-SimpleMatch) patterns, where concatenating the search patters with an | (-join '|') will force you to use an (escaped) regular expression.
To continue on your approach:
(note that in the example, I am using my own settings to search through my script files)
$ProgressFile = '.\ResultsCount.txt'
$SearchRoot = '..\'
$Filter = '*.ps1'
$Searches = #{
'Null' = '.\Null.txt'
'Test' = '.\Test.txt'
'Object' = '.\Object.txt'
}
$Files = Get-ChildItem -Filter $Filter -Path $SearchRoot -Recurse
$Total = $Files.count
$Searches.Values |ForEach-Object { Set-Content -LiteralPath $_ -Value '' }
$i = 0
ForEach ($File in $Files) {
Get-Content -LiteralPath $File.FullName |
Select-String #($Searches.Keys) -AllMatches |ForEach-Object {
$Value = '{0}:{1}:{2}' -f $File.FullName, $_.LineNumber, $_
Add-Content -LiteralPath $Searches[$_.Pattern] -Value $Value
}
'Record {0} of {1}' -f ++$i, $Total |Tee-Object -Append .\ProgressFile.txt
}
Explanations
$Searches = #{ ...
Maps the search patters with the files, you might also use a PSObject list to specify each search (where you could add columns with e.g. context start/end values, etc.)
$Searches.Values |ForEach-Object { Set-Content -LiteralPath $_ -Value '' }
Empties the result files (knowing that they are not part of the main stream you can't use Add-Content)
$i = 0
Unfortunately there is no automatic index that initializes with a foreach loop (yet, see: #13772 Automatic variable for the pipeline index)
Get-Content -LiteralPath $File.FullName
Load the content once into memory
Note1: this is a string array.
Note2: the $Content will be reused each iteration and therefore overwrites the previous one and unloads it from memory
Select-String #($Searches.Keys) -AllMatches |ForEach-Object {
Searches the string array using your (multiple) defined patterns. (you might consider to use the -SimpleMatch parameter if your search strings contain special characters.)
Note: Unfortunately you need to embedded the $Searches.Keys in a array subexpression operator #( ), for details see .Net issue: #56835 Make OrderedDictionaryKeyValueCollection implement IList
$Value = '{0}:{1}:{2}' -f $File.FullName, $_.LineNumber, $_
Build an result output string.
Note: the result of the Select-String does have a (hidden) LineNumber and (matched) Pattern property.
Add-Content -LiteralPath $Searches[$_.Pattern] -Value $Value
Add the result string to the specific mapped output file.
'Record {0} of {1}' -f $i++, $Total |Tee-Object -Append .\ProgressFile.txt
Tee-Object will write the progress to the standard output (display) and also to the specific file.
I am using ConvertTo-Csv to get comma separated output
get-process | convertto-csv -NoTypeInformation -Delimiter ","
It outputs like:
"__NounName","Name","Handles","VM","WS",".....
However I would like to get output without quotes, like
__NounName,Name,Handles,VM,WS....
Here is a way to remove the quotes
get-process | convertto-csv -NoTypeInformation -Delimiter "," | % {$_ -replace '"',''}
But it has a serious drawback if one of the item contains a " it will be removed !
Hmm, I have Powershell 7 preview 1 on my mac, and Export-Csv has a -UseQuotes option that you can set to AsNeeded. :)
I was working on a table today and thought about this very question as I was previewing the CSV file in notepad and decided to see what others had come up with. It seems many have over-complicated the solution.
Here's a real simple way to remove the quote marks from a CSV file generated by the Export-Csv cmdlet in PowerShell.
Create a TEST.csv file with the following data.
"ID","Name","State"
"5","Stephanie","Arizona"
"4","Melanie","Oregon"
"2","Katie","Texas"
"8","Steve","Idaho"
"9","Dolly","Tennessee"
Save As: TEST.csv
Store file contents in a $Test variable
$Test = Get-Content .\TEST.csv
Load $Test variable to see results of the get-content cmdlet
$Test
Load $Test variable again and replace all ( "," ) with a comma, then trim start and end by removing each quote mark
$Test.Replace('","',",").TrimStart('"').TrimEnd('"')
Save/Replace TEST.csv file
$Test.Replace('","',",").TrimStart('"').TrimEnd('"') | Out-File .\TEST.csv -Force -Confirm:$false
Test new file Output with Import-Csv and Get-Content:
Import-Csv .\TEST.csv
Get-Content .\TEST.csv
To Sum it all up, the work can be done with 2 lines of code
$Test = Get-Content .\TEST.csv
$Test.Replace('","',",").TrimStart('"').TrimEnd('"') | Out-File .\TEST.csv -Force -Confirm:$false
I ran into this issue, found this question, but was not satisfied with the answers because they all seem to suffer if the data you are using contains a delimiter, which should remain quoted. Getting rid of the unneeded double quotes is a good thing.
The solution below appears to solve this issue for a general case, and for all variants that would cause issues.
I found this answer elsewhere, Removing quotes from CSV created by PowerShell, and have used it to code up an example answer for the SO community.
Attribution: Credit for the regex, goes 100% to Russ Loski.
Code in a Function, Remove-DoubleQuotesFromCsv
function Remove-DoubleQuotesFromCsv
{
param (
[Parameter(Mandatory=$true)]
[string]
$InputFile,
[string]
$OutputFile
)
if (-not $OutputFile)
{
$OutputFile = $InputFile
}
$inputCsv = Import-Csv $InputFile
$quotedData = $inputCsv | ConvertTo-Csv -NoTypeInformation
$outputCsv = $quotedData | % {$_ -replace `
'\G(?<start>^|,)(("(?<output>[^,"]*?)"(?=,|$))|(?<output>".*?(?<!")("")*?"(?=,|$)))' `
,'${start}${output}'}
$outputCsv | Out-File $OutputFile -Encoding utf8 -Force
}
Test Code
$csvData = #"
id,string,notes,number
1,hello world.,classic,123
2,"a comma, is in here","test data 1",345
3,",a comma, is in here","test data 2",346
4,"a comma, is in here,","test data 3",347
5,"a comma, is in here,","test data 4`r`nwith a newline",347
6,hello world2.,classic,123
"#
$data = $csvData | ConvertFrom-Csv
"`r`n---- data ---"
$data
$quotedData = $data | ConvertTo-Csv -NoTypeInformation
"`r`n---- quotedData ---"
$quotedData
# this regular expression comes from:
# http://www.sqlmovers.com/removing-quotes-from-csv-created-by-powershell/
$fixedData = $quotedData | % {$_ -replace `
'\G(?<start>^|,)(("(?<output>[^,"\n]*?)"(?=,|$))|(?<output>".*?(?<!")("")*?"(?=,|$)))' `
,'${start}${output}'}
"`r`n---- fixedData ---"
$fixedData
$fixedData | Out-File e:\test.csv -Encoding ascii -Force
"`r`n---- e:\test.csv ---"
Get-Content e:\test.csv
Test Output
---- data ---
id string notes number
-- ------ ----- ------
1 hello world. classic 123
2 a comma, is in here test data 1 345
3 ,a comma, is in here test data 2 346
4 a comma, is in here, test data 3 347
5 a comma, is in here, test data 4... 347
6 hello world2. classic 123
---- quotedData ---
"id","string","notes","number"
"1","hello world.","classic","123"
"2","a comma, is in here","test data 1","345"
"3",",a comma, is in here","test data 2","346"
"4","a comma, is in here,","test data 3","347"
"5","a comma, is in here,","test data 4
with a newline","347"
"6","hello world2.","classic","123"
---- fixedData ---
id,string,notes,number
1,hello world.,classic,123
2,"a comma, is in here",test data 1,345
3,",a comma, is in here",test data 2,346
4,"a comma, is in here,",test data 3,347
5,"a comma, is in here,","test data 4
with a newline","347"
6,hello world2.,classic,123
---- e:\test.csv ---
id,string,notes,number
1,hello world.,classic,123
2,"a comma, is in here",test data 1,345
3,",a comma, is in here",test data 2,346
4,"a comma, is in here,",test data 3,347
5,"a comma, is in here,","test data 4
with a newline","347"
6,hello world2.,classic,123
This is pretty similar to the accepted answer but it helps to prevent unwanted removal of "real" quotes.
$delimiter = ','
Get-Process | ConvertTo-Csv -Delimiter $delimiter -NoTypeInformation | foreach { $_ -replace '^"','' -replace "`"$delimiter`"",$delimiter -replace '"$','' }
This will do the following:
Remove quotes that begin a line
Remove quotes that end a line
Replace quotes that wrap a delimiter with the delimiter alone.
Therefore, the only way this would go wrong is if one of the values actually contained not only quotes, but specifically a quote-delimiter-quote sequence, which hopefully should be pretty uncommon.
Once the file is generated, you can run
set-content FILENAME.csv ((get-content FILENAME.csv) -replace '"')
Depending on how pathological (or "full-featured") your CSV data is, one of the posted solutions will already work.
The solution posted by Kory Gill is almost perfect - the only issue remaining is that quotes are removed also for cells containing the line separator \r\n, which is causing issues in many tools.
The solution is adding a newline to the character class expression:
$fixedData = $quotedData | % {$_ -replace `
'\G(?<start>^|,)(("(?<output>[^,"\n]*?)"(?=,|$))|(?<output>".*?(?<!")("")*?"(?=,|$)))' `
,'${start}${output}'}
I wrote this for my needs:
function ConvertTo-Delimited {
[CmdletBinding()]
param(
[Parameter(ValueFromPipeline=$true,Mandatory=$true)]
[psobject[]]$InputObject,
[string]$Delimiter='|',
[switch]$ExcludeHeader
)
Begin {
if ( $ExcludeHeader -eq $false ) {
#(
$InputObject[0].PsObject.Properties | `
Select-Object -ExpandProperty Name
) -Join $Delimiter
}
}
Process {
foreach ($item in $InputObject) {
#(
$item.PsObject.Properties | `
Select-Object Value | `
ForEach-Object {
if ( $null -ne $_.Value ) {$_.Value.ToString()}
else {''}
}
) -Join $Delimiter
}
}
End {}
}
Usage:
$Data = #(
[PSCustomObject]#{
A = $null
B = Get-Date
C = $null
}
[PSCustomObject]#{
A = 1
B = Get-Date
C = 'Lorem'
}
[PSCustomObject]#{
A = 2
B = Get-Date
C = 'Ipsum'
}
[PSCustomObject]#{
A = 3
B = $null
C = 'Lorem Ipsum'
}
)
# with headers
PS> ConvertTo-Delimited $Data
A|B|C
1|7/17/19 9:07:23 PM|Lorem
2|7/17/19 9:07:23 PM|Ipsum
||
# without headers
PS> ConvertTo-Delimited $Data -ExcludeHeader
1|7/17/19 9:08:19 PM|Lorem
2|7/17/19 9:08:19 PM|Ipsum
||
Here's another approach:
Get-Process | ConvertTo-Csv -NoTypeInformation -Delimiter "," |
foreach { $_ -replace '^"|"$|"(?=,)|(?<=,)"','' }
This replaces matches with the empty string, in each line. Breaking down the regex above:
| is like an OR, used to unite the following 4 sub-regexes
^" matches quotes in the beginning of the line
"$ matches quotes in the end of the line
"(?=,) matches quotes that are immediately followed by a comma
(?<=,)" matches quotes that are immediately preceded by a comma
I found that Kory's answer didn't work for the case where the original string included more than one blank field in a row. I.e. "ABC",,"0" was fine but "ABC",,,"0" wasn't handled properly. It stopped replacing quotes after the ",,,". I fixed it by adding "|(?<output>)" near the end of the first parameter, like this:
% {$_ -replace `
'\G(?<start>^|,)(("(?<output>[^,"]*?)"(?=,|$))|(?<output>".*?(?<!")("")*?"(?=,|$))|(?<output>))', `
'${start}${output}'}
I haven't spent much time looking for removing the quotes. But, here is a workaround.
get-process | Export-Csv -NoTypeInformation -Verbose -Path $env:temp\test.csv
$csv = Import-Csv -Path $env:temp\test.csv
This is a quick workaround and there may be a better way to do this.
A slightly modified variant of JPBlanc's answer:
I had an existing csv file which looked like this:
001,002,003
004,005,006
I wanted to export only the first and third column to a new csv file. And for sure I didn't want any quotes ;-)
It can be done like this:
Import-Csv -Path .\source.csv -Delimiter ',' -Header A,B,C | select A,C | ConvertTo-Csv -NoTypeInformation -Delimiter ',' | % {$_ -replace '"',''} | Out-File -Encoding utf8 .\target.csv
Couldn't find an answer to a similar question so I'm posting what I've found here...
For exporting as Pipe Delimited with No Quotes for string qualifiers, use the following:
$objtable | convertto-csv -Delimiter "|" -notypeinformation | select -Skip $headers | % { $_ -replace '"\|"', "|"} | % { $_ -replace '""', '"'} | % { $_ -replace "^`"",''} | % { $_ -replace "`"$",''} | out-file "$OutputPath$filename" -fo -en ascii
This was the only thing I could come up with that could handle quotes and commas within the text; especially things like a quote and comma next to each other at the beginning or ending of a text field.
This function takes a powershell csv object from the pipeline and outputs like convertto-csv but without adding quotes (unless needed).
function convertto-unquotedcsv {
param([Parameter(ValueFromPipeline=$true)]$csv, $delimiter=',', [switch]$noheader=$false)
begin {
$NeedQuotesRex = "($([regex]::escape($delimiter))|[\n\r\t])"
if ($noheader) { $names = #($true) } else { $names = #($false) }
}
process {
$psop = $_.psobject.properties
if (-not $names) {
$names = $psop.name | % {if ($_ -match $NeedQuotesRex) {'"' + $_ + '"'} else {$_}}
$names -join $delimiter # unquoted csv header
}
$values = $psop.value | % {if ($_ -match $NeedQuotesRex) {'"' + $_ + '"'} else {$_}}
$values -join $delimiter # unquoted csv line
}
end {
}
}
$names gets an array of noteproperty names and $values gets an array of notepropery values. It took that special step to output the header. The process block gets the csv object one piece at a time.
Here is a test run
$delimiter = ','; $csvData = #"
id,string,notes,"points per 1,000",number
4,"a delimiter$delimiter is in here,","test data 3",1,348
5,"a comma, is in here,","test data 4`r`nwith a newline",0.5,347
6,hello world2.,classic,"3,000",123
"#
$csvdata | convertfrom-csv | sort number | convertto-unquotedcsv -delimiter $delimiter
id,string,notes,"points per 1,000",number
6,hello world2.,classic,"3,000",123
5,"a comma, is in here,","test data 4
with a newline",0.5,347
4,"a delimiter, is in here,",test data 3,1,348