1 files changed, 217 insertions, 0 deletions
diff --git a/src/diagnostics.rs b/src/diagnostics.rs
new file mode 100644
index 0000000..514818a
--- /dev/null
+++ b/src/diagnostics.rs
@@ -0,0 +1,217 @@
+use miette::{Diagnostic, Error};
+use thiserror::Error;
+
+use crate::{
+	parse::{Attribute, Document, Element, Node},
+	tokenize::{AttributeToken, Token, TokenType},
+};
+
+mod duplicate_attribute_names;
+mod empty_tag_name;
+mod end_tag_with_attributes;
+mod forbidden_top_level_tag;
+mod html_tag;
+mod image_tag;
+mod non_alphabetic_attribute_name;
+mod non_alphabetic_tag_name;
+mod non_self_closing_void_tag;
+mod non_void_self_closing_tag;
+mod self_closing_end_tag;
+mod stray_end_tag;
+mod stray_equal_sign;
+mod too_many_equal_signs;
+mod top_level_character_data;
+mod unclosed_start_tag;
+mod unterminated_comment;
+mod unterminated_tag;
+
+const VOID_TAGS: &[&str] = &[
+	"area", "base", "br", "col", "embed", "hr", "img", "input", "link", "meta", "param", "source",
+	"track", "wbr",
+];
+const ALLOWED_TOP_LEVEL_TAGS: &[&str] = &["import", "const", "block", "head", "body"];
+
+pub struct DiagnosticsContext {
+	source_code: &'static str,
+}
+
+trait TokenDiagnostic: Sized + Send + Sync + 'static {
+	fn handle_token(_: &Token<'_>) -> Option<Self> {
+		None
+	}
+
+	fn handle_attribute_token(_: &AttributeToken<'_>) -> Option<Self> {
+		None
+	}
+}
+
+trait AstDiagnostic: Sized + Send + Sync + 'static {
+	fn handle_document(_: &Document<'_>) -> Option<Self> {
+		None
+	}
+
+	fn handle_node(_: &Node<'_>) -> Option<Self> {
+		None
+	}
+
+	fn handle_element(_: &Element<'_>) -> Option<Self> {
+		None
+	}
+
+	fn handle_attribute(_: &Attribute<'_>) -> Option<Self> {
+		None
+	}
+}
+
+fn maybe_print_diagnostic<D: Diagnostic + Send + Sync + 'static>(
+	ctx: &DiagnosticsContext,
+	diagnostic: Option<D>,
+) {
+	if let Some(diagnostic) = diagnostic {
+		eprintln!(
+			"{}",
+			Error::new(diagnostic).with_source_code(ctx.source_code)
+		);
+	}
+}
+
+#[derive(Debug, Error, Diagnostic)]
+#[error("This file is not valid UTF-8")]
+#[diagnostic(code(NonUtf8), severity(Error))]
+struct NonUtf8;
+
+#[derive(Debug, Error, Diagnostic)]
+#[error("Forbidden NUL character found")]
+#[diagnostic(code(NonUtf8), severity(Error))]
+struct NulCharacter {
+	#[label("Forbidden NUL byte found here")]
+	location: usize,
+}
+
+#[derive(Debug, Error, Diagnostic)]
+#[error("Forbidden control character found")]
+#[diagnostic(code(NonUtf8), severity(Error))]
+struct ControlCharacter {
+	#[label("Control character found")]
+	location: usize,
+}
+
+pub fn run_character_diagnostics(source: &'static [u8]) -> Option<&'static str> {
+	let Ok(source) = str::from_utf8(source) else {
+		eprintln!("{}", Error::new(NonUtf8));
+		return None;
+	};
+	let ctx = DiagnosticsContext {
+		source_code: source,
+	};
+	for (i, char) in source.char_indices() {
+		maybe_print_diagnostic(&ctx, (char == '\0').then_some(NulCharacter { location: i }));
+		maybe_print_diagnostic(
+			&ctx,
+			(char.is_control() && !char.is_ascii_whitespace())
+				.then_some(ControlCharacter { location: i }),
+		);
+	}
+
+	Some(source)
+}
+
+fn run_token_diagnostic<D: Diagnostic + TokenDiagnostic>(
+	ctx: &DiagnosticsContext,
+	tokens: &[Token<'_>],
+) {
+	for token in tokens {
+		maybe_print_diagnostic(ctx, D::handle_token(token));
+
+		let TokenType::Tag {
+			attribute_tokens, ..
+		} = token.ty
+		else {
+			continue;
+		};
+
+		for token in attribute_tokens {
+			maybe_print_diagnostic(ctx, D::handle_attribute_token(token));
+		}
+	}
+}
+
+fn run_ast_diagnostic<D: AstDiagnostic + Diagnostic>(
+	ctx: &DiagnosticsContext,
+	document: &Document<'_>,
+) {
+	maybe_print_diagnostic(ctx, D::handle_document(document));
+	for node in document.children {
+		maybe_print_diagnostic(ctx, D::handle_node(node));
+
+		let Node::Element(element) = node else {
+			continue;
+		};
+		maybe_print_diagnostic(ctx, D::handle_element(element));
+
+		for attribute in element.attributes {
+			maybe_print_diagnostic(ctx, D::handle_attribute(attribute));
+		}
+	}
+}
+
+pub fn run_token_diagnostics(ctx: &DiagnosticsContext, tokens: &[Token<'_>]) {
+	run_token_diagnostic::<non_void_self_closing_tag::NonVoidSelfClosingTag>(ctx, tokens);
+	run_token_diagnostic::<non_self_closing_void_tag::NonSelfClosingVoidTag>(ctx, tokens);
+	run_token_diagnostic::<end_tag_with_attributes::EndTagWithAttributes>(ctx, tokens);
+	run_token_diagnostic::<self_closing_end_tag::SelfClosingEndTag>(ctx, tokens);
+	run_token_diagnostic::<non_alphabetic_tag_name::NonAlphabeticTagName>(ctx, tokens);
+	run_token_diagnostic::<empty_tag_name::EmptyTagName>(ctx, tokens);
+	run_token_diagnostic::<unterminated_tag::UnterminatedTag>(ctx, tokens);
+	run_token_diagnostic::<unterminated_comment::UnterminatedComment>(ctx, tokens);
+	run_token_diagnostic::<non_alphabetic_attribute_name::NonAlphabeticAttributeName>(ctx, tokens);
+	// TODO invalid character reference
+}
+
+pub fn run_ast_diagnostics(ctx: &DiagnosticsContext, document: &Document<'_>) {
+	run_ast_diagnostic::<stray_equal_sign::StrayEqualSign>(ctx, document);
+	run_ast_diagnostic::<too_many_equal_signs::TooManyEqualSigns>(ctx, document);
+	run_ast_diagnostic::<duplicate_attribute_names::DuplicateAttributeNames>(ctx, document);
+	run_ast_diagnostic::<top_level_character_data::TopLevelCharacterData>(ctx, document);
+	run_ast_diagnostic::<forbidden_top_level_tag::ForbiddenTopLevelTag>(ctx, document);
+	run_ast_diagnostic::<stray_end_tag::StrayEndTag>(ctx, document);
+	run_ast_diagnostic::<unclosed_start_tag::UnclosedStartTag>(ctx, document);
+	run_ast_diagnostic::<html_tag::HtmlTag>(ctx, document);
+	run_ast_diagnostic::<image_tag::ImageTag>(ctx, document);
+	// TODO invalid <kuht-var />
+	// TODO invalid <import />
+	// TODO invalid <const />
+	// TODO invalid <block />
+	// TODO deprecated tag
+	// TODO deprecated tag attribute
+	//
+	// TODO forbidden tag in noscript
+	// TODO head tag outside head
+	// TODO html tag in body
+	// TODO heading inside heading
+	// TODO form inside form
+	// TODO button inside button
+	// TODO anchor inside anchor
+	// TODO head inside body
+	// TODO caption, col, colgroup, tbody, tfoot, thead, tr outside table
+	// TODO table inside table
+	// TODO non-hidden input inside table
+	// TODO form inside table
+	// TODO body inside body
+	// TODO col, colgroup, tbody, tfoot, thead, tr inside caption
+	// TODO th, td outside tr
+	// TODO caption, col, colgroup, tr in tbody
+	// TODO caption, col, colgroup, tbody, td, tfoot, th, thead, tr inside table cell
+	// TODO select inside select
+	// TODO input, keygen, textarea inside select
+	// TODO caption, table, tbody, tfoot, thread, tr, td, th inside select
+	//
+	// TODO anchor with invalid href
+	// TODO dt or dd outside dl
+	// TODO li outside ul, ol, menu
+	// TODO invalid dl, ul, menu
+	// TODO invalid attribute values (see microsyntaxes)
+	// TODO aria tags don't match tag semantics
+	// TODO empty or missing image src
+	// TODO missing alt text for image (unless inside figure with caption, or has title, or meta name = "generator")
+}