How to use Group-Object on this? - powershell

I am trying to get all the accounts from $f which do not match the accounts in $table4 into $accounts. But I need to also check if the occupancy number matches or not.
CSV $f:
Account_no |occupant_code
-----------|------------
12345 | 1
67890 | 2
45678 | 3
DataTable $table4
Account_no |occupant_code
-----------|------------
12345 | 1
67890 | 1
45678 | 3
Current code:
$accounts = Import-Csv $f |
select account_no, occupant_code |
where { $table4.account_no -notcontains $_.account_no }
What this needs to do is to check that occupant_code doesn't match, i.e.:
12345: account and occupant from $f and $table4 match; so it's ignored
67890: account matches $table4, but occupancy_code does not match, so it is added to $accounts.
Current result:
Desired result: 67890
I believe I need to use Group-Object, but I do not know how to use that correctly.
I tried:
Import-Csv $f |
select account_no, occupant_code |
Group-Object account_no |
Where-Object { $_.Group.occupant_code -notcontains $table4.occupant_code }

An alternative to Bill's suggestion would be to fill a hashtable with your reference data ($table4) and look up the occupant_code value for each account from $f, assuming that your account numbers are unique:
$ref = #{}
$table4 | ForEach-Object {
$ref[$_.Account_no] = $_.occupant_code
}
$accounts = Import-Csv $f |
Where-Object { $_.occupant_code -ne $ref[$_.Account_no] } |
Select-Object -Expand Account_no

Compare-Object?
csv1.csv:
Account_no,occupant_code
12345,1
67890,2
45678,3
csv2.csv:
Account_no,occupant_code
12345,1
67890,1
45678,3
PowerShell command:
Compare-Object (Import-Csv .\csv1.csv) (Import-Csv .\csv2.csv) -Property occupant_code -PassThru
Output:
Account_no occupant_code SideIndicator
---------- ------------- -------------
67890 1 =>
67890 2 <=

$f | InnerJoin $table4 {$Left.Account_no -eq $Right.Account_no -and $Left.occupant_code -ne $Right.occupant_code} #{Account_no = {$Left.$_}} | Format-Table
Result:
occupant_code Account_no
------------- ----------
{2, 1} 67890
For details see: In Powershell, what's the best way to join two tables into one?

In addition to all the other answers, you might be able to leverage the IndexOf() method on arrays
$services = get-service
$services.name.IndexOf("xbgm")
240
I am on a tablet right now and don't have a handy way to test it, but something along these lines might work for you:
$table4.account_no.IndexOf($_.account_no)
should fetch the index your account_no lives in for $table 4, so you could jam it all into one ugly pipe:
$accounts = Import-Csv $f | select account_no, occupant_code |
where { ($table4.account_no -notcontains $_.account_no) -or ($table4[$table4.account_no.IndexOf($_.account_no)].occupant_code -ne $_.occupant_code) }
An inner join or a normal loop might just be cleaner though, especially if you want to add some other stuff in. Since someone posted an innerjoin, you could try a loop like:
$accounts = new-object System.Collections.ArrayList
$testSet = $table4.account_no
foreach($myThing in Import-Csv $f)
{
if($myThing.account_no -in $testSet )
{
$i = $testSet.IndexOf($myThing.account_no)
if($table4[$i].occupant_code -eq $myThing.occupant_code) {continue}
}
$accounts.add($myThing)
}
Edit for OP, he mentioned $table4 is a data.table
There is probably a much better way to do this, as I haven't used data.table before, but this seems to work fine:
$table = New-Object system.Data.DataTable
$col1 = New-Object system.Data.DataColumn Account_no,([string])
$col2 = New-Object system.Data.DataColumn occupant_code,([int])
$table.columns.add($col1)
$table.columns.add($col2)
$row = $table.NewRow()
$row.Account_no = "12345"
$row.occupant_code = 1
$table.Rows.Add($row)
$row = $table.NewRow()
$row.Account_no = "67890"
$row.occupant_code = 1
$table.Rows.Add($row)
$row = $table.NewRow()
$row.Account_no = "45678"
$row.occupant_code = 3
$table.Rows.Add($row)
$testList = #()
$testlist += [pscustomobject]#{Account_no = "12345"; occupant_code = 1}
$testlist += [pscustomobject]#{Account_no = "67890"; occupant_code = 2}
$testlist += [pscustomobject]#{Account_no = "45678"; occupant_code = 3}
$accounts = new-object System.Collections.ArrayList
$testSet = $table.account_no
foreach($myThing in $testList)
{
if($myThing.account_no -in $testSet )
{
$i = $testSet.IndexOf($myThing.account_no)
if($table.Rows[$i].occupant_code -eq $myThing.occupant_code) {continue}
}
$accounts.add($myThing) | out-null
}
$accounts

Related

How to export two variables into same CSV as joined via PowerShell?

I have a PowerShell script employing poshwsus module like below:
$FileOutput = "C:\WSUSReport\WSUSReport.csv"
$ProcessLog = "C:\WSUSReport\QueryLog2.txt"
$WSUSServers = "C:\WSUSReport\Computers.txt"
$WSUSPort = "8530"
import-module poshwsus
ForEach ($Server in Get-Content $WSUSServers)
{
& connect-poshwsusserver $Server -port $WSUSPort | out-file $ProcessLog -append
$r1 = & Get-PoshWSUSClient | select #{name="Computer";expression={$_.FullDomainName}},#{name="LastUpdated";expression={if ([datetime]$_.LastReportedStatusTime -gt [datetime]"1/1/0001 12:00:00 AM") {$_.LastReportedStatusTime} else {$_.LastSyncTime}}}
$r2 = & Get-PoshWSUSUpdateSummaryPerClient -UpdateScope (new-poshwsusupdatescope) -ComputerScope (new-poshwsuscomputerscope) | Select Computer,NeededCount,DownloadedCount,NotApplicableCount,NotInstalledCount,InstalledCount,FailedCount
}
What I need to do is to export CSV outpout including the results with the columns (like "inner join"):
Computer, NeededCount, DownloadedCount, NotApplicableCount, NotINstalledCount, InstalledCount, FailedCount, LastUpdated
I have tried to use the line below in foreach, but it didn't work as I expected.
$r1 + $r2 | export-csv -NoTypeInformation -append $FileOutput
I appreciate if you may help or advise.
EDIT --> The output I've got:
ComputerName LastUpdate
X A
Y B
X
Y
So no error, first two rows from $r2, last two rows from $r1, it is not joining the tables as I expected.
Thanks!
I've found my guidance in this post: Inner Join in PowerShell (without SQL)
Modified my query accordingly like below, works like a charm.
$FileOutput = "C:\WSUSReport\WSUSReport.csv"
$ProcessLog = "C:\WSUSReport\QueryLog.txt"
$WSUSServers = "C:\WSUSReport\Computers.txt"
$WSUSPort = "8530"
import-module poshwsus
function Join-Records($tab1, $tab2){
$prop1 = $tab1 | select -First 1 | % {$_.PSObject.Properties.Name} #properties from t1
$prop2 = $tab2 | select -First 1 | % {$_.PSObject.Properties.Name} #properties from t2
$join = $prop1 | ? {$prop2 -Contains $_}
$unique1 = $prop1 | ?{ $join -notcontains $_}
$unique2 = $prop2 | ?{ $join -notcontains $_}
if ($join) {
$tab1 | % {
$t1 = $_
$tab2 | % {
$t2 = $_
foreach ($prop in $join) {
if (!$t1.$prop.Equals($t2.$prop)) { return; }
}
$result = #{}
$join | % { $result.Add($_,$t1.$_) }
$unique1 | % { $result.Add($_,$t1.$_) }
$unique2 | % { $result.Add($_,$t2.$_) }
[PSCustomObject]$result
}
}
}
}
ForEach ($Server in Get-Content $WSUSServers)
{
& connect-poshwsusserver $Server -port $WSUSPort | out-file $ProcessLog -append
$r1 = & Get-PoshWSUSClient | select #{name="Computer";expression={$_.FullDomainName}},#{name="LastUpdated";expression={if ([datetime]$_.LastReportedStatusTime -gt [datetime]"1/1/0001 12:00:00 AM") {$_.LastReportedStatusTime} else {$_.LastSyncTime}}}
$r2 = & Get-PoshWSUSUpdateSummaryPerClient -UpdateScope (new-poshwsusupdatescope) -ComputerScope (new-poshwsuscomputerscope) | Select Computer,NeededCount,DownloadedCount,NotApplicableCount,NotInstalledCount,InstalledCount,FailedCount
Join-Records $r1 $r2 | Select Computer,NeededCount,DownloadedCount,NotApplicableCount,NotInstalledCount,InstalledCount,FailedCount, LastUpdated | export-csv -NoTypeInformation -append $FileOutput
}
I think this could be made simpler. Since Select-Object's -Property parameter accepts an array of values, you can create an array of the properties you want to display. The array can be constructed by comparing your two objects' properties and outputting a unique list of those properties.
$selectProperties = $r1.psobject.properties.name | Compare-Object $r2.psobject.properties.name -IncludeEqual -PassThru
$r1,$r2 | Select-Object -Property $selectProperties
Compare-Object by default will output only differences between a reference object and a difference object. Adding the -IncludeEqual switch displays different and equal comparisons. Adding the -PassThru parameter outputs the actual objects that are compared rather than the default PSCustomObject output.

CSV file - count distinct, group by, sum

I have a file that looks like the following;
- Visitor ID,Revenue,Channel,Flight
- 1234,100,Email,BA123
- 2345,200,PPC,BA112
- 456,150,Email,BA456
I need to produce a file that contains;
The count of distinct Visitor IDs (3)
The total revenue (450)
The count of each Channel
Email 2
PPC 2
The count of each Flight
BA123 1
BA112 1
BA456 1
So far I have the following code, however when executing this on the 350MB file, it takes too long and in some cases breaks the memory limit. As I have to run this function on multiple columns, it is going through the file many times. I ideally need to do this in one file pass.
$file = 'log.txt'
function GroupBy($columnName)
{
$objects = Import-Csv -Delimiter "`t" $file | Group-Object $columnName |
Select-Object #{n=$columnName;e={$_.Group[0].$columnName}}, Count
for($i=0;$i -lt $objects.count;$I++) {
$line += $columnName +"|"+$objects[$I]."$columnName" +"|Count|"+ $objects[$I].'Count' + $OFS
}
return $line
}
$finalOutput += GroupBy "Channel"
$finalOutput += GroupBy "Flight"
Write-Host $finalOutput
Any help would be much appreciated.
Thanks,
Craig
The fact that your are importing the CSV again for each column is what is killing your script. Try to do the loading once, then re-use the data. For example:
$data = Import-Csv .\data.csv
$flights = $data | Group-Object Flight -NoElement | ForEach-Object {[PsCustomObject]#{Flight=$_.Name;Count=$_.Count}}
$visitors = ($data | Group-Object "Visitor ID" | Measure-Object).Count
$revenue = ($data | Measure-Object Revenue -Sum).Sum
$channel = $data | Group-Object Channel -NoElement | ForEach-Object {[PsCustomObject]#{Channel=$_.Name;Count=$_.Count}}
You can display the data like this:
"Revenue : $revenue"
"Visitors: $visitors"
$flights | Format-Table -AutoSize
$channel | Format-Table -AutoSize
This will probably work - using hashmaps.
Pros: It will be faster/use less memory.
Cons: It is less readable
by far than Group-Object, and requires more code.
Make it even less memory-hungry: Read the CSV-file line by line
$data = Import-CSV -Path "C:\temp\data.csv" -Delimiter ","
$DistinctVisitors = #{}
$TotalRevenue = 0
$ChannelCount = #{}
$FlightCount = #{}
$data | ForEach-Object {
$DistinctVisitors[$_.'Visitor ID'] = $true
$TotalRevenue += $_.Revenue
if (-not $ChannelCount.ContainsKey($_.Channel)) {
$ChannelCount[$_.Channel] = 0
}
$ChannelCount[$_.Channel] += 1
if (-not $FlightCount.ContainsKey($_.Flight)) {
$FlightCount[$_.Flight] = 0
}
$FlightCount[$_.Flight] += 1
}
$DistinctVisitorsCount = $DistinctVisitors.Keys | Measure-Object | Select-Object -ExpandProperty Count
Write-Output "The count of distinc Visitor IDs $DistinctVisitorsCount"
Write-Output "The total revenue $TotalRevenue"
Write-Output "The Count of each Channel"
$ChannelCount.Keys | ForEach-Object {
Write-Output "$_ $($ChannelCount[$_])"
}
Write-Output "The count of each Flight"
$FlightCount.Keys | ForEach-Object {
Write-Output "$_ $($FlightCount[$_])"
}

Seeking balanced combination of fast, terse, and legible code to add up values from an array of objects

Given the following array of objects:
Email Domain Tally
----- ----- -----
email1#domainA.com domainA.com 4
email1#domainB.com domainB.com 1
email2#domainC.com domainC.com 6
email4#domainA.com domainA.com 1
I'd like to "group by" Domain and add up Tally as I go. The end result would like this:
Domain Tally
------ -----
domainA.com 5
domainB.com 1
domainC.com 6
I have something that works but I feel like it's overly complicated.
$AllTheAddresses = Get-AllTheAddresses
$DomainTally = #()
foreach ($Addy in $AllTheAddresses)
{
if ($DomainTally | Where-Object {$_.RecipientDomain -eq $Addy.RecipientDomain})
{
$DomainTally |
Where-Object {$_.RecipientDomain -eq $Addy.RecipientDomain} |
ForEach-Object {$_.Tally += $Addy.Tally }
}
else
{
$props = #{
RecipientDomain = $Addy.RecipientDomain
Tally = $Addy.Tally
}
$DomainTally += New-Object -TypeName PSObject -Property $props
}
}
In my example, I'm creating the addresses as hashtables, but PowerShell will let you refer to the keys by .Property similar to an object.
If you're truly just summing by the Domain, then it seems like you don't need anything more complicated than a HashTable to create your running total.
The basic summation:
$Tally = #{}
$AllTheAddresses | ForEach-Object {
$Tally[$_.Domain] += $_.Tally
}
Using this sample data...
$AllTheAddresses = #(
#{ Email = "email1#domainA.com"; Domain = "domainA.com"; Tally = 4 };
#{ Email = "email1#domainB.com"; Domain = "domainB.com"; Tally = 1 };
#{ Email = "email1#domainC.com"; Domain = "domainC.com"; Tally = 6 };
#{ Email = "email1#domainA.com"; Domain = "domainA.com"; Tally = 1 }
)
And you get this output:
PS> $tally
Name Value
---- -----
domainC.com 6
domainB.com 1
domainA.com 5
Here is a "PowerShellic" version, notice the piping and flow of the data.
You could of course write this as a one liner (I did originally before I posted the answer here). The 'better' part of this is using the Group-Object and Measure-Object cmdlets. Notice there are no conditionals, again because the example uses the pipeline.
$AllTheAddresses |
Group-Object -Property Domain |
ForEach-Object {
$_ |
Tee-Object -Variable Domain |
Select-Object -Expand Group |
Measure-Object -Sum Tally |
Select-Object -Expand Sum |
ForEach-Object {
New-Object -TypeName PSObject -Property #{
'Domain' = $Domain.Name
'Tally' = $_
}
} |
Select-Object Domain, Tally
}
A more terse version
$AllTheAddresses |
Group Domain |
% {
$_ |
Tee-Object -Variable Domain |
Select -Expand Group |
Measure -Sum Tally |
Select -Expand Sum |
% {
New-Object PSObject -Property #{
'Domain' = $Domain.Name
'Tally' = $_
}
} |
Select Domain, Tally
}
Group-Object is definitely the way to go.
In the interest of terseness:
Get-AllTheAddresses |Group-Object Domain |Select-Object #{N='Domain';E={$_.Name}},#{N='Tally';E={($_.Group.Tally |Measure-Object).Sum}}

Inner Join in PowerShell (without SQL)

How do we make Inner-Join or something a Cross-Join in PowerShell or PowerCLI?
Even though im new to PowerCLI/PowerShell , I do have a basic grasp on them, yet have practically spent 2 days trying to figure this, going through numerous documentations and blogs to no avail.
All I really want to know is if after typing my command
Get-Content File.txt
and getting:
Output1 or Table1 is
Name: Abc
Group: Bad
Policy: Great
Name: redi
Group: Good
Policy: MAD
etc. etc.
100s of these, and obviously more than just the 3 elements of Name, Group, Policy each.
Table2/Output2
Name: Abc
Limit: 10
used: 5
Name: redi
Limit: 20
used: 1
etc. etc.
100s of these.
and like 13 more of these text file tables, all with the "Name" as unique.
How can I combine it into one output at the end using Name with all the other elements?
My most obvious thought was something akin to joins, even if I had to do them 1 at a time, but even that I cant figure out how to do.
Is there anyway to do this in PowerShell itself without me having to go into Python or SQL?
If yes is there a method that is able to combine fields in spots where it's null?
If its not clear what type of result I am hoping for it will look something akin to this:
Name: Abc
Group: Bad
Policy: Great
Limit: 10
used: 5
Name: redi
Group: Good
Policy: MAD
Limit: 20
used: 1
Paweł Dyl provided you a solution
based on your two tables. However you probably need a generic solution where you don't have to specify each property by name yourself.
I would combine each table to a an array. Group the tables on the Name property using the Group-Object cmdlet. Iterate over each group and create a PsObject using the properties:
$table1 = [PSCustomObject]#{ Name = 'Abc'; Group = 'Bad'; Policy = 'Great'}, [PSCustomObject]#{ Name = 'redi'; Group = 'Good'; Policy = 'MAD'}
$table2 = [PSCustomObject]#{ Name = 'Abc'; Limit = '10'; used = '5'}, [PSCustomObject]#{ Name = 'redi'; Limit = '20'; used = '1'}
$allTables = $table1 + $table2
$allTables | group Name | Foreach {
$properties = #{}
$_.Group | Foreach {
$_.PsObject.Properties | Where Name -ne 'Name' | Foreach {
$properties += #{
"$($_.Name)" = "$($_.Value)"
}
}
}
$properties += #{Name = $_.Name}
New-Object PSObject –Property $properties
}
Output:
Group : Bad
Policy : Great
Name : Abc
Limit : 10
used : 5
Group : Good
Policy : MAD
Name : redi
Limit : 20
used : 1
You can use simple loop join as follows:
$table1 = [pscustomobject]#{Name='Abc';Group='Bad';Policy='Great'},[pscustomobject]#{Name='redi';Group='Good ';Policy='MAD'}
$table2 = [pscustomobject]#{Name='Abc';Limit=10;used=5},[pscustomobject]#{Name='redi';Limit=20;used=1}
$table1 | % {
foreach ($t2 in $table2) {
if ($_.Name -eq $t2.Name) {
[pscustomobject]#{Name=$_.Name;Group=$_.Group;Policy=$_.Policy;Limit=$t2.Limit;Used=$t2.Used}
}
}
}
Assuming uniqueness of keys you can also use faster, hashtable approach:
$hashed = $table1 | group Name -AsHashTable
$table2 | % {
$matched = $hashed[$_.Name]
if ($matched) {
[pscustomobject]#{Name=$matched.Name;Group=$matched.Group;Policy=$matched.Policy;Limit=$_.Limit;Used=$_.Used}
}
}
You can also use generic solution and wrap it in function. It matches records by their property names:
function Join-Records($tab1, $tab2){
$prop1 = $tab1 | select -First 1 | % {$_.PSObject.Properties.Name} #properties from t1
$prop2 = $tab2 | select -First 1 | % {$_.PSObject.Properties.Name} #properties from t2
$join = $prop1 | ? {$prop2 -Contains $_}
$unique1 = $prop1 | ?{ $join -notcontains $_}
$unique2 = $prop2 | ?{ $join -notcontains $_}
if ($join) {
$tab1 | % {
$t1 = $_
$tab2 | % {
$t2 = $_
foreach ($prop in $join) {
if (!$t1.$prop.Equals($t2.$prop)) { return; }
}
$result = #{}
$join | % { $result.Add($_,$t1.$_) }
$unique1 | % { $result.Add($_,$t1.$_) }
$unique2 | % { $result.Add($_,$t2.$_) }
[PSCustomObject]$result
}
}
}
}
$table1 = [pscustomobject]#{Name='Abc';Group='Bad';Policy='Great'},
[pscustomobject]#{Name='redi';Group='Good ';Policy='MAD'},
[pscustomobject]#{Name='Not joined';Group='Very bad';Policy='Great'}
$table2 = [pscustomobject]#{Name='Abc';Limit=10;used=5},
[pscustomobject]#{Name='redi';Limit=20;used=1},
[pscustomobject]#{Name='redi';Limit=20;used=2}
#name is only common property, records joined by name
Join-Records $table1 $table2
#example2
$test1 = [pscustomobject]#{A=1;B=1;C='R1'},
[pscustomobject]#{A=1;B=2;C='R2'},
[pscustomobject]#{A=2;B=2;C='R3'}
$test2 = [pscustomobject]#{A=1;B=1;D='R4'},
[pscustomobject]#{A=3;B=2;D='R5'},
[pscustomobject]#{A=4;B=2;D='R6'}
Join-Records $test1 $test2 #joined by two common columns - A and B
You can also cascade calls:
$test1 = [pscustomobject]#{A=1;B=1;C='R1'},
[pscustomobject]#{A=1;B=2;C='R2'},
[pscustomobject]#{A=2;B=2;C='R3'}
$test2 = [pscustomobject]#{A=1;B=1;D='R4'},
[pscustomobject]#{A=3;B=2;D='R5'},
[pscustomobject]#{A=4;B=2;D='R6'}
$test3 = [pscustomobject]#{B=1;E='R7'},
[pscustomobject]#{B=2;E='R8'},
[pscustomobject]#{B=3;E='R9'}
#first join by common A and B, then join result by common B
Join-Records (Join-Records $test1 $test2) $test3
So I found an Answer which was more suitable and it uses the join-Object function which was defined below:
you can access it at https://github.com/RamblingCookieMonster/PowerShell/blob/master/Join-Object.ps1
All I really had to do was Define my outputs as $A and $B and $C and so on, and just
$Join1= Join-Object -Left $A -Right $B -LeftJoinProperty Name - RightJoinProperty Name
made $Join2 then 3 so on until I got it all done
$Join2 = Join-Object -Left $Join1 -Right $C -LeftJoinProperty Name -RightJoinProperty Name
$Join3 = Join-Object -Left $Join2 -Right $D -LeftJoinProperty Name -RightJoinProperty Name
$Join4 = Join-Object -Left $Join3 -Right $E -LeftJoinProperty Name -RightJoinProperty Name
Until I got it all done
$Table1 | Join $Table2 -Using Name
$Table1 | Join $Table2 #Cross Join
See: In Powershell, what's the best way to join two tables into one?

Powershell - Prefix each line of Format-Table with String

I would like to know if there is an easy way of prefixing each line of a powershell table with a String.
For example, if I create an Array using the following code:
$Array = #()
$Object = #{}
$Object.STR_PARAM = "A"
$Object.INT_PARAM = 1
$Array += [PSCustomObject] $Object
$Object = #{}
$Object.STR_PARAM = "B"
$Object.INT_PARAM = 2
$Array += [PSCustomObject] $Object
Calling Format-Table give the following output:
$Array | Format-Table -AutoSize
STR_PARAM INT_PARAM
--------- ---------
A 1
B 2
Instead, I would like to have the following:
$Array | Format-Table-Custom -AutoSize -PrefixString " "
STR_PARAM INT_PARAM
--------- ---------
A 1
B 2
And if possible, I would also like to be able to use the Property parameter like this:
$SimpleFormat = #{Expression={$_.STR_PARAM}; Label="String Param"},
#{Expression={$_.INT_PARAM}; Label="Integer Param"};
$Array | Format-Table-Custom -Property $SimpleFormat -AutoSize -PrefixString "++"
++String Param Integer Param
++------------ -------------
++A 1
++B 2
Any help would be appreciated. Thanks.
You could just use the format expressions directly:
$f = #{Expression={"++" + $_.STR_PARAM}; Label="++String Param"},
#{Expression={$_.INT_PARAM}; Label="Integer Param"};
$Array | Format-Table $f -AutoSize
Output
++String Param Integer Param
-------------- -------------
++A 1
++B 2
Update to use expression and filter
Filter Format-Table-Custom
{
Param
(
[string]
$PrefixString,
[object]
$Property
)
end {
$rows = $input | Format-Table $property -AutoSize | Out-String
$lines = $rows.Split("`n")
foreach ($line in $lines) {
if ($line.Trim().Length -gt 0) {
$PrefixString + $line
}
}
}
}
$f = #{Expression={"--" + $_.STR_PARAM}; Label="--String Param"},
#{Expression={$_.INT_PARAM}; Label="Integer Param"};
$Array | Format-Table-Custom -Property $f -PrefixString "++"
Output
++--String Param Integer Param
++-------------- -------------
++--A 1
++--B 2