feat: overhaul mdbook preprocessor, prettify generated docs, support dummy globals (makspll#377)

# Summary

- Completely overhauls the mdbook preprocessor.
- Adds `dummy` globals to the globals registry for the purposes of
documenting `entity`, `world`, etc.
- Moves all of the manually written docs into the generated docs, and
removes the `WIP` section
- Type and instance lists are formatted better now
- Adds markings in the LAD format for `importance` and `generated` types
- Types in the generated docs are ordered according to their importance,
core types are placed ahead of most types, and user marked `significant`
types are placed even higher
- Non reflected types are supported by the lad builder
- Links in generic `TypedThrough` types are generated at each non
generic type


![image](https://github.com/user-attachments/assets/a4bfa889-771d-4faf-825d-29143851aeb6)

![image](https://github.com/user-attachments/assets/0abadcd9-e0c4-4b4c-a7da-3f46029ad530)

![image](https://github.com/user-attachments/assets/b3fbca0b-1564-4523-bfb9-b88e624c32ae)

![image](https://github.com/user-attachments/assets/dab70e3c-e531-44ec-a1ff-ddba6e12f42c)

Author: makspll

crates/bevy_api_gen/templates/footer.tera

@@ -11,7 +11,8 @@ pub struct {{ "ScriptingPlugin" | prefix_cratename | convert_case(case="upper_ca
 #[script_bindings(
     remote,
     name = "{{ item.ident | convert_case(case="snake") }}_functions",
-    bms_core_path="bevy_mod_scripting_core"
+    bms_core_path="bevy_mod_scripting_core",
+    generated
 )]
 impl {{item.import_path}} {
 {% for function in item.functions %}

crates/bevy_mod_scripting_core/src/bindings/globals/core.rs

@@ -2,10 +2,21 @@
 
 use std::{collections::HashMap, sync::Arc};
 
-use bevy::{app::Plugin, ecs::reflect::AppTypeRegistry};
+use bevy::{
+    app::Plugin,
+    ecs::{entity::Entity, reflect::AppTypeRegistry, world::World},
+};
 use bevy_mod_scripting_derive::script_globals;
 
-use crate::{bindings::{function::from::{Union, Val}, ScriptComponentRegistration, ScriptResourceRegistration, ScriptTypeRegistration, WorldGuard}, docgen::into_through_type_info, error::InteropError};
+use crate::{
+    bindings::{
+        function::from::{Union, Val},
+        ScriptComponentRegistration, ScriptResourceRegistration, ScriptTypeRegistration,
+        WorldGuard,
+    },
+    docgen::into_through_type_info,
+    error::InteropError,
+};
 
 use super::AppScriptGlobalsRegistry;
 
@@ -22,54 +33,67 @@ impl Plugin for CoreScriptGlobalsPlugin {
 
 fn register_static_core_globals(world: &mut bevy::ecs::world::World) {
     let global_registry = world
-            .get_resource_or_init::<AppScriptGlobalsRegistry>()
-            .clone();
-        let type_registry = world
-            .get_resource_or_init::<AppTypeRegistry>()
-            .clone();
-        let mut global_registry = global_registry.write();
-        let type_registry = type_registry.read();
+        .get_resource_or_init::<AppScriptGlobalsRegistry>()
+        .clone();
+    let type_registry = world.get_resource_or_init::<AppTypeRegistry>().clone();
+    let mut global_registry = global_registry.write();
+    let type_registry = type_registry.read();
 
-        // find all reflectable types without generics
-        for registration in type_registry.iter() {
-            if !registration.type_info().generics().is_empty() {
-                continue;
-            }
+    // find all reflectable types without generics
+    for registration in type_registry.iter() {
+        if !registration.type_info().generics().is_empty() {
+            continue;
+        }
 
-            if let Some(global_name) = registration.type_info().type_path_table().ident() {
-                let documentation = "A reference to the type, allowing you to call static methods.";
-                let type_info = registration.type_info();
-                global_registry.register_static_documented_dynamic(
-                    registration.type_id(),
-                    into_through_type_info(type_info),
-                    global_name.into(),
-                    documentation.into(),
-                );
-            }
+        if let Some(global_name) = registration.type_info().type_path_table().ident() {
+            let documentation = "A reference to the type, allowing you to call static methods.";
+            let type_info = registration.type_info();
+            global_registry.register_static_documented_dynamic(
+                registration.type_id(),
+                into_through_type_info(type_info),
+                global_name.into(),
+                documentation.into(),
+            );
         }
+    }
+
+    // register basic globals
+    global_registry.register_dummy::<World>("world", "The current ECS world.");
+    global_registry
+        .register_dummy::<Entity>("entity", "The entity this script is attached to if any.");
+    global_registry.register_dummy::<String>("script_id", "the name/id of this script");
 }
 
-#[script_globals(
-    bms_core_path = "crate",
-    name = "core_globals",
-)]
+#[script_globals(bms_core_path = "crate", name = "core_globals")]
 impl CoreGlobals {
     /// A cache of types normally available through the `world.get_type_by_name` function.
-    /// 
+    ///
     /// You can use this to avoid having to store type references.
-    fn types(guard: WorldGuard) -> Result<HashMap<String, Union<Val<ScriptTypeRegistration>, Union<Val<ScriptComponentRegistration>, Val<ScriptResourceRegistration>>>>, InteropError> {
+    fn types(
+        guard: WorldGuard,
+    ) -> Result<
+        HashMap<
+            String,
+            Union<
+                Val<ScriptTypeRegistration>,
+                Union<Val<ScriptComponentRegistration>, Val<ScriptResourceRegistration>>,
+            >,
+        >,
+        InteropError,
+    > {
         let type_registry = guard.type_registry();
         let type_registry = type_registry.read();
         let mut type_cache = HashMap::<String, _>::default();
-        for registration in type_registry.iter(){
+        for registration in type_registry.iter() {
             if let Some(ident) = registration.type_info().type_path_table().ident() {
                 let registration = ScriptTypeRegistration::new(Arc::new(registration.clone()));
                 let registration = guard.clone().get_type_registration(registration)?;
-                let registration = registration.map_both(Val::from, |u| u.map_both(Val::from, Val::from));
+                let registration =
+                    registration.map_both(Val::from, |u| u.map_both(Val::from, Val::from));
                 type_cache.insert(ident.to_string(), registration);
             }
         }
-        
+
         Ok(type_cache)
     }
-}
+}

crates/bevy_mod_scripting_core/src/bindings/globals/mod.rs

@@ -5,12 +5,15 @@ use super::{
     script_value::ScriptValue,
     WorldGuard,
 };
-use crate::{docgen::{into_through_type_info, typed_through::ThroughTypeInfo}, error::InteropError};
+use crate::{
+    docgen::{into_through_type_info, typed_through::ThroughTypeInfo},
+    error::InteropError,
+};
 use bevy::{ecs::system::Resource, reflect::Typed, utils::hashbrown::HashMap};
 use parking_lot::{RwLock, RwLockReadGuard, RwLockWriteGuard};
 use std::{any::TypeId, borrow::Cow, sync::Arc};
 
-crate::private::export_all_in_modules!{
+crate::private::export_all_in_modules! {
     core
 }
 
@@ -48,10 +51,22 @@ pub struct ScriptGlobal {
     pub type_information: ThroughTypeInfo,
 }
 
+/// A dummy global variable that documents globals set via alternative ways.
+pub struct ScriptGlobalDummy {
+    /// The type ID of the global variable.
+    pub type_id: TypeId,
+    /// Rich type information the global variable.
+    pub type_information: Option<ThroughTypeInfo>,
+
+    /// The documentation for the global dummy variable.
+    pub documentation: Option<Cow<'static, str>>,
+}
+
 /// A registry of global variables that can be exposed to scripts.
 #[derive(Default)]
 pub struct ScriptGlobalsRegistry {
     globals: HashMap<Cow<'static, str>, ScriptGlobal>,
+    dummies: HashMap<Cow<'static, str>, ScriptGlobalDummy>,
 }
 
 impl ScriptGlobalsRegistry {
@@ -85,6 +100,11 @@ impl ScriptGlobalsRegistry {
         self.globals.iter_mut()
     }
 
+    /// Iterates over the dummies in the registry
+    pub fn iter_dummies(&self) -> impl Iterator<Item = (&Cow<'static, str>, &ScriptGlobalDummy)> {
+        self.dummies.iter()
+    }
+
     fn type_erase_maker<
         T: ScriptReturn,
         F: Fn(WorldGuard) -> Result<T, InteropError> + Send + Sync + 'static,
@@ -114,6 +134,26 @@ impl ScriptGlobalsRegistry {
         )
     }
 
+    /// Registers a dummy global into the registry.
+    /// Dummies are not actually exposed to languages but exist purely for the purpose of documentation.
+    /// This can be useful for globals which you cannot expose normally.
+    ///
+    /// Dummy globals are stored as non-static instances, i.e. they're expected to be values not type references.
+    pub fn register_dummy<T: 'static>(
+        &mut self,
+        name: impl Into<Cow<'static, str>>,
+        documentation: impl Into<Cow<'static, str>>,
+    ) {
+        self.dummies.insert(
+            name.into(),
+            ScriptGlobalDummy {
+                documentation: Some(documentation.into()),
+                type_id: TypeId::of::<T>(),
+                type_information: None,
+            },
+        );
+    }
+
     /// Inserts a global into the registry, returns the previous value if it existed.
     ///
     /// This is a version of [`Self::register`] which stores type information regarding the global.
@@ -153,7 +193,7 @@ impl ScriptGlobalsRegistry {
     /// Registers a static global into the registry.
     ///
     /// This is a version of [`Self::register_static`] which stores rich type information regarding the global.
-    pub fn register_static_documented<T: TypedScriptReturn +'static>(
+    pub fn register_static_documented<T: TypedScriptReturn + 'static>(
         &mut self,
         name: Cow<'static, str>,
         documentation: Cow<'static, str>,

crates/bevy_mod_scripting_core/src/bindings/mod.rs

@@ -12,4 +12,5 @@ crate::private::export_all_in_modules! {
     script_system,
     script_value,
     world,
+    type_data
 }

crates/bevy_mod_scripting_core/src/bindings/query.rs

@@ -14,24 +14,28 @@ use bevy::{
 };
 use std::{any::TypeId, collections::VecDeque, sync::Arc};
 
-/// A wrapper around a `TypeRegistration` that provides additional information about the type.
+/// A reference to a type which is not a `Resource` or `Component`.
 ///
-/// This is used as a hook to a rust type from a scripting language. We should be able to easily convert between a type name and a [`ScriptTypeRegistration`].
+/// In general think of this as a handle to a type.
 #[derive(Clone, Reflect)]
 #[reflect(opaque)]
 pub struct ScriptTypeRegistration {
     pub(crate) registration: Arc<TypeRegistration>,
 }
 
 #[derive(Clone, Reflect, Debug)]
-/// A registration for a component type.
+/// A reference to a component type's reflection registration.
+///
+/// In general think of this as a handle to a type.
 pub struct ScriptComponentRegistration {
     pub(crate) registration: ScriptTypeRegistration,
     pub(crate) component_id: ComponentId,
 }
 
 #[derive(Clone, Reflect, Debug)]
-/// A registration for a resource type.
+/// A reference to a resource type's reflection registration.
+///
+/// In general think of this as a handle to a type.
 pub struct ScriptResourceRegistration {
     pub(crate) registration: ScriptTypeRegistration,
     pub(crate) resource_id: ComponentId,
@@ -134,7 +138,25 @@ impl std::fmt::Display for ScriptTypeRegistration {
 
 #[derive(Clone, Default, Reflect)]
 #[reflect(opaque)]
-/// A builder for a query.
+/// The query builder is used to build ECS queries which retrieve spefific components filtered by specific conditions.
+///
+/// For example:
+/// ```rust,ignore
+/// builder.component(componentA)
+///     .component(componentB)
+///     .with(componentC)
+///     .without(componentD)  
+/// ```
+///
+/// Will retrieve entities which:
+/// - Have componentA
+/// - Have componentB
+/// - Have componentC
+/// - Do not have componentD
+///
+/// As well as references to components:
+/// - componentA
+/// - componentB
 pub struct ScriptQueryBuilder {
     pub(crate) components: Vec<ScriptComponentRegistration>,
     with: Vec<ScriptComponentRegistration>,

crates/bevy_mod_scripting_core/src/bindings/reference.rs

@@ -22,8 +22,13 @@ use bevy::{
 };
 use std::{any::TypeId, fmt::Debug};
 
-/// An accessor to a `dyn PartialReflect` struct, stores a base ID of the type and a reflection path
-/// safe to build but to reflect on the value inside you need to ensure aliasing rules are upheld
+/// A reference to an arbitrary reflected instance.
+///
+/// The reference can point to either the ECS, or to the allocator.
+///
+/// References are composed of two parts:
+/// - The base kind, which specifies where the reference points to
+/// - The path, which specifies how to access the value from the base
 #[derive(Debug, Clone, PartialEq, Eq, Reflect)]
 #[reflect(Default, opaque)]
 pub struct ReflectReference {

crates/bevy_mod_scripting_core/src/bindings/type_data.rs

@@ -0,0 +1,42 @@
+//! Contains various `Reflect` type data we use in BMS.
+
+use bevy::reflect::FromType;
+
+/// A marker type to indicate that a type is generated.
+///
+/// To mark a type as generated use:
+/// ```rust,ignore
+/// registry.register_type_data::<T, MarkAsGenerated>();
+/// ```
+#[derive(Clone, Copy)]
+pub struct MarkAsGenerated;
+
+impl<T> FromType<T> for MarkAsGenerated {
+    fn from_type() -> Self {
+        Self
+    }
+}
+
+/// A marker type to indicate that a type is significant.
+///
+/// Significant types appear "before" core types in documentation
+#[derive(Clone, Copy)]
+pub struct MarkAsSignificant;
+
+impl<T> FromType<T> for MarkAsSignificant {
+    fn from_type() -> Self {
+        Self
+    }
+}
+
+/// A marker type to indicate that a type is core to BMS.
+///
+/// core types appear before insignificant types in documentation.
+#[derive(Clone, Copy)]
+pub struct MarkAsCore;
+
+impl<T> FromType<T> for MarkAsCore {
+    fn from_type() -> Self {
+        Self
+    }
+}