xmppftw/xmpp-rs
Watch
1
0
Fork
You've already forked xmpp-rs
 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

minidom, tokio-xmpp: switch xml parsing to rxml

Browse source
This commit is contained in:
Astro 2022-03-22 23:29:25 +01:00
commit d4a5a8247b
33 changed files with 465 additions and 811 deletions
Download patch file Download diff file Expand all files Collapse all files

261
minidom/src/element.rs
View file

@ -17,18 +17,17 @@ use crate::error::{Error, Result};
use crate::namespaces::NSChoice;
use crate::node::Node;
use crate::prefixes::{Namespace, Prefix, Prefixes};
use crate::tree_builder::TreeBuilder;
use std::collections::{btree_map, BTreeMap};
use std::io::Write;
use std::io::{BufRead, Write};
use std::borrow::Cow;
use std::str;
use quick_xml::events::{BytesDecl, BytesEnd, BytesStart, Event};
use quick_xml::Reader as EventReader;
use quick_xml::Writer as EventWriter;
use std::io::BufRead;
use rxml::{EventRead, Lexer, PullDriver, RawParser};
use std::str::FromStr;
@ -84,8 +83,9 @@ pub struct Element {
namespace: String,
/// This is only used when deserializing. If you have to use a custom prefix use
/// `ElementBuilder::prefix`.
prefix: Option<Prefix>,
prefixes: Prefixes,
pub(crate) prefix: Option<Prefix>,
/// Namespace declarations
pub prefixes: Prefixes,
attributes: BTreeMap<String, String>,
children: Vec<Node>,
}
@ -102,8 +102,7 @@ impl FromStr for Element {
type Err = Error;
fn from_str(s: &str) -> Result<Element> {
let mut reader = EventReader::from_str(s);
Element::from_reader(&mut reader)
Element::from_reader(s.as_bytes())
}
}
@ -120,15 +119,8 @@ impl PartialEq for Element {
}
}
fn ensure_no_prefix<S: AsRef<str>>(s: &S) -> Result<()> {
match s.as_ref().split(':').count() {
1 => Ok(()),
_ => Err(Error::InvalidElement),
}
}
impl Element {
fn new<P: Into<Prefixes>>(
pub(crate) fn new<P: Into<Prefixes>>(
name: String,
namespace: String,
prefix: Option<Prefix>,
@ -136,8 +128,6 @@ impl Element {
attributes: BTreeMap<String, String>,
children: Vec<Node>,
) -> Element {
ensure_no_prefix(&name).unwrap();
// TODO: Return Result<Element> instead.
Element {
name,
namespace,
@ -310,123 +300,18 @@ impl Element {
namespace.into().compare(self.namespace.as_ref())
}
/// Parse a document from an `EventReader`.
pub fn from_reader<R: BufRead>(reader: &mut EventReader<R>) -> Result<Element> {
let mut buf = Vec::new();
/// Parse a document from a `BufRead`.
pub fn from_reader<R: BufRead>(reader: R) -> Result<Element> {
let mut tree_builder = TreeBuilder::new();
let mut driver = PullDriver::wrap(reader, Lexer::new(), RawParser::new());
while let Some(event) = driver.read()? {
tree_builder.process_event(event)?;
let mut prefixes = BTreeMap::new();
let root: Element = loop {
let e = reader.read_event(&mut buf)?;
match e {
Event::Empty(ref e) | Event::Start(ref e) => {
break build_element(reader, e, &mut prefixes)?;
}
Event::Eof => {
return Err(Error::EndOfDocument);
}
Event::Comment { .. } => {
return Err(Error::NoComments);
}
Event::Text { .. }
| Event::End { .. }
| Event::CData { .. }
| Event::Decl { .. }
| Event::PI { .. }
| Event::DocType { .. } => (), // TODO: may need more errors
}
};
let mut stack = vec![root];
let mut prefix_stack = vec![prefixes];
loop {
match reader.read_event(&mut buf)? {
Event::Empty(ref e) => {
let mut prefixes = prefix_stack.last().unwrap().clone();
let elem = build_element(reader, e, &mut prefixes)?;
// Since there is no Event::End after, directly append it to the current node
stack.last_mut().unwrap().append_child(elem);
}
Event::Start(ref e) => {
let mut prefixes = prefix_stack.last().unwrap().clone();
let elem = build_element(reader, e, &mut prefixes)?;
stack.push(elem);
prefix_stack.push(prefixes);
}
Event::End(ref e) => {
if stack.len() <= 1 {
break;
}
let prefixes = prefix_stack.pop().unwrap();
let elem = stack.pop().unwrap();
if let Some(to) = stack.last_mut() {
// TODO: check whether this is correct, we are comparing &[u8]s, not &strs
let elem_name = e.name();
let mut split_iter = elem_name.splitn(2, |u| *u == 0x3A);
let possible_prefix = split_iter.next().unwrap(); // Can't be empty.
let opening_prefix = {
let mut tmp: Option<Option<String>> = None;
for (prefix, ns) in prefixes {
if ns == elem.namespace {
tmp = Some(prefix.clone());
break;
}
}
match tmp {
Some(prefix) => prefix,
None => return Err(Error::InvalidPrefix),
}
};
match split_iter.next() {
// There is a prefix on the closing tag
Some(name) => {
// Does the closing prefix match the opening prefix?
match opening_prefix {
Some(prefix) if possible_prefix == prefix.as_bytes() => (),
_ => return Err(Error::InvalidElementClosed),
}
// Does the closing tag name match the opening tag name?
if name != elem.name().as_bytes() {
return Err(Error::InvalidElementClosed);
}
}
// There was no prefix on the closing tag
None => {
// Is there a prefix on the opening tag?
if opening_prefix.is_some() {
return Err(Error::InvalidElementClosed);
}
// Does the opening tag name match the closing one?
if possible_prefix != elem.name().as_bytes() {
return Err(Error::InvalidElementClosed);
}
}
}
to.append_child(elem);
}
}
Event::Text(s) => {
let text = s.unescape_and_decode(reader)?;
if !text.is_empty() {
let current_elem = stack.last_mut().unwrap();
current_elem.append_text_node(text);
}
}
Event::CData(s) => {
let text = s.unescape_and_decode(&reader)?;
if !text.is_empty() {
let current_elem = stack.last_mut().unwrap();
current_elem.append_text_node(text);
}
}
Event::Eof => {
break;
}
Event::Comment(_) => return Err(Error::NoComments),
Event::Decl { .. } | Event::PI { .. } | Event::DocType { .. } => (),
if let Some(root) = tree_builder.root.take() {
return Ok(root);
}
}
Ok(stack.pop().unwrap())
Err(Error::EndOfDocument)
}
/// Output a document to a `Writer`.
@ -822,68 +707,18 @@ impl Element {
})?;
self.children.remove(idx).into_element()
}
}
fn split_element_name<S: AsRef<str>>(s: S) -> Result<(Option<String>, String)> {
let name_parts = s.as_ref().split(':').collect::<Vec<&str>>();
match name_parts.len() {
2 => Ok((Some(name_parts[0].to_owned()), name_parts[1].to_owned())),
1 => Ok((None, name_parts[0].to_owned())),
_ => Err(Error::InvalidElement),
}
}
fn build_element<R: BufRead>(
reader: &EventReader<R>,
event: &BytesStart,
prefixes: &mut BTreeMap<Prefix, Namespace>,
) -> Result<Element> {
let (prefix, name) = split_element_name(str::from_utf8(event.name())?)?;
let mut local_prefixes = BTreeMap::new();
let attributes = event
.attributes()
.map(|o| {
let o = o?;
let key = str::from_utf8(o.key)?.to_owned();
let value = o.unescape_and_decode_value(reader)?;
Ok((key, value))
})
.filter(|o| match *o {
Ok((ref key, ref value)) if key == "xmlns" => {
local_prefixes.insert(None, value.clone());
prefixes.insert(None, value.clone());
false
/// Remove the leading nodes up to the first child element and
/// return it
pub fn unshift_child(&mut self) -> Option<Element> {
while self.children.len() > 0 {
if let Some(el) = self.children.remove(0).into_element() {
return Some(el);
}
Ok((ref key, ref value)) if key.starts_with("xmlns:") => {
local_prefixes.insert(Some(key[6..].to_owned()), value.to_owned());
prefixes.insert(Some(key[6..].to_owned()), value.to_owned());
false
}
_ => true,
})
.collect::<Result<BTreeMap<String, String>>>()?;
let namespace: &String = {
if let Some(namespace) = local_prefixes.get(&prefix) {
namespace
} else if let Some(namespace) = prefixes.get(&prefix) {
namespace
} else {
return Err(Error::MissingNamespace);
}
};
Ok(Element::new(
name,
namespace.clone(),
// Note that this will always be Some(_) as we can't distinguish between the None case and
// Some(None). At least we make sure the prefix has a namespace associated.
Some(prefix),
local_prefixes,
attributes,
Vec::new(),
))
None
}
}
/// An iterator over references to child elements of an `Element`.
@ -1067,9 +902,8 @@ mod tests {
#[test]
fn test_from_reader_simple() {
let xml = "<foo xmlns='ns1'></foo>";
let mut reader = EventReader::from_str(xml);
let elem = Element::from_reader(&mut reader);
let xml = b"<foo xmlns='ns1'></foo>";
let elem = Element::from_reader(&xml[..]);
let elem2 = Element::builder("foo", "ns1").build();
@ -1078,9 +912,8 @@ mod tests {
#[test]
fn test_from_reader_nested() {
let xml = "<foo xmlns='ns1'><bar xmlns='ns1' baz='qxx' /></foo>";
let mut reader = EventReader::from_str(xml);
let elem = Element::from_reader(&mut reader);
let xml = b"<foo xmlns='ns1'><bar xmlns='ns1' baz='qxx' /></foo>";
let elem = Element::from_reader(&xml[..]);
let nested = Element::builder("bar", "ns1").attr("baz", "qxx").build();
let elem2 = Element::builder("foo", "ns1").append(nested).build();
@ -1090,9 +923,8 @@ mod tests {
#[test]
fn test_from_reader_with_prefix() {
let xml = "<foo xmlns='ns1'><prefix:bar xmlns:prefix='ns1' baz='qxx' /></foo>";
let mut reader = EventReader::from_str(xml);
let elem = Element::from_reader(&mut reader);
let xml = b"<foo xmlns='ns1'><prefix:bar xmlns:prefix='ns1' baz='qxx' /></foo>";
let elem = Element::from_reader(&xml[..]);
let nested = Element::builder("bar", "ns1").attr("baz", "qxx").build();
let elem2 = Element::builder("foo", "ns1").append(nested).build();
@ -1102,9 +934,8 @@ mod tests {
#[test]
fn test_from_reader_split_prefix() {
let xml = "<foo:bar xmlns:foo='ns1'/>";
let mut reader = EventReader::from_str(xml);
let elem = Element::from_reader(&mut reader).unwrap();
let xml = b"<foo:bar xmlns:foo='ns1'/>";
let elem = Element::from_reader(&xml[..]).unwrap();
assert_eq!(elem.name(), String::from("bar"));
assert_eq!(elem.ns(), String::from("ns1"));
@ -1118,38 +949,32 @@ mod tests {
#[test]
fn parses_spectest_xml() {
// From: https://gitlab.com/lumi/minidom-rs/issues/8
let xml = r#"
<rng:grammar xmlns:rng="http://relaxng.org/ns/structure/1.0">
let xml = br#"<rng:grammar xmlns:rng="http://relaxng.org/ns/structure/1.0">
<rng:name xmlns:rng="http://relaxng.org/ns/structure/1.0"></rng:name>
</rng:grammar>
"#;
let mut reader = EventReader::from_str(xml);
let _ = Element::from_reader(&mut reader).unwrap();
let _ = Element::from_reader(&xml[..]).unwrap();
}
#[test]
fn does_not_unescape_cdata() {
let xml = "<test xmlns='test'><![CDATA[&apos;&gt;blah<blah>]]></test>";
let mut reader = EventReader::from_str(xml);
let elem = Element::from_reader(&mut reader).unwrap();
let xml = b"<test xmlns='test'><![CDATA[&apos;&gt;blah<blah>]]></test>";
let elem = Element::from_reader(&xml[..]).unwrap();
assert_eq!(elem.text(), "&apos;&gt;blah<blah>");
}
#[test]
fn test_compare_all_ns() {
let xml = "<foo xmlns='foo' xmlns:bar='baz'><bar:meh xmlns:bar='baz' /></foo>";
let mut reader = EventReader::from_str(xml);
let elem = Element::from_reader(&mut reader).unwrap();
let xml = b"<foo xmlns='foo' xmlns:bar='baz'><bar:meh xmlns:bar='baz' /></foo>";
let elem = Element::from_reader(&xml[..]).unwrap();
let elem2 = elem.clone();
let xml3 = "<foo xmlns='foo'><bar:meh xmlns:bar='baz'/></foo>";
let mut reader3 = EventReader::from_str(xml3);
let elem3 = Element::from_reader(&mut reader3).unwrap();
let xml3 = b"<foo xmlns='foo'><bar:meh xmlns:bar='baz'/></foo>";
let elem3 = Element::from_reader(&xml3[..]).unwrap();
let xml4 = "<prefix:foo xmlns:prefix='foo'><bar:meh xmlns:bar='baz'/></prefix:foo>";
let mut reader4 = EventReader::from_str(xml4);
let elem4 = Element::from_reader(&mut reader4).unwrap();
let xml4 = b"<prefix:foo xmlns:prefix='foo'><bar:meh xmlns:bar='baz'/></prefix:foo>";
let elem4 = Element::from_reader(&xml4[..]).unwrap();
assert_eq!(elem, elem2);
assert_eq!(elem, elem3);