Code Scanner

[HighlanderSword]

NO changes made , downloaded the file ,  exported files and ran the code scanner , (and walked thru thru the code to get to section where it is crashing on) is their some other environment stuff that needs to be set before this can be run ?

 

 

[RTFC]

Nope, it should run out of the box. Let me try and see if I can reproduce your issue in a newly-installed test environment...

EDIT: And it does run out of the box, for me, running on v.3.3.14.5. And for me to fix anything, I need to be able to see it break in my own environment. Maybe retry after restarting your OS afresh, or testing on another machine. Not sure what else to suggest, sorry.

Edited July 20, 2020 by RTFC

[HighlanderSword]

same issue on another system, what Version of windows are you running this on

After doing more digging appears to fail in AES on code below

Local $Ret = DllCall("user32.dll", "uint", "CallWindowProc", "ptr", DllStructGetPtr($CodeBuffer) + $AP_Decompress, _
                                                    "ptr", DllStructGetPtr($Output) + 4, _
                                                    "ptr", DllStructGetPtr($Result), _
                                                    "int", 0, _
                                                    "int", 0)

[RTFC]

I'm on W10Pro/x64.

Just a wild guess: you're running in an x64 environment, and it's the test script that's crashing, not CodeScanner.

Edited July 20, 2020 by RTFC
typo

[HighlanderSword]

Yes im running in x64 , but it is codescanner that is crashing

 

when you tested in a new env how did you do that

I created a VM and installed autoit and scite and your codescanner files and had the same issue

Also are you on windows ver 2004 ?

Edited July 20, 2020 by HighlanderSword

[RTFC]

Okay, edit your MCFinclude.au3 _dummycalls() func thus:

Func _dummyCalls()
; prevents MCF:_CreateSingleBuild() from removing the defs as redundant
; this UDF will be fixed-key encrypted

    _MCFCC_Init($CCkeytype,False,True)
    _MCFCC("")
    _MCFCNG("init")

 

[HighlanderSword]

ok and that did  the trick... Can you please explain the modification

[RTFC]

Despite claims to the contrary, Ward's AES machine code does not work on x64; it crashes the system (as you noticed). CodeScanner includes the  MCF library, and the latter includes MCFinclude.au3. MCFinclude initialises both encryption libraries, but if you're running AutoIt in x64, then CodeScanner attempts to run AES's faulty x64 initialisation. So I'm going to patch AES some more so this can no longer happen. Many thanks for alerting me to this issue; it will be fixed in the next bundle release.

[HighlanderSword]

Thanks for getting it to work, couple other thing you might want to be aware of

• #include <D:\Test_includes\__ArrayDisplay.au3> , when code is like that codescanner cant find the include file, changing it to #include "D:\Test_includes\__ArrayDisplay.au3" resolved the issue
• Once Code scanner is done and treeview is displayed double clicking the mouse on a section does not bring up the screen , have to hit the enter key , so for example double click on settings and nothing happens

 

[RTFC]

Thanks, but your first point is actually an AutoIt feature (and the same convention holds in many other programming languages); it's got nothing to do with CodeScanner. #Includes in angled brackets are expected to be stored in AutoIt's default #includes location; and full path references need to be in quotes. Regarding issue 2, I would consider this trivial; learn to live with it.

EDIT: and of course on x64 you'll have to select Bcrypt encryption only in CodeCrypter (under Tab encryption)

Edited July 20, 2020 by RTFC

[Gianni]

Hi rtfc
I'm trying the "codecanner" SW to analyze the list of this small script written just to do some basic test:

Global $Temp = _GetTemp()

If $Temp >= 15 Then
    ConsoleWrite("It's Hot" & @CRLF)
Else
    ConsoleWrite("It's Cold" & @CRLF)
EndIf

Global $ii
While $ii < 10
    For $i = 1 To 10
        ConsoleWrite('.')
    Next
    $ii += 1
WEnd

Func _GetTemp()
    Local $iTemp
    $iTemp = Random(0, 30, 1)
    Return $iTemp
EndFunc   ;==>_GetTemp

I can see that in the output file called "References.txt" each line of the source is described in detail.
But in that script the references to the various "loops" do not have the reference of the two mutual line numbers relating to the mutual "Loop Start" & "Loop End". While that information is contained in the GUI TreeView in the "Results by Subject -> General -> Loops" branch. Are this information present in any other outgoing file? or can they be generated with some run parameters?
In addition, I do not find the references to the lines of the list relating to the conditionally run statements "If Then Else EndIf". Can these also be found in some files or generated with some run parameters?
Thank You in advance.

[RTFC]

Hi Gianni Good questions.

The loop-related data is stored in "loops.txt," which is written out to the CS datadump directory only if internal global parameter $fulldump is true and metacode file writing is off (as it's not necessary for MCF processing). By default in GUI and cmdline mode, this boolean is false. You can either run CodeScanner in cmdline mode with explicit parameter /f (or -f) set, or change the call to "_WriteCSDataDump(True)" in Func _AnalyseCode() to always force full output. (NB this will result in an additional query at runtime whether you wish to inspect these files with explorer when written; just press cancel).

All conditional statements are called "phrases" in this environment. You can find them in file MCF0_phrased.txt after you run CodeCrypter to produce an encrypted version of your script. For that to work (and for Codecrypter to work on your code at all) you need to add #include "MCFinclude.au3" prior to your code, and (re)run CodeSCanner with writeMCF enabled.

In MCF0_phrased, look for "If {phrase", "While {phrase" and/or "Until {phrase"; each line has file and line number references suffices in curly brackets appended.

I never intended CodeScanner to produce separate listings of all conditionals (branching points), but the MCF library is able to produce these (see func _PhraseMCF in MCF.au3). You could hook into the MCF library directly with your code, without jumping through CodeCrypter's hoops (in which case, please consult the MetaCode tutorial pdf here).

Any more issues, don't hesitate to ask.

[Gianni]

Hi @RTFC, Thank you for your answer,

I was able to generate the loops.txt file, but not the MCF0_phrased.txt file
Could you please detail the steps required to generate the MCF0_phrased.txt file as well?

• I added #include "MCFinclude.au3" at the beginning of my Simplest.au3 target file
• I checked that the WriteMetaCode=True entry is present in the CodeScanner.ini file
• I forced the True parameter in the _WriteCSDataDump(True) function present in the _AnalyseCode() portion of the CodeScanner.au3 listing
• so running CodeScanner.au3, this also generated the loops.txt file

Running codecrypter.au3 later I get this alert:

"CodeScanner output file not found:
\....\Simplest.au3.CS_DATA\AU3operators.txt
Are you sure you ran CodeScanner with setting "WriteMetaCode=True",
on target file \....\simplest.au3 before? CodeCrypter requires this function."

the process proceeds anyway, but at the end I can't find the MCF0_phrased.txt file

Where am I doing wrong?
Thanks again for any help

Edited August 31, 2022 by Gianni

[RTFC]

You're not going wrong, I was. CodeScanner was the second AutoIt script I ever wrote, and the first one I ever released (I wrote it to trace/resolve a deeply-nested callstack bug in my first script I just couldn't seem to fix manually; that now seems like a lifetime ago). It's almost a decade old and filled with patches of patches of rookie design mistakes (such as separate, parallel explicit lists of files to process for different cases, which then diverged, which causes "AU3operators.txt" to be written out for MCF processing, but not when writing a full dump). I never noticed this before, so thank you for spotting this. I should really spend some time to overhaul that whole data dump disaster and make it streamlined and consistent for a change. In the meantime, a quick patch would be in CodeScanner's function _WriteCSDataDump, immediately under label "; dump 1-D arrays" to just duplicate line:

"_FileWriteFromArray1D($CS_dumppath & "AU3Functions.txt",$AU3Functions,1)"

and edit it to write out the operators too:

_FileWriteFromArray1D($CS_dumppath & "AU3operators.txt",$AU3operators,1)

Since it always exists, there's no need to check for existence and optionally add an empty dummy file afterwards (immediately below that section).

 

Hint: Instead of using CodeCrypter (which is just a front-end for the MCF library), I would suggest you engage with the latter directly (as intimated earlier), like so:

#include "MCF.au3"

Global $path="<your CS_DATA dump path, including trailing backslash \>"

If _ReadCSDataDump($path)=False Then Exit(-11)

If Not FileExists($path & "MCF0.au3") Then
    If _CreateSingleBuild($path,True)=False Then Exit(-12)
EndIf

If _PhraseMCF($path)=False Then Exit(-13)
ConsoleWrite("Done." & @CRLF)

Easier, no?

Edited August 31, 2022 by RTFC