Regex Custom UDL Parser

Khundian Twitch · Dec 10, 2023, 10:55 PM

Hey everyone, I’m setting Npp up for GECKScript (scripting language for Fallout New Vegas) I have made my own syntax highlight and autocomplete, now I’m setting up the Functionlist. The class block is working great, every line is in the interface between my class blocks… but no names yet.

I’m done writing my regex, it works without problems on a sample document but somehow Npp doesn’t want to accept it. Anyone care to take a look and have some advice?

Need 1 rep poing for posting links, guess I"ll post a sample instead of linking an interactive site. :)

commentExpr="(;.*?$)"

mainExpr="^[\t ]++|(?:Begin)"

<nameExpr expr="^[\t 0-9A-Za-z]+[]!"#$%&'()*+,.\/:;<=>?@[\\_`{|}~^-]+([A-Za-z]+)"

; Vault Level Variables:

Short	bPlayerSafe
Short	bPlayerBlocks
Short	bVaultOpen
Short	bTreasureFound
Short	bSecurityAltInfo
Short	bChristineHelp
Short	bDeanHelp

Int		iTurretsDown
Int		iSecurityState
Int		iElevatorState
Int		iMaintenanceState

; Elijah Variables:

Short	bElijahContacted
Short	bElijahTaunted
Short	bElijahInVault
Short	bElijahPaths
Short	bElijahTurnsSecurityOn
Short	bElijahEntersVault
Short	bElijahArrivesVault
Short	bElijahOpensVault
Short	bElijahFights
Short	bElijahDead
Short	bElijahTrapped
Short	bElijahTrapsPlayer
Short bElijahTeamsWithPlayer
short bPlayerTrapped

Int		iElijahState
Int		iChristineStrikes

Float fAngle

; Operational Variables:
Int		iCount
Float	fTimer

; >>>>---------------- Begin Code;

; >>>>----------------
Begin GameMode

	If (bTreasureFound == 1)
		If (bVaultOpen == 1)																											; If Players have opened the Vault previously.
			If (bElijahContacted == 1)																								; If Players have contacted Elijah - start sequence.
				If (iElijahState == 0)																									; If First Stage of Endgame.
					If (bElijahInVault == 0)																							; If Elijah not yet teleported to Vault.
						NVDLC01ElijahREF.Disable;
						NVDLC01ElijahREF.MoveTo NVDLC01VaultStartMarkerREF;
						SetStage NVDLC01MQ03 48;
						Set bElijahInVault to 1;
						Return;
					Else
						NVDLC01ElijahREF.Enable;																					; Enable Elijah once he's been moved to the Vault.
						Set iElijahState to 1;																							; Ready Elijah for the next stage of events/AI behaviors.
						Set bElijahPaths to 1;																							; Set variable to be checked by Elijah's AI package.
						Return;
					Endif
				Elseif (iElijahState == 1)
					If (iCount == 0)
						NVDLC01ElijahREF.AddScriptPackage NVDLC01VaultElijaIntroPathTravel;

Thanks!

Khundian Twitch · Dec 11, 2023, 1:57 PM

Guess I can add the link now, I received a couple reputation points. :)

This is where I’m at now… “class” block regex added and all lines between the “class” block show up, but can’t get the function names to show up. I’m getting desperate, can anyone shed some light on why its not working?

Thanks!

This is a link to one of the regex’ I’m trying, has good visual representation and can play around with the regex and save a new version on top left of the website if you want. https://regex101.com/r/spMBF6/4

This is a link straight to the debugger, it shows you visually what its doing, and finishes the document without an issue.
https://regex101.com/r/spMBF6/4/debugger

PeterJones · Dec 11, 2023, 1:59 PM

@Khundian-Twitch ,

Regex101 uses a different flavor of regular expression than Notepad++ (which uses the Boost regex library) – so just because something works at that website doesn’t mean it will work in either Notepad++'s search or its Function List parsing.

If you want to test a regex for Notepad++'s Function List parsing, the best thing to do is test it inside Notepad++'s search dialog.

I am hoping that @MAPJe71 will be able to take a look at your definitions and see if he has any insight, because he’s our resident Function List expert.

I’m getting desperate

That is unfortunate for you, as this is a Community of fellow Notepad++ users, none of whom are paid to give you technical support on demand. If someone is able to give you insight, they will … when it is convenient for them.

[link to regex101]

For those who don’t want to follow that link, the current state of the regex appears to be ^(?>[\t\v]+(?<GECKScript>[]!#$%&()*+.\/0-9:;<=>A-Z[a-z{|}^-]+))+

PeterJones · Dec 11, 2023, 2:39 PM

@Khundian-Twitch ,

This one doesn’t use a class, and I am quite certain it doesn’t meet all your needs, but it might give you a working starting point:

<?xml version="1.0" encoding="UTF-8" ?>
<NotepadPlus>
    <functionList>
        <parser
            displayName="KT-UDL"
            id         ="ktudl_fn"
            commentExpr="(;.*?$)"
        >
            <function
                mainExpr="(?-i:^\h*Begin \w+)|(?i:^\h*(If|Elseif).*?$)"
                displayMode="$functionName">
                <functionName>
                    <nameExpr expr="(?i:^\h*Begin \w+)|(?i:^\h*(If|Elseif).*?$)"/>
                </functionName>
            </function>
        </parser>
    </functionList>
</NotepadPlus>

I basically had it find any Begin NameHere or If blah or Elseif blah and call those the “functions”

Looking at your original mainExpr: it’s probably too short to accomplish anything. For function names, the mainExpr must match the entirety of the name (and possibly more). For class ranges, the mainExpr should match all the text from the beginning of the class to the end of the class. My guess is that’s where things were going wrong for you… but since you just gave us pieces of the XML rather than the full definition, I cannot be sure.

Khundian Twitch · Dec 11, 2023, 3:28 PM

@PeterJones
Thanks very much for posting Peter, I had resorted to using it in a hacky way. I just used the regex below in the main expression field and it worked. The regex’ I showed were created with RegexBuddy, Boost Regex version 1.76 and then checked online too.

^[\t\v]+[^\n\r]+

But your solution is more inline with how the feature is supposed to be used. My result looks like this.

Since I wasn’t very hopeful to get some help on this I was already looking how I coould improve my hacky way, but thank you sir, you have given me a much better starting point!

PeterJones · Dec 11, 2023, 3:41 PM

@Khundian-Twitch ,

I had a few more minutes, and was playing around trying to get the classRange to work as well.

The “Begin XXX … End” will define a class named XXX
Things starting with “If” or “ElseIf” will be called “functions” inside that class, but the “if” or “elseif” will not be displayed in the functionList panel

Begin One

    If x
        a
    Elseif y
        b
    Else
        c
    Endif

End

Begin huh

    If something
    Elseif what
    Else
    Endif

End

<?xml version="1.0" encoding="UTF-8" ?>
<NotepadPlus>
    <functionList>
        <parser
            displayName="KT-UDL"
            id         ="ktudl_fn"
            commentExpr="(;.*?$)"
        >
            <function
                mainExpr="(?i:^\h*(If|Elseif).*?$)"
                displayMode="$functionName">
                <functionName>
                    <nameExpr expr="(?i:^\h*(If|Elseif).*?$)"/>
                </functionName>
            </function>
            <classRange
                mainExpr    ="(?s-i)^\h*Begin\b.*?^\h*End\b"
            >
                <className>
                    <nameExpr expr="^\h*Begin\h*\K.*?$" />
                </className>
                <function
                    mainExpr="(?i:^\h*(If|Elseif).*?$)"
                >
                    <functionName>
                        <funcNameExpr expr="(?i:^\h*(If|Elseif)\h*\K.*?$)" />
                    </functionName>
                </function>
            </classRange>
        </parser>
    </functionList>
</NotepadPlus>

… Essentially, when I’m trying to come up with a functionList definition, I pare it down to the absolute simplest I can, try to get just the function names to work, then try to wrap a classRange around that function name. Once I’ve got a minimum working functionList that’s got the main idea of the given language, then I start adding back in restrictions and edge cases one at a time, making sure I have the ability to go back if it stops matching correctly.

Khundian Twitch · Dec 11, 2023, 3:46 PM

@PeterJones

You’re a star! That looks awesome!
If you’re ever in Belgium I owe you some beers! :)