use std::sync::Arc;
use quick_xml::escape::resolve_predefined_entity;
use quick_xml::events::{BytesStart, Event};
use quick_xml::Reader;
use tatara_lisp_eval::ffi::Arity;
use tatara_lisp_eval::{EvalError, Interpreter, MapKey, Value};
use crate::errors::raise_kind;
pub const MALFORMED: &str = "xml_malformed";
pub const REFUSED: &str = "xml_refused";
pub const MAX_DEPTH: usize = 1024;
pub fn install_xml_stdlib<H: 'static>(interp: &mut Interpreter<H>) {
interp.register_fn(
"xml_parse",
Arity::Exact(1),
|args: &[Value], _h: &mut H, span| match &args[0] {
Value::Str(s) => parse(s).map_err(|f| f.raise(span)),
other => Err(EvalError::native_fn(
"xml_parse",
format!("expected a string of XML text, got {}", other.type_name()),
span,
)),
},
);
interp.register_fn(
"xml_render",
Arity::Exact(1),
|args: &[Value], _h: &mut H, span| {
render(&args[0])
.map(|s| Value::Str(Arc::from(s)))
.map_err(|f| f.raise(span))
},
);
}
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Fault {
Malformed(String),
Refused(String),
}
impl Fault {
fn raise(self, span: tatara_lisp::Span) -> EvalError {
match self {
Fault::Malformed(m) => raise_kind(MALFORMED, m, span),
Fault::Refused(m) => raise_kind(REFUSED, m, span),
}
}
}
struct Open {
name: String,
attributes: Vec<(String, String)>,
children: Vec<Value>,
texts: Vec<String>,
}
impl Open {
fn close(self) -> Value {
let text = if self.children.is_empty() {
self.texts.concat()
} else {
self.texts
.into_iter()
.filter(|t| !t.chars().all(char::is_whitespace))
.collect::<String>()
};
element(self.name, self.attributes, self.children, text)
}
}
fn key(name: &str) -> MapKey {
MapKey::Keyword(Arc::from(name))
}
fn element(
name: String,
attributes: Vec<(String, String)>,
children: Vec<Value>,
text: String,
) -> Value {
Value::map([
(key("name"), Value::Str(Arc::from(name))),
(
key("attributes"),
Value::map(
attributes
.into_iter()
.map(|(k, v)| (MapKey::Str(Arc::from(k)), Value::Str(Arc::from(v)))),
),
),
(key("children"), Value::list(children)),
(key("text"), Value::Str(Arc::from(text))),
])
}
fn position(src: &str, byte: u64) -> String {
let end = usize::try_from(byte).unwrap_or(src.len()).min(src.len());
let before = src.get(..end).unwrap_or(src);
let line = before.matches('\n').count() + 1;
let col = before.rsplit('\n').next().map_or(0, |l| l.chars().count()) + 1;
format!("line {line}, column {col}")
}
fn entity_refused(name: &str) -> Fault {
Fault::Refused(format!(
"entity &{name}; is not expanded: blue reads only the five predefined entities and character references, never a DTD entity"
))
}
fn resolve_ref(name: &str) -> Result<String, Fault> {
if let Some(num) = name.strip_prefix('#') {
let code = match num.strip_prefix('x') {
Some(hex) => u32::from_str_radix(hex, 16).ok(),
None => num.parse::<u32>().ok(),
};
return code
.and_then(char::from_u32)
.filter(|c| legal_char(*c))
.map(String::from)
.ok_or_else(|| Fault::Malformed(format!("&{name}; is not a legal character reference")));
}
resolve_predefined_entity(name)
.map(str::to_owned)
.ok_or_else(|| entity_refused(name))
}
fn legal_char(c: char) -> bool {
matches!(c, '\t' | '\n' | '\r') || (c >= ' ' && c != '\u{FFFE}' && c != '\u{FFFF}')
}
fn open(start: &BytesStart<'_>, src: &str, at: u64) -> Result<Open, Fault> {
let name = std::str::from_utf8(start.name().as_ref())
.map_err(|e| Fault::Malformed(e.to_string()))?
.to_owned();
let mut attributes: Vec<(String, String)> = Vec::new();
for attr in start.attributes() {
let attr = attr
.map_err(|e| Fault::Malformed(format!("{e} at {}", position(src, at))))?;
let k = std::str::from_utf8(attr.key.as_ref())
.map_err(|e| Fault::Malformed(e.to_string()))?
.to_owned();
let mut unknown: Option<String> = None;
let v = attr
.unescape_value_with(|ent| {
let hit = resolve_predefined_entity(ent);
if hit.is_none() {
unknown = Some(ent.to_owned());
}
hit
})
.map_err(|e| match unknown.take() {
Some(ent) => entity_refused(&ent),
None => Fault::Malformed(format!("{e} at {}", position(src, at))),
})?
.into_owned();
attributes.push((k, v));
}
Ok(Open {
name,
attributes,
children: Vec::new(),
texts: Vec::new(),
})
}
fn doctype_declares_entity(body: &str) -> bool {
body.contains("<!ENTITY")
}
pub fn parse(src: &str) -> Result<Value, Fault> {
let mut reader = Reader::from_str(src);
reader.config_mut().check_end_names = true;
reader.config_mut().expand_empty_elements = false;
reader.config_mut().check_comments = true;
let mut stack: Vec<Open> = Vec::new();
let mut root: Option<Value> = None;
loop {
let at = reader.buffer_position();
let event = reader.read_event().map_err(|e| {
Fault::Malformed(format!("{e} at {}", position(src, reader.error_position())))
})?;
match event {
Event::Start(start) => {
if root.is_some() && stack.is_empty() {
return Err(Fault::Malformed(format!(
"a second root element at {}",
position(src, at)
)));
}
if stack.len() >= MAX_DEPTH {
return Err(Fault::Refused(format!(
"elements nest deeper than {MAX_DEPTH} at {}",
position(src, at)
)));
}
stack.push(open(&start, src, at)?);
}
Event::Empty(start) => {
if root.is_some() && stack.is_empty() {
return Err(Fault::Malformed(format!(
"a second root element at {}",
position(src, at)
)));
}
let done = open(&start, src, at)?.close();
match stack.last_mut() {
Some(parent) => parent.children.push(done),
None => root = Some(done),
}
}
Event::End(_) => {
let done = stack
.pop()
.ok_or_else(|| Fault::Malformed(format!("an end tag with no start at {}", position(src, at))))?
.close();
match stack.last_mut() {
Some(parent) => parent.children.push(done),
None => root = Some(done),
}
}
Event::Text(t) => {
let text = t
.decode()
.map_err(|e| Fault::Malformed(e.to_string()))?
.into_owned();
match stack.last_mut() {
Some(el) => el.texts.push(text),
None if text.chars().all(char::is_whitespace) => {}
None => {
return Err(Fault::Malformed(format!(
"text outside the root element at {}",
position(src, at)
)))
}
}
}
Event::CData(c) => {
let text = c
.decode()
.map_err(|e| Fault::Malformed(e.to_string()))?
.into_owned();
match stack.last_mut() {
Some(el) => el.texts.push(text),
None => {
return Err(Fault::Malformed(format!(
"CDATA outside the root element at {}",
position(src, at)
)))
}
}
}
Event::GeneralRef(r) => {
let name = r.decode().map_err(|e| Fault::Malformed(e.to_string()))?;
let text = resolve_ref(&name)?;
match stack.last_mut() {
Some(el) => el.texts.push(text),
None => {
return Err(Fault::Malformed(format!(
"a reference outside the root element at {}",
position(src, at)
)))
}
}
}
Event::DocType(d) => {
let body = d.decode().map_err(|e| Fault::Malformed(e.to_string()))?;
if root.is_some() || !stack.is_empty() {
return Err(Fault::Malformed(format!(
"a DOCTYPE after the root element began at {}",
position(src, at)
)));
}
if doctype_declares_entity(&body) {
return Err(Fault::Refused(
"the DOCTYPE declares an entity: blue never expands or fetches DTD entities"
.to_owned(),
));
}
}
Event::Comment(_) | Event::Decl(_) | Event::PI(_) => {}
Event::Eof => break,
}
}
if let Some(unclosed) = stack.last() {
return Err(Fault::Malformed(format!(
"<{}> is never closed",
unclosed.name
)));
}
root.ok_or_else(|| Fault::Malformed("the document has no root element".to_owned()))
}
fn field<'a>(el: &'a Value, name: &str) -> Option<&'a Value> {
match el {
Value::Map(m) => m
.iter()
.find(|(k, _)| matches!(k, MapKey::Keyword(s) | MapKey::Str(s) if &**s == name))
.map(|(_, v)| v),
_ => None,
}
}
fn scalar(v: &Value) -> Option<String> {
match v {
Value::Str(s) | Value::Symbol(s) | Value::Keyword(s) => Some(s.to_string()),
Value::Int(n) => Some(n.to_string()),
Value::Float(n) => Some(n.to_string()),
Value::Bool(b) => Some(b.to_string()),
_ => None,
}
}
fn map_key_text(k: &MapKey) -> Option<String> {
match k {
MapKey::Str(s) | MapKey::Symbol(s) | MapKey::Keyword(s) => Some(s.to_string()),
_ => None,
}
}
fn name_char(c: char, first: bool) -> bool {
c == '_'
|| c == ':'
|| c.is_alphabetic()
|| (!first && (c == '-' || c == '.' || c.is_numeric()))
|| (!c.is_ascii() && !c.is_whitespace() && !c.is_control())
}
fn checked_name(name: &str, what: &str) -> Result<(), Fault> {
let mut chars = name.chars();
let ok = chars.next().is_some_and(|c| name_char(c, true)) && chars.all(|c| name_char(c, false));
if ok {
Ok(())
} else {
Err(Fault::Refused(format!("{what} {name:?} is not an XML name")))
}
}
fn escape_into(out: &mut String, text: &str, in_attribute: bool) -> Result<(), Fault> {
for c in text.chars() {
match c {
'&' => out.push_str("&"),
'<' => out.push_str("<"),
'>' => out.push_str(">"),
'"' if in_attribute => out.push_str("""),
'\t' if in_attribute => out.push_str("	"),
'\n' if in_attribute => out.push_str(" "),
'\r' => out.push_str(" "),
c if legal_char(c) => out.push(c),
c => {
return Err(Fault::Refused(format!(
"U+{:04X} cannot appear in an XML 1.0 document",
u32::from(c)
)))
}
}
}
Ok(())
}
fn render_into(out: &mut String, el: &Value, depth: usize) -> Result<(), Fault> {
if depth > MAX_DEPTH {
return Err(Fault::Refused(format!(
"elements nest deeper than {MAX_DEPTH}"
)));
}
let name = field(el, "name")
.and_then(scalar)
.ok_or_else(|| Fault::Refused(format!("an element is a map with a name; got {}", el.type_name())))?;
checked_name(&name, "element name")?;
out.push('<');
out.push_str(&name);
match field(el, "attributes") {
None | Some(Value::Nil) => {}
Some(Value::Map(attrs)) => {
let mut pairs: Vec<(String, &Value)> = Vec::new();
for (k, v) in attrs.iter() {
let k = map_key_text(k)
.ok_or_else(|| Fault::Refused("an attribute name is text".to_owned()))?;
pairs.push((k, v));
}
pairs.sort_by(|a, b| a.0.cmp(&b.0));
for (k, v) in pairs {
checked_name(&k, "attribute name")?;
let v = scalar(v).ok_or_else(|| {
Fault::Refused(format!("attribute {k} is not text: {}", v.type_name()))
})?;
out.push(' ');
out.push_str(&k);
out.push_str("=\"");
escape_into(out, &v, true)?;
out.push('"');
}
}
Some(other) => {
return Err(Fault::Refused(format!(
"attributes is a map; got {}",
other.type_name()
)))
}
}
let text = match field(el, "text") {
None | Some(Value::Nil) => String::new(),
Some(v) => scalar(v)
.ok_or_else(|| Fault::Refused(format!("text is text; got {}", v.type_name())))?,
};
let children: Vec<Value> = match field(el, "children") {
None | Some(Value::Nil) => Vec::new(),
Some(Value::List(xs)) => xs.iter().cloned().collect(),
Some(other) => {
return Err(Fault::Refused(format!(
"children is a list; got {}",
other.type_name()
)))
}
};
if text.is_empty() && children.is_empty() {
out.push_str("/>");
return Ok(());
}
out.push('>');
escape_into(out, &text, false)?;
for child in &children {
render_into(out, child, depth + 1)?;
}
out.push_str("</");
out.push_str(&name);
out.push('>');
Ok(())
}
pub fn render(el: &Value) -> Result<String, Fault> {
let mut out = String::new();
render_into(&mut out, el, 0)?;
Ok(out)
}
#[cfg(test)]
mod tests {
use super::*;
fn get<'a>(el: &'a Value, k: &str) -> &'a Value {
field(el, k).unwrap_or_else(|| panic!("no {k} in {el:?}"))
}
fn text(el: &Value) -> String {
scalar(get(el, "text")).unwrap_or_default()
}
fn kids(el: &Value) -> Vec<Value> {
match get(el, "children") {
Value::List(xs) => xs.iter().cloned().collect(),
other => panic!("children {other:?}"),
}
}
#[test]
fn an_empty_document_is_malformed_not_nil() {
assert!(matches!(parse(""), Err(Fault::Malformed(_))));
assert!(matches!(parse(" \n"), Err(Fault::Malformed(_))));
}
#[test]
fn a_parsed_document_renders_and_parses_back_to_itself() {
let src = "<?xml version=\"1.0\"?><!DOCTYPE r><r a=\"1 & <2> "q"\" ns:b=\"x y\"><c>t & u</c><c/><![CDATA[<raw>&]]></r>";
let once = render(&parse(src).expect("parses")).expect("renders");
let again = render(&parse(&once).expect("reparses")).expect("renders again");
assert_eq!(once, again);
assert_eq!(
once,
"<r a=\"1 & <2> "q"\" ns:b=\"x y\"><raw>&<c>t & u</c><c/></r>"
);
}
#[test]
fn values_are_unescaped_and_prefixes_kept() {
let el = parse("<x:r xmlns:x=\"urn:example\" a=\"<AB\">1 > 0</x:r>").expect("parses");
assert_eq!(scalar(get(&el, "name")).as_deref(), Some("x:r"));
let Value::Map(attrs) = get(&el, "attributes") else { panic!() };
assert_eq!(attrs.get(&MapKey::Str(Arc::from("a"))).and_then(scalar).as_deref(), Some("<AB"));
assert_eq!(attrs.get(&MapKey::Str(Arc::from("xmlns:x"))).and_then(scalar).as_deref(), Some("urn:example"));
assert_eq!(text(&el), "1 > 0");
}
#[test]
fn whitespace_between_elements_is_not_text() {
let el = parse("<a>\n <b> x </b>\n <c/>\n</a>").expect("parses");
assert_eq!(text(&el), "");
assert_eq!(kids(&el).len(), 2);
assert_eq!(text(&kids(&el)[0]), " x ");
}
#[test]
fn malformed_documents_are_refused_by_kind() {
for bad in [
"<a>",
"<a></b>",
"<a><b></a></b>",
"</a>",
"<a/><b/>",
"text<a/>",
"<a x=1/>",
"<a x=\"1\" x=\"2\"/>",
"<a>�</a>",
] {
assert!(matches!(parse(bad), Err(Fault::Malformed(_))), "{bad}: {:?}", parse(bad));
}
}
#[test]
fn external_and_internal_entities_are_never_expanded() {
let xxe = "<?xml version=\"1.0\"?><!DOCTYPE r [<!ENTITY x SYSTEM \"file:///etc/passwd\">]><r>&x;</r>";
assert!(matches!(parse(xxe), Err(Fault::Refused(_))));
let laughs = "<!DOCTYPE r [<!ENTITY a \"ha\"><!ENTITY b \"&a;&a;&a;&a;&a;&a;&a;&a;&a;&a;\">]><r>&b;</r>";
assert!(matches!(parse(laughs), Err(Fault::Refused(_))));
assert!(matches!(parse("<r>&x;</r>"), Err(Fault::Refused(_))));
assert!(matches!(parse("<r a=\"&x;\"/>"), Err(Fault::Refused(_))));
let unreferenced = "<!DOCTYPE r [<!ENTITY x SYSTEM \"file:///etc/passwd\">]><r/>";
assert!(matches!(parse(unreferenced), Err(Fault::Refused(_))));
let declared_only = "<!DOCTYPE issues [<!ELEMENT issues (issue*)><!ATTLIST issues v CDATA \"\">]><issues/>";
assert!(parse(declared_only).is_ok());
}
#[test]
fn nesting_beyond_the_bound_is_refused() {
let deep = format!("{}{}", "<a>".repeat(MAX_DEPTH + 1), "</a>".repeat(MAX_DEPTH + 1));
assert!(matches!(parse(&deep), Err(Fault::Refused(_))));
let ok = format!("{}{}", "<a>".repeat(MAX_DEPTH), "</a>".repeat(MAX_DEPTH));
assert!(parse(&ok).is_ok());
}
#[test]
fn render_escapes_and_refuses_what_xml_cannot_hold() {
let el = element(
"r".into(),
vec![("q".into(), "a\"<&>\t".into())],
vec![],
"x < y & z".into(),
);
assert_eq!(
render(&el).expect("renders"),
"<r q=\"a"<&>	\">x < y & z</r>"
);
let plain = element("r".into(), vec![], vec![], "a\"b\n\tc".into());
assert_eq!(render(&plain).expect("renders"), "<r>a\"b\n\tc</r>");
let bad = element("r x".into(), vec![], vec![], String::new());
assert!(matches!(render(&bad), Err(Fault::Refused(_))));
let ctl = element("r".into(), vec![], vec![], "\u{1}".into());
assert!(matches!(render(&ctl), Err(Fault::Refused(_))));
assert!(matches!(render(&Value::Int(1)), Err(Fault::Refused(_))));
}
#[test]
fn the_primitives_raise_typed_kinds_in_blue() {
let err = crate::run(r#"xml_parse("<a>")"#).expect_err("must raise");
assert!(err.to_string().contains(MALFORMED), "{err}");
let err = crate::run(r#"xml_parse("<r>&x;</r>")"#).expect_err("must raise");
assert!(err.to_string().contains(REFUSED), "{err}");
let v = crate::run(r#"xml_render(xml_parse("<a b=\"1\"><c/></a>"))"#)
.expect("runs")
.value;
assert_eq!(scalar(&v).as_deref(), Some("<a b=\"1\"><c/></a>"));
}
}