///|
pub fn html_to_sexp(source : String) -> Result[String, String] {
let document = @html5.parse(source)
render_html_fragment(document)
}
///|
fn render_html_fragment(document : @html5.Document) -> Result[String, String] {
render_html_fragment_terms(document, document.get_children(0)).map(terms => {
terms.join("\n")
})
}
///|
fn render_html_fragment_terms(
document : @html5.Document,
node_ids : Array[Int],
) -> Result[Array[String], String] {
let terms : Array[String] = []
append_html_fragment_terms(document, terms, node_ids).map(_ => terms)
}
///|
fn render_html_fragment_node_terms(
document : @html5.Document,
node_id : Int,
) -> Result[Array[String], String] {
if node_id == document.document_element {
return render_html_document_element_terms(document, node_id)
}
render_html_node(document, node_id).map(optional_term_to_terms)
}
///|
fn render_html_document_element_terms(
document : @html5.Document,
node_id : Int,
) -> Result[Array[String], String] {
match document.get_node(node_id) {
Some(@html5.ElementNode(element)) => {
let terms : Array[String] = []
for child_id in element.children {
match
append_html_fragment_terms(
document,
terms,
html_document_fragment_child_ids(document, child_id),
) {
Ok(_) => ()
Err(error) => return Err(error)
}
}
Ok(terms)
}
_ => Err("internal error: document root is not an html element")
}
}
///|
fn append_html_fragment_terms(
document : @html5.Document,
terms : Array[String],
node_ids : Array[Int],
) -> Result[Unit, String] {
for node_id in node_ids {
match render_html_fragment_node_terms(document, node_id) {
Ok(node_terms) => append_terms(terms, node_terms)
Err(error) => return Err(error)
}
}
Ok(())
}
///|
fn html_document_fragment_child_ids(
document : @html5.Document,
child_id : Int,
) -> Array[Int] {
if child_id == document.head_element || child_id == document.body_element {
document.get_children(child_id)
} else {
[child_id]
}
}
///|
fn render_html_node(
document : @html5.Document,
node_id : Int,
) -> Result[String?, String] {
match document.get_node(node_id) {
Some(@html5.TextNode(text)) => render_html_text_term(text)
Some(@html5.CommentNode(text)) => Ok(Some(render_html_comment_term(text)))
Some(@html5.ElementNode(element)) =>
render_html_element_term(document, node_id, element)
Some(@html5.DocumentTypeNode(..)) => Ok(None)
Some(@html5.DocumentNode(children~))
| Some(@html5.DocumentFragmentNode(children~)) =>
render_html_node_list(document, children)
None => Err("internal error: missing html node #" + node_id.to_string())
}
}
///|
fn render_html_node_list(
document : @html5.Document,
node_ids : Array[Int],
) -> Result[String?, String] {
collect_rendered_html_terms(document, node_ids).map(join_non_empty_terms)
}
///|
fn render_html_text_term(text : String) -> Result[String?, String] {
if !text.is_empty() && !starts_with_whitespace(text) {
return Ok(Some(escape_sexp_text(text)))
}
Ok(Some(render_raw_text_form(text)))
}
///|
fn render_html_comment_term(text : String) -> String {
let normalized = if text.is_blank() { "" } else { text.trim().to_owned() }
if normalized.is_empty() {
"(% )"
} else {
"(% " + escape_sexp_text(normalized) + ")"
}
}
///|
fn render_html_element_term(
document : @html5.Document,
node_id : Int,
element : @html5.Element,
) -> Result[String?, String] {
let builder = StringBuilder::new()
builder.write_char('(')
builder.write_string(element.tag_name)
match append_rendered_html_attributes(builder, element.attributes) {
Ok(_) => ()
Err(error) => return Err(error)
}
match
append_rendered_child_terms(
builder,
document,
html_element_child_ids(document, node_id, element),
) {
Ok(_) => ()
Err(error) => return Err(error)
}
builder.write_char(')')
Ok(Some(builder.to_string()))
}
///|
fn append_rendered_html_attributes(
builder : StringBuilder,
attrs : Array[(String, String)],
) -> Result[Unit, String] {
for attr in attrs {
let (name, value) = attr
builder.write_char(' ')
match render_html_attribute(name, value) {
Ok(attr_term) => builder.write_string(attr_term)
Err(error) => return Err(error)
}
}
Ok(())
}
///|
fn append_rendered_child_terms(
builder : StringBuilder,
document : @html5.Document,
node_ids : Array[Int],
) -> Result[Unit, String] {
collect_rendered_html_terms(document, node_ids).map(child_terms => {
for term in child_terms {
builder.write_char(' ')
builder.write_string(term)
}
})
}
///|
fn html_element_child_ids(
document : @html5.Document,
node_id : Int,
element : @html5.Element,
) -> Array[Int] {
match document.get_template_content(node_id) {
id if id != @html5.NO_NODE => document.get_children(id)
_ => element.children
}
}
///|
fn render_html_attribute(
name : String,
value : String,
) -> Result[String, String] {
if value.is_empty() {
return Ok("(:" + name + ")")
}
if starts_with_whitespace(value) {
return Err(
"unsupported html attribute: leading whitespace value on " + name,
)
}
Ok("(:" + name + " " + escape_sexp_text(value) + ")")
}
///|
fn collect_rendered_html_terms(
document : @html5.Document,
node_ids : Array[Int],
) -> Result[Array[String], String] {
let terms : Array[String] = []
for node_id in node_ids {
match render_html_node(document, node_id) {
Ok(Some(term)) => terms.push(term)
Ok(None) => ()
Err(error) => return Err(error)
}
}
Ok(terms)
}
///|
fn append_terms(target : Array[String], terms : Array[String]) -> Unit {
target.append(terms)
}
///|
fn starts_with_whitespace(text : String) -> Bool {
text.get_char(0).map_or(false, ch => ch.is_whitespace())
}
///|
fn render_raw_text_form(text : String) -> String {
"(# " + escape_sexp_text(text) + ")"
}
///|
fn optional_term_to_terms(term : String?) -> Array[String] {
match term {
Some(term) => [term]
None => []
}
}
///|
fn join_non_empty_terms(terms : Array[String]) -> String? {
if terms.is_empty() {
None
} else {
Some(terms.join("\n"))
}
}
///|
fn escape_sexp_text(text : String) -> String {
let chars = text.to_array()
let builder = StringBuilder::new()
let mut index = 0
while index < chars.length() {
let ch = chars[index]
if ch == '\\' {
if index + 1 < chars.length() {
let next = chars[index + 1]
if next == '(' || next == ')' {
builder.write_char('\\')
}
}
builder.write_char(ch)
} else if ch == '(' || ch == ')' {
builder.write_char('\\')
builder.write_char(ch)
} else {
builder.write_char(ch)
}
index = index + 1
}
builder.to_string()
}