From: Andrea Cornell <anders@acorn.pw>
Date: Sat, 23 Sep 2023 03:09:45 +0000 (-0400)
Subject: code jumble
X-Git-Url: https://jcornell.net/gitweb/gitweb.cgi?a=commitdiff_plain;h=44267a0b1cdb299cf4bb695ad5639f95449be10a;p=counting.git

code jumble

this is so that Jakob can use UniqHtmlElement for testing the number parsing
---

diff --git a/sharedmodel/src/html.rs b/sharedmodel/src/html.rs
new file mode 100644
index 0000000..a585349
--- /dev/null
+++ b/sharedmodel/src/html.rs
@@ -0,0 +1,16 @@
+use std::borrow::Borrow;
+
+#[derive(Debug)]
+pub enum HtmlNode<T, U> {
+	Text(T),
+	Element(U)
+}
+
+pub trait HtmlElement: Sized {
+	type Text1: AsRef<str>;
+	type Text2: AsRef<str>;
+	type ContentsIter: DoubleEndedIterator<Item=HtmlNode<Self::Text1, Self>>;
+	fn name(&self) -> &str;
+	fn contents(self) -> Self::ContentsIter;
+	fn text(&self) -> Self::Text2;
+}
diff --git a/sharedmodel/src/html_html5ever.rs b/sharedmodel/src/html_html5ever.rs
new file mode 100644
index 0000000..9afc890
--- /dev/null
+++ b/sharedmodel/src/html_html5ever.rs
@@ -0,0 +1,200 @@
+use std::cell::{Cell, RefCell};
+use std::rc::{Rc, Weak};
+use std::borrow::Cow;
+
+use html5ever::tree_builder::{TreeSink, ElementFlags};
+use html5ever::{QualName, ExpandedName, LocalNameStaticSet, Attribute};
+use html5ever::tendril::{StrTendril, TendrilSink};
+use html5ever::interface::{NodeOrText, QuirksMode};
+use html5ever::driver::ParseOpts;
+use string_cache::Atom;
+
+use crate::html::{HtmlNode, HtmlElement};
+
+type RcNode<'a> = HtmlNode<StrTendril, RcHtmlElement<'a>>;
+
+pub fn parse_html(html: &str) -> UniqHtmlElement {
+    let mut parser = html5ever::parse_fragment(RcHtmlElement(Rc::new(HtmlElementInner::root())), ParseOpts::default(), QualName::new(None, Atom::from("http://www.w3.org/1999/xhtml"), Atom::from("body")), Vec::new());
+	parser.process(StrTendril::from(html));
+	parser.finish()
+}
+
+// #[derive(Debug)]
+struct HtmlElementInner<'a> {
+	name: QualName,
+	parent: Cell<Weak<HtmlElementInner<'a>>>,
+	contents: RefCell<Vec<RcNode<'a>>>
+}
+
+impl HtmlElementInner<'_> {
+	fn root() -> Self {
+		let name = QualName::new(None, Atom::from(""), Atom::from(""));
+		let parent = Cell::new(Weak::new());
+		let contents = RefCell::new(Vec::new());
+		HtmlElementInner { name, parent, contents }
+	}
+}
+
+//#[derive(Clone, Debug)]
+#[derive(Clone)]
+struct RcHtmlElement<'a>(Rc<HtmlElementInner<'a>>);
+
+fn nodeortext_to_htmlnode<'a>(v: NodeOrText<RcHtmlElement<'a>>) -> RcNode<'a> {
+	match v {
+		NodeOrText::AppendNode(h) => HtmlNode::Element(h),
+		NodeOrText::AppendText(s) => HtmlNode::Text(s)
+	}
+}
+
+fn node_eq_elem<'a>(node: &RcNode<'a>, elem: &RcHtmlElement<'a>) -> bool {
+	match node {
+		HtmlNode::Element(elem1) if Rc::ptr_eq(&elem.0, &elem1.0) => true,
+		_ => false
+	}
+}
+
+impl<'a> TreeSink for RcHtmlElement<'a> {
+	type Handle = Self;
+	type Output = UniqHtmlElement;
+	fn finish(self) -> UniqHtmlElement {
+		// let HtmlElementInner { name, parent, contents } = Rc::try_unwrap(self.0).map_err(|_| ()).expect("no refs left around");
+		let HtmlElementInner { ref name, ref contents, .. } = *self.0;
+		let name = name.local.clone();
+		let contents = contents.borrow_mut().drain(..).map(|node| {
+			match node {
+				HtmlNode::Text(s) => HtmlNode::Text(s),
+				HtmlNode::Element(e) => HtmlNode::Element(e.finish())
+			}
+		}).collect();
+		UniqHtmlElement { name, contents }
+	}
+
+	fn parse_error(&mut self, msg: Cow<'static, str>) {
+		println!("{}", msg);
+		panic!("how do parse??");
+	}
+
+	fn get_document(&mut self) -> Self::Handle {
+		self.clone()
+	}
+
+	fn elem_name<'b>(&'b self, target: &'b Self::Handle) -> ExpandedName<'b> {
+		target.0.name.expanded()
+	}
+
+	fn create_element(&mut self, name: QualName, attrs: Vec<Attribute>, flags: ElementFlags) -> Self::Handle {
+		assert_eq!(flags.template, false);
+		assert_eq!(flags.mathml_annotation_xml_integration_point, false);
+		let contents = RefCell::new(Vec::new());
+		let parent = Cell::new(Weak::new());
+		// handily, we never use attribute data. Toss it
+		RcHtmlElement(Rc::new(HtmlElementInner { name, parent, contents }))
+	}
+
+	fn create_comment(&mut self, text: StrTendril) -> Self::Handle {
+		panic!("HTML parser encountered a comment");
+	}
+
+	fn create_pi(&mut self, target: StrTendril, data: StrTendril) -> Self::Handle {
+		panic!("HTML parser encountered a processing instruction");
+	}
+
+	fn append(&mut self, parent: &Self::Handle, child: NodeOrText<Self::Handle>) {
+		let node = nodeortext_to_htmlnode(child);
+		if let HtmlNode::Element(ref elem) = node {
+			elem.0.parent.set(Rc::downgrade(&parent.0));
+		}
+		parent.0.contents.borrow_mut().push(node);
+	}
+
+	fn append_based_on_parent_node(&mut self, element: &Self::Handle, prev_element: &Self::Handle, child: NodeOrText<Self::Handle>) {
+		panic!("Not clear what this is for or if it needs to be implemented");
+	}
+
+	fn append_doctype_to_document(&mut self, name: StrTendril, public_id: StrTendril, system_id: StrTendril) {
+		panic!("This can likely be a no-op, just not sure if it gets called for these fragments");
+	}
+
+	fn get_template_contents(&mut self, target: &Self::Handle) -> Self::Handle {
+		panic!("HTML parser encountered a <template> element")
+	}
+
+	fn same_node(&self, x: &Self::Handle, y: &Self::Handle) -> bool {
+		Rc::ptr_eq(&x.0, &y.0)
+	}
+
+	fn set_quirks_mode(&mut self, mode: QuirksMode) {}
+
+	fn append_before_sibling(&mut self, sibling: &Self::Handle, new_node: NodeOrText<Self::Handle>) {
+		let parent_weak = sibling.0.parent.take();
+		let parent = parent_weak.upgrade().expect("parent is not dropped");
+		let mut children = parent.contents.borrow_mut();
+		sibling.0.parent.set(parent_weak);
+		let pos = children.iter().position(|node| node_eq_elem(node, sibling)).expect("sibling is child of parent");
+		children.insert(pos, nodeortext_to_htmlnode(new_node));
+	}
+
+	fn add_attrs_if_missing(&mut self, target: &Self::Handle, attrs: Vec<Attribute>) {}
+
+	fn remove_from_parent(&mut self, target: &Self::Handle) {
+		let parent = target.0.parent.replace(Weak::new()).upgrade().expect("target has parent");
+		let mut children = parent.contents.borrow_mut();
+		let pos = children.iter().position(|node| node_eq_elem(node, target)).expect("sibling is child of parent");
+		children.remove(pos);
+	}
+
+	fn reparent_children(&mut self, node: &Self::Handle, new_parent: &Self::Handle) {
+		new_parent.0.contents.borrow_mut().extend(node.0.contents.borrow_mut().drain(..));
+	}
+}
+
+
+#[derive(Debug)]
+pub struct UniqHtmlElement {
+	name: Atom<LocalNameStaticSet>,
+	contents: Vec<HtmlNode<StrTendril, UniqHtmlElement>>
+}
+
+/*fn node_text(node: &HtmlNode<StrTendril, UniqHtmlElement>) -> Cow<str> {
+	match node {
+		HtmlNode::Text(s) => Cow::Borrowed(s.as_ref()),
+		HtmlNode::Element(e) => e.text()
+	}
+}*/
+
+fn element_text_inner(buf: &mut Option<StrTendril>, elem: &UniqHtmlElement) {
+	for node in &elem.contents {
+		match node {
+			HtmlNode::Text(s) => match buf {
+				None => { *buf = Some(s.clone()); },
+				Some(tendril) => tendril.push_tendril(&s)
+			},
+			HtmlNode::Element(e) => element_text_inner(buf, &e)
+		}
+	}
+}
+
+impl HtmlElement for UniqHtmlElement {
+	type Text1 = StrTendril;
+	type Text2 = StrTendril;
+	type ContentsIter = <Vec<HtmlNode<Self::Text1, Self>> as IntoIterator>::IntoIter;
+	fn name(&self) -> &str {
+		&*self.name
+	}
+	fn contents(self) -> Self::ContentsIter {
+		self.contents.into_iter()
+	}
+	fn text(&self) -> StrTendril {
+		let mut result = None;
+		element_text_inner(&mut result, self);
+		result.unwrap_or(StrTendril::new())
+/*		let mut init = self.contents.get(0).map(node_text).unwrap_or(Cow::Borrowed(""));
+		for node in &self.contents[1..] {
+			let text = node_text(node);
+			if text.len() > 0 {
+				init.to_mut().push_str(&text);
+			}
+		}
+		init*/
+	}
+}
diff --git a/sharedmodel/src/html_nodes.rs b/sharedmodel/src/html_nodes.rs
deleted file mode 100644
index a585349..0000000
--- a/sharedmodel/src/html_nodes.rs
+++ /dev/null
@@ -1,16 +0,0 @@
-use std::borrow::Borrow;
-
-#[derive(Debug)]
-pub enum HtmlNode<T, U> {
-	Text(T),
-	Element(U)
-}
-
-pub trait HtmlElement: Sized {
-	type Text1: AsRef<str>;
-	type Text2: AsRef<str>;
-	type ContentsIter: DoubleEndedIterator<Item=HtmlNode<Self::Text1, Self>>;
-	fn name(&self) -> &str;
-	fn contents(self) -> Self::ContentsIter;
-	fn text(&self) -> Self::Text2;
-}
diff --git a/sharedmodel/src/lib.rs b/sharedmodel/src/lib.rs
index 7a10431..c0430fd 100644
--- a/sharedmodel/src/lib.rs
+++ b/sharedmodel/src/lib.rs
@@ -1,5 +1,7 @@
 mod live_event_buffer;
-mod html_nodes;
+mod html;
+mod html_html5ever;
 
 pub use live_event_buffer::LiveUpdateBuffer;
-pub use html_nodes::{HtmlNode, HtmlElement};
+pub use html::{HtmlNode, HtmlElement};
+pub use html_html5ever::{parse_html};
diff --git a/sharedmodel/src/live_event_buffer.rs b/sharedmodel/src/live_event_buffer.rs
index dc4e36e..52f7014 100644
--- a/sharedmodel/src/live_event_buffer.rs
+++ b/sharedmodel/src/live_event_buffer.rs
@@ -1,5 +1,5 @@
 use uuid::Uuid;
-use crate::html_nodes::HtmlElement;
+use crate::HtmlElement;
 
 pub struct LiveUpdateMeta {
 	id: Uuid,
diff --git a/sharedmodel/src/main.rs b/sharedmodel/src/main.rs
index 37a8bc4..b8ee3b4 100644
--- a/sharedmodel/src/main.rs
+++ b/sharedmodel/src/main.rs
@@ -1,205 +1,11 @@
-use std::cell::{Cell, RefCell};
-use std::rc::{Rc, Weak};
-use std::borrow::Cow;
-
-use html5ever::tree_builder::{TreeSink, ElementFlags};
-use html5ever::{QualName, ExpandedName, LocalNameStaticSet, Attribute};
-use html5ever::tendril::{StrTendril, TendrilSink};
-use html5ever::interface::{NodeOrText, QuirksMode};
-use html5ever::driver::ParseOpts;
-use string_cache::Atom;
-
-use sharedmodel::{HtmlNode, HtmlElement};
+use sharedmodel::{parse_html, HtmlElement};
 
 fn main() {
-    let mut x = html5ever::parse_fragment(RcHtmlElement(Rc::new(HtmlElementInner::root())), ParseOpts::default(), QualName::new(None, Atom::from("http://www.w3.org/1999/xhtml"), Atom::from("body")), Vec::new());
 	//let html = "<p>This is <em>a <a href=\"#\">paragraph</a></em>. There are</p><ul><li>nested</li><li>tags, <strong>especially in <em>here</em></strong></li></ul>";
 	//let html = "<p>A paragraph</p>";
 	let html = "<section><p>Can you handle this</p><table><tr><th>a</th><th>b</th></tr><tr><td>1</td><td>2</td></tr><tr><td>3</td><td>4</td></tr></table></section>";
-	x.process(StrTendril::from(html));
-	let tr = x.finish();
+	let tr = parse_html(html);
 	println!("{:?}", tr);
 	println!("{}", tr.text())
 }
 
-type RcNode<'a> = HtmlNode<StrTendril, RcHtmlElement<'a>>;
-
-// #[derive(Debug)]
-struct HtmlElementInner<'a> {
-	name: QualName,
-	parent: Cell<Weak<HtmlElementInner<'a>>>,
-	contents: RefCell<Vec<RcNode<'a>>>
-}
-
-impl HtmlElementInner<'_> {
-	fn root() -> Self {
-		let name = QualName::new(None, Atom::from(""), Atom::from(""));
-		let parent = Cell::new(Weak::new());
-		let contents = RefCell::new(Vec::new());
-		HtmlElementInner { name, parent, contents }
-	}
-}
-
-//#[derive(Clone, Debug)]
-#[derive(Clone)]
-struct RcHtmlElement<'a>(Rc<HtmlElementInner<'a>>);
-
-fn nodeortext_to_htmlnode<'a>(v: NodeOrText<RcHtmlElement<'a>>) -> RcNode<'a> {
-	match v {
-		NodeOrText::AppendNode(h) => HtmlNode::Element(h),
-		NodeOrText::AppendText(s) => HtmlNode::Text(s)
-	}
-}
-
-fn node_eq_elem<'a>(node: &RcNode<'a>, elem: &RcHtmlElement<'a>) -> bool {
-	match node {
-		HtmlNode::Element(elem1) if Rc::ptr_eq(&elem.0, &elem1.0) => true,
-		_ => false
-	}
-}
-
-impl<'a> TreeSink for RcHtmlElement<'a> {
-	type Handle = Self;
-	type Output = UniqHtmlElement;
-	fn finish(self) -> UniqHtmlElement {
-		// let HtmlElementInner { name, parent, contents } = Rc::try_unwrap(self.0).map_err(|_| ()).expect("no refs left around");
-		let HtmlElementInner { ref name, ref contents, .. } = *self.0;
-		let name = name.local.clone();
-		let contents = contents.borrow_mut().drain(..).map(|node| {
-			match node {
-				HtmlNode::Text(s) => HtmlNode::Text(s),
-				HtmlNode::Element(e) => HtmlNode::Element(e.finish())
-			}
-		}).collect();
-		UniqHtmlElement { name, contents }
-	}
-
-	fn parse_error(&mut self, msg: Cow<'static, str>) {
-		println!("{}", msg);
-		panic!("how do parse??");
-	}
-
-	fn get_document(&mut self) -> Self::Handle {
-		self.clone()
-	}
-
-	fn elem_name<'b>(&'b self, target: &'b Self::Handle) -> ExpandedName<'b> {
-		target.0.name.expanded()
-	}
-
-	fn create_element(&mut self, name: QualName, attrs: Vec<Attribute>, flags: ElementFlags) -> Self::Handle {
-		assert_eq!(flags.template, false);
-		assert_eq!(flags.mathml_annotation_xml_integration_point, false);
-		let contents = RefCell::new(Vec::new());
-		let parent = Cell::new(Weak::new());
-		// handily, we never use attribute data. Toss it
-		RcHtmlElement(Rc::new(HtmlElementInner { name, parent, contents }))
-	}
-
-	fn create_comment(&mut self, text: StrTendril) -> Self::Handle {
-		panic!("HTML parser encountered a comment");
-	}
-
-	fn create_pi(&mut self, target: StrTendril, data: StrTendril) -> Self::Handle {
-		panic!("HTML parser encountered a processing instruction");
-	}
-
-	fn append(&mut self, parent: &Self::Handle, child: NodeOrText<Self::Handle>) {
-		let node = nodeortext_to_htmlnode(child);
-		if let HtmlNode::Element(ref elem) = node {
-			elem.0.parent.set(Rc::downgrade(&parent.0));
-		}
-		parent.0.contents.borrow_mut().push(node);
-	}
-
-	fn append_based_on_parent_node(&mut self, element: &Self::Handle, prev_element: &Self::Handle, child: NodeOrText<Self::Handle>) {
-		panic!("Not clear what this is for or if it needs to be implemented");
-	}
-
-	fn append_doctype_to_document(&mut self, name: StrTendril, public_id: StrTendril, system_id: StrTendril) {
-		panic!("This can likely be a no-op, just not sure if it gets called for these fragments");
-	}
-
-	fn get_template_contents(&mut self, target: &Self::Handle) -> Self::Handle {
-		panic!("HTML parser encountered a <template> element")
-	}
-
-	fn same_node(&self, x: &Self::Handle, y: &Self::Handle) -> bool {
-		Rc::ptr_eq(&x.0, &y.0)
-	}
-
-	fn set_quirks_mode(&mut self, mode: QuirksMode) {}
-
-	fn append_before_sibling(&mut self, sibling: &Self::Handle, new_node: NodeOrText<Self::Handle>) {
-		let parent_weak = sibling.0.parent.take();
-		let parent = parent_weak.upgrade().expect("parent is not dropped");
-		let mut children = parent.contents.borrow_mut();
-		sibling.0.parent.set(parent_weak);
-		let pos = children.iter().position(|node| node_eq_elem(node, sibling)).expect("sibling is child of parent");
-		children.insert(pos, nodeortext_to_htmlnode(new_node));
-	}
-
-	fn add_attrs_if_missing(&mut self, target: &Self::Handle, attrs: Vec<Attribute>) {}
-
-	fn remove_from_parent(&mut self, target: &Self::Handle) {
-		let parent = target.0.parent.replace(Weak::new()).upgrade().expect("target has parent");
-		let mut children = parent.contents.borrow_mut();
-		let pos = children.iter().position(|node| node_eq_elem(node, target)).expect("sibling is child of parent");
-		children.remove(pos);
-	}
-
-	fn reparent_children(&mut self, node: &Self::Handle, new_parent: &Self::Handle) {
-		new_parent.0.contents.borrow_mut().extend(node.0.contents.borrow_mut().drain(..));
-	}
-}
-
-
-#[derive(Debug)]
-struct UniqHtmlElement {
-	name: Atom<LocalNameStaticSet>,
-	contents: Vec<HtmlNode<StrTendril, UniqHtmlElement>>
-}
-
-/*fn node_text(node: &HtmlNode<StrTendril, UniqHtmlElement>) -> Cow<str> {
-	match node {
-		HtmlNode::Text(s) => Cow::Borrowed(s.as_ref()),
-		HtmlNode::Element(e) => e.text()
-	}
-}*/
-
-fn element_text_inner(buf: &mut Option<StrTendril>, elem: &UniqHtmlElement) {
-	for node in &elem.contents {
-		match node {
-			HtmlNode::Text(s) => match buf {
-				None => { *buf = Some(s.clone()); },
-				Some(tendril) => tendril.push_tendril(&s)
-			},
-			HtmlNode::Element(e) => element_text_inner(buf, &e)
-		}
-	}
-}
-
-impl HtmlElement for UniqHtmlElement {
-	type Text1 = StrTendril;
-	type Text2 = StrTendril;
-	type ContentsIter = <Vec<HtmlNode<Self::Text1, Self>> as IntoIterator>::IntoIter;
-	fn name(&self) -> &str {
-		&*self.name
-	}
-	fn contents(self) -> Self::ContentsIter {
-		self.contents.into_iter()
-	}
-	fn text(&self) -> StrTendril {
-		let mut result = None;
-		element_text_inner(&mut result, self);
-		result.unwrap_or(StrTendril::new())
-/*		let mut init = self.contents.get(0).map(node_text).unwrap_or(Cow::Borrowed(""));
-		for node in &self.contents[1..] {
-			let text = node_text(node);
-			if text.len() > 0 {
-				init.to_mut().push_str(&text);
-			}
-		}
-		init*/
-	}
-}