How to search an object for a value? - powershell

Question

Let's say you have a giant object - one which may or may not have nested arrays / objects,
# Assuming 'user1' exists in the current domain
$obj = Get-ADUser 'user1' -Properties *
and I want to search that object for the string SMTP case-insensitively...
What I tried
$obj | Select-String "SMTP"
But it does not work because the match is inside a nested Collection... to be concise, it sits inside the property $obj.proxyAddresses.
If I run $obj.proxyAddress.GetType() it returns:
IsPublic IsSerial Name BaseType
-------- -------- ---- --------
True False ADPropertyValueCollection System.Collections.CollectionBase
What's the best way to go about this? I know you could loop through the properties and look for it manually using wildcard matching or .Contains(), but I'd prefer a built in solution.
Thus, it would be a grep for objects and not only strings.

Answer 1

Here's one solution. It can be very slow depending on what depth you search to; but a depth of 1 or 2 works well for your scenario:
function Find-ValueMatchingCondition {
Param (
[Parameter(Mandatory = $true, ValueFromPipeline = $true)]
[PSObject]$InputObject
,
[Parameter(Mandatory = $true)]
[ScriptBlock]$Condition
,
[Parameter()]
[Int]$Depth = 10
,
[Parameter()]
[string]$Name = 'InputObject'
,
[Parameter()]
[System.Management.Automation.PSMemberTypes]$PropertyTypesToSearch = ([System.Management.Automation.PSMemberTypes]::Properties)
)
Process {
if ($InputObject -ne $null) {
if ($InputObject | Where-Object -FilterScript $Condition) {
New-Object -TypeName 'PSObject' -Property #{Name=$Name;Value=$InputObject}
}
#also test children (regardless of whether we've found a match
if (($Depth -gt 0) -and -not ($InputObject.GetType().IsPrimitive -or ($InputObject -is 'System.String'))) {
[string[]]$members = Get-Member -InputObject $InputObject -MemberType $PropertyTypesToSearch | Select-Object -ExpandProperty Name
ForEach ($member in $members) {
$InputObject."$member" | Where-Object {$_ -ne $null} | Find-ValueMatchingCondition -Condition $Condition -Depth ($Depth - 1) -Name $member | ForEach-Object {$_.Name = ('{0}.{1}' -f $Name, $_.Name);$_}
}
}
}
}
}
Get-AdUser $env:username -Properties * `
| Find-ValueMatchingCondition -Condition {$_ -like '*SMTP*'} -Depth 2
Example Results:
Value Name
----- ----
smtp:SomeOne#myCompany.com InputObject.msExchShadowProxyAddresses
SMTP:some.one#myCompany.co.uk InputObject.msExchShadowProxyAddresses
smtp:username#myCompany.com InputObject.msExchShadowProxyAddresses
smtp:some.one#myCompany.mail.onmicrosoft.com InputObject.msExchShadowProxyAddresses
smtp:SomeOne#myCompany.com InputObject.proxyAddresses
SMTP:some.one#myCompany.co.uk InputObject.proxyAddresses
smtp:username#myCompany.com InputObject.proxyAddresses
smtp:some.one#myCompany.mail.onmicrosoft.com InputObject.proxyAddresses
SMTP:some.one#myCompany.mail.onmicrosoft.com InputObject.targetAddress
Explanation
Find-ValueMatchingCondition is a function which takes a given object (InputObject) and tests each of its properties against a given condition, recursively.
The function is divided into two parts. The first part is the testing of the input object itself against the condition:
if ($InputObject | Where-Object -FilterScript $Condition) {
New-Object -TypeName 'PSObject' -Property #{Name=$Name;Value=$InputObject}
}
This says, where the value of $InputObject matches the given $Condition then return a new custom object with two properties; Name and Value. Name is the name of the input object (passed via the function's Name parameter), and Value is, as you'd expect, the object's value. If $InputObject is an array, each of the values in the array is assessed individually. The name of the root object passed in is defaulted as "InputObject"; but you can override this value to whatever you like when calling the function.
The second part of the function is where we handle recursion:
if (($Depth -gt 0) -and -not ($InputObject.GetType().IsPrimitive -or ($InputObject -is 'System.String'))) {
[string[]]$members = Get-Member -InputObject $InputObject -MemberType $PropertyTypesToSearch | Select-Object -ExpandProperty Name
ForEach ($member in $members) {
$InputObject."$member" | Where-Object {$_ -ne $null} | Find-ValueMatchingCondition -Condition $Condition -Depth ($Depth - 1) -Name $member | ForEach-Object {$_.Name = ('{0}.{1}' -f $Name, $_.Name);$_}
}
}
The If statement checks how deep we've gone into the original object (i.e. since each of an objects properties may have properties of their own, to a potentially infinite level (since properties may point back to the parent), it's best to limit how deep we can go. This is essentially the same purpose as the ConvertTo-Json's Depth parameter.
The If statement also checks the object's type. i.e. for most primitive types, that type holds the value, and we're not interested in their properties/methods (primitive types don't have any properties, but do have various methods, which may be scanned depending on $PropertyTypeToSearch). Likewise if we're looking for -Condition {$_ -eq 6} we wouldn't want all strings of length 6; so we don't want to drill down into the string's properties. This filter could likely be improved further to help ignore other types / we could alter the function to provide another optional script block parameter (e.g. $TypeCondition) to allow the caller to refine this to their needs at runtime.
After we've tested whether we want to drill down into this type's members, we then fetch a list of members. Here we can use the $PropertyTypesToSearch parameter to change what we search on. By default we're interested in members of type Property; but we may want to only scan those of type NoteProperty; especially if dealing with custom objects. See https://learn.microsoft.com/en-us/dotnet/api/system.management.automation.psmembertypes?view=powershellsdk-1.1.0 for more info on the various options this provides.
Once we've selected what members/properties of the input object we wish to inspect, we fetch each in turn, ensure they're not null, then recurse (i.e. call Find-ValueMatchingCondition). In this recursion, we decrement $Depth by one (i.e. since we've already gone down 1 level & we stop at level 0), and pass the name of this member to the function's Name parameter.
Finally, for any returned values (i.e. the custom objects created by part 1 of the function, as outlined above), we prepend the $Name of our current InputObject to the name of the returned value, then return this amended object. This ensures that each object returned has a Name representing the full path from the root InputObject down to the member matching the condition, and gives the value which matched.

Answer 2

Note: This answer contains background information and offers a quick-and-dirty approach that requires no custom functionality.
For a more more thorough, systematic approach based on reflection via a custom function, see JohnLBevan's helpful answer.
Select-String operates on strings, and when it coerces an input object of a different type to a string, it essentially calls .ToString() on it, which often yields generic representations such as the mere type name and typically not an enumeration of the properties.
Note that an object's .ToString() representation is not the same as PowerShell's default output to the console, which is much richer.
If all you're looking for is to find a substring in the for-display string representation of an object, you can pipe to Out-String -Stream before piping to Select-String:
$obj | Out-String -Stream | Select-String "SMTP"
Out-String creates a string representation that is the same as what renders to the console by default (it uses PowerShell's output-formatting system); adding -Stream emits that representation line by line, whereas by default a single, multi-line string is emitted.
Note: Recent versions of PowerShell come with convenience function oss, which wraps Out-String -Stream:
$obj | oss | Select-String "SMTP"
Of course, this method will only work if the for-display representation actually shows the data of interest - see caveats below.
That said, searching in the for-display representations is arguably what Select-String should do by default - see GitHub issue #10726
Caveats:
If the formatted representation happens to be tabular and your search string is a property name, the value of interest may be on the next line.
You can address this by forcing a list-style display - where each property occupies a line of its own (both name and value) - as follows:
$obj | Format-List | Out-String -Stream | Select-String "SMTP"
If you anticipate multi-line property values, you can use Select-String's -Context parameter to include lines surrounding a match, such as -Context 0,1 to also output the line after a match.
If you know that the values of interest are in a collection-valued property, you can use $FormatEnumerationLimit = -1 to force listing of all elements (by default, only the first 4 elements are displayed).
Caveat: As of PowerShell Core 6.1.0, $FormatEnumerationLimit is only effective if set in the global scope - see this GitHub issue.
However, once you hit the need to set preference variable $FormatEnumerationLimit, it's time to consider the more thorough solution based on a custom function in John's answer.
Values may get truncated in the representation, because Out-String assumes a fixed line width; you can use -Width to change that, but be careful with large numbers, because tabular representations then use the full width for every output line.