StrResolve: String Variables Resolving Plugin (for masktools expressions)

[martin53]

I was always bothered that especially masktools expr="..." strings were tedious to read and manage as soon as externally calculated variables are involved.

E.g. when gamma is a float variable like
gamma = 1.2
and you want to use gamma in a mt_lut expression
you need to write
gamma = 1.2
"(( x/255 ) ^ " + gamma.string + ") * 255"
\.mt_polish

... which becomes soon hard to read with a bit more sophisticated expressions, especially with parentheses and quotes counting.

Here you'll find a string helper plugin StrResolve (http://www.mediafire.com/file/wu6zexhssd0y77i/170310_StrResolve_v050.zip).
It takes a string and parses words prefixed by a prefix string if they are existing AviSynth variables in the scope of the plugin. If so, it retrieves their value, converts them to string if neccessary and inserts this in the place of the word. Otherwise an error is issued.
The benefit over the previous solution "...".ReplaceStr("gamma", gamma.string) is that it is universal, i.e. it recognizes all variables, regardless of their names, and needs only be called once in case of multiple variables.

With the plugin, it's

gamma = 1.2
expr = "(( x/255 ) ^ $gamma) * 255"
\.StrResolve.mt_polish

i.e. you just use AviSynth variables much like constants and operators inside the expression string.

The documentation is in an example AVS file in the ZIP.

v.0.5.0, follows the semver (semantic versioning) standard and other plugin writers are encouraged to adopt this, and copy the compile time formatter to automate yyyy/mm/dd-comparable compile time documentation.

Also, it features an # inline comment removal and multi line->single line reformatting, to make masktools' expressions even more human readable.

Finally, I could not resist and included a feature to not only insert values from variables, but also enter script snippets between curly brace twins: {{...}} which are executed with Eval("...") and the result is also inserted in the place of the code. This should enable most everything inside expression strings.

Edit March 10, 2017: v0.5.0, as described above. Testers welcome, probably not yet bug free. :rolleyes: Is the name still OK or are there better proposals?

[real.finder]

just download and see there are no dll, just source code

[raffriff42]

I embed variables in expressions like this... s = "((x/255)*$a)^(1/$g)*255+$b"
\ .StrReplace (https://forum.doom9.org/showthread.php?p=1540504#post154050)("$a", String(gain))
\ .StrReplace("$g", String(gamma))
\ .StrReplace("$b", String(bias))
\ .mt_polish

[martin53]

Currently the masktools reserved words are case insentitively not resolved.
It would be possible to e.g. not resolve x, but to check if X can be resolved, if that is preferred. That way, lowercase letters could be used to identify masktools placeholders and reserved words, uppercase letters to identify variables (or vice versa, but only one alternative).

(EDIT 2017/03/10 no more true for current version)

[TheFluff]

I'm probably a fool for asking this, but whyyyyyyy. Why, doom9 forums poster martin53? Why did you do it? Why did you rethink that which had no need of being rethunk? Why did you write code? Much like you shouldn't reinvent wheels (especially not square ones), you shouldn't solve solved problems. I have an old friend who is hurt and confused by your actions, and I think you should apologize to him. His name is sprintf, and I really don't think he's ever done anything to hurt you.

If you had just made a plugin that was nothing more than a thin wrapper around sprintf, he would have been good to you. He is old, but he's reliable! Many hundreds of millions - maybe even billions - of users have trusted him to do the right thing in the past. If you let him solve your problems, you could have had many nice things that people have enjoyed for many years, now. Formatting floats with arbitrary precision! Converting floats to integers! Re-ordering arguments! All kinds of nifty stuff! The language interpreter itself could handle mistyped variable names! You could separate data from code! You wouldn't even have had to do anything, my old buddy would have handled it all for you! Why did you have to betray him like this?

(it's actually really weird that avisynth doesn't actually seem to have have sprintf in the language already - it's even exposed as a function in the public api. there's a dumb one-variable sprintf in the String() function but idgi, why would you even do that and not have the real thing)

[martin53]

Very practical M53, maybe would be nice to have that in the avs+ core.

I have to admit that there is maybe no other use for this little thingy than masktools' expr string preparation. - Maybe xyremap though I should pay respect for it's reserved words too in this case.

Eval("...") and GScript("...") directly work on interpreter level and I currently can't imagine a use case where it is useful to resolve a variable name before interpreting the surrounding script source text.

(I have the hidden thought in my mind that the core functionality is transferred to mt_lut...()'s processing when it has no clue about the next token, and it was a finger exercise to not only post a suggestion, but give a fully working implementation of the proposal. But I don't want pinterf to just copy and include it: rather would I fork his Github and then make a pull request with it for the addition as another finger exercise :) )

[StainlessS]

It kinda looks sexy to have arg names embedded in the string.
I really do dislike the string concatenation stuff and the String() use as usually used in AVS, I think I did look at fluffy suggested
stuff in API and did not like the 4KB limitation, so RT_String() dont use it (neither do RT_SubTitle nor RT_WriteFile nor RT_DebugF),
but for most small case use I guess the fluffy one is correct, and the API would quite suffice.

[martin53]

RT_String() vs. StrResolve() functionalities are very different. Taking Raffriff42's example
"((x/255*gain)^(1/gamma)*255+bias)",

that would make the doubtlessly nice approach
gain = 1.0
gamma = 1.0
bias = 1.0
expr = "((x/255*%f)^(1/%f)*255+%f)"
\.RT_String(gain, gamma, bias)
compared to
gain = 1.0
gamma = 1.0
bias = 1.0
expr = "((x/255*gain)^(1/gamma)*255+bias)"
\.StrResolve

StrResolve checks for existing variables in its scope, which is much different even from parsing alone. Can sprintf() be used as a parser (OK, aside from the format string by itself)? I'd be glad to learn.

Btw. everyone is free to convert float variables into strings in advance. Then StrResolve takes the strings, of course. See AviSynth's String(float / int [, string format_string]) function. Just change
gain = 1.0
to gain = (1.0).String("%.10f")
but don't forget AviSynth stores in 32 bits! 420.3 is wrong in the 5th digit after the decimal point already.

[TheFluff]

oookay, you asked for it.

StrResolve checks for existing variables in its scope, which is much different even from parsing alone. Can sprintf() be used as a parser (OK, aside from the format string by itself)? I'd be glad to learn.
It checks for variables in its scope and SILENTLY DOES THE WRONG THING IF THEY ARE NOT DEFINED. Your software is installing delayed bugs. In your particular use case that you've hardcoded yourself into (why bind so hard to this exact masktools version? why do you hate robust software?) you're unlikely to not notice this, however, since you've neatly swept the problem under the rpn compiler's carpet, but still, it's so backwards (and did I mention other versions? what if someone adds reserved words? how enjoyable do you find the idea of keeping a repository of five different versions of this? masktools is a very widely forked piece of software). Handle your own darn errors, son!

Did you read the part where I said the language interpreter could do this for you? Not only does it also consider global variables, it will do the right thing and immediately detect your bugs for you by throwing an exception if you pass a variable that doesn't exist. Have you people never heard of the concept of a "type error"? Speaking of type errors, your implementation also does the wrong thing silently with things that aren't strings.

And what do you even mean by a "parser"?

but don't forget AviSynth stores in 32 bits! 420.3 is wrong in the 5th digit after the decimal point already.
Yes, very good, rounding is a separate problem and 32-bit float is bad. We're making progress. Unfortunately though string() still only takes one argument and doing the conversion yourself for five variables is still obnoxious. And you're even invoking it yourself internally, so why not accept a format string while you're at it?

If you want variable interpolation in strings, write Perl, for chrissakes, but not even Perl accepts random words as identifiers without a marker of some kind, and it has delimiters and escaping for this kind of thing.

Really, if your goal is to make rpn with embedded variables readable, what you want is a templating engine that supports inline comments. Five people probably wrote one in Javascript this week - don't get left behind! If you don't want to be that hipster, an sprintf implementation that supports named variables (such things exist! mindblowing, huh!) would be an acceptable substitute.

[StainlessS]

Martin53, I had not downloaded your source, when I had tried, I got a 'MediaFire is having Problems' or similar type message, downloaded now and had a peek.

Its a pleasure to see that you are such a tidy coder (so much tidier and less messy than that DavidHorman, lest wise that what he tells us, we never get to actually see if he tells true on that :) ).

Anyways, the fluffy one again seems to have cast his evil spell, and perhaps rightly so (dont you just hate it when that happens).

If you want to put out a mod of RT_String, you are welcome, perhaps you could improve and convert to use some of the M$ more secure
(though less portable) library functions. I've ripped out the 'esc' arg, just allowed escape sequences in data strings which I'm quite sure
that you would not want.

There are likely some things that are not needed for MaskTools type use, (eg %X hex), take them out if you wish or no harm to leave in,
the Sprintf thng is only really of use for devs in creating AVSValue strings for return to the user, not really suited for use by the user
in a String() type function.
Here supplied code allows parsing of the user supplied format string and args list.
If you improve on it, I'm bound to steal some of it back off you.

I Suggest the name FString(), but your call.

AVSValue __cdecl RT_String(AVSValue args, void* user_data, IScriptEnvironment* env) {
char *myName="RT_String: ";
const char *s = args[0].AsString(); // text
int arrsz = args[1].ArraySize();

enum {
CHICKEN=64
};

// what size buffer we need ?
int i,mem=strlen(s) + 1 + CHICKEN;
for(i=0;i<arrsz;++i) {
if(args[1][i].IsString()) {
const char *st=args[1][i].AsString();
mem += strlen(st) + 1 + CHICKEN;
} else {
mem += 8 + CHICKEN; // no particular reason why so big, just chicken factor.
}
}

char *pbuf = new char[(mem+1)*2];
if(pbuf==NULL)
env->ThrowError("%sCannot allocate memory",myName);

char *ptem=pbuf+(mem+1); // temp buffer

const unsigned char * r= (const unsigned char *)s;
char *p=pbuf;
int c,ix=0;
int t=0;

// Parse text and insert variables
while(c=*r) {
if(c=='%') {
++r;
if(*r=='\0') {
*p++ ='%';
} else if(*r=='%') {
*p++=*r++; // replace escaped double % with single
} else {
if(ix>=arrsz) {
delete [] pbuf;
env->ThrowError("%sExpecting data arg (%d)",myName,ix+1);
}
char *tp=ptem;
*tp++='%';
if(*r=='-' || *r=='+' || *r=='0' || *r==' ' || *r=='#') // flags
*tp++=*r++;
if(*r=='*') { // int holds length
t=args[1][ix].IsBool() ?1: \
args[1][ix].IsString() ?2: \
args[1][ix].IsInt() ?3: \
args[1][ix].IsFloat() ?4: \
0;
if(t!=3) {
delete [] pbuf;
env->ThrowError("%sUnsupported data type, Expecting Width as Int (%d)",myName,ix+1);
}
tp+=sprintf(tp,"%d",args[1][ix].AsInt()); // EDIT: FIXED: WAS IsInt
++r; // skip '*'
++ix; // next data
} else {
while(*r>='0' && *r<='9') {
*tp++ = *r++;
}
}
if(*r=='.') {
*tp++ = *r++; // precision prefix
if(*r=='*') { // int holds length
t=args[1][ix].IsBool() ?1: \
args[1][ix].IsString() ?2: \
args[1][ix].IsInt() ?3: \
args[1][ix].IsFloat() ?4: \
0;
if(t!=3) {
delete [] pbuf;
env->ThrowError("%sUnsupported data type, Expecting Precision as Int (%d)",myName,ix+1);
}
tp+=sprintf(tp,"%d",args[1][ix].AsInt());
++r; // skip '*'
++ix; // next data
} else {
while(*r>='0' && *r<='9') {
*tp++ = *r++;
}
}
}
t=args[1][ix].IsBool() ?1: \
args[1][ix].IsString() ?2: \
args[1][ix].IsInt() ?3: \
args[1][ix].IsFloat() ?4: \
0;
// type
if( (*r=='c' ) || (*r=='C' ) || // char as int
(*r=='d' || *r=='i') || // int
(*r=='o' || *r=='u' || *r=='x' || *r=='X')) { // unsigned int
if(t!=3) {
delete [] pbuf;
env->ThrowError("%sType='%c', Expecting Int data (%d)",myName,*r,ix+1);
}
*tp++=*r++;
*tp='\0';
p+=sprintf(p,ptem,args[1][ix].AsInt());
++ix; // next data
} else if(*r=='e' || *r=='E' || *r=='f' || *r=='g' || *r=='G') { // double
if(t!=4&&t!=3) {
delete [] pbuf;
env->ThrowError("%sType='%c', Expecting Float (%d)",myName,*r,ix+1);
}
*tp++=*r++;
*tp='\0';
p+=sprintf(p,ptem,args[1][ix].AsFloat());
++ix; // next data
} else if((*r=='s')||(*r=='S')) { // string
if(t!=2&&t!=1) {
delete [] pbuf;
env->ThrowError("%sType='s', Expecting String (%d)",myName,ix+1);
}
*tp++=*r++;
*tp='\0';
if(t==1) { // Bool
p+=sprintf(p,ptem,args[1][ix].AsBool()?"True":"False");
} else { // String
p+=sprintf(p,ptem,args[1][ix].AsString());
}
++ix; // next data
} else {
delete [] pbuf;
env->ThrowError("%sUnknown format type '%c' (%d)",myName,*r,ix+1); // eg Clip
}
}
} else if(c == '\\') {
++r;
c=*r;
// abfnrtv
switch (c) {
case '\0' : *p++='\\'; break; // copy single backslash at end of string
case '\\' : *p++=*r++; break; // replace double backslash with single backslash
case 'n' : ++r; *p++='\n'; break;
case 'r' : ++r; *p++='\r'; break;
case 't' : ++r; *p++='\t'; break;
case 'v' : ++r; *p++='\v'; break;
case 'f' : ++r; *p++='\f'; break;
case 'b' : ++r; *p++='\b'; break;
case 'a' : ++r; *p++='\a'; break;
default : *p++='\\'; *p++=*r++; break; // anything else we copy backslash and whatever follows
}
} else {
*p++=*r++;
}
}
*p=0; // nul term

if(ix<arrsz) {
delete [] pbuf;
env->ThrowError("%sUnexpected data arg (%d)",myName,ix+1);
}

AVSValue ret = env->SaveString(pbuf,p-pbuf);
delete [] pbuf;
return ret;
}



extern "C" __declspec(dllexport) const char* __stdcall AvisynthPluginInit2(IScriptEnvironment* env) {
env->AddFunction("RT_String", "s.*",String, 0);
return "`RT_String' RT_String plugin";
}


Do with it as you will, some might like a small stringy thingy :)

EDIT: BUGFIX IN BLUE

[feisty2]

It's weird tho, cuz I'm also a fan of the c++ programming language and somehow StainlessS' code always looks like machine code to me..

enum {
CHICKEN = 64
};

seriously????? That's the way you abuse enums?
Can't you just make it readable, make it c++ like

constexpr auto chicken = 64;

?????????

Edit: you people are responsible for all the c++ haters out there, it's all that weird abusing shit that scared lots of people away

[wonkey_monkey]

How about an option to only replace variables when they are prefixed with $? Or just allow an optional user-specified prefix.

I will probably code something like this into future versions of xyremap/pixel_rpn etc. Strangely I had need of exactly such a thing for the first time just two days ago, though I didn't see the opportunity for doing it in this way, so I had a lot of string(...)s.

[martin53]

How about an option to only replace variables when they are prefixed with $? Or just allow an optional user-specified prefix.
Funny, exactly the same thing came to my mind last night. v0.3 is on mediafire, see 1st post.
Pls read the AVS file.

- Selectable prefix char added, defaults to ':' (EDIT: default "$" in later versions) which I judge unobtrusive and is not yet used in conflicting manners in today's RPN. unfortunately I named the parameter 'trig' instead of the obvious 'prefix' :o And if the majority of users prefers '$' I'll change it.
- additional switch 'strict' added to adapt to user's intention of behaviour.

[martin53]

(1) hardcoded ... this exact masktools version ... what if someone adds reserved words ...

(2) Did you read the part where I said the language interpreter could do this for you? Not only does it also consider global variables, it will do the right thing and immediately detect your bugs for you by throwing an exception if you pass a variable that doesn't exist.

(3) "type error"


Some reasonable objections here. It's fine to discuss proposals among people, makes them better in the end.

(1) though I originally had a 'clean' RPN string in my mind, similar to a script source text, and personally don't like script languages that require prefix characters for variables, there are indeed good reasons to vote for a prefix; mostly the fact that the writer's intention is known, and an error can be issued at all.

(2) I don't see how the script language interpreter can be of help here; maybe I lack some knowledge. The env->GetVar() function does consider local and global variables. The plugin now throws an exception by default, too. In the specific case of masktools' expressions, this is of limited value, because masktools itself silently breaks RPNs it cannot interpret.
Regarding sprintf() once more: I looked at a stackoverflow.com example with named placeholders. But that approach will always have two occurences of the varying value: the placeholder and the actual variable. For someone like me that's fine in a text string (my mind is able to process that), but I don't like it in a program language source text or in a complex math formula. There I prefer the direct occurence of surrogates, like the most famous x variable.
y = sin(x) <-> %y = sin(%x), $y, $x <-> %f = sin(%f), $y, $x

(3) many programming languages offer automatic type conversion as a convenience, I wish AviSynth would.

[pinterf]

Discuss and clean up the idea, I was thinking about similar things, but in connection of the possible integration of GScriptclip in avs+ core, but without specifying the parameter list string in a separate parameter. A more general method is always welcomed.

[martin53]

These 0x20...0x7e ASCII characters are currently not used in RPN formulas:

$,:;\_`{}

$ is already used in some other programming languages as a prefix for variable names. In AviSynth, it is the hex number prefix. It looks like AviSynth hex numbers always allow a leading zero, even if additionally prepended. This would be a possible and intuitive exact distinction from names of variables. (current RPN interpreters can not interpret hex numbers, but they might learn to)

: though not directly used in RPN, it is part of the ternary Elvis operator. "x<16?y:x" is a valid infix expression and collides with variable name x. With given infix formulas, StrResolve must be executed before mt_polish (or mt_polish must learn how to transform the expressions, which I don't prefer). Although it is possible to write "x<16?y: x" instead, the mandatory space character is an unelegant syntactical requirement.

, and ; these characters typically are found at the end of expressions or separators, it feels unusual to use them as prefixes.

\ could be found in path strings, which is currently not an issue with RPN, but maybe restricts the function otherwise.

_ not nice because it is also a valid variable name character. Variable X outside the expression is _X inside, _X outside is __X inside etc. - confusing.

` is not yet used and does work. Is it present on many language keyboard layouts. One reason not to use it might be the fact that it seems to be last character that can be used as an escaping character one day, because it is yet unused. I think it is even already used for escaping or so in linux?

{} Curly braces come as a pair. It looks too ugly to write {x for variable x, and also to write }x. I had thought of setting {x} as syntax, but prefer a single prefix, because it's shorter.

Conclusion
I think I shall set $ the standard prefix character and rename the parameter to prefix in the next plugin version. Objections welcome.

[wonkey_monkey]

$,:;\_`{}

For what it's worth, : is used in my RPN compiler as a shortcut for "else" and also as a prefix for labels. I think I've used \ as a reverse divide (to avoid having to include a swap), as well, and \ is problematic in other ways, since it's a prefix for control characters (it doesn't appear in forum comments if you type it between speech marks: "").

ternary Elvis

Autocorrect?

I think $ is the obvious choice, and I'm not just saying that because I write a lot of PHP.

[raffriff42]

It can help to use source-clip names in expressions too, especially for mt_lutxyz:# http://avisynth.nl/index.php/Color_conversions
#
# G = Y - 2*(U-128)*(1-Kb)*Kb/Kg - 2*(V-128)*(1-Kr)*Kr/Kg
#
sG = ("$Y-(2*($U-128)*(1-$Kb)*$Kb/$Kg)-(2*($V-128)*(1-$Kr)*$Kr/$Kg)")
\ .StrReplace("$Y", "x")
\ .StrReplace("$U", "y")
\ .StrReplace("$V", "z")
\ .StrReplace("$Kr", sKr) [* 0.2126 for PC.709 *]
\ .StrReplace("$Kg", sKg) [* 0.7152 *]
\ .StrReplace("$Kb", sKb) [* 0.0722 *]
\ .mt_polish
G = mt_lutxyz(Y, U, V, sG)

Ternary Elvis, that's great! loveme ? now : never

[TheFluff]

(2) I don't see how the script language interpreter can be of help here

If your function takes input by the way of arguments:
value = doSomething(argument)
then you automatically get name and type checking for free (name checking as in checking that whatever variable/function identifier you may be passing is declared in the current scope, at compile time). Avisynth functions can actually have typed signatures, y'know. Not really relevant for a sprintf-like thing though because you kinda want to support an arbitrary number of arguments with arbitrary types.

If you on the other hand take input by inspecting the current state of the calling scope, you need to do this name and type checking on your own. I haven't looked at your latest version but you previously used this implementation:
int error = -1;
if (!isReservedWord(token))
error = env->GetVarDef(token, AVSValue(-1)).AsInt(); // returns -1 when var in token doesn't exist
if (error == -1) { // the token is reserved or not a variable
// append token as is to output string
strncat_s(resolvedString, 4096, token, strlen(token));
}
else {
...
which makes you unable to distinguish between a variable set to -1 and a name error because GetVarDef would return AVSValue(-1) in both those cases. If you want to check for undefined, check for undefined, not for negative integer one. That still doesn't let you distinguish between a declared variable with an undefined value and a name error, though. You can't actually get working name checking via the API.

[martin53]

If your function takes input by the way of arguments:
value = doSomething(argument)
then you automatically get name and type checking.

... on the other hand ...
env->GetVarDef(token, AVSValue(-1)).AsInt();

makes you unable to distinguish between a variable set to -1 and a name error because GetVarDef would return AVSValue(-1) in both those cases. ... You can't actually get working name checking via the API.

Thank you for the explanation!

@ALL
The described behaviour sounds weird, but I must confirm: StrResolve is currently unable to resolve variables with value -1 because although they exist, the API function GetVarDef() - remember, the function which is made to tell a plugin if a variable exists - lies.

Well, if the script writer needs to enter the variables that are used in the infix or RPN formula as additional parameters to the plugin, there is almost no ergonomic benefit left. Let's see if at least AviSynth+ can be improved here etc.
EDIT: This was clarified later: GetVarDef() isn't made to tell if a variable exists, but avoid an exception if it does not exist. Still, by two calls with different defaults, the plugin can make a decision if the variable really exists (both calls return the same value).

Different topic: I'm also a little concerned about the general version control (or better, its complete absence) in current worldwide spread plugins and scripts, and did improve the VersionInfo function to a state where I think I can suggest other plugin writers to copy it and use it as the least standard.
Published scripts should then get the version info and string-compare it to a literal minimum version. The string has a free 1st part (for authors who love version 0.0.0.01.alpha), proposedly terminated by a space character, then a YYYY/MM/YY hh:mm:ss compile time stamp which is calculated automatically for those autors typically forgetting to increment the manually maintained version string.
I encourage script authors do do such version compare in scripts they publish for others, to enhance the overall user experience with AviSynth.

[TheFluff]

@ALL
The described behaviour sounds weird, but I must confirm: StrResolve is currently unable to resolve variables with value -1 because although they exist, the API function GetVarDef() - remember, the function which is made to tell a plugin if a variable exists - lies.

It doesn't lie, you've just misunderstood what the intended behavior is (not that weird, the documentation isn't super helpful). It's not for checking if a variable is declared or not - in fact, it cannot actually be used do that*. What it actually does is the exact same thing as GetVar, except that if the given identifier hasn't been declared in the current scope, it does not throw an exception, but instead returns the provided default value. If you just want to check if the variable is set to anything at all, pass AVSValue() as the second argument (which gives you an AVSValue with type "void", or undefined) and check if the value you get back is .Defined().

That doesn't actually tell you if the variable is set to anything useful, though. That's an entirely separate question.

It's also worth noting here that in Avisynth script, functions are not a first class type and reside in a separate namespace (c.f. env->FunctionExists).

Different topic: I'm also a little concerned about the general version control (or better, its complete absence) in current worldwide spread plugins and scripts, and did improve the VersionInfo function to a state where I think I can suggest other plugin writers to copy it and use it as the least standard.

There is a standard for semantic versioning that everyone uses. It's called semver (http://semver.org/), which stands for "semantic versioning". How convenient is that?


* Actually, when I thought about it I can think of a way to do name checking, but you'd need two calls to GetVarDef and I'm not sure if it even works. Passing AVSValue() as the default won't let you distinguish between "variable exists in scope but has the undefined value" and "variable does not exist", so you'd need to check for something else as well after testing for undefined. For example (not tested):
!(env->GetVarDef("foo", AVSValue()).Defined()) && env->GetVarDef("foo", AVSValue(true)).AsBool()
(might need a default arg on that .AsBool(), not sure, and I dunno if that's an avs+ extension either)

[Gavino]

You can't actually get working name checking via the API.
Not via GetVarDef() you can't (or at least not unless you use your trick of calling it twice with different defaults), but you can just use plain old GetVar() and check for the NotFound exception which was the standard way of accessing variables before v2.6.

Actually, it's perhaps not surprising that martin53 misunderstood the intention of GetVarDef() - his code (including the comment) is taken directly from the bad example (http://avisynth.nl/index.php/Filter_SDK/Cplusplus_API#GetVarDef.2C_v6) in the wiki, which suggests its purpose is to check for name existence, when in fact it is provided merely as a convenient alternative to GetVar() for the cases when a default value can be used if the variable doesn't exist.

[TheFluff]

Not via GetVarDef() you can't (or at least not unless you use your trick of calling it twice with different defaults), but you can just use plain old GetVar() and check for the NotFound exception which was the standard way of accessing variables before v2.6.
Yes, you can do that. I didn't mention it, though, because to accomplish it you need to summon Great Evil (in the form of inexplicable crashes and memory leaks) by attempting to catch a C++ exception thrown from a different DLL with a (potentially) different runtime environment. In this case that might actually work and not even leak a noticeable amount of memory, which might lead people to think that it is actually a thing you can reasonably do, and that would be very bad. It took a very, very long time to get Avisynth plugin developers to even realize how awful it was and even today I don't think it's well known - people have just gotten used to bizarre crashes.

(env->ThrowException() from a plugin is safe; it's thrown from within the Avisynth DLL and also handled there. Plain throw isn't - you must never allow an exception thrown from somewhere in your DLL to escape. Attempting to catch exceptions thrown by the Avisynth environment from within your plugin isn't safe either.)

e: see http://avisynth.nl/index.php/AviSynth%2B#AviSynth_Plugin_Writing_Tips - the part about exceptions applies to all avisynth versions and in fact to all C++ DLL's in general.

Actually, it's perhaps not surprising that martin53 misunderstood the intention of GetVarDef() - his code (including the comment) is taken directly from the bad example (http://avisynth.nl/index.php/Filter_SDK/Cplusplus_API#GetVarDef.2C_v6) in the wiki, which suggests its purpose is to check for name existence, when in fact it is provided merely as a convenient alternative to GetVar() for the cases when a default value can be used if the variable doesn't exist.

Ah, that would make a lot of sense.

[martin53]

This seems to work
bool test1 = env->GetVarDef(strToken, AVSValue(false)).AsBool();
bool test2 = env->GetVarDef(strToken, AVSValue(true)).AsBool();
if (test1 != test2)
env->ThrowError("...");

plugin is updated, but I lack the time to publish the update today

P.S. would you like to be allowed to write

"""
x 255 / // normalize x to 0...1
$gain * // apply gain
1 $gamma / // prepare 1/gamma
^ // apply power operator
255 * // rescale to full range
$bias + // and finally add bias
""".StrResolve

and get a valid RPN string with the c++ comments removed and linebreaks replaced by spaces?

[wonkey_monkey]

and get a valid RPN string with the c++ comments removed and linebreaks replaced by spaces?

Those should perhaps be optional features (pixel_rpn also allows # to declare comments, so if they weren't stripped and newlines were replaced with spaces, things would break). Or is this intended only to ever be guaranteed compatible with masktools? Which is fair enough, can't go around imagining what everyone will ever do with RPN.

Does masktools already treat linebreaks as normal whitespace? If so then I don't see a need for that replacement, since the string will work with newlines anyway.

[martin53]

Does masktools already treat linebreaks as normal whitespace? If so then I don't see a need for that replacement, since the string will work with newlines anyway.
No, masktools fails with linebreaks. I made the comment separator a string parameter and asked pinterf, if he agrees to define new names for #B and #F, so we can use '#' as compatible inline comment separator.

EDIT: Default comment separator is now #, pinterf agreed to change the two operators

[martin53]

Just a post to point out the new edit in post #1

[StainlessS]

Source code in this earlier post was erroneous (fixed in post):- https://forum.doom9.org/showthread.php?p=1799906#post1799906

Alternative source and dll posted here:- https://forum.doom9.org/showthread.php?p=1808978#post1808978