VBA dynamically populate nested data structure - vba

Question

I have a few SQL tables, some which are linked, that I would like to query once and store locally in a single variable. I can't predict the length of the data ahead of time so I need a dynamic data structure.
Example data I'm querying:
Table 1
NameA
Red
Green
Blue
Table 2
NameA NameB
Red A
Red B
Red C
Blue D
Blue E
Green F
Table 3
NameA NameC
Red One
Blue Two
Blue Three
Blue Four
Blue Five
Green Six
Green Seven
I need to be able to filter and access NameB and NameC based on NameA values. I would prefer a nested dictionary structure where I could query like below:
Table1("0") 'will equal "Red"
Table2("Red")("0") 'will equal "A"
Table2("Blue")("1") 'will equal "E"
Table3("Green")("1") 'will equal "Seven"
'note: point here is data structure, not order of results
I have tried using VBA's nested dictionaries but have been unable to get around the lack of a "deep copy" function. One algorithm I wrote:
With SqlQueryResult
i = 0
Do Until .EOF
Call Table1.Add(CStr(i), .Fields(0).Value)
i = i + 1
.MoveNext
Loop
End With
For Each key In Table1.Keys
SqlQueryResult = GetResultsFromQuery(SELECT NameB WHERE NameA = Table1(key))
With SqlQueryResult
i = 0
Do Until .EOF
Call TempDict.Add(CStr(i), .Fields(0).Value)
i = i + 1
.MoveNext
Loop
End With
Set Table2(Table1(key)) = TempDict
TempDict.RemoveAll
Next key
Unfortunately assigning a Dict to another Dict only sets a reference and doesn't actually copy over data -- when I delete TempDict, the nested data from Table2 is also removed.
I also can't have a new dictionary per "branch" in the nest structure as I need this data to be available at a module-level scope, and therefore need to define these in the top of the module before program execution.
I've looked at multi-dimentional dynamic arrays - these can't be assigned to a parent structure like a dictionary. I also can't predict the size of each of these tables, e.g. Table1 might be 5/20/100/etc in size, Red may have 2/5/100/etcetc results in Table 2, Blue have 1/20/etcetc results in Table 2. Redim only works on a single dimension in an array.
I've had a brief look at Collections as well, and I am not sure these are viable.
I don't have much experience with classes and I would rather avoid a very involved process - I want it to be easy to add linked and unliked (i.e. data linked to Table 1, like Table 2 and 3, vs stand-alone data not related to any other table) to this program should I need to in the future. (My benchmark for "easy" is a pandas dataframe in python).

Answer 1

A simple wrapper class for scripting dictionaries which implements a clone method. This should work fine with primitive datatypes.
Option Explicit
Private Type State
Dict As scripting.Dictionary
End Type
Private s As State
Private Sub Class_Initialize()
Set s.Dict = New scripting.Dictionary
End Sub
Public Function Clone()
Dim myClone As scripting.Dictionary
Set myClone = New scripting.Dictionary
Dim myKey As Variant
For Each myKey In s.Dict
myClone.Add myKey, s.Dict.Item(myKey)
Next
Set Clone = myClone
End Function
Public Property Get Item(ByVal Key As Variant) As Variant
Item = s.Dict.Item(Key)
End Property
Public Property Set Item(ByVal Key As Variant, ByVal Value As Variant)
s.Dict.Item(Key) = Value
End Property
Public Sub Add(ByVal Key As Variant, ByVal Item As Variant)
s.Dict.Add Key, Item
End Sub
You will now be able to say
Set Table2.Item(Table1.Item(key)) = TempDict.Clone