Suggested Map UDF

[Melba23]

This was a place holder post to spark interest - much better implementations than my initial thoughts found later in the thread.

M23

Edited July 31, 2014 by Melba23
Removed code

[Zedna]

Maybe _FileWriteToMap should be rather _FileWriteFromMap?

[Melba23]

Zedna,

Sorry - typo.

M23

[guinness]

A couple of points for the examples.

• Global should be changed to Local.
• Variables undeclared and fail Au3Check with additional settings.
• They should be encased in an example function as well. Similar to the Map* functions.
• Run Tidy on both the examples and includes.
• Remove EOLs at the end of the Map.au3 and just leave a single CRLF.

 

PS I can make these changes myself once committed.

With _MapToString() I feel there should be consistency between that function and _ArrayToString(). For example I would expect it to print {Array} as it does in _MapToString(). I am guessing people will start to expect _ArrayToString() to have the same output when an array or map is used.

#include <MsgBoxConstants.au3>
#include <Map.au3>

Local $mMap[], $aTemp = [0, 1, 2, 3, 4, 5]
$mMap.Tom = "Tom value"
$mMap.Dick = $aTemp
$mMap.Harry = "Harry value"

Local $sMap = _MapToString($mMap)
MsgBox($MB_SYSTEMMODAL, "Map string", $sMap)
ClipPut($sMap)

Local $aArray = ["Tom value", $aTemp, "Harry value"]
Local $sArray = _ArrayToString($aArray) ; Tom value|{Array}|Harry value
MsgBox($MB_SYSTEMMODAL, "Array string", $sArray)

In _MapToArray() I feel the addition of quotes should be left to the user as this function should be a straight map to array function only. If the user wants quotes then it's up to them to add this cosmetic feature.

Great work by the way. It's just I have managed to sit down today and go through the proposed UDF in full.

[jpm]

Small remark

I don't know where the Map name come from.

Anyway I suggest the naming of such variable start with "m" as opposed to "a" for Array

[Melba23]

guinness,

Thanks for the encouraging words. I will look at changing the Array functions to return the same sort of {---} content as the Map ones when dealing with internal arrays/maps.

jpm,

  Quote

I don't know where the Map name come from

Jon decided >here (private forum) - and the $m prefix comes from there too.

And there are lots of $aMap array variables in the UDF as the $mMap maps are converted for manipulation/iteration/etc.

M23

[jpm]

  On 7/28/2014 at 10:48 AM, Melba23 said:

guinness,

Thanks for the encouraging words. I will look at changing the Array functions to return the same sort of {---} content as the Map ones when dealing with internal arrays/maps.

jpm,

Jon decided >here (private forum) - and the $m prefix comes from there too.

And there are lots of $aMap array variables in the UDF as the $mMap maps are converted for manipulation/iteration/etc.

M23

Thanks for pointing the post

$m must be added to Wiki

[jvanegmond]

I understand the appeal of converting the map into an array and then reusing existing array functions to do the hard work. For _MapDisplay this makes a lot of sense, considering how much heavy lifting Array.au3 does. However, considering maps may hold millions of items converting the entire structure into an array can take several seconds*. Functions _MapMaxKey, _MapMinKey, _MapMaxValue, _MapMinValue, _MapFindAll, _MapSearch and _MapToString probably should not be calling _MapToArray.

* Preliminary speedtest with latest beta. 100K elements converted into array: 63 seconds. (Might be a problem with map implementation though)

Edit #100: Couldn't sleep. Code may not be the best.

I changed the map iteration stuff to For $vKey In MapKeys($mMap) everywhere. It looks better and sets a good example.

Changelog:

 - _MapFindAll: Removed _MapToArray on the input

 - _MapFindAll: Do not declare empty array as the maximum-size of the input (e.g. $avResult[uBound($aMap)] ) but use a sexy new map to solve that $avResult[] (we have lists now, everyone!). Kind of a shame I had to Return _MapToArray($avResult) at the end.

 - _MapMaxKey: Removed _MapToArray on the input

 - _MapMinKey: Removed _MapToArray on the input

 - _MapSearch: Removed _MapToArray on the input

 - _MapToArray: Code cleanup.

 - _MapToString: Code cleanup. 

Here is a diff: http://www.mergely.com/J2OQdvns/ (P.S. We need better tools.)

Map.au3

Edited July 28, 2014 by Manadar

[guinness]

Perhaps a common function set should be created between Array.au3 and Map.au3? To reduce duplicate code. Personally I prefer tidy and easy to maintain code first, then consider the speed factor afterwards.

[jvanegmond]

  On 7/28/2014 at 8:12 PM, guinness said:

Perhaps a common function set should be created between Array.au3 and Map.au3? To reduce duplicate code. Personally I prefer tidy and easy to maintain code first, then consider the speed factor afterwards.

Forgotten this from my last post.

 

I broke _MapFindAll and _MapSearch. And maybe the _MapMin / _MapMax as well. The code for comparisons need to be shared between Array and Map in a different way than converting to arrays and calling array libraries.

 

MapSearch documentation has reference to $iCase?

 

_MapSearch ( Const ByRef $mMap, $vValue [, $iCompare = 0 ] )


###Parameters###
@@ParamTable@@
$mMap
    The Map to search
$vValue
    The specified value for which to search
$iCompare
    [optional]
        0 Casting of variables to the same type (default), "string" = 0, "" = 0 or "0" = 0 match (If $iCase = 0)

Edit: And it also would be really awesome if _ArrayDisplay took a map and just displayed it no questions asked. Check _MapFindAll for why. I'm expecting to see this pattern way more often in the future. Basically it's the "If you make a map with just integers keys, you have a list" realization.

Edited July 28, 2014 by Manadar

[Zedna]

@Manadar

I think _MapSearch() should return something (null?) or SetError when nothing is found.

http://www.mergely.com/J2OQdvns/

Edited July 28, 2014 by Zedna

[jvanegmond]

  On 7/28/2014 at 9:57 PM, Zedna said:

@Manadar

I think _MapSearch() should return something (null?) or SetError when nothing is found.

http://www.mergely.com/J2OQdvns/

 

I think it should be like this

Func _MapSearch(Const ByRef $mMap, $vSearch, $iCompare = 0)

    If Not IsMap($mMap) Then Return SetError(1, 0, -1)
    If $iCompare = Default Then $iCompare = 0

    For $vKey In MapKeys($mMap)
        $vValue = $mMap[$vKey]

        If $vValue == $vSearch Then
            Local $aResult[]
            $aResult[$vKey] = $vValue
            Return $aResult
        EndIf
    Next

    Return Null

EndFunc   ;==>_MapSearch

You either get null if nothing is found, or you get the whole key/value pair in a map. Not finding what you were looking for in a map is not really an error condition so no @error here.

Edited July 29, 2014 by Manadar

[czardas]

I think you should remove the $iCompare parameter.

Edited July 29, 2014 by czardas

[guinness]

  On 7/29/2014 at 7:08 AM, Manadar said:

I think it should be like this

Not finding what you were looking for in a map is not really an error condition so no @error here.

Couldn't agree more.

[guinness]

Manadar,

In  _MapToString() you are checking if a key is a map or an array, though they can only be integers or strings. Also the function wasn't returning anything as you were only updating the $sOutput variable if it was a map or an array.

Also _MapSearch() isn't using the $iCompare variable.

Local $mMap[]
For $i = 0 To 19
    $mMap[$i] = $i
Next
ConsoleWrite(_MapToString($mMap) & @CRLF)

Func _MapToString(Const ByRef $mMap, $sDelim_Item = "|", $sDelim_Row = @CRLF)
    If $sDelim_Item = Default Then $sDelim_Item = "|"
    If $sDelim_Row = Default Then $sDelim_Row = @CRLF
    If Not IsMap($mMap) Then Return SetError(1, 0, -1)

    Local $sRet = "", _
            $vValue = 0

    For $vKey In MapKeys($mMap)
        $vValue = $mMap[$vKey]
        $sRet &= $vKey & $sDelim_Item
        __MapToString_Print($sRet, $vValue)
        $sRet &= $sDelim_Row
    Next

    $sRet = StringTrimRight($sRet, StringLen($sDelim_Row)) ; Strip the last delimiter.
    Return $sRet
EndFunc   ;==>_MapToString

Func __MapToString_Print(ByRef $sOutput, ByRef $vValue)
    Switch VarGetType($vValue)
        Case "Array"
            $sOutput &= "{Array}"
        Case "Map"
            $sOutput &= "{Map}"
        Case Else
            $sOutput &= $vValue
    EndSwitch
EndFunc   ;==>__MapToString_Print

[jvanegmond]

Thanks guinness, czardas. Was just working on it and included your feedback.

 

Changelog:

- New: _MapKeysToArray

- New: _MapValuesToArray

- Fixed: _MapToString / _MapToClip not working

- Changed: Added $iCase to _MapSearch and _MapFindAll

- Changed: Split comparison logic into new function __MapSearchComparison. (needs testing) I did not update Array.au3 because _ArraySearch code is a mess. : (

- Changed: _ArrayDisplay will call _MapDisplay when it receives a map

- Changed: _MapSearch returns a key/value pair or null

- Removed: _MapToArray $bString parameter to quote keys (and put this in _MapDisplay since it is highly specific for the purpose of displaying a map)

 

Diff: http://mergely.com/hUuFirz8/

 

 

 

I'm not sure if we should include _MapRenameKey. Worst case scenario it will end up making two whole array copies. That may surprise someone who uses the library therefore it's a leaky abstraction.

 

Apart from maybe Array.au3 rewrites I don't expect much more changes to Map.au3

Map UDF and Examples.zip

Edited July 29, 2014 by Manadar

[jpm]

Hi,

I don't like to have _ArrayDisplay() to call _MapDisplay() as other Array functions will need to retrieve an extra include for nothing.

_ArrayDisplay should display array. Perhaps _MapDisplay() can display one or the other as the syntax are very similar.

If we want not to include all Array.au3 in Map.au3 we can isolate _ArrayDisplay() in a new include

Cheers

Jpm

[Mat]

How achievable is a Collection.au3 include that is used by both Map.au3 and Array.au3 Manadar?

[TypeIt]

  On 7/29/2014 at 2:21 PM, Mat said:

How achievable is a Collection.au3 include that is used by both Map.au3 and Array.au3 Manadar?

Should that contain other functions can operate on both types?

Not every include file needs to be public.

Edited July 29, 2014 by TypeIt

[Mat]

  On 7/29/2014 at 2:35 PM, TypeIt said:

Should that contain other functions can operate on both types?

Not every include file needs to be public.

Where possible, yes, in most cases it won't save us anything though. _ArrayDisplay is a special case in that the code is very long, so the actual array processing is a relatively small part of it.