ZScript is a new (circa 2017) scripting language that has sprung from the ceasing of ZDoom and the subsequent reprisal of GZDoom as mainline. It is similar to Java, though it has many deficiencies, oddities and other such issues. Despite this, it is still the most powerful Doom modding tool since straight up source editing, and will likely stay that way for a while until Eternity Engine inevitably becomes competition-worthy with scripting additions.
This documentation serves as an introduction to and informal specification of the ZScript language from a programmer's viewpoint. It should also be useful for non-programmers looking for specifics on the inner workings of the language and more information on the functions and properties provided to it.
ZScript runs in a virtual machine much like ACS, although because it is *not* compiled to bytecode and uses an object-oriented structure, the virtual machine is far more complex, and also therefore quite a bit slower. ZScript may only be read from source files by the engine, which has several benefits as well as detriments. It is the opinion of the author that this is a bad solution, but the author will refrain from going on a several-paragraph tirade about why bytecode is always better than source, even if it is an optional component.
In any case, here we are. This documentation will detail all aspects of ZScript, from the language and type system to the API and finer details. This document is distributed under the [CC0 public domain license](LICENSE.txt) in the hope that it is useful reference and serves as a solid basis for further writings. This document was originally written by Alison Sanderson (Marrub.) Attribution is encouraged but not required.
This document's syntaxes are written in a specific way to be easy to read but still close enough to a formal syntax that, for instance, someone writing a parser could do so off of this document. Here is a legend describing all syntax element spellings:
| Spelling | Meaning |
| -------- | ------- |
| Keyword | Any keyword with spaces around it is spelled as-is. |
| Symbol | Any symbol with spaces around it is spelled as-is, the whitespace is only for clarity and may be omitted. |
| `Syntax` | A syntax element defined by this document. Spelled as according to its grammar. |
| `Syntax...` | A syntax element of which there may be any amount of. Spelled as according to its grammar. |
| `Syntax{N}` | A syntax element of which there may be exactly N amount of. Spelled as according to its grammar. |
| `$[` and `]$` | An optional syntax element, which may be omitted by the user. |
| `"text"` | Any string literal, contents do not necessarily have to be what is inside unless explicitly stated. |
2018-11-18 08:44:41 -08:00
Translation Unit
================
Full ZScript files are referred to as "translation units." This terminology comes from the C standard, and refers simply to the entirety of a ZScript source file. ZScript files are looked for in lumps named `zscript` with any extension. The standard extension is `.txt`, but `.zsc` and `.zs` are common as well. The author of this documentation prefers `.zsc`.
The base translation unit `zscript` may start with a version directive, then followed by any number of top-level definitions and `#include` directives. Included translation units may not have version directives.
All keywords and identifiers in ZScript are case insensitive.
Versions
========
A version directive may be placed at the very beginning of a ZScript file, the syntax being:
```
version "num"
```
Where `num` is the ZScript version to use. By default ZScript is `version "2.3"`, the original ZScript specification. This old version is not supported by this documentation and it is highly encouraged to always use the latest version of ZScript. The minimum version supported by this documentation is 3.0.
Top-level
=========
A ZScript file can have one of several things at the top level of the file, following a version directive:
- Class definitions
- Structure definitions
- Enumeration definitions
- Constant definitions
- Include directives
Class definitions
=================
A class defines an object type within ZScript, and is most of what you'll be creating within the language.
All classes inherit from other classes. The base class can be set within the class header, but if it is not the class will automatically inherit from Object.
Classes are subject to Scoping. They are also implicitly reference values, and therefore can be null. Use `new` to instantiate a new class object.
Classes that inherit from Actor can replace other actors when spawned in maps, and can also be used freely in `DECORATE`. Actors have states, which will not be explained in this document as they are already well-documented on the ZDoom wiki.
Property definitions are used within classes to define defaultable attributes on actors. They are not valid on classes not derived from Actor.
When registered, a property will be available in the `default` block as `ClassName.PropertyName`. Properties can be given multiple members to initialize.
Property definitions take the form `property Identifier : Member $[ , Member]$... ;` (where `Member` is an identifier naming any member in the current class.)
2018-11-18 08:44:41 -08:00
Properties defined in ZScript are usable from `DECORATE`.
Flag definitions are used within classes to define defaultable boolean flags on actors. They are not valid on classes not derived from Actor.
When registered, a flag will be available in the `default` block as `CLASSNAME.FLAGNAME`, as well as a member as `bFLAGNAME`.
Each flag operates on a singular bit of any integer member of the class. The integer must be exactly 32 bits, though if it is signed or not does not matter. This means each backing integer can hold exactly 32 flags each (assuming no duplicated flags,) and more will require another one to be added. (Internally, the `Actor` flags are currently up to over 8 backing integers.)
A flag's backing integer may not be `meta`, although it may be `readonly`, `private`, or any other access modifier. The generated flag member will be publicly visible regardless of the backing integer's visibility.
Flag definitions take the form `flagdef Identifier : Member , Number ;` where `Number` is the bit in `Member` to use, starting from `0` and ending at `31`.
Flags defined in ZScript are usable from `DECORATE`.
2018-11-18 08:44:41 -08:00
## Default blocks
Default blocks are used on classes derived from Actor to create an overridable list of defaults to properties, allowing for swift creation of flexible actor types.
In `DECORATE`, this is everything that isn't in the `states` block, but in ZScript, for syntax flexibility purposes, it must be enclosed in a block with `default` at the beginning, formed:
These are the same as `DECORATE`, but states that do not have function blocks require terminating semicolons. Double quotes around `#` and `-` are no longer required. State blocks can be subject to Action Scoping with the syntax `states(Scope)`.
A state definition block has the syntax:
```
states $[ ( Scope ) ]$
{
$[State-or-label...]$
}
```
State-or-label either defines a state label or a state itself, with the syntax of:
Where `Char` is any ASCII character, `Number-or-random` is one of:
```
Number
random ( Number , Number )
```
`State-option` is one of:
```
bright
fast
slow
nodelay
canraise
offset ( Number , Number )
light ( String $[ , String]$... )
```
And finally, `State-func` is one of:
```
;
Identifier ( Argument-list ) ;
{ $[Statement...]$ }
```
The first will attach no action function to the state. The second will attach the specified action function with the specified arguments, and the third will create an anonymous action function and attach it.
2018-11-18 08:44:41 -08:00
Structure definitions
=====================
A structure is an object type that does not inherit from Object and is not always (though occasionally is) a reference type, unlike classes. Structures marked as `native` are passed by-reference to and from the engine in an implicit pseudo-type `Pointer<T>`, and `null` can be passed in their place. Also note that this means the engine can return `null` structures. Non-native structures cannot be passed as arguments or returned normally.
Structures are preferred for basic compound data types that do not need to be instanced and are often used as a way of generalizing code. They cannot be returned from functions.
| `clearscope` | Structure has Data scope. Default. |
| `native` | Structure is from the engine. Only usable internally. |
| `play` | Structure has Play scope. |
| `ui` | Structure has UI scope. |
| `version ( "ver" )` | Restricted to ZScript version `ver` or higher. |
2018-11-18 08:44:41 -08:00
## Structure content
Structure contents are an optional list of various things logically contained within the structure, including:
- Member declarations
- Method definitions
- Enumeration definitions
- Constant definitions
Enumeration definitions
=======================
An enumeration is a list of named numbers, which by default will be incremental from 0. By default they decay to the type `int`, but the default decay type can be set manually.
Optionally followed by a semicolon. `Integer-type` in this context is any valid integral type name.
Enumerators can either be incremental (from the last enumerator or 0 if there is none) or explicitly set with the syntax `Identifier = Expression`. Enumerators must be followed by a comma unless it is the end of the list.
Constants are not assignable. Their type is inferred from their value, so if you wish for them to have a specific type, you must cast the value to that type.
Static array definitions
=========================
Similar to constants, static arrays are named values, but for an array. They are created with the syntax:
Static arrays cannot be multi-dimensional, unlike normal fixed-size arrays.
2018-11-18 08:44:41 -08:00
Include directives
==================
Include directives include other files to be processed by the ZScript compiler, allowing you to organize and separate code into different files. Their syntax is simple:
```
#include "filename"
```
Note that included filenames will conflict with other mods. If two mods have a file named `zscript/MyCoolClasses.zsc` and both include it, expecting to get different files, the engine will fail to load with a script error.
To avoid this, it is suggested to place your ZScript code under a uniquely named sub-folder.
Types
=====
ZScript has several categories of types: Integer types, floating-point (decimal) types, strings, vectors, names, classes, et al. There are a wide variety of ways to use these types, as well as a wide variety of places they are used.
Types determine what kind of value an object stores, how it acts within an expression, etc. All objects, constants and enumerations have a type. Argument lists use types to ensure a function is used properly.
Most basic types have methods attached to them, and both integer and floating-point type names have symbols accessible from them. See the API section for more information.
## Integers
Integer types are basic integral numbers. They include:
| Name | Usable as argument | Bits | Lowest value | Highest value |
The `string` type is a mutable, garbage-collected string reference type. Strings are not structures or classes, however there are methods attached to the type, detailed in the API section.
The `name` type is an indexed string. While their contents are the same as a string, their actual value is merely an integer which can be compared far quicker than a string. Names are used for many internal purposes such as damage type names. Strings are implicitly cast to names.
The `color` type can be converted from a string using the `X11RGB` lump or a hex color in the format `#RRGGBB`, or with either `color(R, G, B)` or `color(A, R, G, B)`.
## Vectors
| Name | Usable as argument |
| ---- | :----------------: |
| `vector2` | Yes |
| `vector3` | Yes |
There are two vector types in ZScript, `vector2` and `vector3`, which hold two and three members, respectively. Their members can be accessed through `x`, `y` and (for `vector3`,) `z`. `vector3` can additionally get the X and Y components as a `vector2` with `xy`.
Vectors can use many operators and even have special ones to themselves. See the Expressions and Operators section for more information.
Fixed-size arrays take the form `Type[size]`. They hold `size` number of `Type` elements, which can be accessed with the array access operator. Multi-dimensional arrays are also supported. Note that this kind of type can also be declared in variable names themselves.
Dynamically sized arrays take the form `array<Type>`, and hold an arbitrary number of `Type` elements, which can be accessed with the array access operator.
Dynamic-sized arrays do not have their lifetime scoped to their current block, so:
Class type references are used to describe a concrete *type* rather than an object. They simply take the form `class`, and can be restrained to descendants of a type with the syntax `class<Type>`. Strings are implicitly cast to class type references.
Types prefixed with `@` are native pointers to objects (as opposed to objects placed directly in the structure's data.) This is not usable in user code.
2018-11-18 08:44:41 -08:00
A type name that is within a specific scope can be accessed by prefixing it with a `.`. The type `.MyClass.MySubStructure` will resolve to the type `MySubStructure` contained within `MyClass`.
A read-only type, as its name implies, may only be read from, and is effectively immutable. They take the form `readonly<Type>`. Do note that this is separate from the member declaration flag.
## Other types
| Name | Usable as argument | Description |
| ---- | :----------------: | ----------- |
| `bool` | Yes | Holds one of two values: `true` or `false`. |
| `sound` | Yes | Holds a sound reference. |
| `spriteid` | Yes | Holds a sprite reference. |
| `state` | Yes | A reference to an actor state. |
| `statelabel` | Yes | The name of an actor state. |
| `textureid` | Yes | Holds a texture reference. |
| `void` | No | Alias for `None`. Unknown purpose, likely implementation error. |
| `voidptr` | No | A pointer to a real memory address. Implementation detail. |
Strings will implicitly convert to `sound` and `statelabel`.
Variable names can have an array's size on them, instead of on the type, or none. Variable names are formed as either:
```
Identifier
Identifier Array-size
```
## Array size
Array sizes can be multi-dimensional or automatically sized, so all of the following syntaxes are available:
```
[ ]
[ Expression ] $[Array-size...]$
```
2018-11-18 08:44:41 -08:00
Expressions and Operators
=========================
## Literals
Much like C or most other programming languages, ZScript has object literals, including string literals, integer literals, float literals, name literals, boolean literals, and the null pointer.
### String literals
String literals take the same form as in C:
```
"text here"
```
String literals have character escapes, which are formed with a backslash and a character. Character escapes include:
| Spelling | Output |
| -------- | ------ |
| `\"` | A literal `"`. |
| `\\` | A literal `\`. |
| `\` followed by newline | Concatenates the next line with this one. |
To quote [cppreference](https://en.cppreference.com/w/cpp/language/escape), "of the octal escape sequences, `\0` is the most useful because it represents the terminating null character in null-terminated strings."
String literals, also like C and C++, will be concatenated when put directly next to each other. For example, this:
```
"text 1" "text 2"
```
Will be parsed as a single string literal with the text `"text 1text 2"`.
### Class type literals
Class type literals take the same form as string literals, but do note that they are not the same.
### Name literals
Name literals are similar to string literals, though they use apostrophes instead of quote marks:
```
'text here'
```
They do not concatenate like string literals, and do not have character escapes.
### Integer literals
Integer literals are formed similarly to C. They may take one of three forms, and be typed `uint` or `int` based on whether there is a `u` or `U` at the end or not.
The parser also supports an optional `l`/`L` suffix as in C, though it does not actually do anything, and it is advised you do not use it for potential forward compatibility purposes.
Integer literals can be in the basic base-10/decimal form:
```
1234567890 // int
500u // uint
```
Base-16/hexadecimal form, which may use upper- or lower-case decimals and `0x` prefix, depending on user preference:
```
0x123456789ABCDEF0
0XaBcDeF0 // don't do this, please.
0x7fff
0x7FFFFFFF
```
And, base-8/octal form, prefixed with a `0`:
```
0777
0414444
```
### Float literals
Float literals, much like integer literals, are formed similarly to C, but they do not support hex-float notation. Float literals support exponent notation.
The parser supports an optional `f`/`F` suffix as in C, though it does not actually do anything, and it is advised you do not use it for potential forward compatibility purposes.
Float literals can be formed in a few ways:
```
0.5 //=> 0.5
.5 //=> 0.5
1. //=> 1.0
```
And with exponents:
```
0.5e+2 //=> 50
50e-2 //=> 0.5
```
### Boolean literals
The two boolean literals are spelled `false` and `true`, and much like C, can decay to the integer literals `0` and `1`.
### Null pointer
The null pointer literal is spelled `null` and represents an object that does not exist in memory. Like C, it is not equivalent to the integer literal `0`, and is more similar to C++'s `nullptr`.
## Expressions
Expressions contain literals or other such *expressions* of objects, including arithmetic and various conditions.
### Primary expressions
Basic expressions, also known as primary expressions, can be one of:
- An identifier for a constant or variable.
- The `Super` keyword.
- Any object literal.
- A vector literal.
- An expression in parentheses.
Identifiers work as you expect, they reference a variable or constant. The `Super` keyword references the parent type or any member within it.
#### Vector literals
Vector literals are not under object literals as they are not constants in the same manner as other literals, since they contain expressions within them. As such, they are expressions, not proper value-based literals. They can be formed with:
All components must have type `double`, except in the second grammar where `X` is `vector2`.
2018-11-18 08:44:41 -08:00
### Postfix expressions
Postfix expressions are affixed at the end of an expression, and are used for a large variety of things, although the actual amount of postfix expressions is small:
Function calls may name arguments which have defaults with the syntax `Identifier : Expression`, possibly skipping over other defaulted arguments. After the first named defaultable argument, all other arguments must be named as well.
2018-11-18 08:44:41 -08:00
### Unary expressions
Unary expressions are affixed at the beginning of an expression. The simplest example of a unary expression is the negation operator, `-`, as in `-500`. Unary expressions include:
| `++ A` | Pre-increment. This adds 1 to the object and evaluates as the resulting value. |
| `-- A` | Pre-decrement. This subtracts 1 from the object and evaluates as the resulting value. |
| `~ A` | Bitwise negation. Flips all bits in an integer. |
| `+ A` | Affirmation. Does not actually do anything. |
| `alignof A` | Evaluates the alignment of the type of an expression. Unknown purpose. |
| `sizeof A` | Evaluates the size of the type of an expression. Unknown purpose. |
2018-11-18 08:44:41 -08:00
### Binary expressions
Binary expressions operate on two expressions, and are the most common kind of expression. They are used inline like regular math syntax, i.e. `1 + 1`. Binary expressions include:
| `A % B` | Division remainder, also known as "modulus." Unlike C, this works on floats, too. |
| `A ** B` | Exponent ("power of.") |
| `A << B` | Left bitwise shift. |
| `A >> B` | Right bitwise shift. |
| `A >>> B` | Right unsigned bitwise shift. |
| `A cross B` | Vector cross-product. |
| `A dot B` | Vector dot-product. |
| `A .. B` | Concatenation, creates a string from two values. |
| `A < B` | `true` if `A` is less than `B`. |
| `A > B` | `true` if `A` is greater than `B`. |
| `A <= B` | `true` if `A` is less than or equal to `B`. |
| `A >= B` | `true` if `A` is greater than or equal to `B`. |
| `A == B` | `true` if `A` is equal to `B`. |
| `A != B` | `true` if `A` is not equal to `B`. |
| `A ~== B` | `true` if `A` is approximately equal to `B`. For strings this is a case-insensitive comparison, for floats and vectors this checks if the difference between the two is smaller than ε. |
| `A && B` | `true` if `A` and `B` are both `true`. |
| `A \|\| B` | `true` if `A` or `B` is `true`. |
| `A is "B"` | `true` if `A`'s type is equal to or a descendant of `B`. |
| `A <>= B` | Signed difference between `A` and `B`. |
Assignment expressions are a subset of binary expressions which *are never constant expressions*. They assign a value to another value, as one might guess.
The ternary expression is formed `A ? B : C`, and will evaluate to `B` if `A` is `true`, or `C` if it is `false`.
2018-11-18 08:44:41 -08:00
Statements
==========
All functions are made up of a list of *statements* enclosed with left and right braces, which in and of itself is a statement called a *compound statement*, or *block*.
Note that the statement list is optional, so an empty compound statement `{}` is entirely valid.
## Expression statements
An expression statement is the single most common type of statement in just about any programming language. In ZScript, exactly like C and C++, an expression statement is simply formed with any expression followed by a semicolon. Function calls and variable assignments are expressions, for instance, so it is quite clear why they are common.
A switch statement takes an expression of integer or name type and selects a labeled statement to run. They work the same as in C and ACS:
```
switch ( Expression ) Statement
```
2018-11-18 08:44:41 -08:00
## Loop statements
ZScript has five loop statements, `for`, `while`, `until`, `do while` and `do until`. `for`, `while` and `do while` work the same as in C, C++ and ACS, while `until` and `do until` do the inverse of `while` and `do while`.
The `for` loop takes a limited statement and two optional expressions: The statement for when the loop begins (which is scoped to the loop,) one expression for checking if the loop should break, and one which is executed every time the loop iterates. Its syntax is:
```
for ( $[Expression-or-Local-variable-statement]$ ; $[Expression]$ ; $[Expression]$ ) Statement
```
2018-11-18 08:44:41 -08:00
The `while` loop simply takes one expression for checking if the loop should break, equivalent to `for(; a;)`.
The `until` loop is equivalent to `while(!a)`. Their syntax are:
```
while ( Expression ) Statement
until ( Expression ) Statement
```
2018-11-18 08:44:41 -08:00
`do while` and `do until` will only check the expression after the first iteration is complete. The `do while` and `do until` loops are formed as such:
As in C, there are three control flow statements that manipulate where the program will execute statements next, which are available contextually. They are `continue`, `break` and `return`.
`break` is available in loop statements and switch statements, and will break out of the containing statement early:
```
break ;
```
`return` is available in functions. If the function does not return any values, it may only be spelled `return;` and will simply exit the function early. If the function does return values, it takes a comma-separated list for each value returned:
Local variable statements are formed in one of 2 ways. The `let` keyword can be used to automatically determine the type of the variable from the initializer, while the regular syntax uses an explicit type, and initialization is optional.
Note that the surrounding brackets are literal and not an optional element.
2018-11-18 08:44:41 -08:00
## Static array statements
Static arrays can be defined normally as a statement.
## Null statements
A null statement does nothing, and is formed `;`. It is similar to an empty compound statement.
Member declarations
===================
Member declarations define data within a structure or class that can be accessed directly within methods of the object (or its derived classes,) or indirectly from instances of it with the member access operator.
| `meta` | Member is read-only static class data. Only really useful on actors, since these can be set via properties on them. |
| `native` | Member is from the engine. Only usable internally. |
| `play` | Member has Play scope. |
| `private` | Member is not visible to any class but this one. |
| `protected` | Member is not visible to any class but this one and any descendants of it. |
| `readonly` | Member is not writable. |
| `transient` | Member is not saved into save games. Required for unserializable objects and recommended for UI context objects. |
| `ui` | Member has UI scope. |
| `version ( "ver" )` | Restricted to ZScript version `ver` or higher. |
2018-11-18 08:44:41 -08:00
Method definitions
==================
Method definitions define functions within a structure or class that can be accessed directly within other methods of the object (or its derived classes,) or indirectly from instances of it with the member access operator.
Methods marked as `virtual` may have their functionality overridden by derived classes, and in those overrides one can use the `Super` keyword to call the parent function.
If `const` is placed after the function signature and before the function body, the method will not be allowed to modify any members in the object instance it's being called on.
The keyword `void` can be used in place of the return type (or type list) to have a method which does not have any return value. Similarly, one can place `void` where the argument list might be, although this is redundant as having no argument list at all is allowed.
2018-11-18 08:44:41 -08:00
Arguments of methods may only be of certain types due to technical limitations. See the type table for a list of which are usable and which are not.
All methods which are not `static` have an implicit `self` parameter which refers to this object, although if you wish to refer to a member of `self`, you do not need to reference it directly, as it is already implicitly in scope.
Method arguments must all have a name and type, and optionally the last arguments in the list may have a default value, (*Version 3.3.0 and newer*) except in functions marked `override`. The syntax is:
ZScript includes an extra method type for descendents of `Actor` called *actions*, which are intended to be run from actor states and give extra information to the function. Action functions change the meaning of the `self` parameter and pass in `invoker` and `stateinfo` parameters as well. `stateinfo` refers to the `State` which this action was called from. Here is a chart for the meanings of the `self` and `invoker` variables under each scope:
| Scope | `self` meaning | `invoker` meaning |
| ----- | -------------- | ----------------- |
| None | The actor this function operates on, ambiguous in some contexts | State owner |
