Features & slight refector

Cargo.toml

@@ -1,11 +1,37 @@
 [package]
-name = "rust_learn"
+name = "nml"
 version = "0.1.0"
 edition = "2021"
+license = "GNU AGPL3"
 
-# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+[[bin]]
+name = "nml"
+path = "src/main.rs"
+
+[[bin]]
+name = "nmlls"
+path = "src/server.rs"
+
+[profile.profiling]
+inherits = "release"
+debug = true
 
 [dependencies]
-crossbeam-utils = "0.8.19"
-files = "2.2.3"
+ariadne = "0.4.1"
+dashmap = "6.0.1"
+downcast-rs = "1.2.1"
+getopts = "0.2.21"
+lazy_static = "1.5.0"
+lsp-server = "0.7.6"
+lsp-types = "0.97.0"
+mlua = { version = "0.9.9", features = ["lua54", "vendored"] }
 regex = "1.10.3"
+rusqlite = "0.31.0"
+rust-crypto = "0.2.36"
+serde = "1.0.204"
+serde_json = "1.0.120"
+syntect = "5.2.0"
+tokio = { version = "1.38.1", features = ["macros", "rt-multi-thread", "io-std"]}
+
+tower-lsp = "0.20.0"
+unicode-segmentation = "1.11.0"

src/cache/cache.rs

@@ -0,0 +1,97 @@
+use std::{error::Error, path::PathBuf};
+
+use rusqlite::{types::FromSql, Connection, Params, ToSql};
+
+struct Cache {
+	con: Connection
+}
+
+impl Cache {
+    fn new(file: PathBuf) -> Result<Self, String> {
+		match Connection::open(file)
+		{
+			Err(e) => return Err(format!("Could not connect to cache database: {}", e.to_string())),
+			Ok(con) => Ok(Self { con })
+		}
+    }
+}
+
+pub enum CachedError<E>
+{
+	SqlErr(rusqlite::Error),
+	GenErr(E)
+}
+
+pub trait Cached
+{
+	type Key;
+	type Value;
+
+	/// SQL Query to create the cache table
+	/// Note: You must use `IF NOT EXIST`
+	fn sql_table() -> &'static str;
+
+	/// SQL Get query
+	fn sql_get_query() -> &'static str;
+
+	/// SQL insert query
+	fn sql_insert_query() -> &'static str;
+
+	fn key(&self) -> <Self as Cached>::Key;
+
+	fn init(con: &mut Connection) -> Result<(), rusqlite::Error>
+	{
+		con.execute(<Self as Cached>::sql_table(), ())
+			.map(|_| ())
+	}
+
+	fn cached<E, F>(&self, con: &mut Connection, f: F)
+		-> Result<<Self as Cached>::Value, CachedError<E>>
+	where
+		<Self as Cached>::Key: ToSql,
+		<Self as Cached>::Value: FromSql + ToSql,
+		F: FnOnce(&Self) -> Result<<Self as Cached>::Value, E>,
+    {
+		let key = self.key();
+
+		// Find in cache
+		let mut query = match con.prepare(<Self as Cached>::sql_get_query())
+		{
+			Ok(query) => query,
+			Err(e) => return Err(CachedError::SqlErr(e))
+		};
+
+		let value = query.query_row([&key], |row|
+		{
+			Ok(row.get_unwrap::<_, <Self as Cached>::Value>(0))
+		}).ok();
+
+		if let Some(value) = value
+		{
+			// Found in cache
+			return Ok(value)
+		}
+		else
+		{
+			// Compute a value
+			let value = match f(&self)
+			{
+				Ok(val) => val,
+				Err(e) => return Err(CachedError::GenErr(e))
+			};
+
+			// Try to insert
+			let mut query = match con.prepare(<Self as Cached>::sql_insert_query())
+			{
+				Ok(query) => query,
+				Err(e) => return Err(CachedError::SqlErr(e))
+			};
+
+			match query.execute((&key, &value))
+			{
+				Ok(_) => Ok(value),
+				Err(e) => Err(CachedError::SqlErr(e))
+			}
+		}
+    }
+}

src/cache/mod.rs

@@ -0,0 +1 @@
+pub mod cache;

src/compiler/compiler.rs

@@ -0,0 +1,153 @@
+use std::{cell::{RefCell, RefMut}, rc::Rc};
+
+
+use rusqlite::Connection;
+
+use crate::document::{document::Document, variable::Variable};
+
+#[derive(Clone, Copy)]
+pub enum Target
+{
+	HTML,
+	LATEX,
+}
+
+pub struct Compiler
+{
+	target: Target,
+	cache: Option<RefCell<Connection>>,
+}
+
+impl Compiler
+{
+	pub fn new(target: Target, db_path: Option<String>) -> Self {
+		let cache = match db_path
+		{
+			None => None,
+			Some(path) => {
+				match Connection::open(path)
+				{
+					Err(e) => panic!("Cannot connect to database: {e}"),
+					Ok(con) => Some(con),
+				}
+			}
+		};
+		Self {
+			target,
+			cache: cache.map(|con| RefCell::new(con)),
+		}
+	}
+
+	pub fn target(&self) -> Target
+	{
+		self.target
+	}
+
+	pub fn cache(&self) -> Option<RefMut<'_, Connection>>
+	{
+		self.cache
+			.as_ref()
+			.map(RefCell::borrow_mut)
+	}
+
+	pub fn sanitize<S: AsRef<str>>(&self, str: S) -> String {
+		match self.target
+		{
+			Target::HTML => str.as_ref()
+				.replace("&", "&amp;")
+				.replace("<", "&lt;")
+				.replace(">", "&gt;")
+				.replace("\"", "&quot;"),
+			_ => todo!("Sanitize not implemented")
+		}
+	}
+
+	pub fn header(&self, document: &Document) -> String
+	{
+		pub fn get_variable_or_error(document: &Document, var_name: &'static str) -> Option<Rc<dyn Variable>>
+		{
+			document.get_variable(var_name)
+				.and_then(|(_, var)| Some(var))
+				.or_else(|| {
+					println!("Missing variable `{var_name}` in {}", document.source().name());
+					None
+				})
+		}
+
+		let mut result = String::new();
+		match self.target()
+		{
+			Target::HTML => {
+				result += "<!DOCTYPE HTML><html><head>";
+				result += "<meta charset=\"UTF-8\">";
+				if let Some(page_title) = get_variable_or_error(document, "html.page_title")
+				{
+					result += format!("<title>{}</title>", self.sanitize(page_title.to_string())).as_str();
+				}
+
+				if let Some((_, css)) = document.get_variable("html.css")
+				{
+					result += format!("<link rel=\"stylesheet\" href=\"{}\">", self.sanitize(css.to_string())).as_str();
+				}
+				result += "</head><body>";
+
+				// TODO: TOC
+				// TODO: Author, Date, Title, Div
+			},
+			Target::LATEX => {
+
+			},
+		}
+		result
+	}
+
+	pub fn footer(&self, _document: &Document) -> String
+	{
+		let mut result = String::new();
+		match self.target()
+		{
+			Target::HTML => {
+				result += "</body></html>";
+			},
+			Target::LATEX => {
+
+			},
+		}
+		result
+	}
+	
+	pub fn compile(&self, document: &Document) -> String
+	{
+		let mut out = String::new();
+		let borrow = document.content.borrow();
+
+		// Header
+		out += self.header(document).as_str();
+		
+		// Body
+		for i in 0 .. borrow.len()
+		{
+			let elem = &borrow[i];
+			//let prev = match i
+			//{
+			//	0 => None,
+			//	_ => borrow.get(i-1),
+			//};
+			//let next = borrow.get(i+1);
+
+			match elem.compile(self, document)
+			{
+				Ok(result) => {
+					//println!("Elem: {}\nCompiled to: {result}", elem.to_string());
+					out.push_str(result.as_str())
+				},
+				Err(err) => println!("Unable to compile element: {err}\n{}", elem.to_string())
+			}
+		}
+
+		// Footer
+		out += self.footer(document).as_str();
+
+		out
+	}
+}

src/compiler/mod.rs

@@ -0,0 +1 @@
+pub mod compiler;

src/document/document.rs

@@ -0,0 +1,210 @@
+use std::cell::{Ref, RefCell, RefMut};
+use std::collections::hash_map::HashMap;
+use std::rc::Rc;
+
+use crate::parser::source::Source;
+
+use super::element::Element;
+use super::variable::Variable;
+
+
+#[derive(Debug)]
+pub struct Scope {
+    /// List of all referenceable elements in current scope.
+    /// All elements in this should return a non empty 
+	pub referenceable: HashMap<String, usize>,
+	pub variables: HashMap<String, Rc<dyn Variable>>,
+}
+
+impl Scope {
+	fn new() -> Self {
+		Self {
+			referenceable: HashMap::new(),
+			variables: HashMap::new(),
+		}
+	}
+
+    pub fn merge(&mut self, other: &mut Scope, merge_as: &String, ref_offset: usize)
+    {
+        match merge_as.is_empty()
+        {
+            true => {
+                // References
+                self.referenceable.extend(other.referenceable.drain()
+                    .map(|(name, idx)|
+                        (name, idx+ref_offset)));
+
+                // Variables
+                self.variables.extend(other.variables.drain()
+                    .map(|(name, var)|
+                        (name, var)));
+            },
+            false => {
+                // References
+                self.referenceable.extend(other.referenceable.drain()
+                    .map(|(name, idx)|
+                        (format!("{merge_as}.{name}"), idx+ref_offset)));
+
+                // Variables
+                self.variables.extend(other.variables.drain()
+                    .map(|(name, var)|
+                        (format!("{merge_as}.{name}"), var)));
+            }
+        }
+    }
+}
+
+#[derive(Debug)]
+pub struct Document<'a> {
+	source: Rc<dyn Source>,
+	parent: Option<&'a Document<'a>>, /// Document's parent
+
+	// FIXME: Render these fields private
+	pub content: RefCell<Vec<Box<dyn Element>>>,
+	pub scope: RefCell<Scope>,
+}
+
+impl<'a> Document<'a>  {
+	pub fn new(source: Rc<dyn Source>, parent: Option<&'a Document<'a>>) -> Self
+	{
+		Self {
+			source: source,
+			parent: parent,
+			content: RefCell::new(Vec::new()),
+			scope: RefCell::new(Scope::new()),
+		}
+	}
+
+	pub fn source(&self) -> Rc<dyn Source> { self.source.clone() }
+
+    pub fn parent(&self) -> Option<&Document> { self.parent }
+
+    /// Push an element [`elem`] to content. [`in_paragraph`] is true if a paragraph is active
+	pub fn push(&self, elem: Box<dyn Element>)
+	{
+        // Add index of current element to scope's reference table
+        if let Some(referenceable) = elem.as_referenceable()
+        {
+            // Only add if referenceable holds a reference
+            if let Some(ref_name) = referenceable.reference_name()
+            {
+                self.scope.borrow_mut().referenceable.insert(ref_name.clone(), self.content.borrow().len());
+            }
+        }
+
+		self.content.borrow_mut().push(elem);
+	}
+
+	pub fn last_element<T: Element>(&self, recurse: bool) -> Option<Ref<'_, T>>
+	{
+		let elem = Ref::filter_map(self.content.borrow(),
+		|content| content.last()
+			.and_then(|last| last.downcast_ref::<T>())).ok();
+
+		if elem.is_some() || !recurse { return elem }
+
+		match self.parent
+		{
+			None => None,
+			Some(parent) => parent.last_element(true),
+		}
+	}
+
+	pub fn last_element_mut<T: Element>(&self, recurse: bool) -> Option<RefMut<'_, T>>
+	{
+		let elem = RefMut::filter_map(self.content.borrow_mut(),
+		|content| content.last_mut()
+			.and_then(|last| last.downcast_mut::<T>())).ok();
+
+		if elem.is_some() || !recurse { return elem }
+
+		match self.parent
+		{
+			None => None,
+			Some(parent) => parent.last_element_mut(true),
+		}
+	}
+
+	pub fn get_reference(&self, ref_name: &str) -> Option<(&Document<'a>, std::cell::Ref<'_, Box<dyn Element>>)> {
+		match self.scope.borrow().referenceable.get(ref_name) {
+			// Return if found
+			Some(elem) => {
+                return Some((&self,
+                    std::cell::Ref::map(self.content.borrow(),
+                    |m| &m[*elem])))
+            },
+
+			// Continue search recursively
+			None => match self.parent {
+				Some(parent) => return parent.get_reference(ref_name),
+
+				// Not found
+				None => return None,
+			}
+		}
+	}
+
+    pub fn add_variable(&self, variable: Rc<dyn Variable>)
+    {
+        self.scope.borrow_mut().variables.insert(
+            variable.name().to_string(),
+            variable);
+    }
+
+    pub fn get_variable<S: AsRef<str>>(&self, name: S) -> Option<(&Document<'a>, Rc<dyn Variable>)>
+    {
+        match self.scope.borrow().variables.get(name.as_ref())
+        {
+            Some(variable) => {
+                return Some((&self, variable.clone()));
+            },
+
+			// Continue search recursively
+            None => match self.parent {
+                Some(parent) => return parent.get_variable(name),
+
+                // Not found
+                None => return None,
+            }
+        }
+    }
+
+	pub fn remove_variable<S: AsRef<str>>(&self, name: S) -> Option<(&Document<'a>, Rc<dyn Variable>)>
+	{
+        match self.scope.borrow_mut().variables.remove(name.as_ref())
+        {
+            Some(variable) => {
+                return Some((&self, variable.clone()));
+            },
+
+			// Continue search recursively
+            None => match self.parent {
+                Some(parent) => return parent.remove_variable(name),
+
+                // Not found
+                None => return None,
+            }
+        }
+	}
+
+    /// Merges [`other`] into [`self`]
+    pub fn merge(&self, other: Document, merge_as: Option<&String>)
+    {
+		match merge_as
+		{
+			Some(merge_as)	=> self.scope.borrow_mut()
+				.merge(
+					&mut *other.scope.borrow_mut(),
+					merge_as,
+					self.content.borrow().len()+1),
+			_ => {},
+		}
+
+        // Content
+        self.content.borrow_mut().extend((other.content.borrow_mut())
+            .drain(..)
+            .map(|value| value));
+    }
+}
+
+

src/document/element.rs

@@ -0,0 +1,95 @@
+use std::str::FromStr;
+
+use downcast_rs::{impl_downcast, Downcast};
+use crate::{compiler::compiler::Compiler, parser::source::Token};
+
+use super::document::Document;
+
+#[derive(Debug, PartialEq, Eq, Clone)]
+pub enum ElemKind {
+    /// An invisible element (i.e comment)
+    Invisible,
+    /// Special elements don't trigger special formatting events
+    Special,
+    /// Inline elements don't break paragraphing
+    Inline,
+    /// Block elements are outside of paragraphs
+    Block,
+}
+
+impl FromStr for ElemKind {
+    type Err = String;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+		match s
+		{
+			"invisible" => Ok(ElemKind::Invisible),
+			"special"   => Ok(ElemKind::Special),
+			"inline"    => Ok(ElemKind::Inline),
+			"block"     => Ok(ElemKind::Block),
+			_ => Err(format!("Unknown ElemKind: {s}"))
+		}
+    }
+}
+
+pub trait Element: Downcast
+{
+    /// Gets the element defined location i.e token without filename
+    fn location(&self) -> &Token;
+
+    fn kind(&self) -> ElemKind;
+
+    /// Get the element's name
+	fn element_name(&self) -> &'static str;
+
+    /// Outputs element to string for debug purposes
+    fn to_string(&self) -> String;
+
+    fn as_referenceable(&self) -> Option<&dyn ReferenceableElement> { None }
+
+    /// Compiles element
+    fn compile(&self, compiler: &Compiler, document: &Document) -> Result<String, String>;
+}
+impl_downcast!(Element);
+
+impl core::fmt::Debug for dyn Element
+{
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.to_string())
+    }
+}
+
+pub trait ReferenceableElement : Element {
+    /// Reference name
+	fn reference_name(&self) -> Option<&String>;
+}
+
+#[derive(Debug)]
+pub struct Text
+{
+    location: Token,
+	content: String,
+}
+
+impl Text
+{
+	pub fn new(location: Token, content: String) -> Text
+	{
+		Text {
+            location: location,
+			content: content
+		}
+	}
+}
+
+impl Element for Text
+{
+    fn location(&self) -> &Token { &self.location }
+    fn kind(&self) -> ElemKind { ElemKind::Inline }
+	fn element_name(&self) -> &'static str { "Text" }
+    fn to_string(&self) -> String { format!("{self:#?}") }
+
+    fn compile(&self, compiler: &Compiler, _document: &Document) -> Result<String, String> {
+        Ok(compiler.sanitize(self.content.as_str()))
+    }
+}

src/document/mod.rs

@@ -0,0 +1,3 @@
+pub mod document;
+pub mod element;
+pub mod variable;

src/document/variable.rs

@@ -0,0 +1,146 @@
+use std::{path::PathBuf, rc::Rc};
+use crate::parser::{parser::Parser, source::{Source, Token, VirtualSource}};
+use super::{document::Document, element::Text};
+
+
+// TODO enforce to_string(from_string(to_string())) == to_string()
+pub trait Variable
+{
+	fn location(&self) -> &Token;
+
+	fn name(&self) -> &str;
+	/// Parse variable from string, returns an error message on failure
+	fn from_string(&mut self, str: &str) -> Option<String>;
+
+	/// Converts variable to a string
+	fn to_string(&self) -> String;
+
+    fn parse<'a>(&self, location: Token, parser: &dyn Parser, document: &'a Document);
+}
+
+impl core::fmt::Debug for dyn Variable
+{
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+		write!(f, "{}{{{}}}", self.name(), self.to_string())
+    }
+}
+
+#[derive(Debug)]
+pub struct BaseVariable
+{
+	location: Token,
+    name: String,
+    value: String,
+}
+
+impl BaseVariable {
+    pub fn new(location: Token, name: String, value: String) -> Self {
+        Self { location, name, value }
+    }
+}
+
+impl Variable for BaseVariable
+{
+	fn location(&self) -> &Token { &self.location }
+
+    fn name(&self) -> &str { self.name.as_str() }
+
+    fn from_string(&mut self, str: &str) -> Option<String> {
+        self.value = str.to_string();
+        None
+    }
+
+    fn to_string(&self) -> String { self.value.clone() }
+
+    fn parse<'a>(&self, _location: Token, parser: &dyn Parser, document: &'a Document) {
+		let source = Rc::new(VirtualSource::new(
+				self.location().clone(),
+			self.name().to_string(),
+			self.to_string()));
+
+		parser.parse_into(source, document);
+	}
+}
+
+#[derive(Debug)]
+pub struct PathVariable
+{
+	location: Token,
+    name: String,
+    path: PathBuf,
+}
+
+impl PathVariable
+{
+    pub fn new(location: Token, name: String, path: PathBuf) -> Self {
+        Self { location, name, path }
+    }
+}
+
+impl Variable for PathVariable
+{
+	fn location(&self) -> &Token { &self.location }
+
+    fn name(&self) -> &str { self.name.as_str() }
+
+    fn from_string(&mut self, str: &str) -> Option<String> {
+        self.path = PathBuf::from(std::fs::canonicalize(str).unwrap());
+        None
+    }
+
+    fn to_string(&self) -> String { self.path.to_str().unwrap().to_string() }
+
+    fn parse<'a>(&self, location: Token, parser: &dyn Parser, document: &'a Document){
+		// TODO: Avoid copying the location twice...
+		// Maybe create a special VirtualSource where the `content()` method
+		// calls `Variable::to_string()`
+		let source = Rc::new(VirtualSource::new(
+			location.clone(),
+			self.name().to_string(),
+			self.to_string()));
+
+        parser.push(document, Box::new(Text::new(
+			Token::new(0..source.content().len(), source),
+			self.to_string()
+        )));
+    }
+}
+
+/*
+struct ConfigVariable<T>
+{
+	value: T,
+	name: String,
+
+	desc: String,
+	validator: Box<dyn Fn(&Self, &T) -> Option<&String>>,
+}
+
+impl<T> ConfigVariable<T>
+{
+	fn description(&self) -> &String { &self.desc }
+}
+
+impl<T> Variable for ConfigVariable<T>
+where T: FromStr + Display
+{
+	fn name(&self) -> &str { self.name.as_str() }
+
+	/// Parse variable from string, returns an error message on failure
+	fn from_string(&mut self, str: &str) -> Option<String> {
+		match str.parse::<T>()
+		{
+			Ok(value) => {
+				(self.validator)(self, &value).or_else(|| {
+                    self.value = value;
+                    None
+                })
+			},
+			Err(_) => return Some(format!("Unable to parse `{str}` into variable `{}`", self.name))
+		}
+	}
+
+	/// Converts variable to a string
+	fn to_string(&self) -> String { self.value.to_string() }
+}
+*/

src/elements/code.rs

@@ -0,0 +1,390 @@
+use std::{collections::HashMap, ops::Range, rc::Rc, sync::Once};
+
+use ariadne::{Fmt, Label, Report, ReportKind};
+use crypto::{digest::Digest, sha2::Sha512};
+use regex::{Captures, Regex};
+use syntect::{easy::HighlightLines, highlighting::ThemeSet, parsing::SyntaxSet};
+
+use crate::{cache::cache::{Cached, CachedError}, compiler::compiler::{Compiler, Target}, document::{document::Document, element::{ElemKind, Element}}, parser::{parser::Parser, rule::RegexRule, source::{Source, Token}, util::{self, Property, PropertyParser}}};
+use lazy_static::lazy_static;
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+enum CodeKind
+{
+	FullBlock,
+	MiniBlock,
+	Inline,
+}
+
+#[derive(Debug)]
+struct Code
+{
+	location: Token,
+	block: CodeKind,
+	language: String,
+	name: Option<String>,
+	code: String,
+	theme: Option<String>,
+	line_offset: usize,
+}
+
+impl Code {
+    fn new(location: Token, block: CodeKind, language: String, name: Option<String>, code: String, theme: Option<String>, line_offset: usize) -> Self {
+        Self { location, block, language, name, code, theme, line_offset }
+    }
+
+	fn highlight_html(&self, compiler: &Compiler) -> Result<String, String>
+	{
+		lazy_static! {
+			static ref syntax_set : SyntaxSet = SyntaxSet::load_defaults_newlines();
+			static ref theme_set : ThemeSet = ThemeSet::load_defaults();
+		}
+		let syntax = match syntax_set.find_syntax_by_name(self.language.as_str())
+		{
+			Some(syntax) => syntax,
+			None => return Err(format!("Unable to find syntax for language: {}", self.language))
+		};
+
+		let theme_string = match self.theme.as_ref()
+		{
+			Some(theme) => theme.as_str(),
+			None => "base16-ocean.dark",
+		};
+		let mut h = HighlightLines::new(syntax, &theme_set.themes[theme_string]);
+
+		let mut result = String::new();
+		if self.block == CodeKind::FullBlock
+		{
+			result += "<div class=\"code-block\">";
+			if let Some(name) = &self.name
+			{
+				result += format!("<div class=\"code-block-title\">{}</div>",
+					compiler.sanitize(name.as_str())).as_str();
+			}
+
+			result += format!("<div class=\"code-block-content\"><table cellspacing=\"0\">").as_str();
+			for (line_id, line) in self.code.split(|c| c == '\n').enumerate()
+			{
+				result += "<tr><td class=\"code-block-gutter\">";
+
+				// Line number
+				result += format!("<pre><span>{}</span></pre>", line_id+self.line_offset).as_str();
+
+				// Code
+				result += "</td><td class=\"code-block-line\"><pre>";
+				match h.highlight_line(line, &syntax_set)
+				{
+					Err(e) => return Err(format!("Error highlighting line `{line}`: {}", e.to_string())),
+					Ok(regions) => {
+						match syntect::html::styled_line_to_highlighted_html(&regions[..], syntect::html::IncludeBackground::No)
+						{
+							Err(e) => return Err(format!("Error highlighting code: {}", e.to_string())),
+							Ok(highlighted) => result += if highlighted.is_empty() { "<br>" } else { highlighted.as_str() }
+						}
+					}
+				}
+				result += "</pre></td></tr>";
+			}
+
+			result += "</table></div></div>";
+		}
+		else if self.block == CodeKind::MiniBlock
+		{
+			result += "<div class=\"code-block\"><div class=\"code-block-content\"><table cellspacing=\"0\">";
+
+			for line in self.code.split(|c| c == '\n')
+			{
+				result += "<tr><td class=\"code-block-line\"><pre>";
+				// Code
+				match h.highlight_line(line, &syntax_set)
+				{
+					Err(e) => return Err(format!("Error highlighting line `{line}`: {}", e.to_string())),
+					Ok(regions) => {
+						match syntect::html::styled_line_to_highlighted_html(&regions[..], syntect::html::IncludeBackground::No)
+						{
+							Err(e) => return Err(format!("Error highlighting code: {}", e.to_string())),
+							Ok(highlighted) => result += if highlighted.is_empty() { "<br>" } else { highlighted.as_str() }
+						}
+					}
+				}
+				result += "</pre></td></tr>";
+			}
+			result += "</table></div></div>";
+		}
+		else if self.block == CodeKind::Inline
+		{
+			result += "<a class=\"inline-code\"><code>";
+			match h.highlight_line(self.code.as_str(), &syntax_set)
+			{
+				Err(e) => return Err(format!("Error highlighting line `{}`: {}", self.code, e.to_string())),
+				Ok(regions) => {
+					match syntect::html::styled_line_to_highlighted_html(&regions[..], syntect::html::IncludeBackground::No)
+					{
+						Err(e) => return Err(format!("Error highlighting code: {}", e.to_string())),
+						Ok(highlighted) => result += highlighted.as_str()
+					}
+				}
+			}
+			result += "</code></a>";
+		}
+
+		Ok(result)
+	}
+}
+
+impl Cached for Code
+{
+    type Key = String;
+    type Value = String;
+
+    fn sql_table() -> &'static str {
+		"CREATE TABLE IF NOT EXISTS cached_code (
+				digest	     TEXT PRIMARY KEY,
+				highlighted  BLOB NOT NULL);"
+    }
+
+    fn sql_get_query() -> &'static str {
+		"SELECT highlighted FROM cached_code WHERE digest = (?1)"
+    }
+
+    fn sql_insert_query() -> &'static str {
+		"INSERT INTO cached_code (digest, highlighted) VALUES (?1, ?2)"
+    }
+
+    fn key(&self) -> <Self as Cached>::Key {
+		let mut hasher = Sha512::new();
+		hasher.input((self.block as usize).to_be_bytes().as_slice());
+		hasher.input((self.line_offset as usize).to_be_bytes().as_slice());
+		self.theme.as_ref().map(|theme| hasher.input(theme.as_bytes()));
+		self.name.as_ref().map(|name| hasher.input(name.as_bytes()));
+		hasher.input(self.language.as_bytes());
+		hasher.input(self.code.as_bytes());
+
+		hasher.result_str()
+    }
+}
+
+impl Element for Code {
+    fn location(&self) -> &Token { &self.location }
+
+    fn kind(&self) -> ElemKind { if self.block == CodeKind::Inline { ElemKind::Inline } else { ElemKind::Block } }
+
+    fn element_name(&self) -> &'static str { "Code Block" }
+
+    fn to_string(&self) -> String { format!("{self:#?}") }
+
+    fn compile(&self, compiler: &Compiler, _document: &Document)
+		-> Result<String, String> {
+
+		match compiler.target()
+		{
+			Target::HTML => {
+				static CACHE_INIT : Once = Once::new();
+				CACHE_INIT.call_once(|| if let Some(mut con) = compiler.cache() {
+					if let Err(e) = Code::init(&mut con)
+					{
+						eprintln!("Unable to create cache table: {e}");
+					}
+				});
+
+				if let Some(mut con) = compiler.cache()
+				{
+					match self.cached(&mut con, |s| s.highlight_html(compiler))
+					{
+						Ok(s) => Ok(s),
+						Err(e) => match e
+						{
+							CachedError::SqlErr(e) => Err(format!("Querying the cache failed: {e}")),
+							CachedError::GenErr(e) => Err(e)
+						}
+					}
+				}
+				else
+				{
+					self.highlight_html(compiler)
+				}
+			}
+			Target::LATEX => { todo!("") }
+		}
+    }
+}
+
+pub struct CodeRule {
+	re: [Regex; 2],
+	properties: PropertyParser,
+}
+
+impl CodeRule {
+	pub fn new() -> Self {
+		let mut props = HashMap::new();
+		props.insert("line_offset".to_string(),
+			Property::new(
+				true,
+				"Line number offset".to_string(),
+					Some("1".to_string())));
+		Self {
+			re: [
+				Regex::new(r"(?:^|\n)```(?:\[((?:\\.|[^\\\\])*?)\])?(.*?)(?:,(.*))?\n((?:\\(?:.|\n)|[^\\\\])*?)```").unwrap(),
+				Regex::new(r"``(?:\[((?:\\.|[^\[\]\\])*?)\])?(?:(.*)(?:\n|,))?((?:\\(?:.|\n)|[^\\\\])*?)``").unwrap(),
+			],
+			properties: PropertyParser::new(props)
+		}
+	}
+}
+
+impl RegexRule for CodeRule
+{
+    fn name(&self) -> &'static str { "Code" }
+
+    fn regexes(&self) -> &[regex::Regex] { &self.re }
+
+    fn on_regex_match(&self, index: usize, parser: &dyn Parser, document: &Document, token: Token, matches: Captures)
+		-> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>> {
+		let mut reports = vec![];
+
+		let properties = match matches.get(1)
+		{
+			None => match self.properties.default() {
+				Ok(properties) => properties,
+				Err(e) => {
+					reports.push(
+						Report::build(ReportKind::Error, token.source(), token.start())
+						.with_message("Invalid code")
+						.with_label(
+							Label::new((token.source().clone(), token.range.clone()))
+							.with_message(format!("Code is missing properties: {e}"))
+							.with_color(parser.colors().error))
+						.finish());
+						return reports;
+				},
+			}
+			Some(props) => {
+				let processed = util::process_escaped('\\', "]",
+					props.as_str().trim_start().trim_end());
+				match self.properties.parse(processed.as_str())
+				{
+					Err(e) => {
+						reports.push(
+							Report::build(ReportKind::Error, token.source(), props.start())
+							.with_message("Invalid Code Properties")
+							.with_label(
+								Label::new((token.source().clone(), props.range()))
+								.with_message(e)
+								.with_color(parser.colors().error))
+							.finish());
+						return reports;
+					}
+					Ok(properties) => properties
+				}
+			}
+		};
+		
+		let code_lang = match matches.get(2)
+		{
+			None => "Plain Text".to_string(),
+			Some(lang) => {
+				let code_lang = lang.as_str().trim_end().trim_start().to_string();
+				if code_lang.is_empty()
+				{
+					reports.push(
+						Report::build(ReportKind::Error, token.source(), lang.start())
+						.with_message("Missing code language")
+						.with_label(
+							Label::new((token.source().clone(), lang.range()))
+							.with_message("No language specified")
+							.with_color(parser.colors().error))
+						.finish());
+
+					return reports;
+				}
+				
+				// TODO: validate language
+
+				code_lang
+			}
+		};
+
+		let mut code_content = if index == 0
+		{ util::process_escaped('\\',"```", matches.get(4).unwrap().as_str()) }
+		else
+		{ util::process_escaped('\\',"``", matches.get(3).unwrap().as_str()) };
+		if code_content.bytes().last() == Some('\n' as u8) // Remove newline
+		{
+			code_content.pop();
+		}
+
+		if code_content.is_empty()
+		{
+			reports.push(
+				Report::build(ReportKind::Error, token.source(), token.start())
+				.with_message("Missing code content")
+				.with_label(
+					Label::new((token.source().clone(), token.range.clone()))
+					.with_message("Code content cannot be empty")
+					.with_color(parser.colors().error))
+				.finish());
+			return reports;
+		}
+
+		let theme = document.get_variable("code.theme")
+			.and_then(|(_doc, var)| Some(var.to_string()));
+
+		if index == 0 // Block
+		{
+			let code_name = matches.get(3)
+				.and_then(|name| {
+					let code_name = name.as_str().trim_end().trim_start().to_string();
+					(!code_name.is_empty()).then_some(code_name)
+				});
+			let line_offset = match properties.get("line_offset",
+				|prop, value| value.parse::<usize>().map_err(|e| (prop, e)))
+			{
+				Ok((_prop, offset)) => offset,
+				Err((prop, e)) => {
+					reports.push(
+						Report::build(ReportKind::Error, token.source(), token.start())
+						.with_message("Invalid Code Property")
+						.with_label(
+							Label::new((token.source().clone(), token.start()+1..token.end()))
+							.with_message(format!("Property `line_offset: {}` cannot be converted: {}",
+									prop.fg(parser.colors().info),
+									e.fg(parser.colors().error)))
+							.with_color(parser.colors().warning))
+						.finish());
+						return reports;
+				}
+			};
+
+			parser.push(document, Box::new(
+					Code::new(
+						token.clone(),
+						CodeKind::FullBlock,
+						code_lang,
+						code_name,
+						code_content,
+						theme,
+						line_offset
+					)
+			));
+		}
+		else // Maybe inline
+		{
+			let block = if code_content.contains('\n') { CodeKind::MiniBlock }
+			else { CodeKind::Inline };
+
+			parser.push(document, Box::new(
+					Code::new(
+						token.clone(),
+						block,
+						code_lang,
+						None,
+						code_content,
+						theme,
+						1,
+					)
+			));
+		}
+
+		reports
+    }
+}

src/elements/comment.rs

@@ -0,0 +1,81 @@
+use regex::{Captures, Regex};
+use crate::parser::{parser::Parser, rule::RegexRule, source::{Source, Token}};
+use ariadne::{Report, Label, ReportKind};
+use crate::{compiler::compiler::Compiler, document::{document::Document, element::{ElemKind, Element}}};
+use std::{ops::Range, rc::Rc};
+
+#[derive(Debug)]
+pub struct Comment {
+    location: Token,
+	content: String,
+}
+
+impl Comment
+{
+    pub fn new(location: Token, content: String ) -> Self {
+        Self { location: location, content }
+    }
+}
+
+impl Element for Comment
+{
+    fn location(&self) -> &Token { &self.location }
+    fn kind(&self) -> ElemKind { ElemKind::Invisible }
+    fn element_name(&self) -> &'static str { "Comment" }
+    fn to_string(&self) -> String { format!("{self:#?}") }
+    fn compile(&self, _compiler: &Compiler, _document: &Document)
+		-> Result<String, String> {
+		Ok("".to_string())
+    }
+}
+
+pub struct CommentRule {
+	re: [Regex; 1],
+}
+
+impl CommentRule {
+	pub fn new() -> Self {
+		Self { re: [Regex::new(r"\s*::(.*)").unwrap()] }
+	}
+}
+
+impl RegexRule for CommentRule {
+	fn name(&self) -> &'static str { "Comment" }
+
+	fn regexes(&self) -> &[Regex] { &self.re }
+
+	fn on_regex_match(&self, _: usize, parser: &dyn Parser, document: &Document, token: Token, matches: Captures)
+		-> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>> {
+		let mut reports = vec![];
+
+		let content = match matches.get(1)
+		{
+			None => panic!("Unknown error"),
+			Some(comment) => {
+				let trimmed = comment.as_str().trim_start().trim_end().to_string();
+				if trimmed.is_empty()
+				{
+					reports.push(
+						Report::build(ReportKind::Warning, token.source(), comment.start())
+						.with_message("Empty comment")
+						.with_label(
+							Label::new((token.source(), comment.range()))
+							.with_message("Comment is empty")
+							.with_color(parser.colors().warning))
+						.finish());
+				}
+
+				trimmed
+			}
+		};
+		
+        parser.push(document, Box::new(
+            Comment::new(
+				token.clone(),
+                content
+            )
+        ));
+
+        return reports;
+	}
+}

src/elements/import.rs

@@ -0,0 +1,155 @@
+use regex::Regex;
+use crate::parser::{parser::{Parser, ReportColors}, rule::RegexRule, source::{Source, SourceFile, Token}};
+use ariadne::{Report, Fmt, Label, ReportKind};
+use crate::document::document::Document;
+use std::{ops::Range, rc::Rc};
+
+use super::paragraph::Paragraph;
+
+pub struct ImportRule {
+	re: [Regex; 1],
+}
+
+impl ImportRule {
+	pub fn new() -> Self {
+		Self {
+            re: [Regex::new(r"(?:^|\n)@import(?:\[(.*)\])?[^\S\r\n]+(.*)").unwrap()],
+        }
+	}
+
+    pub fn validate_name(_colors: &ReportColors, name: &str) -> Result<String, String>
+    {
+        Ok(name.to_string())
+    }
+
+    pub fn validate_as(_colors: &ReportColors, as_name: &str) -> Result<String, String>
+    {
+        // TODO: Use variable name validation rules
+        Ok(as_name.to_string())
+    }
+}
+
+impl RegexRule for ImportRule {
+	fn name(&self) -> &'static str { "Import" }
+
+	fn regexes(&self) -> &[Regex] { &self.re }
+
+	fn on_regex_match(&self, _: usize, parser: &dyn Parser, document: &Document, token: Token, matches: regex::Captures) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>
+	{
+		let mut result = vec![];
+
+        // Path
+        let import_file = match matches.get(2)
+        {
+            Some(name) => {
+                match ImportRule::validate_name(parser.colors(), name.as_str())
+                {
+                    Err(msg) => {
+                        result.push(
+                            Report::build(ReportKind::Error, token.source(), name.start())
+                            .with_message("Invalid name for import")
+                            .with_label(
+                                Label::new((token.source(), name.range()))
+                                .with_message(format!("Import name `{}` is invalid. {msg}",
+                                        name.as_str().fg(parser.colors().highlight)))
+                                .with_color(parser.colors().error))
+                            .finish());
+
+                        return result;
+                    },
+                    Ok(filename) => {
+                        let meta = match std::fs::metadata(filename.as_str())
+                        {
+                            Err(_) => {
+                                result.push(
+                                    Report::build(ReportKind::Error, token.source(), name.start())
+                                    .with_message("Invalid import path")
+                                    .with_label(
+                                        Label::new((token.source(), name.range()))
+                                        .with_message(format!("Unable to access file `{}`",
+                                                filename.fg(parser.colors().highlight)))
+                                        .with_color(parser.colors().error))
+                                    .finish());
+                                    return result;
+                            },
+                            Ok(meta) => meta
+                        };
+
+                        if !meta.is_file()
+                        {
+                            result.push(
+                                Report::build(ReportKind::Error, token.source(), name.start())
+                                .with_message("Invalid import path")
+                                .with_label(
+                                    Label::new((token.source(), name.range()))
+                                    .with_message(format!("Path `{}` is not a file!",
+                                            filename.fg(parser.colors().highlight)))
+                                    .with_color(parser.colors().error))
+                                .finish());
+                            return result;
+                        }
+
+                        filename
+                    },
+                }
+            }
+            _ => panic!("Invalid name for import")
+        };
+
+        // [Optional] import as
+        let import_as = match matches.get(1)
+        {
+            Some(as_name) => {
+                match ImportRule::validate_as(parser.colors(), as_name.as_str())
+                {
+                    Ok(as_name) => as_name,
+                    Err(msg) => {
+                        result.push(
+                            Report::build(ReportKind::Error, token.source(), as_name.start())
+                            .with_message("Invalid name for import as")
+                            .with_label(
+                                Label::new((token.source(), as_name.range()))
+                                .with_message(format!("Canot import `{import_file}` as `{}`. {msg}",
+                                        as_name.as_str().fg(parser.colors().highlight)))
+                                .with_color(parser.colors().error))
+                            .finish());
+
+                        return result;
+                    },
+                }
+            }
+            _ => "".to_string()
+        };
+
+        let import = match SourceFile::new(import_file, Some(token.clone()))
+		{
+			Ok(import) => Rc::new(import),
+			Err(path) => {
+                result.push(
+					Report::build(ReportKind::Error, token.source(), token.start())
+					.with_message("Unable to read file content")
+					.with_label(
+						Label::new((token.source(), token.range))
+						.with_message(format!("Failed to read content from path `{path}`"))
+						.with_color(parser.colors().error))
+					.finish());
+				return result;
+			}
+		};
+
+		// TODO
+        let import_doc = parser.parse(import, Some(&document));
+
+        document.merge(import_doc, Some(&import_as));
+
+		// Close paragraph
+		if document.last_element::<Paragraph>(false).is_some()
+		{
+			parser.push(document, Box::new(Paragraph::new(
+				Token::new(token.end()..token.end(), token.source())
+			)));
+		}
+
+        return result;
+	}
+}

src/elements/link.rs

@@ -0,0 +1,149 @@
+use regex::Regex;
+use crate::parser::{parser::Parser, rule::RegexRule, source::{Source, Token}, util};
+use ariadne::{Report, Fmt, Label, ReportKind};
+use crate::{compiler::compiler::{Compiler, Target}, document::{document::Document, element::{ElemKind, Element}}};
+use std::{ops::Range, rc::Rc};
+
+#[derive(Debug)]
+pub struct Link {
+    location: Token,
+	name: String, // Link name
+	url: String, // Link url
+}
+
+impl Link
+{
+    pub fn new(location: Token, name: String, url: String) -> Self {
+        Self { location: location, name, url }
+    }
+}
+
+impl Element for Link
+{
+    fn location(&self) -> &Token { &self.location }
+    fn kind(&self) -> ElemKind { ElemKind::Inline }
+    fn element_name(&self) -> &'static str { "Link" }
+    fn to_string(&self) -> String { format!("{self:#?}") }
+    fn compile(&self, compiler: &Compiler, _document: &Document) -> Result<String, String> {
+        match compiler.target()
+        {
+            Target::HTML => {
+                Ok(format!("<a href=\"{}\">{}</a>",
+                    compiler.sanitize(self.url.as_str()),
+                    compiler.sanitize(self.name.as_str()),
+                ))
+            },
+            Target::LATEX => {
+                Ok(format!("\\href{{{}}}{{{}}}",
+                    compiler.sanitize(self.url.as_str()),
+                    compiler.sanitize(self.name.as_str()),
+                ))
+            },
+        }
+    }
+}
+
+pub struct LinkRule {
+	re: [Regex; 1],
+}
+
+impl LinkRule {
+	pub fn new() -> Self {
+		Self { re: [Regex::new(r"(?:^|\n)```(.*?)(?:,(.*))?\n((?:\\.|[^\[\]\\])*?)```").unwrap()] }
+	}
+}
+
+impl RegexRule for LinkRule {
+	fn name(&self) -> &'static str { "Link" }
+
+	fn regexes(&self) -> &[Regex] { &self.re }
+
+	fn on_regex_match(&self, _: usize, parser: &dyn Parser, document: &Document, token: Token, matches: regex::Captures) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>
+	{
+		let mut result = vec![];
+        let link_name = match matches.get(1)
+        {
+            Some(name) => {
+                if name.as_str().is_empty()
+                {
+					result.push(
+						Report::build(ReportKind::Error, token.source(), name.start())
+						.with_message("Empty link name")
+						.with_label(
+							Label::new((token.source().clone(), name.range()))
+							.with_message("Link name is empty")
+							.with_color(parser.colors().error))
+						.finish());
+					return result;
+                }
+				// TODO: process into separate document...
+                let text_content = util::process_text(document, name.as_str());
+
+                if text_content.as_str().is_empty()
+                {
+					result.push(
+						Report::build(ReportKind::Error, token.source(), name.start())
+						.with_message("Empty link name")
+						.with_label(
+							Label::new((token.source(), name.range()))
+							.with_message(format!("Link name is empty. Once processed, `{}` yields `{}`",
+                                name.as_str().fg(parser.colors().highlight),
+                                text_content.as_str().fg(parser.colors().highlight),
+                                ))
+							.with_color(parser.colors().error))
+						.finish());
+					return result;
+                }
+				text_content
+            },
+            _ => panic!("Empty link name"),
+        };
+
+        let link_url = match matches.get(2)
+        {
+            Some(url) => {
+                if url.as_str().is_empty()
+                {
+					result.push(
+						Report::build(ReportKind::Error, token.source(), url.start())
+						.with_message("Empty link url")
+						.with_label(
+							Label::new((token.source(), url.range()))
+							.with_message("Link url is empty")
+							.with_color(parser.colors().error))
+						.finish());
+					return result;
+                }
+                let text_content = util::process_text(document, url.as_str());
+
+                if text_content.as_str().is_empty()
+                {
+					result.push(
+						Report::build(ReportKind::Error, token.source(), url.start())
+						.with_message("Empty link url")
+						.with_label(
+							Label::new((token.source(), url.range()))
+							.with_message(format!("Link url is empty. Once processed, `{}` yields `{}`",
+                                url.as_str().fg(parser.colors().highlight),
+                                text_content.as_str().fg(parser.colors().highlight),
+                                ))
+							.with_color(parser.colors().error))
+						.finish());
+					return result;
+                }
+				text_content
+            },
+            _ => panic!("Empty link url"),
+        };
+
+        parser.push(document, Box::new(
+            Link::new(
+				token.clone(),
+                link_name,
+                link_url
+            )
+        ));
+
+        return result;
+	}
+}

src/elements/list.rs

@@ -0,0 +1,335 @@
+use std::{any::Any, cell::Ref, ops::Range, rc::Rc};
+
+use crate::{compiler::compiler::{Compiler, Target}, document::{document::Document, element::{ElemKind, Element}}, parser::{parser::Parser, rule::Rule, source::{Cursor, Source, Token, VirtualSource}}};
+use ariadne::{Label, Report, ReportKind};
+use regex::Regex;
+
+use super::paragraph::Paragraph;
+
+#[derive(Debug)]
+pub struct ListEntry {
+	location: Token,
+	numbering: Vec<(bool, usize)>,
+	content: Vec<Box<dyn Element>>,
+
+	// TODO bullet_maker : FnMut<...>
+}
+
+impl ListEntry {
+	pub fn new(location: Token, numbering: Vec<(bool, usize)>, content: Vec<Box<dyn Element>>) -> Self {
+		Self { location, numbering, content }
+	}
+}
+
+#[derive(Debug)]
+pub struct List
+{
+	location: Token,
+	entries: Vec<ListEntry>
+}
+
+impl List
+{
+	pub fn new(location: Token) -> Self
+	{
+		Self
+		{
+			location,
+			entries: Vec::new()
+		}
+	}
+
+	pub fn push(&mut self, entry: ListEntry)
+	{
+		self.location.range = self.location.start()..entry.location.end();
+		self.entries.push(entry);
+	}
+}
+
+impl Element for List
+{
+    fn location(&self) -> &Token { &self.location }
+
+    fn kind(&self) -> ElemKind { ElemKind::Block }
+
+    fn element_name(&self) -> &'static str { "List" }
+
+    fn to_string(&self) -> String { format!("{self:#?}") }
+
+    fn compile(&self, compiler: &Compiler, document: &Document) -> Result<String, String> {
+		match compiler.target()
+		{
+			Target::HTML => {
+				let mut result = String::new();
+
+				//TODO: Do something about indexing
+				let mut current_list: Vec<bool> = vec![];
+				let mut match_stack = |result: &mut String, target: &Vec<(bool, usize)>| {
+
+					// Find index after which current_list and target differ
+					let mut match_idx = 0usize;
+					for i in 0..current_list.len()
+					{
+						if i >= target.len() || current_list[i] != target[i].0 { break }
+						else { match_idx = i+1; }
+					}
+
+					// Close until same match
+					for _ in match_idx..current_list.len()
+					{
+						result.push_str(["</ul>", "</ol>"][current_list.pop().unwrap() as usize]);
+					}
+
+					// Open
+					for i in match_idx..target.len()
+					{
+						result.push_str(["<ul>", "<ol>"][target[i].0 as usize]);
+						current_list.push(target[i].0);
+					}
+				};
+
+				match self.entries.iter()
+					.try_for_each(|ent|
+					{
+						match_stack(&mut result, &ent.numbering);
+						result.push_str("<li>");
+						match ent.content.iter().enumerate()
+							.try_for_each(|(idx, elem)| { 
+								match elem.compile(compiler, document) {
+									Err(e) => Err(e),
+									Ok(s) => { result.push_str(s.as_str()); Ok(()) }
+								}
+							})
+						{
+							Err(e) => Err(e),
+							_ => {
+								result.push_str("</li>");
+								Ok(())
+							}
+						}
+					})
+				{
+					Err(e) => return Err(e),
+					_ => {}
+				}
+				match_stack(&mut result, &Vec::<(bool, usize)>::new());
+
+				Ok(result)
+			}
+			Target::LATEX => Err("Unimplemented compiler".to_string())
+		}
+    }
+}
+
+/*
+impl Element for ListEntry
+{
+	fn location(&self) -> &Token { &self.location }
+	fn kind(&self) -> ElemKind { ElemKind::Inline }
+	fn element_name(&self) -> &'static str { "List" }
+	fn to_string(&self) -> String { format!("{self:#?}") }
+	fn compile(&self, compiler: &Compiler) -> Result<String, String> {
+		lazy_static! {
+			static ref STATE_NAME : &'static str = "list.state";
+			static ref LIST_OPEN : [&'static str; 2] = ["<ul>", "<ol>"];
+			static ref LIST_CLOSE : [&'static str; 2] = ["</ul>", "</ol>"];
+		}
+
+		// TODO: State.shouldpreserve?
+		// Called upon every element
+		//let state = compiler.get_state_mut::<ListState, _>(*STATE_NAME)
+		//.or_else(|| {
+		//	compiler.insert_state(STATE_NAME.to_string(), Box::new(ListState(Vec::new())) as Box<dyn Any>);
+		//	compiler.get_state_mut::<ListState, _>(*STATE_NAME)
+		//}).unwrap();
+
+		match compiler.target()
+		{
+			Target::HTML => {
+				let mut result = String::new();
+
+				//TODO: Do something about indexing
+				//&self.numbering.iter()
+				//	.zip(&state.0)
+				//	.for_each(|((wants_numbered, _), is_numbered)|
+				//	{
+				//		
+				//	});
+
+				result.push_str("<li>");
+				match self.content.iter()
+					.try_for_each(|ent| match ent.compile(compiler) {
+						Err(e) => Err(e),
+						Ok(s) => Ok(result.push_str(s.as_str())),
+					})
+				{
+					Err(e) => return Err(e),
+					_ => {}
+				}
+				result.push_str("</li>");
+				//result.push_str(LIST_OPEN[self.numbered as usize]);
+				//self.entries.iter()
+				//	.for_each(|(_index, entry)|
+				//		result.push_str(format!("<li>{}</li>", compiler.compile(entry)).as_str()));
+				//result.push_str(LIST_CLOSE[self.numbered as usize]);
+				Ok(result)
+			}
+			Target::LATEX => Err("Unimplemented compiler".to_string())
+		}
+	}
+}
+*/
+
+pub struct ListRule
+{
+	start_re: Regex,
+	continue_re: Regex
+}
+
+impl ListRule {
+	pub fn new() -> Self {
+		Self {
+			start_re: Regex::new(r"(?:^|\n)(?:[^\S\r\n]+)([*-]+).*").unwrap(),
+			continue_re: Regex::new(r"(?:^|\n)([^\S\r\n]+).*").unwrap(),
+		}
+
+	}
+
+	fn parse_depth(depth: &str, document: &Document) -> Vec<(bool, usize)>
+	{
+		let mut parsed = vec![];
+		let prev_entry = document.last_element::<List>(true)
+			.and_then(|list| Ref::filter_map(list, |m| m.entries.last() ).ok() )
+			.and_then(|entry| Ref::filter_map(entry, |e| Some(&e.numbering)).ok() );
+
+		let mut continue_match = true;
+		depth.chars().enumerate().for_each(|(idx, c)|
+		{
+			let number = prev_entry.as_ref()
+				.and_then(|v| {
+					if !continue_match { return None }
+					let numbered = c == '-';
+
+					match v.get(idx)
+					{
+						None => None,
+						Some((prev_numbered, prev_idx)) => {
+							if *prev_numbered != numbered { continue_match = false; None } // New depth
+							else if idx+1 == v.len() { Some(prev_idx+1) } // Increase from previous
+							else { Some(*prev_idx) } // Do nothing
+						}
+					}
+				})
+				.or(Some(0usize))
+				.unwrap();
+
+			match c
+			{
+				'*' => parsed.push((false, number)),
+				'-' => parsed.push((true, number)),
+				_ => panic!("Unimplemented")
+			}
+		});
+
+		return parsed;
+	}
+}
+
+impl Rule for ListRule
+{
+	fn name(&self) -> &'static str { "List" }
+
+	fn next_match(&self, cursor: &Cursor) -> Option<(usize, Box<dyn Any>)> {
+		self.start_re.find_at(cursor.source.content(), cursor.pos)
+			.map_or(None,
+			|m| Some((m.start(), Box::new([false;0]) as Box<dyn Any>)) )
+	}
+
+	fn on_match<'a>(&self, parser: &dyn Parser, document: &'a Document<'a>, cursor: Cursor, _match_data: Option<Box<dyn Any>>) -> (Cursor, Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>) {
+		let mut reports = vec![];
+		let content = cursor.source.content();
+		let (end_cursor, numbering, source) = match self.start_re.captures_at(content, cursor.pos) {
+			None => panic!("Unknown error"),
+			Some(caps) => {
+				let mut end_pos = caps.get(0).unwrap().end();
+
+				let mut spacing = None; // Spacing used to continue list entry
+				loop {
+					// If another entry starts on the next line, don't continue matching
+					match self.next_match(&cursor.at(end_pos))
+					{
+						Some((pos, _)) => {
+							if pos == end_pos { break }
+						}
+						None => {},
+					}
+
+					// Continue matching as current entry
+					match self.continue_re.captures_at(content, end_pos) {
+						None => break,
+						Some(continue_caps) => {
+							if continue_caps.get(0).unwrap().start() != end_pos { break }
+
+							// Get the spacing
+							let cap_spacing = continue_caps.get(1).unwrap();
+							match &spacing {
+								None => spacing = Some(cap_spacing.range()),
+								Some(spacing) => 'some: {
+									if content[cap_spacing.range()] == content[spacing.clone()] { break 'some }
+
+									reports.push(
+										Report::build(ReportKind::Warning, cursor.source.clone(), continue_caps.get(1).unwrap().start())
+										.with_message("Invalid list entry spacing")
+										.with_label(
+											Label::new((cursor.source.clone(), cap_spacing.range()))
+											.with_message("Spacing for list entries must match")
+											.with_color(parser.colors().warning))
+										.with_label(
+											Label::new((cursor.source.clone(), spacing.clone()))
+											.with_message("Previous spacing")
+											.with_color(parser.colors().warning))
+										.finish());
+								},
+							}
+							end_pos = continue_caps.get(0).unwrap().end();
+						}
+					}
+				}
+
+				let start_pos = caps.get(1).unwrap().end();
+				let source = VirtualSource::new(
+					Token::new(start_pos..end_pos, cursor.source.clone()),
+					"List Entry".to_string(),
+					content.as_str()[start_pos..end_pos].to_string(),
+				);
+
+				(cursor.at(end_pos),
+				ListRule::parse_depth(caps.get(1).unwrap().as_str(), document),
+				source)
+			},
+		};
+
+        let parsed_entry = parser.parse(Rc::new(source), Some(&document));
+		let mut parsed_paragraph = parsed_entry.last_element_mut::<Paragraph>(false).unwrap(); // Extract content from paragraph
+		let entry = ListEntry::new(
+			Token::new(cursor.pos..end_cursor.pos, cursor.source.clone()),
+			numbering,
+			std::mem::replace(&mut parsed_paragraph.content, Vec::new())
+		);
+
+		// Ger previous list, if none insert a new list
+		let mut list = match document.last_element_mut::<List>(false)
+		{
+			Some(last) => last,
+			None => {
+				parser.push(document,
+					Box::new(List::new(
+							Token::new(cursor.pos..end_cursor.pos, cursor.source.clone()))));
+				document.last_element_mut::<List>(false).unwrap()
+			}
+		};
+		list.push(entry);
+
+		(end_cursor, reports)
+	}
+}

src/elements/mod.rs

@@ -0,0 +1,13 @@
+pub mod registrar;
+pub mod comment;
+pub mod paragraph;
+pub mod variable;
+pub mod import;
+pub mod script;
+pub mod list;
+pub mod style;
+pub mod section;
+pub mod link;
+pub mod code;
+pub mod tex;
+pub mod raw;

src/elements/paragraph.rs

@@ -0,0 +1,127 @@
+use std::{any::Any, ops::Range, rc::Rc};
+
+use ariadne::Report;
+use regex::Regex;
+
+use crate::{compiler::compiler::{Compiler, Target}, document::{document::Document, element::{ElemKind, Element}}, parser::{parser::Parser, rule::Rule, source::{Cursor, Source, Token}}};
+
+// TODO: Full refactor
+// Problem is that document parsed from other sources i.e by variables
+// are not merged correctly into existing paragraph
+// A solution would be to use the "(\n){2,}" regex to split paragraph, which would reduce the work needed for process_text
+// Another fix would be to keep parsing (recursively) into the same document (like previous version)
+// The issue is that this would break the current `Token` implementation
+// Which would need to be reworked
+#[derive(Debug)]
+pub struct Paragraph
+{
+    location: Token,
+	pub content: Vec<Box<dyn Element>>
+}
+
+impl Paragraph
+{
+    pub fn new(location: Token) -> Self {
+        Self { location, content: Vec::new() }
+    }
+
+	pub fn is_empty(&self) -> bool { self.content.is_empty() }
+
+	pub fn push(&mut self, elem: Box<dyn Element>)
+	{
+		if elem.location().source() == self.location().source()
+		{
+			self.location.range = self.location.start() .. elem.location().end();
+		}
+		self.content.push(elem);
+	}
+
+	pub fn find_back<P: FnMut(&&Box<dyn Element + 'static>) -> bool>(&self, mut predicate: P)
+		-> Option<&Box<dyn Element>> {
+		self.content.iter().rev()
+			.find(predicate)
+	}
+}
+
+impl Element for Paragraph
+{
+    fn location(&self) -> &Token { &self.location }
+
+    fn kind(&self) -> ElemKind { ElemKind::Special }
+
+    fn element_name(&self) -> &'static str { "Paragraph" }
+
+    fn to_string(&self) -> String { format!("{:#?}", self)  }
+
+    fn compile(&self, compiler: &Compiler, document: &Document) -> Result<String, String> {
+		if self.content.is_empty() { return Ok(String::new()) }
+
+        match compiler.target()
+        {
+            Target::HTML => {
+				let mut result = String::new();
+				//if prev.is_none() || prev.unwrap().downcast_ref::<Paragraph>().is_none()
+				{ result.push_str("<p>"); }
+				//else
+				//{ result.push_str(" "); }
+
+				let err = self.content.iter().try_for_each(|elem| {
+					match elem.compile(compiler, document)
+					{
+						Err(e) => return Err(e),
+						Ok(content) => { result.push_str(content.as_str()); Ok(()) },
+					}
+				});
+				//if next.is_none() || next.unwrap().downcast_ref::<Paragraph>().is_none()
+				{ result.push_str("</p>"); }
+
+				match err
+				{
+					Err(e) => Err(e),
+					Ok(()) => Ok(result),
+				}
+            }
+            Target::LATEX => todo!("Unimplemented compiler")
+        }
+    }
+}
+
+pub struct ParagraphRule
+{
+	re: Regex,
+}
+
+impl ParagraphRule {
+    pub fn new() -> Self {
+        Self {
+			re: Regex::new(r"\n{2,}").unwrap()
+		}
+    }
+}
+
+impl Rule for ParagraphRule
+{
+    fn name(&self) -> &'static str { "Paragraphing" }
+
+    fn next_match(&self, cursor: &Cursor) -> Option<(usize, Box<dyn Any>)> {
+		self.re.find_at(cursor.source.content(), cursor.pos)
+			.and_then(|m| Some((m.start(), Box::new([false;0]) as Box<dyn Any>)) )
+    }
+
+    fn on_match(&self, parser: &dyn Parser, document: &Document, cursor: Cursor, _match_data: Option<Box<dyn Any>>)
+		-> (Cursor, Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>) {
+
+		let end_cursor = match self.re.captures_at(cursor.source.content(), cursor.pos)
+		{
+			None => panic!("Unknown error"),
+			Some(capture) =>
+				cursor.at(capture.get(0).unwrap().end()-1)
+		};
+
+		parser.push(document, Box::new(Paragraph::new(
+			Token::new(cursor.pos..end_cursor.pos, cursor.source.clone())
+		)));
+
+		(end_cursor, Vec::new())
+    }
+}

src/elements/raw.rs

@@ -0,0 +1,164 @@
+use regex::{Captures, Regex};
+use crate::{compiler::compiler::Compiler, document::element::{ElemKind, Element}, parser::{parser::Parser, rule::RegexRule, source::{Source, Token}, util::{self, Property, PropertyParser}}};
+use ariadne::{Fmt, Label, Report, ReportKind};
+use crate::document::document::Document;
+use std::{collections::HashMap, ops::Range, rc::Rc, str::FromStr};
+
+#[derive(Debug)]
+struct Raw {
+	location: Token,
+	kind: ElemKind,
+	content: String,
+}
+
+impl Raw {
+    fn new(location: Token, kind: ElemKind, content: String) -> Self {
+        Self { location, kind, content }
+    }
+}
+
+impl Element for Raw {
+    fn location(&self) -> &Token { &self.location }
+    fn kind(&self) -> ElemKind { self.kind.clone() }
+
+    fn element_name(&self) -> &'static str { "Raw" }
+
+    fn to_string(&self) -> String { format!("{self:#?}") }
+
+    fn compile(&self, compiler: &Compiler, _document: &Document) -> Result<String, String> {
+		Ok(self.content.clone())
+    }
+}
+
+pub struct RawRule {
+	re: [Regex; 1],
+	properties: PropertyParser,
+}
+
+impl RawRule {
+	pub fn new() -> Self {
+		let mut props = HashMap::new();
+		props.insert("kind".to_string(),
+			Property::new(
+				true,
+				"Element display kind".to_string(),
+					Some("inline".to_string())));
+		Self {
+            re: [
+				Regex::new(r"\{\?(?:\[((?:\\.|[^\[\]\\])*?)\])?(?:((?:\\.|[^\\\\])*?)(\?\}))?").unwrap()
+			],
+			properties: PropertyParser::new(props)
+        }
+	}
+}
+
+impl RegexRule for RawRule
+{
+    fn name(&self) -> &'static str { "Raw" }
+
+    fn regexes(&self) -> &[regex::Regex] { &self.re }
+
+    fn on_regex_match(&self, _index: usize, parser: &dyn Parser, document: &Document, token: Token, matches: Captures)
+		-> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>> {
+		let mut reports = vec![];
+
+		let raw_content = match matches.get(2)
+		{
+			// Unterminated
+			None => {
+				reports.push(
+					Report::build(ReportKind::Error, token.source(), token.start())
+					.with_message("Unterminated Raw Code")
+					.with_label(
+						Label::new((token.source().clone(), token.range.clone()))
+						.with_message(format!("Missing terminating `{}` after first `{}`",
+								"?}".fg(parser.colors().info),
+								"{?".fg(parser.colors().info)))
+						.with_color(parser.colors().error))
+					.finish());
+					return reports;
+			}
+			Some(content) => {
+				let processed = util::process_escaped('\\', "?}",
+					content.as_str().trim_start().trim_end());
+
+				if processed.is_empty()
+				{
+					reports.push(
+						Report::build(ReportKind::Warning, token.source(), content.start())
+						.with_message("Empty Raw Code")
+						.with_label(
+							Label::new((token.source().clone(), content.range()))
+							.with_message("Raw code is empty")
+							.with_color(parser.colors().warning))
+						.finish());
+				}
+				processed
+			}
+		};
+
+		let properties = match matches.get(1)
+		{
+			None => match self.properties.default() {
+				Ok(properties) => properties,
+				Err(e) => {
+					reports.push(
+						Report::build(ReportKind::Error, token.source(), token.start())
+						.with_message("Invalid Raw Code")
+						.with_label(
+							Label::new((token.source().clone(), token.range.clone()))
+							.with_message(format!("Raw code is missing properties: {e}"))
+							.with_color(parser.colors().error))
+						.finish());
+						return reports;
+				},
+			}		
+			Some(props) => {
+				let processed = util::process_escaped('\\', "]",
+					props.as_str().trim_start().trim_end());
+				match self.properties.parse(processed.as_str())
+				{
+					Err(e) => {
+						reports.push(
+							Report::build(ReportKind::Error, token.source(), props.start())
+							.with_message("Invalid Raw Code Properties")
+							.with_label(
+								Label::new((token.source().clone(), props.range()))
+								.with_message(e)
+								.with_color(parser.colors().error))
+							.finish());
+						return reports;
+					}
+					Ok(properties) => properties
+				}
+			}
+		};
+
+		let raw_kind : ElemKind = match properties.get("kind",
+			|prop, value| ElemKind::from_str(value.as_str()).map_err(|e| (prop, e)))
+		{
+			Ok((_prop, kind)) => kind,
+			Err((prop, e)) => {
+				reports.push(
+					Report::build(ReportKind::Error, token.source(), token.start())
+					.with_message("Invalid Raw Code Property")
+					.with_label(
+						Label::new((token.source().clone(), token.range.clone()))
+						.with_message(format!("Property `kind: {}` cannot be converted: {}",
+								prop.fg(parser.colors().info),
+								e.fg(parser.colors().error)))
+						.with_color(parser.colors().warning))
+					.finish());
+					return reports;
+			}
+		};
+
+		parser.push(document, Box::new(Raw::new(
+			token.clone(),
+			raw_kind,
+			raw_content
+		)));
+
+        reports
+    }
+}

src/elements/registrar.rs

@@ -0,0 +1,22 @@
+use crate::parser::parser::Parser;
+
+use super::{code::CodeRule, comment::CommentRule, import::ImportRule, link::LinkRule, list::ListRule, paragraph::ParagraphRule, raw::RawRule, script::ScriptRule, section::SectionRule, style::StyleRule, tex::TexRule, variable::{VariableRule, VariableSubstitutionRule}};
+
+
+pub fn register<P: Parser>(parser: &mut P)
+{
+	parser.add_rule(Box::new(CommentRule::new()), None);
+	parser.add_rule(Box::new(ParagraphRule::new()), None);
+    parser.add_rule(Box::new(ImportRule::new()), None);
+    parser.add_rule(Box::new(ScriptRule::new()), None);
+    parser.add_rule(Box::new(VariableRule::new()), None);
+    parser.add_rule(Box::new(VariableSubstitutionRule::new()), None);
+    parser.add_rule(Box::new(RawRule::new()), None);
+    parser.add_rule(Box::new(ListRule::new()), None);
+    parser.add_rule(Box::new(CodeRule::new()), None);
+    parser.add_rule(Box::new(TexRule::new()), None);
+
+    parser.add_rule(Box::new(StyleRule::new()), None);
+    parser.add_rule(Box::new(SectionRule::new()), None);
+    parser.add_rule(Box::new(LinkRule::new()), None);
+}

src/elements/script.rs

@@ -0,0 +1,201 @@
+use regex::{Captures, Regex};
+use crate::{document::element::Text, lua::kernel::{Kernel, KernelHolder}, parser::{parser::{Parser, ReportColors}, rule::RegexRule, source::{Source, Token, VirtualSource}, util}};
+use ariadne::{Fmt, Label, Report, ReportKind};
+use crate::document::document::Document;
+use std::{ops::Range, rc::Rc};
+
+pub struct ScriptRule
+{
+	re: [Regex; 2],
+	eval_kinds: [(&'static str, &'static str); 2]
+}
+
+impl ScriptRule {
+	pub fn new() -> Self {
+		Self {
+            re: [
+				Regex::new(r"(?:^|\n)@<(?:(.*)\n?)((?:\\.|[^\[\]\\])*?)(?:\n?)>@").unwrap(),
+				Regex::new(r"%<([^\s[:alpha:]])?(?:\[(.*?)\])?((?:\\.|[^\[\]\\])*?)(?:\n?)>%").unwrap()
+			],
+			eval_kinds: [
+				("", "Eval to text"),
+				("!", "Eval and parse"),
+			]
+        }
+	}
+
+	fn validate_kernel_name(colors: &ReportColors, name: &str)
+		-> Result<String, String> {
+		let trimmed = name.trim_end().trim_start();
+		if trimmed.is_empty() { return Ok("main".to_string()) }
+		else if trimmed.find(|c: char| c.is_whitespace()).is_some() { 
+			return Err(format!("Kernel name `{}` contains whitespaces",
+					trimmed.fg(colors.highlight)))
+		}
+
+		Ok(trimmed.to_string())
+	}
+
+	fn validate_kind(&self, colors: &ReportColors, kind: &str)
+		-> Result<usize, String> {
+		match self.eval_kinds.iter().position(|(kind_symbol, _)| kind == *kind_symbol)
+		{
+			Some(id) => Ok(id),
+			None => Err(format!("Unable to find eval kind `{}`. Available kinds:{}",
+					kind.fg(colors.highlight),
+					self.eval_kinds.iter().fold(String::new(), |out, (symbol, name)| {
+						out + format!("\n - '{symbol}' => {name}").as_str()
+					})))
+		}
+	}
+}
+
+impl RegexRule for ScriptRule
+{
+    fn name(&self) -> &'static str { "Script" }
+
+    fn regexes(&self) -> &[regex::Regex] { &self.re }
+
+    fn on_regex_match(&self, index: usize, parser: &dyn Parser, document: &Document, token: Token, matches: Captures)
+		-> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>> {
+		let mut reports = vec![];
+
+		let kernel_name = match matches.get(if index == 0 {1} else {2}) {
+			None => "main".to_string(),
+			Some(name) => {
+				match ScriptRule::validate_kernel_name(parser.colors(), name.as_str())
+				{
+					Ok(name) => name,
+					Err(e) => {
+						reports.push(
+							Report::build(ReportKind::Error, token.source(), name.start())
+							.with_message("Invalid kernel name")
+							.with_label(
+								Label::new((token.source(), name.range()))
+								.with_message(e)
+								.with_color(parser.colors().error))
+							.finish());
+						return reports;
+					}
+				}
+			}
+		};
+		let kernel_name = matches.get(if index == 0 {1} else {2})
+			.and_then(|name| {
+				let trimmed = name.as_str().trim_start().trim_end();
+				(!trimmed.is_empty()).then_some(trimmed)
+			})
+			.unwrap_or("main");
+		let kernel = parser.get_kernel(kernel_name).unwrap_or_else(|| {
+			parser.insert_kernel(kernel_name.to_string(), Kernel::new())
+		});
+
+		let kernel_data = matches.get(if index == 0 {2} else {3})
+			.and_then(|code| { 
+				let trimmed = code.as_str().trim_start().trim_end();
+				(!trimmed.is_empty()).then_some((trimmed, code.range()))
+			}).or_else(|| {
+				reports.push(
+					Report::build(ReportKind::Warning, token.source(), token.start())
+					.with_message("Invalid kernel code")
+					.with_label(
+						Label::new((token.source(), token.start()+1..token.end()))
+						.with_message("Kernel code is empty")
+						.with_color(parser.colors().warning))
+					.finish());
+
+				None
+			});
+
+		if kernel_data.is_none() { return reports; }
+
+		let (kernel_content, kernel_range) = kernel_data.unwrap();
+		let source = Rc::new(VirtualSource::new(
+			Token::new(kernel_range, token.source()),
+			format!("{}#{}:lua_kernel@{kernel_name}", token.source().name(), matches.get(0).unwrap().start()),
+			util::process_escaped('\\', ">@", kernel_content)
+		)) as Rc<dyn Source>;
+		
+		let chunk = kernel.lua.load(source.content())
+			.set_name(kernel_name);
+		if index == 0 // @< ... >@ -> Exec
+		{
+			match chunk.exec()
+			{
+				Ok(_) => {},
+				Err(e) => {
+					reports.push(
+						Report::build(ReportKind::Error, source.clone(), 0)
+						.with_message("Invalid kernel code")
+						.with_label(
+							Label::new((source.clone(), 0..source.content().len()))
+							.with_message(format!("Kernel execution failed:\n{}", e.to_string()))
+							.with_color(parser.colors().error))
+						.finish());
+						}
+			}
+		}
+		else if index == 1 // %< ... >% -> Eval
+		{
+			let kind = match matches.get(1) {
+				None => 0,
+				Some(kind) => {
+					match self.validate_kind(parser.colors(), kind.as_str())
+					{
+						Ok(kind) => kind,
+						Err(msg) => {
+							reports.push(
+								Report::build(ReportKind::Error, token.source(), kind.start())
+								.with_message("Invalid kernel code kind")
+								.with_label(
+									Label::new((token.source(), kind.range()))
+									.with_message(msg)
+									.with_color(parser.colors().error))
+								.finish());
+							return reports;
+						}
+					}
+				}
+			};
+
+			match chunk.eval::<String>()
+			{
+				Ok(result) => {
+					if kind == 0 // Eval to text
+					{
+						if !result.is_empty()
+						{
+							parser.push(document, Box::new(Text::new(
+							Token::new(1..source.content().len(), source.clone()),
+							util::process_text(document, result.as_str()),
+							)));
+						}
+					}
+					else if kind == 1 // Eval and Parse
+					{
+						let parse_source = Rc::new(VirtualSource::new(
+							Token::new(0..source.content().len(), source.clone()),
+							format!("parse({})", source.name()),
+							result
+						)) as Rc<dyn Source>;
+						//println!("SRC={parse_source:#?}, {}", parse_source.content());
+
+						parser.parse_into(parse_source, document);
+					}
+				},
+				Err(e) => {
+					reports.push(
+						Report::build(ReportKind::Error, source.clone(), 0)
+						.with_message("Invalid kernel code")
+						.with_label(
+							Label::new((source.clone(), 0..source.content().len()))
+							.with_message(format!("Kernel evaluation failed:\n{}", e.to_string()))
+							.with_color(parser.colors().error))
+						.finish());
+				}
+			}
+		}
+
+        reports
+    }
+}

src/elements/section.rs

@@ -0,0 +1,208 @@
+use regex::Regex;
+use crate::{compiler::compiler::Target, parser::{parser::Parser, rule::RegexRule, source::{Source, Token}}};
+use ariadne::{Report, Fmt, Label, ReportKind};
+use crate::{compiler::compiler::Compiler, document::{document::Document, element::{ElemKind, Element, ReferenceableElement}}};
+use std::{ops::Range, rc::Rc};
+
+#[derive(Debug)]
+pub struct Section {
+    location: Token,
+	title: String, // Section title
+	depth: usize, // Section depth
+	kind: u8, // Section kind, e.g numbered, unnumbred, ...
+	reference: Option<String>, // Section reference name
+}
+
+impl Section
+{
+    pub fn new(location: Token, title: String, depth: usize, kind: u8, reference: Option<String>) -> Self {
+        Self { location: location, title, depth, kind, reference }
+    }
+}
+
+impl Element for Section
+{
+    fn location(&self) -> &Token { &self.location }
+    fn kind(&self) -> ElemKind { ElemKind::Block }
+    fn element_name(&self) -> &'static str { "Section" }
+    fn to_string(&self) -> String { format!("{self:#?}") }
+    fn as_referenceable(&self) -> Option<&dyn ReferenceableElement> { Some(self) }
+    fn compile(&self, compiler: &Compiler, _document: &Document) -> Result<String, String> {
+        match compiler.target()
+        {
+            Target::HTML => {
+                Ok(format!("<h{0}>{1}</h{0}>",
+                        self.depth,
+                        compiler.sanitize(self.title.as_str())))
+            }
+            Target::LATEX => Err("Unimplemented compiler".to_string())
+        }
+    }
+}
+
+impl ReferenceableElement for Section
+{
+    fn reference_name(&self) -> Option<&String> { self.reference.as_ref() }
+}
+
+pub struct SectionRule {
+	re: [Regex; 1],
+}
+
+impl SectionRule {
+	pub fn new() -> Self {
+		Self { re: [Regex::new(r"(?:^|\n)(#{1,})(?:\{(.*)\})?((\*|\+){1,})?(.*)").unwrap()] }
+	}
+}
+
+pub mod section_kind
+{
+	pub const NONE : u8 = 0x00;
+	pub const NO_TOC : u8 = 0x01;
+	pub const NO_NUMBER : u8 = 0x02;
+}
+
+impl RegexRule for SectionRule {
+	fn name(&self) -> &'static str { "Section" }
+
+	fn regexes(&self) -> &[Regex] { &self.re }
+
+	fn on_regex_match(&self, _: usize, parser: &dyn Parser, document: &Document, token: Token, matches: regex::Captures) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>
+	{
+		let mut result = vec![];
+		let section_depth = match matches.get(1)
+		{
+			Some(depth) => {
+				if depth.len() > 6 {
+					result.push(
+					Report::build(ReportKind::Error, token.source(), depth.start())
+						.with_message("Invalid section depth")
+						.with_label(
+							Label::new((token.source(), depth.range()))
+							.with_message(format!("Section is of depth {}, which is greather than {} (maximum depth allowed)",
+                            depth.len().fg(parser.colors().info),
+                            6.fg(parser.colors().info)))
+							.with_color(parser.colors().error))
+						.finish());
+						return result;
+				}
+
+				depth.len()
+			},
+			_ => panic!("Empty section depth"),
+		};
+
+		// [Optional] Reference name
+		let section_refname = matches.get(2).map_or_else(|| None,
+			|refname| {
+				// Check for duplicate reference
+				if let Some((ref_doc, reference)) = document.get_reference(refname.as_str())
+				{
+					result.push(
+						Report::build(ReportKind::Warning, token.source(), refname.start())
+						.with_message("Duplicate reference name")
+						.with_label(
+							Label::new((token.source(), refname.range()))
+							.with_message(format!("Reference with name `{}` is already defined in `{}`",
+									refname.as_str().fg(parser.colors().highlight),
+									ref_doc.source().name().as_str().fg(parser.colors().highlight)))
+							.with_message(format!("`{}` conflicts with previously defined reference to {}",
+									refname.as_str().fg(parser.colors().highlight),
+									reference.element_name().fg(parser.colors().highlight)))
+							.with_color(parser.colors().warning))
+                        .with_label(
+                            Label::new((ref_doc.source(), reference.location().start()+1..reference.location().end() ))
+                            .with_message(format!("`{}` previously defined here",
+                                refname.as_str().fg(parser.colors().highlight)))
+							.with_color(parser.colors().warning))
+						.with_note(format!("Previous reference was overwritten"))
+						.finish());
+				}
+			Some(refname.as_str().to_string())
+		});
+
+		// Section kind
+		let section_kind = match matches.get(3)
+		{
+			Some(kind) => {
+				match kind.as_str() {
+					"*+" | "+*" => section_kind::NO_NUMBER | section_kind::NO_TOC,
+					"*" => section_kind::NO_NUMBER,
+					"+" => section_kind::NO_TOC,
+					"" => section_kind::NONE,
+					_ => {
+						result.push(
+							Report::build(ReportKind::Error, token.source(), kind.start())
+							.with_message("Invalid section numbering kind")
+							.with_label(
+								Label::new((token.source(), kind.range()))
+								.with_message(format!("Section numbering kind must be a combination of `{}` for unnumbered, and `{}` for non-listing; got `{}`",
+										"*".fg(parser.colors().info),
+										"+".fg(parser.colors().info),
+										kind.as_str().fg(parser.colors().highlight)))
+								.with_color(parser.colors().error))
+								.with_help(format!("Leave empty for a numbered listed section"))
+							.finish());
+						return result;
+					}
+				}
+			}
+			_ => section_kind::NONE,
+		};
+
+		// Spacing + Section name
+		let section_name = match matches.get(5)
+		{
+			Some(name) => {
+				let split = name.as_str().chars()
+					.position(|c| !c.is_whitespace())
+					.unwrap_or(0);
+
+				let section_name = &name.as_str()[split..];
+				if section_name.is_empty() // No name
+				{
+					result.push(
+						Report::build(ReportKind::Error, token.source(), name.start())
+						.with_message("Missing section name")
+						.with_label(
+							Label::new((token.source(), name.range()))
+							.with_message("Sections require a name before line end")
+							.with_color(parser.colors().error))
+						.finish());
+					return result;
+				}
+
+				// No spacing
+				if split == 0
+				{
+					result.push(
+						Report::build(ReportKind::Warning, token.source(), name.start())
+						.with_message("Missing section spacing")
+						.with_label(
+							Label::new((token.source(), name.range()))
+							.with_message("Sections require at least one whitespace before the section's name")
+							.with_color(parser.colors().warning))
+                        .with_help(format!("Add a space before `{}`", section_name.fg(parser.colors().highlight)))
+						.finish());
+					return result;
+				}
+
+
+				section_name.to_string()
+			},
+			_ => panic!("Empty section name")
+		};
+
+        parser.push(document, Box::new(
+            Section::new(
+				token.clone(),
+                section_name,
+                section_depth,
+                section_kind,
+                section_refname
+            )
+        ));
+
+        return result;
+	}
+}

src/elements/style.rs

@@ -0,0 +1,185 @@
+use regex::{Captures, Regex};
+use crate::{compiler::compiler::{Compiler, Target}, document::element::{ElemKind, Element}, parser::{parser::Parser, rule::RegexRule, source::{Source, Token}, state::State}};
+use ariadne::{Fmt, Label, Report, ReportKind};
+use crate::document::document::Document;
+use crate::parser::state::Scope;
+use std::{cell::RefCell, ops::Range, rc::Rc};
+use lazy_static::lazy_static;
+
+use super::paragraph::Paragraph;
+
+#[derive(Debug)]
+pub struct Style {
+	location: Token,
+	kind: usize,
+	close: bool,
+}
+
+impl Style
+{
+	pub fn new(location: Token, kind: usize, close: bool) -> Self {
+		Self { location, kind, close }
+	}
+}
+
+impl Element for Style
+{
+	fn location(&self) -> &Token { &self.location }
+	fn kind(&self) -> ElemKind { ElemKind::Inline }
+	fn element_name(&self) -> &'static str { "Section" }
+	fn to_string(&self) -> String { format!("{self:#?}") }
+	fn compile(&self, compiler: &Compiler, _document: &Document) -> Result<String, String> {
+		match compiler.target()
+		{
+			Target::HTML => {
+				Ok([
+					// Bold
+					"<b>", "</b>",
+					// Italic
+					"<i>", "</i>",
+					// Underline
+					"<u>", "</u>",
+					// Code
+					"<em>", "</em>",
+				][self.kind*2 + self.close as usize].to_string())
+			}
+			Target::LATEX => Err("Unimplemented compiler".to_string())
+		}
+	}
+}
+
+struct StyleState
+{
+	toggled: [Option<Token>; 4]
+}
+
+impl StyleState {
+	const NAMES : [&'static str; 4] = ["Bold", "Italic", "Underline", "Code"];
+
+	fn new() -> Self {
+		Self { toggled: [None, None, None, None] }
+	}
+}
+
+impl State for StyleState
+{
+	fn scope(&self) -> Scope { Scope::PARAGRAPH }
+
+	fn on_remove<'a>(&self, parser: &dyn Parser, document: &Document) -> Vec<Report<'a, (Rc<dyn Source>, Range<usize>)>> {
+		let mut result = Vec::new();
+		self.toggled
+			.iter()
+			.zip(StyleState::NAMES)
+			.for_each(|(token, name)|
+		{
+			if token.is_none() { return } // Style not enabled
+			let token = token.as_ref().unwrap();
+
+			//let range = range.as_ref().unwrap();
+
+			//let active_range = range.start .. paragraph.location().end()-1;
+
+			let paragraph = document.last_element::<Paragraph>(false).unwrap();
+			let paragraph_end = paragraph.content.last()
+				.and_then(|last| Some((last.location().source(), last.location().end()-1 .. last.location().end())))
+				.unwrap();
+
+			// TODO: Allow style to span multiple documents if they don't break paragraph.
+			result.push(
+				Report::build(ReportKind::Error, token.source(), token.start())
+				.with_message("Unterminated style")
+				//.with_label(
+				//	Label::new((document.source(), active_range.clone()))
+				//	.with_order(0)
+				//	.with_message(format!("Style {} is not terminated before the end of paragraph",
+				//	name.fg(parser.colors().info)))
+				//	.with_color(parser.colors().error))
+				.with_label(
+					Label::new((token.source(), token.range.clone()))
+					.with_order(1)
+					.with_message(format!("Style {} starts here",
+					name.fg(parser.colors().info)))
+					.with_color(parser.colors().info))
+				.with_label(
+					Label::new(paragraph_end)
+					.with_order(1)
+					.with_message(format!("Paragraph ends here"))
+					.with_color(parser.colors().info))
+				.with_note("Styles cannot span multiple documents (i.e @import)")
+				.finish());
+		});
+
+		return result;
+	}
+}
+
+pub struct StyleRule {
+	re: [Regex; 4],
+}
+
+impl StyleRule {
+	pub fn new() -> Self {
+		Self {
+			re: [
+				// Bold
+				Regex::new(r"\*\*").unwrap(),
+				// Italic
+				Regex::new(r"\*").unwrap(),
+				// Underline
+				Regex::new(r"__").unwrap(),
+				// Code
+				Regex::new(r"`").unwrap()
+			]
+		}
+	}
+}
+
+lazy_static! {
+	static ref STATE_NAME : String = "elements.style".to_string();
+}
+
+impl RegexRule for StyleRule
+{
+	fn name(&self) -> &'static str { "Style" }
+
+	fn regexes(&self) -> &[regex::Regex] { &self.re }
+
+	fn on_regex_match(&self, index: usize, parser: &dyn Parser, document: &Document, token: Token, _matches: Captures) -> Vec<Report<(Rc<dyn Source>, Range<usize>)>> {
+		let result = vec![];
+
+		let query = parser.state().query(&STATE_NAME);
+		let state = match query
+		{
+			Some(state) => state,
+			None => { // Insert as a new state
+				match parser.state_mut().insert(STATE_NAME.clone(), Rc::new(RefCell::new(StyleState::new())))
+				{
+					Err(_) => panic!("Unknown error"),
+					Ok(state) => state,
+				}
+			}
+		};
+
+		if let Some(style_state) = state
+			.borrow_mut()
+			.as_any_mut()
+			.downcast_mut::<StyleState>()
+		{
+			style_state.toggled[index] = style_state.toggled[index].clone().map_or(Some(token.clone()), |_| None);
+			parser.push(document, Box::new(
+				Style::new(
+					token.clone(),
+					index,
+					!style_state.toggled[index].is_some()
+				)
+			));
+		}
+		else
+		{
+			panic!("Invalid state at `{}`", STATE_NAME.as_str());
+		}
+
+		return result;
+	}
+
+}

src/elements/tex.rs

@@ -0,0 +1,263 @@
+use std::{io::{Read, Write}, ops::Range, process::{Command, Stdio}, rc::Rc, sync::Once};
+
+use ariadne::{Fmt, Label, Report, ReportKind};
+use crypto::{digest::Digest, sha2::Sha512};
+use regex::{Captures, Regex};
+
+use crate::{cache::cache::{Cached, CachedError}, compiler::compiler::{Compiler, Target}, document::{document::Document, element::{ElemKind, Element}}, parser::{parser::Parser, rule::RegexRule, source::{Source, Token}, util}};
+
+#[derive(Debug, PartialEq, Eq)]
+enum TexKind
+{
+	Block,
+	Inline,
+}
+
+impl From<&TexKind> for ElemKind
+{
+    fn from(value: &TexKind) -> Self {
+		match value {
+			TexKind::Inline => ElemKind::Inline,
+			_ => ElemKind::Block
+		}
+    }
+}
+
+#[derive(Debug)]
+struct Tex
+{
+	location: Token,
+	block: TexKind,
+	env: String,
+	tex: String,
+	caption: Option<String>,
+}
+
+impl Tex {
+    fn new(location: Token, block: TexKind, env: String, tex: String, caption: Option<String>) -> Self {
+        Self { location, block, env, tex, caption }
+    }
+
+	fn format_latex(fontsize: &String, preamble: &String, tex: &String) -> FormattedTex
+	{
+		FormattedTex(format!(r"\documentclass[{}pt,preview]{{standalone}}
+{}
+\begin{{document}}
+\begin{{preview}}
+{}
+\end{{preview}}
+\end{{document}}",
+		fontsize, preamble, tex))
+	}
+}
+
+struct FormattedTex(String);
+
+impl FormattedTex
+{
+	/// Renders latex to svg
+	fn latex_to_svg(&self, exec: &String, fontsize: &String) -> Result<String, String>
+	{
+		print!("Rendering LaTex `{}`... ", self.0);
+		let process = match Command::new(exec)
+			.arg("--fontsize").arg(fontsize)
+			.stdout(Stdio::piped())
+			.stdin(Stdio::piped())
+			.spawn()
+			{
+				Err(e) => return Err(format!("Could not spawn `{exec}`: {}", e)),
+				Ok(process) => process
+			};
+
+		if let Err(e) = process.stdin.unwrap().write_all(self.0.as_bytes())
+		{
+			panic!("Unable to write to `latex2svg`'s stdin: {}", e);
+		}
+
+		let mut result = String::new();
+		match process.stdout.unwrap().read_to_string(&mut result)
+		{
+			Err(e) => panic!("Unable to read `latex2svg` stdout: {}", e),
+			Ok(_) => {}
+		}
+		println!("Done!");
+
+		Ok(result)
+	}
+}
+
+impl Cached for FormattedTex
+{
+    type Key = String;
+    type Value = String;
+
+    fn sql_table() -> &'static str {
+		"CREATE TABLE IF NOT EXISTS cached_tex (
+				digest TEXT PRIMARY KEY,
+				svg    BLOB NOT NULL);"
+    }
+
+    fn sql_get_query() -> &'static str {
+		"SELECT svg FROM cached_tex WHERE digest = (?1)"
+    }
+
+    fn sql_insert_query() -> &'static str {
+		"INSERT INTO cached_tex (digest, svg) VALUES (?1, ?2)"
+    }
+
+    fn key(&self) -> <Self as Cached>::Key {
+		let mut hasher = Sha512::new();
+		hasher.input(self.0.as_bytes());
+
+		hasher.result_str()
+    }
+}
+
+impl Element for Tex {
+    fn location(&self) -> &Token { &self.location }
+
+    fn kind(&self) -> ElemKind { (&self.block).into() }
+
+    fn element_name(&self) -> &'static str { "LaTeX" }
+
+    fn to_string(&self) -> String { format!("{self:#?}") }
+
+    fn compile(&self, compiler: &Compiler, document: &Document)
+		-> Result<String, String> {
+
+		match compiler.target() {
+			Target::HTML => {
+				static CACHE_INIT : Once = Once::new();
+				CACHE_INIT.call_once(|| if let Some(mut con) = compiler.cache() {
+					if let Err(e) = FormattedTex::init(&mut con)
+					{
+						eprintln!("Unable to create cache table: {e}");
+					}
+				});
+
+				let exec = document.get_variable(format!("tex.{}.exec", self.env))
+					.map_or("latex2svg".to_string(), |(_, var)| var.to_string());
+				// FIXME: Because fontsize is passed as an arg, verify that it cannot be used to execute python/shell code
+				let fontsize = document.get_variable(format!("tex.{}.fontsize", self.env))
+					.map_or("12".to_string(), |(_, var)| var.to_string());
+				let preamble = document.get_variable(format!("tex.{}.preamble", self.env))
+					.map_or("".to_string(), |(_, var)| var.to_string());
+				let prepend = if self.block == TexKind::Inline { "".to_string() }
+				else
+				{
+					document.get_variable(format!("tex.{}.block_prepend", self.env))
+						.map_or("".to_string(), |(_, var)| var.to_string()+"\n")
+				};
+
+				let latex = match self.block
+				{
+					TexKind::Inline => Tex::format_latex(
+						&fontsize,
+						&preamble,
+						&format!("${{{}}}$", self.tex)),
+					_ => Tex::format_latex(
+						&fontsize,
+						&preamble,
+						&format!("{prepend}{}", self.tex))
+				};
+
+				if let Some(mut con) = compiler.cache()
+				{
+					match latex.cached(&mut con, |s| s.latex_to_svg(&exec, &fontsize))
+					{
+						Ok(s) => Ok(s),
+						Err(e) => match e
+						{
+							CachedError::SqlErr(e) => Err(format!("Querying the cache failed: {e}")),
+							CachedError::GenErr(e) => Err(e)
+						}
+					}
+				}
+				else
+				{
+					latex.latex_to_svg(&exec, &fontsize)
+				}
+			}
+			_ => todo!("Unimplemented")
+		}
+    }
+}
+
+pub struct TexRule {
+	re: [Regex; 2],
+}
+
+impl TexRule {
+	pub fn new() -> Self {
+		Self {
+			re: [
+				Regex::new(r"\$\|(?:\[(.*)\])?(?:((?:\\.|[^\\\\])*?)\|\$)?").unwrap(),
+				Regex::new(r"\$(?:\[(.*)\])?(?:((?:\\.|[^\\\\])*?)\$)?").unwrap(),
+			],
+		}
+	}
+}
+
+impl RegexRule for TexRule
+{
+    fn name(&self) -> &'static str { "Tex" }
+
+    fn regexes(&self) -> &[regex::Regex] { &self.re }
+
+    fn on_regex_match(&self, index: usize, parser: &dyn Parser, document: &Document, token: Token, matches: Captures)
+		-> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>> {
+		let mut reports = vec![];
+
+		let tex_env = matches.get(1)
+			.and_then(|env| Some(env.as_str().trim_start().trim_end()))
+			.and_then(|env| (!env.is_empty()).then_some(env))
+			.unwrap_or("main");
+
+		let tex_content = match matches.get(2)
+		{
+			// Unterminated `$`
+			None => {
+				reports.push(
+					Report::build(ReportKind::Error, token.source(), token.start())
+					.with_message("Unterminated Tex Code")
+					.with_label(
+						Label::new((token.source().clone(), token.range.clone()))
+						.with_message(format!("Missing terminating `{}` after first `{}`",
+							["|$", "$"][index].fg(parser.colors().info),
+							["$|", "$"][index].fg(parser.colors().info)))
+						.with_color(parser.colors().error))
+					.finish());
+				return reports;
+			}
+			Some(content) => {
+				let processed = util::process_escaped('\\', ["|$", "$"][index],
+					content.as_str().trim_start().trim_end());
+
+				if processed.is_empty()
+				{
+					reports.push(
+						Report::build(ReportKind::Warning, token.source(), content.start())
+						.with_message("Empty Tex Code")
+						.with_label(
+							Label::new((token.source().clone(), content.range()))
+							.with_message("Tex code is empty")
+							.with_color(parser.colors().warning))
+						.finish());
+				}
+				processed
+			}
+		};
+
+		// TODO: Caption
+
+		parser.push(document, Box::new(Tex::new(
+			token,
+			if index == 1 { TexKind::Inline } else { TexKind::Block },
+			tex_env.to_string(),
+			tex_content,
+			None,
+		)));
+
+		reports
+    }
+}

src/elements/variable.rs

@@ -0,0 +1,329 @@
+use regex::Regex;
+use crate::parser::{parser::{Parser, ReportColors}, rule::RegexRule, source::{Source, Token}};
+use ariadne::{Report, Fmt, Label, ReportKind};
+use crate::document::{document::Document, variable::{BaseVariable, PathVariable, Variable}};
+use std::{ops::Range, rc::Rc};
+
+pub struct VariableRule {
+	re: [Regex; 1],
+    kinds: Vec<(String, String)>,
+}
+
+impl VariableRule {
+	pub fn new() -> Self {
+		Self {
+            re: [Regex::new(r"(?:^|\n)@([^[:alpha:]])?(.*)=((?:\\\n|.)*)").unwrap()],
+            kinds: vec![
+                ("".into(), "Regular".into()),
+                ("'".into(), "Path".into())
+            ]
+        }
+	}
+
+
+    pub fn make_variable(&self, colors: &ReportColors, location: Token, kind: usize, name: String, value: String) -> Result<Rc<dyn Variable>, String>
+    {
+        match self.kinds[kind].0.as_str()
+        {
+            "" => {
+                Ok(Rc::new(BaseVariable::new(location, name, value)))
+            }
+            "'" => {
+                match std::fs::canonicalize(value.as_str()) // TODO: not canonicalize
+                {
+                    Ok(path) => Ok(Rc::new(PathVariable::new(location, name, path))),
+                    Err(e) => Err(format!("Unable to canonicalize path `{}`: {}",
+                            value.fg(colors.highlight),
+                            e.to_string()))
+                }
+            }
+            _ => panic!("Unhandled variable kind")
+        }
+    }
+
+    // Trim and check variable name for validity
+    pub fn validate_name<'a>(colors: &ReportColors, original_name: &'a str) -> Result<&'a str, String>
+    {
+        let name = original_name.trim_start().trim_end();
+        if name.contains("%")
+        {
+            return Err(format!("Name cannot contain '{}'",
+                "%".fg(colors.info)));
+        }
+        return Ok(name);
+    }
+
+    pub fn validate_value(_colors: &ReportColors, original_value: &str) -> Result<String, String>
+    {
+		let mut escaped = 0usize;
+		let mut result = String::new();
+		for c in original_value.trim_start().trim_end().chars() {
+			if c == '\\' { escaped += 1 }
+			else if c == '\n' {
+				match escaped {
+					0 => return Err("Unknown error wile capturing variable".to_string()),
+					// Remove '\n'
+					1 => {},
+					// Insert '\n'
+					_ => {
+						result.push(c);
+						(0..escaped-2).for_each(|_| result.push('\\'));
+					}
+				}
+				escaped = 0;
+			}
+			else {
+				(0..escaped).for_each(|_| result.push('\\'));
+				escaped = 0;
+				result.push(c);
+			}
+		}
+		(0..escaped).for_each(|_| result.push('\\'));
+
+		Ok(result)
+    }
+}
+
+impl RegexRule for VariableRule {
+	fn name(&self) -> &'static str { "Variable" }
+
+	fn regexes(&self) -> &[Regex] { &self.re }
+
+	fn on_regex_match(&self, _: usize, parser: &dyn Parser, document: &Document, token: Token, matches: regex::Captures) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>
+	{
+		let mut result = vec![];
+        // [Optional] variable kind
+		let var_kind = match matches.get(1)
+		{
+			Some(kind) => {
+                // Find kind
+                let r = self.kinds.iter().enumerate().find(|(_i, (ref char, ref _name))| {
+                    char == kind.as_str() });
+
+                // Unknown kind specified
+                if r.is_none()
+                {
+                    result.push(
+                        Report::build(ReportKind::Error, token.source(), kind.start())
+                        .with_message("Unknown variable kind")
+                        .with_label(
+                            Label::new((token.source(), kind.range()))
+                            .with_message(format!("Variable kind `{}` is unknown",
+                                    kind.as_str().fg(parser.colors().highlight)))
+                            .with_color(parser.colors().error))
+                        .with_help(format!("Leave empty for regular variables. Available variable kinds:{}",
+                            self.kinds.iter().skip(1).fold("".to_string(), |acc, (char, name)| {
+                                acc + format!("\n - `{}` : {}",
+                                    char.fg(parser.colors().highlight),
+                                    name.fg(parser.colors().info)).as_str()
+                            })))
+                        .finish());
+
+                    return result;
+                }
+
+                r.unwrap().0
+			}
+			None => 0,
+		};
+
+        let var_name = match matches.get(2)
+        {
+            Some(name) => {
+                match VariableRule::validate_name(&parser.colors(), name.as_str())
+                {
+                    Ok(var_name) => var_name,
+                    Err(msg) => {
+                        result.push(
+                            Report::build(ReportKind::Error, token.source(), name.start())
+                            .with_message("Invalid variable name")
+                            .with_label(
+                                Label::new((token.source(), name.range()))
+                                .with_message(format!("Variable name `{}` is not allowed. {msg}",
+                                        name.as_str().fg(parser.colors().highlight)))
+                                .with_color(parser.colors().error))
+                            .finish());
+
+                        return result;
+                    },
+                }
+            },
+            _ => panic!("Unknown variable name")
+        };
+
+        let var_value = match matches.get(3)
+        {
+            Some(value) => {
+                match VariableRule::validate_value(&parser.colors(), value.as_str())
+                {
+                    Ok(var_value) => var_value,
+                    Err(msg ) => {
+                        result.push(
+                            Report::build(ReportKind::Error, token.source(), value.start())
+                            .with_message("Invalid variable value")
+                            .with_label(
+                                Label::new((token.source(), value.range()))
+                                .with_message(format!("Variable value `{}` is not allowed. {msg}",
+                                        value.as_str().fg(parser.colors().highlight)))
+                                .with_color(parser.colors().error))
+                            .finish());
+
+                        return result;
+                    }
+                }
+            }
+            _ => panic!("Invalid variable value")
+        };
+
+        match self.make_variable(&parser.colors(), token.clone(), var_kind, var_name.to_string(), var_value)
+        {
+            Ok(variable) => document.add_variable(variable),
+            Err(msg) => {
+                let m = matches.get(0).unwrap();
+                result.push(
+                    Report::build(ReportKind::Error, token.source(), m.start())
+                    .with_message("Unable to create variable")
+                    .with_label(
+                        Label::new((token.source(), m.start()+1 .. m.end() ))
+                        .with_message(format!("Unable to create variable `{}`. {}",
+                                var_name.fg(parser.colors().highlight),
+                                msg))
+                        .with_color(parser.colors().error))
+                    .finish());
+
+                return result;
+            }
+        }
+
+        return result;
+	}
+}
+
+pub struct VariableSubstitutionRule
+{
+	re: [Regex; 1],
+}
+
+impl VariableSubstitutionRule {
+	pub fn new() -> Self {
+		Self {
+            re: [Regex::new(r"%(.*?)%").unwrap()],
+        }
+	}
+}
+
+impl RegexRule for VariableSubstitutionRule
+{
+    fn name(&self) -> &'static str { "Variable Substitution" }
+
+    fn regexes(&self) -> &[regex::Regex] { &self.re }
+
+    fn on_regex_match(&self, _index: usize, parser: &dyn Parser, document: &Document, token: Token, matches: regex::Captures) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>> {
+		let mut result = vec![];
+
+        let variable = match matches.get(1)
+        {
+            Some(name) => {
+                // Empty name
+                if name.as_str().is_empty()
+                {
+                    result.push(
+                        Report::build(ReportKind::Error, token.source(), name.start())
+                        .with_message("Empty variable name")
+                        .with_label(
+                            Label::new((token.source(), matches.get(0).unwrap().range()))
+                            .with_message(format!("Missing variable name for substitution"))
+                            .with_color(parser.colors().error))
+                        .finish());
+
+                    return result;
+                }
+                // Leading spaces
+                else if name.as_str().trim_start() != name.as_str()
+                {
+                    result.push(
+                        Report::build(ReportKind::Error, token.source(), name.start())
+                        .with_message("Invalid variable name")
+                        .with_label(
+                            Label::new((token.source(), name.range()))
+                            .with_message(format!("Variable names contains leading spaces"))
+                            .with_color(parser.colors().error))
+                        .with_help("Remove leading spaces")
+                        .finish());
+
+                    return result;
+                }
+                // Trailing spaces
+                else if name.as_str().trim_end() != name.as_str()
+                {
+                    result.push(
+                        Report::build(ReportKind::Error, token.source(), name.start())
+                        .with_message("Invalid variable name")
+                        .with_label(
+                            Label::new((token.source(), name.range()))
+                            .with_message(format!("Variable names contains trailing spaces"))
+                            .with_color(parser.colors().error))
+                        .with_help("Remove trailing spaces")
+                        .finish());
+
+                    return result;
+                }
+				// Invalid name
+                match VariableRule::validate_name(&parser.colors(), name.as_str())
+                {
+					Err(msg) =>
+					{
+						result.push(
+							Report::build(ReportKind::Error, token.source(), name.start())
+							.with_message("Invalid variable name")
+							.with_label(
+								Label::new((token.source(), name.range()))
+								.with_message(msg)
+								.with_color(parser.colors().error))
+							.finish());
+
+						return result;
+					}
+					_ => {},
+                }
+				
+				// Get variable
+                match document.get_variable(name.as_str())
+                {
+                    None => {
+                        result.push(
+                            Report::build(ReportKind::Error, token.source(), name.start())
+                            .with_message("Unknown variable name")
+                            .with_label(
+                                Label::new((token.source(), name.range()))
+                                .with_message(format!("Unable to find variable with name: `{}`",
+                                        name.as_str().fg(parser.colors().highlight)))
+                                .with_color(parser.colors().error))
+                            .finish());
+                        return result;
+                    }
+                    Some((_, var)) => var,
+                }
+            },
+            _ => panic!("Unknown error")
+        };
+
+		variable.parse(token, parser, document);
+		//let parsed = variable.parse(
+		//	token,
+		//	parser,
+		//	document
+		//);
+		////document.merge(parsed, None);
+		//parsed.content.borrow_mut()
+		//	.drain(..)
+		//	.for_each(|elem| parser.push(document, elem));
+		//parser.push(document, )
+
+		// TODO: Full rework of document
+		// parser shound parse into previous document, and not into a new document
+		// This should prevent having to sue `recurse: bool` in the last_element getters
+
+        return result;
+    }
+}

src/files.rs

@@ -1,3 +0,0 @@
-pub mod file;
-pub mod cursor;
-pub mod token;

src/files/cursor.rs

@@ -1,28 +0,0 @@
-use super::file::File;
-
-#[derive(Debug)]
-pub struct Cursor<'a>
-{
-    pub file: &'a File,
-    pub content: String,
-    pub position: usize,
-}
-
-impl<'a> Cursor<'a>
-{
-    pub fn new(_file: &'a File) -> Result<Cursor<'a>, std::io::Error>
-    {
-        let _content = match std::fs::read_to_string(&_file.path)
-        {
-            Ok(content) => content,
-            Err(error) => return Err(error),
-        };
-
-        Ok(Cursor
-        {
-            file: _file,
-            content: _content,
-            position: 0usize,
-        })
-    }
-}

src/files/file.rs

@@ -1,17 +0,0 @@
-use std::path::Path;
-
-#[derive(Debug)]
-pub struct File
-{
-    pub path: String,
-}
-
-impl File
-{
-    pub fn new(_path: String) -> File
-    {
-        File {
-            path: _path,
-        }
-    }
-}

src/files/token.rs

@@ -1,30 +0,0 @@
-use super::file::File;
-use super::cursor::Cursor;
-
-pub struct Token<'a>
-{
-    file: &'a File,
-    start: usize,
-    len: usize,
-}
-
-impl<'a> Token<'a>
-{
-    pub fn new(_file: &'a File, _start: usize, _len: usize) -> Token<'a>
-    {
-        Token {
-            file: _file,
-            start: _start,
-            len: _len,
-        }
-    }
-
-    pub fn from(cursor: &'a Cursor, mat: regex::Match<'a>) -> Token<'a>
-    {
-        Token {
-            file: cursor.file,
-            start: cursor.position,
-            len: mat.len(),
-        }
-    }
-}

src/lsp/mod.rs

@@ -0,0 +1 @@
+pub mod parser;

src/lsp/parser.rs

@@ -0,0 +1,30 @@
+use std::{cell::RefCell, collections::HashMap};
+
+use crate::{elements::registrar::register, lua::kernel::Kernel, parser::{rule::Rule, state::StateHolder}};
+
+struct LSParser
+{
+	rules: Vec<Box<dyn Rule>>,
+
+	// Parser state
+	pub state: RefCell<StateHolder>,
+	//pub kernels: RefCell<HashMap<String, Kernel>>,
+}
+
+impl LSParser {
+	pub fn default() -> Self
+	{
+		let mut parser = LSParser {
+			rules: vec![],
+			state: RefCell::new(StateHolder::new()),
+			//kernels: RefCell::new(HashMap::new()),
+		};
+
+		// TODO: Main kernel
+		//register(&mut parser);
+
+		parser
+	}
+}
+
+

src/lua/kernel.rs

@@ -0,0 +1,21 @@
+use std::cell::RefMut;
+
+use mlua::Lua;
+
+pub struct Kernel
+{
+	pub lua: Lua,
+}
+
+impl Kernel {
+    pub fn new() -> Self {
+        Self { lua: Lua::new() }
+    }
+}
+
+pub trait KernelHolder
+{
+	fn get_kernel(&self, name: &str) -> Option<RefMut<'_, Kernel>>;
+
+	fn insert_kernel(&self, name: String, kernel: Kernel) -> RefMut<'_, Kernel>;
+}

src/lua/mod.rs

@@ -0,0 +1 @@
+pub mod kernel;

src/main.rs

@@ -1,110 +1,110 @@
+#![feature(char_indices_offset)]
+mod document;
+mod compiler;
 mod parser;
-use self::parser::rule::SyntaxRule;
-use self::parser::section::SectionRule;
-mod files;
-use self::files::file::File;
-use self::files::cursor::Cursor;
-mod syntax;
-use syntax::element::Element;
-use syntax::element::Text;
+mod elements;
+mod lua;
+mod cache;
+
+use std::{env, rc::Rc};
+
+use compiler::compiler::Compiler;
+use getopts::Options;
+use parser::{langparser::LangParser, parser::Parser};
+
+use crate::parser::source::SourceFile;
+extern crate getopts;
+
+fn print_usage(program: &str, opts: Options) {
+    let brief = format!("Usage: {} -i FILE [options]", program);
+    print!("{}", opts.usage(&brief));
+}
+
+fn print_version()
+{
+	print!("NML -- Not a Markup Language
+Copyright (c) 2024
+NML is licensed under the GNU Affero General Public License version 3 (AGPLv3),
+under the terms of the Free Software Foundation <https://www.gnu.org/licenses/agpl-3.0.en.html>.
+
+This program is free software; you may modify and redistribute it.
+There is NO WARRANTY, to the extent permitted by law.
+
+NML version: 0.4\n");
+}
 
 fn main() {
-    let file = File::new(String::from("./test.nml"));
-    let mut cursor = Cursor::new(&file).unwrap();
-    cursor.position = 5;
+    let args: Vec<String> = env::args().collect();
+    let program = args[0].clone();
 
-    let rule_se = SectionRule::new();
-    let (token, res) = rule_se.on_match(&cursor).unwrap();
-    println!("{}", res.elements.len());
+	let mut opts = Options::new();
+	opts.optopt("i", "", "Input file", "FILE");
+	opts.optopt("d", "database", "Cache database location", "PATH");
+	opts.optmulti("z", "debug", "Debug options", "OPTS");
+	opts.optflag("h", "help", "Print this help menu");
+	opts.optflag("v", "version", "Print program version and licenses");
 
-
-    /*
-    let re_sections = regex::Regex::new(r"(?:^|\n)(#{1,})(\*|\+)((?:\t| ){0,})(.*)").unwrap();
-
-    //let mut validators = Vec::<Box<dyn GroupValidator>>::new();
-    let f = File::new(Box::new(std::path::Path::new("./test.nml")));
-    let content = std::fs::read_to_string(*f.path).unwrap();
-    
-    let grammar = vec![re_sections];
-    let mut positions = [0usize; 1];
-
-    let mut i = 0;
-    while i < content.len()
+	let matches = match opts.parse(&args[1..]) {
+		Ok(m) => { m }
+		Err(f) => { panic!("{}", f.to_string()) }
+	};
+	if matches.opt_present("v")
 	{
-        // Update every positions
-        for k in 0..grammar.len()
+		print_version();
+		return;
+	}
+	if matches.opt_present("h") {
+		print_usage(&program, opts);
+		return;
+	}
+	if !matches.opt_present("i") {
+		print_usage(&program, opts);
+		return;
+	}
+
+	let input = matches.opt_str("i").unwrap();
+	let debug_opts = matches.opt_strs("z");
+	let db_path = matches.opt_str("d");
+
+	let parser = LangParser::default();
+
+	// Parse
+	let source = SourceFile::new(input.to_string(), None).unwrap();
+	let doc = parser.parse(Rc::new(source), None);
+
+	if debug_opts.contains(&"ast".to_string())
 	{
-            let rule = &grammar[k];
-            let position = &mut positions[k];
-            if *position == std::usize::MAX { continue };
+		println!("-- BEGIN AST DEBUGGING --");
+		doc.content.borrow().iter().for_each(|elem| {
+			println!("{}", (elem).to_string())
+		});
+		println!("-- END AST DEBUGGING --");
+	}
 
-            match rule.find_at(&content, i)
+
+	if debug_opts.contains(&"ref".to_string())
 	{
-                Some(mat) => *position = mat.start(),
-                None => *position = std::usize::MAX,
+		println!("-- BEGIN REFERENCES DEBUGGING --");
+		let sc = doc.scope.borrow();
+		sc.referenceable.iter().for_each(|(name, pos)| {
+			println!(" - {name}: `{:#?}`", doc.content.borrow()[*pos]);
+		});
+		println!("-- END REFERENCES DEBUGGING --");
 	}
-            println!("{position}");
-        }
-
-        // Gets closest match
-        let mut next_position = std::usize::MAX;
-        let mut closest_match = std::usize::MAX;
-        for k in 0..grammar.len()
+	if debug_opts.contains(&"var".to_string())
 	{
-            if positions[k] >= next_position { continue; }
-
-            next_position = positions[k];
-            closest_match = k;
+		println!("-- BEGIN VARIABLES DEBUGGING --");
+		let sc = doc.scope.borrow();
+		sc.variables.iter().for_each(|(_name, var)| {
+			println!(" - `{:#?}`", var);
+		});
+		println!("-- END VARIABLES DEBUGGING --");
 	}
 
-        println!("Unmatched: {}", &content[i..next_position]);
 
-        // No matches left
-        if closest_match == std::usize::MAX
-        {
-            println!("Done");
-            break;
-        }
-
-        // Extract matches from rule
-        i = next_position; // Set to begining of match
-        let mat = &grammar[closest_match].captures_at(&content, i).unwrap(); // Capture match
-        for m in 0..mat.len()
-        {
-            match mat.get(m)
-            {
-                Some(s) => { 
-                    println!("Group {m}: `{}`", s.as_str());
-                },
-                None => println!("Group {m}: None"),
-            }
-        }
-
-        i += mat.get(0).unwrap().len(); // Add match length
-        println!("Left={}", &content[i..]);
-        println!("pos={i}");
-
-        let mut s = String::new();
-        std::io::stdin().read_line(&mut s).expect("Did not enter a correct string");
-    }
-    */
-    
-
-    
-    /*
-    validators.push(Box::new(StringValidator::new("Depth".to_string(), |_group| -> ValidationStatus {
-        ValidationStatus::Ok()
-    })));
-    validators.push(Box::new(StringValidator::new("Index Type".to_string(), |group| -> ValidationStatus {
-        match group
-        {
-            "" => ValidationStatus::Ok(),
-            "*" => ValidationStatus::Ok(),
-            _ => ValidationStatus::Error("")
-        }
-        ValidationStatus::Ok()
-    })));
-    */
-    //let _sec_rule = SyntaxRule::new("Section".to_string(), r"(?m)(?:^|\n)(#{1,})(\\*|\\+)((?:\t| ){0,})(.*)", validators).unwrap();
+	let compiler = Compiler::new(compiler::compiler::Target::HTML, db_path);
+	let out = compiler.compile(&doc);
+
+	std::fs::write("a.html", out).unwrap();
 }
 

src/parser.rs

@@ -1,3 +0,0 @@
-pub mod rule;
-pub mod section;
-

src/parser/langparser.rs

@@ -0,0 +1,277 @@
+use std::{cell::{RefCell, RefMut}, collections::{HashMap, HashSet}, ops::Range, rc::Rc};
+
+use ariadne::{Label, Report};
+
+use crate::{document::{document::Document, element::{ElemKind, Element, Text}}, elements::{paragraph::Paragraph, registrar::register}, lua::kernel::{Kernel, KernelHolder}, parser::source::{SourceFile, VirtualSource}};
+
+use super::{parser::{Parser, ReportColors}, rule::Rule, source::{Cursor, Source, Token}, state::StateHolder, util};
+
+/// Parser for the language
+pub struct LangParser
+{
+	rules: Vec<Box<dyn Rule>>,
+	colors: ReportColors,
+
+	// Parser state
+	pub err_flag: RefCell<bool>,
+	pub state: RefCell<StateHolder>,
+	pub kernels: RefCell<HashMap<String, Kernel>>,
+}
+
+impl LangParser
+{
+	pub fn default() -> Self
+	{
+		let mut s = Self {
+			rules: vec![],
+			colors: ReportColors::with_colors(),
+			err_flag: RefCell::new(false),
+			state: RefCell::new(StateHolder::new()),
+			kernels: RefCell::new(HashMap::new()),
+		};
+		register(&mut s);
+		s.kernels.borrow_mut()
+			.insert("main".to_string(), Kernel::new());
+		s
+	}
+
+	fn handle_reports<'a>(&self, _source: Rc<dyn Source>, reports: Vec<Report<'a, (Rc<dyn Source>, Range<usize>)>>)
+	{
+		for mut report in reports
+		{
+			let mut sources: HashSet<Rc<dyn Source>> = HashSet::new();
+			fn recurse_source(sources: &mut HashSet<Rc<dyn Source>>, source: Rc<dyn Source>) {
+				sources.insert(source.clone());
+				match source.location()
+				{
+					Some(parent) => {
+						let parent_source = parent.source();
+						if sources.get(&parent_source).is_none()
+						{
+							recurse_source(sources, parent_source);
+						}
+					}
+					None => {}
+				}
+			}
+
+			report.labels.iter().for_each(|label| {
+				recurse_source(&mut sources, label.span.0.clone());
+			});
+
+			let cache = sources.iter()
+				.map(|source| (source.clone(), source.content().clone()))
+				.collect::<Vec<(Rc<dyn Source>, String)>>();
+
+			cache.iter()
+				.for_each(|(source, _)| {
+					if let Some (location) = source.location()
+					{
+						if let Some(_s) = source.downcast_ref::<SourceFile>()
+						{
+							report.labels.push(
+								Label::new((location.source(), location.start()+1 .. location.end()))
+								.with_message("In file included from here")
+								.with_order(-1)
+							);
+						};
+
+						if let Some(_s) = source.downcast_ref::<VirtualSource>()
+						{
+							let start = location.start() + (location.source().content().as_bytes()[location.start()] == '\n' as u8)
+								.then_some(1)
+								.unwrap_or(0);
+							report.labels.push(
+								Label::new((location.source(), start .. location.end()))
+								.with_message("In evaluation of")
+								.with_order(-1)
+							);
+						};
+					}
+				});
+			report.eprint(ariadne::sources(cache)).unwrap()
+		}
+	}
+}
+
+impl Parser for LangParser
+{
+	fn colors(&self) -> &ReportColors { &self.colors }
+
+	fn rules(&self) -> &Vec<Box<dyn Rule>> { &self.rules }
+	fn add_rule(&mut self, rule: Box<dyn Rule>, after: Option<&'static str>)
+	{
+		// Error on duplicate rule
+		let rule_name = (*rule).name();
+		self.rules.iter().for_each(|rule| {
+			if (*rule).name() != rule_name { return; }
+		
+			panic!("Attempted to introduce duplicate rule: `{rule_name}`");
+		});
+
+		match after
+		{
+			Some(name) => {
+				let before = self.rules.iter()
+					.enumerate()
+					.find(|(_pos, r)| (r).name() == name);
+
+				match before
+				{
+					Some((pos, _)) => self.rules.insert(pos+1, rule),
+					_ => panic!("Unable to find rule named `{name}`, to insert rule `{}` after it", rule.name())
+				}
+			}
+			_ => self.rules.push(rule)
+		}
+	}
+
+	fn state(&self) -> std::cell::Ref<'_, StateHolder> { self.state.borrow() }
+	fn state_mut(&self) -> std::cell::RefMut<'_, StateHolder> { self.state.borrow_mut() }
+
+	/// Add an [`Element`] to the [`Document`]
+	fn push<'a>(&self, doc: &'a Document<'a>, elem: Box<dyn Element>)
+	{
+		if elem.kind() == ElemKind::Inline || elem.kind() == ElemKind::Invisible
+		{
+			let mut paragraph = doc.last_element_mut::<Paragraph>(false)
+				.or_else(|| {
+					doc.push(Box::new(Paragraph::new(elem.location().clone())));
+					doc.last_element_mut::<Paragraph>(false)
+				}).unwrap();
+
+			paragraph.push(elem);
+		}
+		else
+		{
+			// Process paragraph events
+			if doc.last_element_mut::<Paragraph>(false)
+				.is_some_and(|_| true)
+			{
+				self.handle_reports(doc.source(),
+				self.state_mut().on_scope_end(self, &doc, super::state::Scope::PARAGRAPH));
+			}
+
+			doc.push(elem);
+		}
+	}
+
+	fn parse<'a>(&self, source: Rc<dyn Source>, parent: Option<&'a Document<'a>>) -> Document<'a>
+	{
+		let doc = Document::new(source.clone(), parent);
+		let mut matches = Vec::new();
+		for _ in 0..self.rules.len() {
+			matches.push((0usize, None));
+		}
+
+		let content = source.content();
+		let mut cursor = Cursor::new(0usize, doc.source()); // Cursor in file
+
+		if parent.is_some() // Terminate parent's paragraph state
+		{
+			self.handle_reports(parent.as_ref().unwrap().source(),
+			self.state_mut().on_scope_end(self, parent.as_ref().unwrap(), super::state::Scope::PARAGRAPH));
+		}
+	
+		loop
+		{
+			let (rule_pos, rule, match_data) = self.update_matches(&cursor, &mut matches);
+
+			// Unmatched content
+			let text_content = util::process_text(&doc, &content.as_str()[cursor.pos..rule_pos.pos]);
+			if !text_content.is_empty()
+			{
+				self.push(&doc, Box::new(Text::new(
+					Token::new(cursor.pos..rule_pos.pos, source.clone()),
+					text_content
+				)));
+			}
+
+			if let Some(rule) = rule
+			{
+
+				// Rule callback
+				let (new_cursor, reports) = (*rule).on_match(self, &doc, rule_pos, match_data);
+
+				self.handle_reports(doc.source(), reports);
+
+				// Advance
+				cursor = new_cursor;
+			}
+			else // No rules left
+			{
+				break;
+			}
+		}
+
+		// State
+		self.handle_reports(doc.source(),
+			self.state_mut().on_scope_end(self, &doc, super::state::Scope::DOCUMENT));
+		
+		return doc;
+	}
+
+	fn parse_into<'a>(&self, source: Rc<dyn Source>, document: &'a Document<'a>)
+	{
+		let mut matches = Vec::new();
+		for _ in 0..self.rules.len() {
+			matches.push((0usize, None));
+		}
+
+		let content = source.content();
+		let mut cursor = Cursor::new(0usize, source.clone());
+
+		loop
+		{
+			let (rule_pos, rule, match_data) = self.update_matches(&cursor, &mut matches);
+
+			// Unmatched content
+			let text_content = util::process_text(&document, &content.as_str()[cursor.pos..rule_pos.pos]);
+			if !text_content.is_empty()
+			{
+				self.push(&document, Box::new(Text::new(
+							Token::new(cursor.pos..rule_pos.pos, source.clone()),
+							text_content
+				)));
+			}
+
+			if let Some(rule) = rule
+			{
+				// Rule callback
+				let (new_cursor, reports) = (*rule).on_match(self, &document, rule_pos, match_data);
+
+				self.handle_reports(document.source(), reports);
+
+				// Advance
+				cursor = new_cursor;
+			}
+			else // No rules left
+			{
+				break;
+			}
+		}
+
+		// State
+		//self.handle_reports(source.clone(),
+		//	self.state_mut().on_scope_end(&self, &document, super::state::Scope::DOCUMENT));
+		
+		//return doc;
+	}
+}
+
+impl KernelHolder for LangParser
+{
+    fn get_kernel(&self, name: &str)
+		-> Option<RefMut<'_, Kernel>> {
+		RefMut::filter_map(self.kernels.borrow_mut(),
+		|map| map.get_mut(name)).ok()
+    }
+
+    fn insert_kernel(&self, name: String, kernel: Kernel)
+		-> RefMut<'_, Kernel> {
+			//TODO do not get
+		self.kernels.borrow_mut()
+			.insert(name.clone(), kernel);
+		self.get_kernel(name.as_str()).unwrap()
+    }
+}

src/parser/mod.rs

@@ -0,0 +1,6 @@
+pub mod source;
+pub mod parser;
+pub mod langparser;
+pub mod rule;
+pub mod state;
+pub mod util;

src/parser/parser.rs

@@ -0,0 +1,126 @@
+use std::any::Any;
+use std::cell::{Ref, RefCell, RefMut};
+use std::collections::{HashMap, HashSet};
+use std::ops::Range;
+use std::rc::Rc;
+use unicode_segmentation::UnicodeSegmentation;
+
+use super::rule::Rule;
+use super::source::{Cursor, Source};
+use super::state::StateHolder;
+use crate::document::document::Document;
+use crate::document::element::Element;
+use ariadne::Color;
+use crate::lua::kernel::{Kernel, KernelHolder};
+
+pub struct ReportColors
+{
+	pub error: Color,
+	pub warning: Color,
+	pub info: Color,
+	pub highlight: Color,
+}
+
+impl ReportColors {
+	pub fn with_colors() -> Self {
+		Self {
+			error: Color::Red,
+			warning: Color::Yellow,
+			info: Color::BrightBlue,
+			highlight: Color::BrightMagenta,
+		}
+	}
+
+	pub fn without_colors() -> Self {
+		Self {
+			error: Color::Primary,
+			warning: Color::Primary,
+			info: Color::Primary,
+			highlight: Color::Primary,
+		}
+	}
+}
+
+pub trait Parser: KernelHolder
+{
+	/// Gets the colors for formatting errors
+	///
+	/// When colors are disabled, all colors should resolve to empty string
+	fn colors(&self) -> &ReportColors;
+
+	fn rules(&self) -> &Vec<Box<dyn Rule>>;
+	fn add_rule(&mut self, rule: Box<dyn Rule>, after: Option<&'static str>);
+
+	fn state(&self) -> Ref<'_, StateHolder>;
+	fn state_mut(&self) -> RefMut<'_, StateHolder>;
+
+	// Update [`matches`] and returns the position of the next matched rule.
+	// If rule is empty, it means that there are no rules left to parse (i.e
+	// end of document).
+	fn update_matches(&self, cursor: &Cursor, matches: &mut Vec<(usize, Option<Box<dyn Any>>)>)
+		-> (Cursor, Option<&Box<dyn Rule>>, Option<Box<dyn Any>>) 
+	{
+		// Update matches
+		// TODO: Trivially parellalizable
+		self.rules().iter().zip(matches.iter_mut()).for_each(
+			|(rule, (matched_at, match_data))| {
+			// Don't upate if not stepped over yet
+			if *matched_at > cursor.pos { return }
+
+			(*matched_at, *match_data) = match rule.next_match(cursor)
+			{
+				None => (usize::MAX, None),
+				Some((mut pos, mut data)) =>
+				{
+					// Check if escaped
+					while pos != usize::MAX
+					{
+						let content = cursor.source.content().as_str();
+						let mut graphemes = content[0 .. pos].graphemes(true);
+						let mut escaped = false;
+						'inner: loop
+						{
+							let g = graphemes.next_back();
+							if !g.is_some() || g.unwrap() != "\\" { break 'inner; }
+
+							escaped = !escaped;
+						}
+						if !escaped { break; }
+
+						// Find next potential match
+						(pos, data) = match rule.next_match(&cursor.at(pos+1)) {
+							Some((new_pos, new_data)) => (new_pos, new_data),
+							None => (usize::MAX, data) // Stop iterating
+						}
+					}
+
+					(pos, (pos != usize::MAX).then_some(data))
+				}
+			}
+		});
+
+		// Get winning match
+		let (winner, (next_pos, _match_data)) = matches.iter()
+			.enumerate()
+			.min_by_key(|(_, (pos, _match_data))| pos).unwrap();
+		if *next_pos == usize::MAX // No rule has matched
+		{
+			let content = cursor.source.content();
+			// No winners, i.e no matches left
+			return (cursor.at(content.len()), None, None);
+		}
+
+		(cursor.at(*next_pos),
+		Some(&self.rules()[winner]),
+		std::mem::replace(&mut matches[winner].1, None))
+	}
+
+	/// Add an [`Element`] to the [`Document`]
+	fn push<'a>(&self, doc: &'a Document<'a>, elem: Box<dyn Element>);
+
+	/// Parse [`Source`] into a new [`Document`]
+	fn parse<'a>(&self, source: Rc<dyn Source>, parent: Option<&'a Document<'a>>) -> Document<'a>;
+
+	/// Parse [`Source`] into an already existing [`Document`]
+	fn parse_into<'a>(&self, source: Rc<dyn Source>, document: &'a Document<'a>);
+}

src/parser/rule.rs

@@ -1,51 +1,103 @@
-use regex::Captures;
-use super::super::syntax::element::Element;
-use super::super::files::cursor::Cursor;
-use super::super::files::token::Token;
+use super::parser::Parser;
+use super::source::{Cursor, Source, Token};
+use ariadne::Report;
+use crate::document::document::Document;
 
-pub struct RuleResult
-{
-    length: usize,
-    pub elements: Vec<Box<dyn Element>>,
+use std::any::Any;
+use std::ops::Range;
+use std::rc::Rc;
+
+pub trait Rule {
+	/// Returns rule's name
+	fn name(&self) -> &'static str;
+	/// Finds the next match starting from [`cursor`]
+	fn next_match(&self, cursor: &Cursor) -> Option<(usize, Box<dyn Any>)>;
+	/// Callback when rule matches
+	fn on_match(&self, parser: &dyn Parser, document: &Document, cursor: Cursor, match_data: Option<Box<dyn Any>>) -> (Cursor, Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>);
 }
 
-impl RuleResult
-{
-    pub fn new(_length: usize, elem: Box<dyn Element>) -> RuleResult
-    {
-        RuleResult
-        {
-            length: _length,
-            elements: vec![elem],
-        }
-    }
-}
-
-#[derive(Debug)]
-pub struct RuleError<'a>
-{
-    // where: token
-	cursor: &'a Cursor<'a>,
-    mat: Option<regex::Match<'a>>,
-    message: String,
-}
-
-impl<'a> RuleError<'a>
-{
-    pub fn new(_cursor: &'a Cursor<'a>, _match: Option<regex::Match<'a>>, _message: String) -> RuleError<'a>
-    {
-		RuleError
-		{
-			cursor: _cursor,
-			mat: _match,
-			message: _message,
-		}
-    }
-}
-
-pub trait SyntaxRule
+/*
+pub trait RegexRule: Rule
 {
 	fn name(&self) -> &'static str;
-    fn next_match<'a>(&self, cursor: &'a Cursor) -> Option<usize>;
-	fn on_match<'a>(&self, cursor: &'a Cursor) -> Result<(Token<'a>, RuleResult), RuleError<'a>>;
+
+	/// Returns the rule's regex
+	fn regex(&self) -> &regex::Regex;
+	/// Callback on regex rule match
+	fn on_regex_match<'a>(&self, parser: &Parser, document: &Document, token: Token<'a>, matches: regex::Captures) -> Vec<Report<'a, (String, Range<usize>)>>;
+}
+
+impl<T: RegexRule> Rule for T {
+    fn name(&self) -> &'static str { RegexRule::name(self) }
+
+	/// Finds the next match starting from [`cursor`]
+	fn next_match<'a>(&self, cursor: &'a Cursor) -> Option<usize>
+	{
+		let re = self.regex();
+
+		let content = cursor.file.content.as_ref().unwrap();
+		match re.find_at(content.as_str(), cursor.pos)
+		{
+			Some(m) => Some(m.start()),
+			None => None,
+		}
+	}
+
+	fn on_match<'a>(&self, parser: &Parser, document: &Document, cursor: Cursor<'a>) -> (Cursor<'a>, Vec<Report<'a, (String, Range<usize>)>>)
+	{
+		let content = cursor.file.content.as_ref().unwrap();
+		let matches = self.regex().captures_at(content.as_str(), cursor.pos).unwrap();
+		let token = Token::new(cursor.pos, matches.get(0).unwrap().len(), cursor.file);
+
+		let token_end = token.end();
+		(cursor.at(token_end), self.on_regex_match(parser, document, token, matches))
+	}
+}
+*/
+
+pub trait RegexRule
+{
+	fn name(&self) -> &'static str;
+
+	/// Returns the rule's regexes
+	fn regexes(&self) -> &[regex::Regex];
+
+	/// Callback on regex rule match
+	fn on_regex_match(&self, index: usize, parser: &dyn Parser, document: &Document, token: Token, matches: regex::Captures) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>;
+}
+
+impl<T: RegexRule> Rule for T {
+	fn name(&self) -> &'static str { RegexRule::name(self) }
+
+	/// Finds the next match starting from [`cursor`]
+	fn next_match(&self, cursor: &Cursor)
+		-> Option<(usize, Box<dyn Any>)> {
+		let content = cursor.source.content();
+        let mut found: Option<(usize, usize)> = None;
+        self.regexes().iter().enumerate().for_each(|(id, re)| {
+            if let Some(m) = re.find_at(content.as_str(), cursor.pos)
+            {
+                found = found
+					.and_then(|(f_pos, f_id)|
+						if f_pos > m.start() { Some((m.start(), id)) } else { Some((f_pos, f_id)) } )
+					.or(Some((m.start(), id)));
+            }
+        });
+
+        return found.map(|(pos, id)|
+			(pos, Box::new(id) as Box<dyn Any>));
+	}
+
+	fn on_match(&self, parser: &dyn Parser, document: &Document, cursor: Cursor, match_data: Option<Box<dyn Any>>)
+		-> (Cursor, Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>) {
+		let content = cursor.source.content();
+		let index = unsafe { match_data.unwrap_unchecked().downcast::<usize>().unwrap_unchecked() };
+		let re = &self.regexes()[*index];
+
+		let captures = re.captures_at(content.as_str(), cursor.pos).unwrap();
+		let token = Token::new(captures.get(0).unwrap().range(), cursor.source.clone());
+
+		let token_end = token.end();
+		return (cursor.at(token_end), self.on_regex_match(*index, parser, document, token, captures));
+	}
 }

src/parser/section.rs

@@ -1,151 +0,0 @@
-use regex::Regex;
-use super::rule::{RuleResult, RuleError, SyntaxRule};
-use super::super::files::cursor::Cursor;
-use super::super::files::token::Token;
-use super::super::syntax::element::{Element, ReferenceableElement};
-
-pub mod SectionKind 
-{
-    pub const NONE : u8 = 0x00;
-    pub const NO_TOC : u8 = 0x01;
-    pub const NO_NUMBER : u8 = 0x02;
-}
-
-pub struct Section
-{
-    title: String,
-    reference: Option<String>,
-    section_kind: u8,
-    depth: usize,
-}
-
-impl Section
-{
-    pub fn new<'h>(_title: String, _reference: Option<String>, kind: u8, _depth: usize) -> Section
-    {
-        Section {
-            title: _title,
-            reference: _reference,
-            section_kind: kind,
-            depth: _depth,
-        }
-    }
-}
-
-impl Element for Section
-{
-    fn element_name(&self) -> &'static str { "Section" }
-}
-
-impl ReferenceableElement for Section
-{
-    fn reference_name(&self) -> Option<&String> { self.reference.as_ref() }
-}
-
-
-// TODO: Single file for grammar + element, and add `Rule` suffix for rules
-pub struct SectionRule
-{
-	regex: Regex,
-}
-
-impl SectionRule
-{
-    pub fn new() -> SectionRule
-    {
-		SectionRule
-		{
-			regex: regex::Regex::new(r"(?:^|\n)(#{1,})(\{.*\})?((?:\*|\+){0,})?((?:\t| ){0,})(.*)").unwrap()
-		}
-    }
-}
-
-impl SyntaxRule for SectionRule
-{
-	fn name(&self) -> &'static str { "Section" }
-
-    fn next_match<'a>(&self, cursor: &'a Cursor) -> Option<usize>
-    {
-		match self.regex.find_at(&cursor.content, cursor.position)
-		{
-			Some(m) => Some(m.start()),
-			None => None
-		}
-    }
-
-	fn on_match<'a>(&self, cursor: &'a Cursor) -> Result<(Token<'a>, RuleResult), RuleError<'a>>
-	{
-        let m = self.regex.captures_at(&cursor.content, cursor.position).unwrap(); // Capture match
-		
-		let section_depth = match m.get(1)
-        {
-            Some(depth) => {
-				if depth.len() > 6
-				{
-					return Err(RuleError::new(&cursor, m.get(1),
-					format!("Section depth must not be greater than 6, got `{}` (depth: {})", depth.as_str(), depth.len())))
-				}
-
-				depth.len()
-            }
-            _ => return Err(RuleError::new(&cursor, m.get(1), String::from("Empty section depth")))
-        };
-
-		// Spacing
-		match m.get(4)
-		{
-			Some(spacing) => {
-				if spacing.as_str().is_empty() || !spacing.as_str().chars().all(|c| c == ' ' || c == '\t')
-				{
-					return Err(RuleError::new(&cursor, m.get(4),
-					format!("Sections require spacing made of spaces or tab before the section's title, got: `{}`", spacing.as_str())))
-				}
-			}
-			_ => return Err(RuleError::new(&cursor, m.get(4),
-			String::from("Sections require spacing made of spaces or tab before the section's title")))
-		}
-
-		let section_refname = match m.get(2)
-		{
-			Some(reference) => {
-				// TODO: Validate reference name
-				// TODO: After parsing, check for duplicate references
-				Some(String::from(reference.as_str()))
-			}
-			_ => None
-		};
-
-		let section_kind = match m.get(3)
-		{
-			Some(kind) => {
-				match kind.as_str() {
-					"*+" => SectionKind::NO_NUMBER | SectionKind::NO_TOC,
-					"*" => SectionKind::NO_NUMBER,
-					"+" => SectionKind::NO_TOC,
-					"" => SectionKind::NONE,
-					_ => return Err(RuleError::new(&cursor, m.get(3),
-					format!("Section kind must be either `*` for unnumbered, `+` to hide from TOC or `*+`, got `{}`. Leave empty for normal sections", kind.as_str())))
-				}
-			}
-			_ => SectionKind::NONE,
-		};
-
-		let section_title = match m.get(5) {
-			Some(title) => match title.as_str() {
-				"" => return Err(RuleError::new(&cursor, m.get(5),
-				String::from("Sections require a non-empty title"))),
-				_ => String::from(title.as_str())
-			}
-			_ => return Err(RuleError::new(&cursor, m.get(5),
-			String::from("Sections require a non-empty title")))
-		};
-
-        let section = Box::new(Section::new(
-            section_title,
-            section_refname,
-            section_kind,
-            section_depth));
-
-        Ok((Token::from(cursor, m.get(0).unwrap()), RuleResult::new(m.get(0).unwrap().len(), section)))
-	}
-}

src/parser/source.rs

@@ -0,0 +1,177 @@
+use std::{fs, ops::Range, rc::Rc};
+use core::fmt::Debug;
+
+use downcast_rs::{impl_downcast, Downcast};
+
+/// Trait for source content
+pub trait Source: Downcast
+{
+    /// Gets the source's location
+    fn location(&self) -> Option<&Token>;
+    /// Gets the source's name
+    fn name(&self) -> &String;
+    /// Gets the source's content
+    fn content(&self) -> &String;
+}
+impl_downcast!(Source);
+
+impl core::fmt::Display for dyn Source
+{
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+		write!(f, "{}", self.name())
+    }
+}
+
+impl core::fmt::Debug for dyn Source
+{
+	// TODO
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+		write!(f, "Source{{{}}}", self.name())
+    }
+}
+
+impl std::cmp::PartialEq for dyn Source
+{
+    fn eq(&self, other: &Self) -> bool {
+    	self.name() == other.name()
+    }
+}
+
+impl std::cmp::Eq for dyn Source {}
+
+impl std::hash::Hash for dyn Source
+{
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+		self.name().hash(state)
+    }
+}
+
+pub struct SourceFile
+{
+    location: Option<Token>,
+    path: String,
+    content: String,
+}
+
+
+impl SourceFile
+{
+    // TODO: Create a SourceFileRegistry holding already loaded files to avoid reloading them
+    pub fn new(path: String, location: Option<Token>) -> Result<Self, String>
+    {
+		match fs::read_to_string(&path)
+		{
+			Err(_) => return Err(String::from(format!("Unable to read file content: `{}`", path))),
+			Ok(content) => Ok(Self{
+                location,
+                path,
+                content,
+            }),
+		}
+    }
+}
+
+impl Source for SourceFile
+{
+    fn location(&self) -> Option<&Token> { self.location.as_ref() }
+    fn name(&self) -> &String { &self.path }
+    fn content(&self) -> &String { &self.content }
+}
+
+pub struct VirtualSource
+{
+	location: Token,
+    name: String,
+    content: String,
+}
+
+impl VirtualSource
+{
+    pub fn new(location: Token, name: String, content: String) -> Self
+    {
+        Self { location, name, content }
+    }
+}
+
+impl Source for VirtualSource
+{
+    fn location(&self) -> Option<&Token> { Some(&self.location) }
+    fn name(&self) -> &String { &self.name }
+    fn content(&self) -> &String { &self.content }
+}
+
+#[derive(Debug)]
+pub struct Cursor
+{
+	pub pos: usize,
+	pub source: Rc<dyn Source>,
+}
+
+impl Cursor {
+	pub fn new(pos: usize, source: Rc<dyn Source>) -> Self {
+		Self { pos, source }
+	}
+
+    /// Creates [`cursor`] at [`new_pos`] in the same [`file`]
+	pub fn at(&self, new_pos: usize) -> Self
+	{
+		Self {
+			pos: new_pos,
+			source: self.source.clone(),
+		}
+	}
+}
+
+impl Clone for Cursor
+{
+	fn clone(&self) -> Self {
+		Self {
+			pos: self.pos,
+			source: self.source.clone(),
+		}
+	}
+
+	fn clone_from(&mut self, source: &Self) {
+		*self = source.clone()
+	}
+}
+
+#[derive(Debug, Clone)]
+pub struct Token
+{
+    pub range: Range<usize>,
+	source: Rc<dyn Source>,
+}
+
+impl Token
+{
+	pub fn new(range: Range<usize>, source: Rc<dyn Source>) -> Self {
+		Self { range, source }
+	}
+
+	pub fn source(&self) -> Rc<dyn Source>
+	{
+		return self.source.clone()
+	}
+
+    /// Construct Token from a range
+    pub fn from(start: &Cursor, end: &Cursor) -> Self
+    {
+        assert!(Rc::ptr_eq(&start.source, &end.source));
+
+        Self {
+            range: start.pos .. end.pos,
+            source: start.source.clone()
+        }
+    }
+
+	pub fn start(&self) -> usize
+	{
+		return self.range.start;
+	}
+
+	pub fn end(&self) -> usize
+	{
+		return self.range.end;
+	}
+}

src/parser/state.rs

@@ -0,0 +1,87 @@
+use std::{cell::RefCell, collections::HashMap, ops::Range, rc::Rc};
+
+use ariadne::Report;
+use downcast_rs::{impl_downcast, Downcast};
+
+use crate::document::document::Document;
+
+use super::{parser::Parser, source::Source};
+
+/// Scope for state objects
+#[derive(PartialEq, PartialOrd)]
+pub enum Scope
+{
+	/// Global state
+	GLOBAL = 0,
+	/// Document-local state
+	DOCUMENT = 1,
+	/// Paragraph-local state
+	/// NOTE: Even though paragraph may span across multiple documents,
+	/// a paragraph-local state should be removed when importing a new document
+	PARAGRAPH = 2,
+}
+
+pub trait State: Downcast
+{
+	/// Returns the state's [`Scope`]
+	fn scope(&self) -> Scope;
+
+	/// Callback called when state goes out of scope
+	fn on_remove<'a>(&self, parser: &dyn Parser, document: &Document) -> Vec<Report<'a, (Rc<dyn Source>, Range<usize>)>>;
+}
+impl_downcast!(State);
+
+/// Object owning all the states
+pub struct StateHolder
+{
+	data: HashMap<String, Rc<RefCell<dyn State>>>
+}
+
+impl StateHolder
+{
+	pub fn new() -> Self {
+		Self {
+			data: HashMap::new(),
+		}
+	}
+
+	// Attempts to push [`state`]. On collision, returns an error with the already present state
+	pub fn insert(&mut self, name: String, state: Rc<RefCell<dyn State>>) -> Result<Rc<RefCell<dyn State>>, Rc<RefCell<dyn State>>>
+	{
+		match self.data.insert(name, state.clone())
+		{
+			Some(state) => Err(state),
+			_ => Ok(state)
+		}
+	}
+
+	pub fn query(&self, name: &String) -> Option<Rc<RefCell<dyn State>>>
+	{
+		self.data
+			.get(name)
+			.map_or(None, |st| Some(st.clone()))
+	}
+
+	pub fn on_scope_end(&mut self, parser: &dyn Parser, document: &Document, scope: Scope) -> Vec<Report<'_, (Rc<dyn Source>, Range<usize>)>>
+	{
+		let mut result = vec![];
+		
+		self.data
+			.retain(|_name, state|
+			{
+				if state.borrow().scope() >= scope
+				{
+					state.borrow().on_remove(parser, document)
+						.drain(..)
+						.for_each(|report| result.push(report));
+					false
+				}
+				else
+				{
+					true
+				}
+			});
+
+		return result;
+	}
+}

src/parser/util.rs

@@ -0,0 +1,343 @@
+use std::collections::HashMap;
+
+use unicode_segmentation::UnicodeSegmentation;
+
+use crate::{document::{document::Document, element::ElemKind}, elements::paragraph::Paragraph};
+
+/// Processes text for escape characters and paragraphing
+pub fn process_text(document: &Document, content: &str) -> String
+{
+	let mut escaped = false;
+	let mut newlines = 0usize; // Consecutive newlines
+	//println!("Processing: [{content}]");
+	let processed = content
+		.grapheme_indices(true)
+		.fold((String::new(), None),
+		|(mut out, prev), (_pos, g)| {
+		if newlines != 0 && g != "\n"
+		{
+			newlines = 0;
+
+			// Add a whitespace if necessary
+			match out.chars().last()
+			{
+				Some(c) => {
+					// NOTE: \n is considered whitespace, so previous codepoint can be \n
+					// (Which can only be done by escaping it)
+					if !c.is_whitespace() || c == '\n'
+					{
+						out += " ";
+					}
+				}
+				None => {
+					if document.last_element::<Paragraph>(false)
+						.and_then(|par| par.find_back(|e| e.kind() != ElemKind::Invisible)
+						.and_then(|e| Some(e.kind() == ElemKind::Inline)))
+							.unwrap_or(false)
+					{
+						out += " ";
+					}
+				} // Don't output anything
+			}
+		}
+
+		// Output grapheme literally when escaped
+		if escaped
+		{
+			escaped = false;
+			return (out + g, Some(g));
+		}
+		// Increment newlines counter
+		else if g == "\n"
+		{
+			newlines += 1;
+			return (out, Some(g));
+		}
+		// Determine if escaped
+		else if g == "\\"
+		{
+			escaped = !escaped;
+			return (out, Some(g));
+		}
+		// Whitespaces
+		else if g.chars().count() == 1 && g.chars().last().unwrap().is_whitespace()
+		{
+			// Content begins with whitespace
+			if prev.is_none()
+			{
+				if document.last_element::<Paragraph>(false).is_some()
+				{
+					return (out+g, Some(g));
+				}
+				else
+				{
+					return (out, Some(g));
+				}
+			}
+			// Consecutive whitespaces are converted to a single whitespace
+			else if prev.unwrap().chars().count() == 1 &&
+					prev.unwrap().chars().last().unwrap().is_whitespace()
+			{
+				return (out, Some(g));
+			}
+		}
+
+		return (out + g, Some(g));
+	}).0.to_string();
+
+	return processed;
+}
+
+/// Processed a string and escapes a single token out of it
+/// Escaped characters other than the [`token`] will be not be treated as escaped
+///
+/// # Example
+/// ```
+/// assert_eq!(process_escaped('\\', "%", "escaped: \\%, also escaped: \\\\\\%, untouched: \\a"),
+/// "escaped: %, also escaped: \\%, untouched \\a");
+/// ```
+pub fn process_escaped<S: AsRef<str>>(escape: char, token: &'static str, content: S) -> String
+{
+	let mut processed = String::new();
+	let mut escaped = 0;
+	let mut token_it = token.chars().peekable();
+	for c in content.as_ref().chars()
+		.as_str()
+			.trim_start()
+			.trim_end()
+			.chars()
+			{
+				if c == escape
+				{
+					escaped += 1;
+				}
+				else if escaped % 2 == 1 && token_it.peek().map_or(false, |p| *p == c)
+				{
+					let _ = token_it.next();
+					if token_it.peek() == None
+					{
+						(0..((escaped-1)/2))
+							.for_each(|_| processed.push(escape));
+						escaped = 0;
+						token_it = token.chars().peekable();
+						processed.push_str(token);
+					}
+				}
+				else
+				{
+					if escaped != 0
+					{
+						// Add untouched escapes
+						(0..escaped).for_each(|_| processed.push('\\'));
+						token_it = token.chars().peekable();
+						escaped = 0;
+					}
+					processed.push(c);
+				}
+			}
+	// Add trailing escapes
+	(0..escaped).for_each(|_| processed.push('\\'));
+
+	processed
+}
+
+#[derive(Debug)]
+pub struct Property
+{
+	required: bool,
+	description: String,
+	default: Option<String>,
+}
+
+impl Property {
+    pub fn new(required: bool, description: String, default: Option<String>) -> Self {
+        Self { required, description, default }
+    }
+}
+
+impl core::fmt::Display for Property
+{
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+		match self.default.as_ref()
+		{
+			None => write!(f, "{} {}", 
+				["[Opt]", "[Req]"][self.required as usize],
+				self.description),
+			Some(default) => write!(f, "{} {} (Deafult: {})",
+				["[Opt]", "[Req]"][self.required as usize],
+				self.description,
+				default)
+		}
+    }
+}
+
+#[derive(Debug)]
+pub struct PropertyMap<'a>
+{
+	pub(crate) properties: HashMap<String, (&'a Property, String)>
+}
+
+impl<'a> PropertyMap<'a> {
+    pub fn new() -> Self {
+        Self { properties: HashMap::new() }
+    }
+
+	pub fn get<T, Error, F: FnOnce(&'a Property, &String) -> Result<T, Error>>(&self, name: &str, f: F)
+		-> Result<(&'a Property, T), Error> {
+		let (prop, value) = self.properties.get(name).unwrap();
+
+		f(prop, value).and_then(|value| Ok((*prop, value)))
+	}
+}
+
+pub struct PropertyParser {
+	properties: HashMap<String, Property>,
+}
+
+impl PropertyParser {
+    pub fn new(properties: HashMap<String, Property>) -> Self {
+        Self { properties }
+    }
+
+	/// Attempts to build a default propertymap
+	///
+	/// Returns an error if at least one [`Property`] is required and doesn't provide a default
+	pub fn default(&self) -> Result<PropertyMap<'_>, String> {
+		let mut properties = PropertyMap::new();
+
+		for (name, prop) in &self.properties
+		{
+			match (prop.required, prop.default.as_ref())
+			{
+    			(true, None) => return Err(format!("Missing property `{name}` {prop}")),
+				(false, None) => {},
+    			(_, Some(default)) => {
+						properties.properties.insert(
+						name.clone(),
+						(prop, default.clone())
+					);
+				}
+			}
+		}
+
+		Ok(properties)
+	}
+
+	/// Parses properties string "prop1=value1, prop2 = val\,2" -> {prop1: value1, prop2: val,2}
+	///
+	/// # Key-value pair
+	///
+	/// Property names/values are separated by a single '=' that cannot be escaped.
+	/// Therefore names cannot contain the '=' character.
+	///
+	/// # Example
+	///
+	/// ```
+	/// let properties = HashMap::new();
+	/// properties.insert("width", Property::new(true, "Width of the element in em", None));
+	///
+	/// let parser = PropertyParser::new(properties);
+	/// let pm = parser.parse("width=15").unwrap();
+	///
+	/// assert!(pm.get("width", |_, val| val.parse::<i32>()) == Ok(15));
+	/// ```
+	/// # Return value
+	///
+	/// Returns the parsed property map, or an error if either:
+	///  * A required property is missing
+	///  * An unknown property is present
+	///  * A duplicate property is present
+	///
+	/// Note: Only ',' inside values can be escaped, other '\' are treated literally
+	pub fn parse(&self, content: &str) -> Result<PropertyMap<'_>, String> {
+		let mut properties = PropertyMap::new();
+		let mut try_insert = |name: &String, value: &String|
+			-> Result<(), String> {
+			let trimmed_name = name.trim_end().trim_start();
+			let trimmed_value = value.trim_end().trim_start();
+			let prop = match self.properties.get(trimmed_name)
+			{
+				None => return Err(format!("Unknown property name: `{trimmed_name}` (with value: `{trimmed_value}`). Valid properties are:\n{}",
+					self.properties.iter().fold(String::new(),
+					|out, (name, prop)| out + format!(" - {name}: {prop}\n").as_str()))),
+				Some(prop) => prop
+			};
+
+			if let Some((_, previous)) = properties.properties.insert(
+				trimmed_name.to_string(),
+				(prop, trimmed_value.to_string()))
+			{
+				return Err(format!("Duplicate property `{trimmed_name}`, previous value: `{previous}` current value: `{trimmed_value}`"))
+			}
+
+			Ok(())
+		};
+
+		let mut in_name = true;
+		let mut name = String::new();
+		let mut value = String::new();
+		let mut escaped = 0usize;
+		for c in content.chars()
+		{
+			if c == '\\'
+			{
+				escaped += 1;
+			}
+			else if c == '=' && in_name
+			{
+				in_name = false;
+				(0..escaped).for_each(|_| name.push('\\'));
+				escaped = 0;
+			}
+			else if c == ',' && !in_name
+			{
+				if escaped % 2 == 0 // Not escaped
+				{
+					(0..escaped/2).for_each(|_| value.push('\\'));
+					escaped = 0;
+					in_name = true;
+
+					if let Err(e) = try_insert(&name, &value) {
+						return Err(e)
+					}
+					name.clear();
+					value.clear();
+				}
+				else
+				{
+					(0..(escaped-1)/2).for_each(|_| value.push('\\'));
+					value.push(',');
+					escaped = 0;
+				}
+			}
+			else
+			{
+				if in_name {
+					(0..escaped).for_each(|_| name.push('\\'));
+					name.push(c)
+				}
+				else {
+					(0..escaped).for_each(|_| value.push('\\'));
+					value.push(c)
+				}
+				escaped = 0;
+			}
+		}
+		if !in_name && value.trim_end().trim_start().is_empty()
+		{
+			return Err("Expected a value after last `=`".to_string())
+		}
+		else if name.is_empty() || value.is_empty()
+		{
+			return Err("Expected non empty property list.".to_string());
+		}
+
+		if let Err(e) = try_insert(&name, &value) {
+			return Err(e)
+		}
+
+		// TODO: Missing properties
+
+		Ok(properties)
+	}
+}

src/server.rs

@@ -0,0 +1,236 @@
+#![feature(char_indices_offset)]
+mod document;
+mod compiler;
+mod parser;
+mod elements;
+mod lua;
+mod cache;
+mod lsp;
+
+use std::collections::HashMap;
+use std::rc::Rc;
+use std::sync::Arc;
+
+use dashmap::DashMap;
+use document::variable::Variable;
+use tower_lsp::jsonrpc::Result;
+use tower_lsp::lsp_types::*;
+use tower_lsp::{Client, LanguageServer, LspService, Server};
+
+#[derive(Debug)]
+struct Backend {
+    client: Client,
+	document_map: DashMap<String, String>,
+	//variables: DashMap<String, HashMap<String, Arc<dyn Variable + Send + Sync + 'static>>>,
+}
+
+#[derive(Debug)]
+struct TextDocumentItem {
+	uri: Url,
+	text: String,
+	version: i32,
+}
+
+impl Backend {
+	async fn on_change(&self, params: TextDocumentItem) {
+		self.document_map
+			.insert(params.uri.to_string(), params.text.clone());
+		let ParserResult {
+			ast,
+			parse_errors,
+			semantic_tokens,
+		} = parse(&params.text);
+		let diagnostics = parse_errors
+			.into_iter()
+			.filter_map(|item| {
+				let (message, span) = match item.reason() {
+					chumsky::error::SimpleReason::Unclosed { span, delimiter } => {
+						(format!("Unclosed delimiter {}", delimiter), span.clone())
+					}
+					chumsky::error::SimpleReason::Unexpected => (
+						format!(
+							"{}, expected {}",
+							if item.found().is_some() {
+								"Unexpected token in input"
+							} else {
+								"Unexpected end of input"
+							},
+							if item.expected().len() == 0 {
+								"something else".to_string()
+							} else {
+								item.expected()
+									.map(|expected| match expected {
+										Some(expected) => expected.to_string(),
+										None => "end of input".to_string(),
+									})
+								.collect::<Vec<_>>()
+									.join(", ")
+							}
+						),
+						item.span(),
+						),
+						chumsky::error::SimpleReason::Custom(msg) => (msg.to_string(), item.span()),
+				};
+
+				|| -> Option<Diagnostic> {
+					// let start_line = rope.try_char_to_line(span.start)?;
+					// let first_char = rope.try_line_to_char(start_line)?;
+					// let start_column = span.start - first_char;
+					let start_position = offset_to_position(span.start, &rope)?;
+					let end_position = offset_to_position(span.end, &rope)?;
+					// let end_line = rope.try_char_to_line(span.end)?;
+					// let first_char = rope.try_line_to_char(end_line)?;
+					// let end_column = span.end - first_char;
+					Some(Diagnostic::new_simple(
+							Range::new(start_position, end_position),
+							message,
+					))
+				}()
+			})
+		.collect::<Vec<_>>();
+
+		self.client
+			.publish_diagnostics(params.uri.clone(), diagnostics, Some(params.version))
+			.await;
+
+		if let Some(ast) = ast {
+			self.ast_map.insert(params.uri.to_string(), ast);
+		}
+		// self.client
+		//     .log_message(MessageType::INFO, &format!("{:?}", semantic_tokens))
+		//     .await;
+		self.semantic_token_map
+			.insert(params.uri.to_string(), semantic_tokens);
+		}
+}
+
+#[tower_lsp::async_trait]
+impl LanguageServer for Backend {
+    async fn initialize(&self, _: InitializeParams) -> Result<InitializeResult> {
+		Ok(InitializeResult {
+			server_info: None,
+			capabilities: ServerCapabilities {
+				text_document_sync: Some(TextDocumentSyncCapability::Kind(
+						TextDocumentSyncKind::FULL,
+				)),
+				completion_provider: Some(CompletionOptions {
+					resolve_provider: Some(false),
+					trigger_characters: Some(vec!["%".to_string()]),
+					work_done_progress_options: Default::default(),
+					all_commit_characters: None,
+					completion_item: None,
+				}),
+				semantic_tokens_provider: Some(
+                    SemanticTokensServerCapabilities::SemanticTokensRegistrationOptions(
+                        SemanticTokensRegistrationOptions {
+                            text_document_registration_options: {
+                                TextDocumentRegistrationOptions {
+                                    document_selector: Some(vec![DocumentFilter {
+                                        language: Some("nml".to_string()),
+                                        scheme: Some("file".to_string()),
+                                        pattern: None,
+                                    }]),
+                                }
+                            },
+                            semantic_tokens_options: SemanticTokensOptions {
+                                work_done_progress_options: WorkDoneProgressOptions::default(),
+                                legend: SemanticTokensLegend {
+                                    token_types: vec![SemanticTokenType::COMMENT, SemanticTokenType::MACRO],
+                                    token_modifiers: vec![],
+                                },
+                                range: None, //Some(true),
+                                full: Some(SemanticTokensFullOptions::Bool(true)),
+                            },
+                            static_registration_options: StaticRegistrationOptions::default(),
+                        },
+                    ),
+                ),
+				..ServerCapabilities::default()
+			},
+		})
+    }
+
+    async fn initialized(&self, _: InitializedParams) {
+        self.client
+            .log_message(MessageType::INFO, "server initialized!")
+            .await;
+    }
+
+    async fn shutdown(&self) -> Result<()> {
+        Ok(())
+    }
+
+	async fn did_open(&self, params: DidOpenTextDocumentParams) {
+		self.client
+			.log_message(MessageType::INFO, "file opened!")
+			.await;
+		self.on_change(TextDocumentItem {
+			uri: params.text_document.uri,
+			text: params.text_document.text,
+			version: params.text_document.version,
+		})
+		.await
+	}
+
+	async fn did_change(&self, mut params: DidChangeTextDocumentParams) {
+		self.on_change(TextDocumentItem {
+			uri: params.text_document.uri,
+			text: std::mem::take(&mut params.content_changes[0].text),
+			version: params.text_document.version,
+		})
+		.await
+	}
+
+    async fn completion(&self, params: CompletionParams) -> Result<Option<CompletionResponse>> {
+        let uri = params.text_document_position.text_document.uri;
+        let position = params.text_document_position.position;
+        let completions = || -> Option<Vec<CompletionItem>> {
+            let mut ret = Vec::with_capacity(0);
+
+            Some(ret)
+        }();
+        Ok(completions.map(CompletionResponse::Array))
+    }
+
+	async fn semantic_tokens_full(
+		&self,
+		params: SemanticTokensParams,
+	) -> Result<Option<SemanticTokensResult>> {
+		let uri = params.text_document.uri.to_string();
+		self.client
+			.log_message(MessageType::LOG, "semantic_token_full")
+			.await;
+		let semantic_tokens = || -> Option<Vec<SemanticToken>> {
+			let semantic_tokens = vec![
+				SemanticToken {
+					delta_line: 1,
+					delta_start: 2,
+					length: 5,
+					token_type: 1,
+					token_modifiers_bitset: 0,
+				}
+			];
+			Some(semantic_tokens)
+		}();
+		if let Some(semantic_token) = semantic_tokens {
+			return Ok(Some(SemanticTokensResult::Tokens(SemanticTokens {
+				result_id: None,
+				data: semantic_token,
+			})));
+		}
+		Ok(None)
+	}
+}
+
+#[tokio::main]
+async fn main() {
+    let stdin = tokio::io::stdin();
+    let stdout = tokio::io::stdout();
+
+    let (service, socket) = LspService::new(
+		|client|
+			Backend {
+				client
+			});
+    Server::new(stdin, stdout, socket).serve(service).await;
+}

src/syntax.rs

@@ -1 +0,0 @@
-pub mod element;

src/syntax/document.rs

@@ -1,4 +0,0 @@
-struct Document
-{
-    
-}

src/syntax/element.rs

@@ -1,30 +0,0 @@
-pub trait Element
-{
-    fn element_name(&self) -> &'static str;
-    fn token(&'a self) -> Token<'a>
-}
-
-pub trait ReferenceableElement : Element
-{
-    fn reference_name(&self) -> Option<&String>;
-}
-
-pub struct Text
-{
-    content: String,
-}
-
-impl Text
-{
-    pub fn new<'h>(_content: &'h str) -> Text
-    {
-        Text {
-            content: String::from(_content)
-        }
-    }
-}
-
-impl Element for Text
-{
-    fn element_name(&self) -> &'static str { "Text" }
-}