minidom, tokio-xmpp: switch xml parsing to rxml
This commit is contained in:
parent
3901068717
commit
d4a5a8247b
33 changed files with 465 additions and 811 deletions
|
|
@ -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['>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['>blah<blah>]]></test>";
|
||||
let elem = Element::from_reader(&xml[..]).unwrap();
|
||||
assert_eq!(elem.text(), "'>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);
|
||||
|
|
|
|||
|
|
@ -20,21 +20,12 @@ pub enum Error {
|
|||
/// An error from quick_xml.
|
||||
XmlError(::quick_xml::Error),
|
||||
|
||||
/// An UTF-8 conversion error.
|
||||
Utf8Error(::std::str::Utf8Error),
|
||||
|
||||
/// An I/O error, from std::io.
|
||||
IoError(::std::io::Error),
|
||||
/// Error from rxml parsing
|
||||
ParserError(rxml::Error),
|
||||
|
||||
/// An error which is returned when the end of the document was reached prematurely.
|
||||
EndOfDocument,
|
||||
|
||||
/// An error which is returned when an element is closed when it shouldn't be
|
||||
InvalidElementClosed,
|
||||
|
||||
/// An error which is returned when an elemet's name contains more colons than permitted
|
||||
InvalidElement,
|
||||
|
||||
/// An error which is returned when an element being serialized doesn't contain a prefix
|
||||
/// (be it None or Some(_)).
|
||||
InvalidPrefix,
|
||||
|
|
@ -42,9 +33,6 @@ pub enum Error {
|
|||
/// An error which is returned when an element doesn't contain a namespace
|
||||
MissingNamespace,
|
||||
|
||||
/// An error which is returned when a comment is to be parsed by minidom
|
||||
NoComments,
|
||||
|
||||
/// An error which is returned when a prefixed is defined twice
|
||||
DuplicatePrefix,
|
||||
}
|
||||
|
|
@ -53,14 +41,10 @@ impl StdError for Error {
|
|||
fn cause(&self) -> Option<&dyn StdError> {
|
||||
match self {
|
||||
Error::XmlError(e) => Some(e),
|
||||
Error::Utf8Error(e) => Some(e),
|
||||
Error::IoError(e) => Some(e),
|
||||
Error::ParserError(e) => Some(e),
|
||||
Error::EndOfDocument => None,
|
||||
Error::InvalidElementClosed => None,
|
||||
Error::InvalidElement => None,
|
||||
Error::InvalidPrefix => None,
|
||||
Error::MissingNamespace => None,
|
||||
Error::NoComments => None,
|
||||
Error::DuplicatePrefix => None,
|
||||
}
|
||||
}
|
||||
|
|
@ -70,21 +54,12 @@ impl std::fmt::Display for Error {
|
|||
fn fmt(&self, fmt: &mut std::fmt::Formatter) -> std::fmt::Result {
|
||||
match self {
|
||||
Error::XmlError(e) => write!(fmt, "XML error: {}", e),
|
||||
Error::Utf8Error(e) => write!(fmt, "UTF-8 error: {}", e),
|
||||
Error::IoError(e) => write!(fmt, "IO error: {}", e),
|
||||
Error::ParserError(e) => write!(fmt, "XML parser error: {}", e),
|
||||
Error::EndOfDocument => {
|
||||
write!(fmt, "the end of the document has been reached prematurely")
|
||||
}
|
||||
Error::InvalidElementClosed => {
|
||||
write!(fmt, "the XML is invalid, an element was wrongly closed")
|
||||
}
|
||||
Error::InvalidElement => write!(fmt, "the XML element is invalid"),
|
||||
Error::InvalidPrefix => write!(fmt, "the prefix is invalid"),
|
||||
Error::MissingNamespace => write!(fmt, "the XML element is missing a namespace",),
|
||||
Error::NoComments => write!(
|
||||
fmt,
|
||||
"a comment has been found even though comments are forbidden"
|
||||
),
|
||||
Error::DuplicatePrefix => write!(fmt, "the prefix is already defined"),
|
||||
}
|
||||
}
|
||||
|
|
@ -96,15 +71,9 @@ impl From<::quick_xml::Error> for Error {
|
|||
}
|
||||
}
|
||||
|
||||
impl From<::std::str::Utf8Error> for Error {
|
||||
fn from(err: ::std::str::Utf8Error) -> Error {
|
||||
Error::Utf8Error(err)
|
||||
}
|
||||
}
|
||||
|
||||
impl From<::std::io::Error> for Error {
|
||||
fn from(err: ::std::io::Error) -> Error {
|
||||
Error::IoError(err)
|
||||
impl From<rxml::Error> for Error {
|
||||
fn from(err: rxml::Error) -> Error {
|
||||
Error::ParserError(err)
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -83,6 +83,7 @@ pub mod error;
|
|||
mod namespaces;
|
||||
pub mod node;
|
||||
mod prefixes;
|
||||
pub mod tree_builder;
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests;
|
||||
|
|
|
|||
|
|
@ -10,16 +10,16 @@
|
|||
|
||||
use crate::element::Element;
|
||||
use crate::error::{Error, ParserError, Result};
|
||||
use crate::tree_builder::TreeBuilder;
|
||||
|
||||
use bytes::BytesMut;
|
||||
use quick_xml::Reader as EventReader;
|
||||
use std::cell::RefCell;
|
||||
use rxml::{PushDriver, RawParser};
|
||||
use std::str;
|
||||
|
||||
/// Parser
|
||||
#[derive(Debug)]
|
||||
pub struct Parser {
|
||||
buffer: RefCell<BytesMut>,
|
||||
driver: PushDriver<RawParser>,
|
||||
tree_builder: TreeBuilder,
|
||||
state: ParserState,
|
||||
}
|
||||
|
||||
|
|
@ -90,14 +90,17 @@ impl Parser {
|
|||
/// Creates a new Parser
|
||||
pub fn new() -> Parser {
|
||||
Parser {
|
||||
buffer: RefCell::new(BytesMut::new()),
|
||||
driver: PushDriver::default(),
|
||||
tree_builder: TreeBuilder::new(),
|
||||
state: ParserState::Empty,
|
||||
}
|
||||
}
|
||||
|
||||
/// Feed bytes to the parser.
|
||||
pub fn feed(&mut self, bytes: BytesMut) -> Result<()> {
|
||||
self.buffer.borrow_mut().unsplit(bytes);
|
||||
self.driver.feed(bytes);
|
||||
bytes.clear();
|
||||
|
||||
let state = match self.state {
|
||||
ParserState::Empty => {
|
||||
// TODO: Try splitting xml prolog and stream header
|
||||
|
|
|
|||
|
|
@ -13,9 +13,7 @@
|
|||
use crate::element::Element;
|
||||
use crate::error::Error;
|
||||
|
||||
use quick_xml::Reader;
|
||||
|
||||
const TEST_STRING: &'static str = r#"<root xmlns="root_ns" a="b" xml:lang="en">meow<child c="d"/><child xmlns="child_ns" d="e" xml:lang="fr"/>nya</root>"#;
|
||||
const TEST_STRING: &'static [u8] = br#"<root xmlns="root_ns" a="b" xml:lang="en">meow<child c="d"/><child xmlns="child_ns" d="e" xml:lang="fr"/>nya</root>"#;
|
||||
|
||||
fn build_test_tree() -> Element {
|
||||
let mut root = Element::builder("root", "root_ns")
|
||||
|
|
@ -36,9 +34,8 @@ fn build_test_tree() -> Element {
|
|||
|
||||
#[test]
|
||||
fn reader_works() {
|
||||
let mut reader = Reader::from_str(TEST_STRING);
|
||||
assert_eq!(
|
||||
Element::from_reader(&mut reader).unwrap(),
|
||||
Element::from_reader(TEST_STRING).unwrap(),
|
||||
build_test_tree()
|
||||
);
|
||||
}
|
||||
|
|
@ -143,7 +140,7 @@ fn writer_works() {
|
|||
{
|
||||
root.write_to(&mut writer).unwrap();
|
||||
}
|
||||
assert_eq!(String::from_utf8(writer).unwrap(), TEST_STRING);
|
||||
assert_eq!(writer, TEST_STRING);
|
||||
}
|
||||
|
||||
#[test]
|
||||
|
|
@ -153,7 +150,10 @@ fn writer_with_decl_works() {
|
|||
{
|
||||
root.write_to_decl(&mut writer).unwrap();
|
||||
}
|
||||
let result = format!(r#"<?xml version="1.0" encoding="utf-8"?>{}"#, TEST_STRING);
|
||||
let result = format!(
|
||||
r#"<?xml version="1.0" encoding="utf-8"?>{}"#,
|
||||
String::from_utf8(TEST_STRING.to_owned()).unwrap()
|
||||
);
|
||||
assert_eq!(String::from_utf8(writer).unwrap(), result);
|
||||
}
|
||||
|
||||
|
|
@ -348,8 +348,7 @@ fn two_elements_with_same_arguments_different_order_are_equal() {
|
|||
|
||||
#[test]
|
||||
fn namespace_attributes_works() {
|
||||
let mut reader = Reader::from_str(TEST_STRING);
|
||||
let root = Element::from_reader(&mut reader).unwrap();
|
||||
let root = Element::from_reader(TEST_STRING).unwrap();
|
||||
assert_eq!("en", root.attr("xml:lang").unwrap());
|
||||
assert_eq!(
|
||||
"fr",
|
||||
|
|
@ -424,7 +423,7 @@ fn namespace_inherited_prefixed2() {
|
|||
fn fail_comments() {
|
||||
let elem: Result<Element, Error> = "<foo xmlns='ns1'><!-- bar --></foo>".parse();
|
||||
match elem {
|
||||
Err(Error::NoComments) => (),
|
||||
Err(_) => (),
|
||||
_ => panic!(),
|
||||
};
|
||||
}
|
||||
|
|
@ -432,20 +431,16 @@ fn fail_comments() {
|
|||
#[test]
|
||||
fn xml_error() {
|
||||
match "<a xmlns='ns1'></b>".parse::<Element>() {
|
||||
Err(crate::error::Error::XmlError(_)) => (),
|
||||
Err(crate::error::Error::ParserError(rxml::Error::NotWellFormed(
|
||||
rxml::error::WFError::ElementMismatch,
|
||||
))) => (),
|
||||
err => panic!("No or wrong error: {:?}", err),
|
||||
}
|
||||
|
||||
match "<a xmlns='ns1'></".parse::<Element>() {
|
||||
Err(crate::error::Error::XmlError(_)) => (),
|
||||
err => panic!("No or wrong error: {:?}", err),
|
||||
}
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn invalid_element_error() {
|
||||
match "<a:b:c>".parse::<Element>() {
|
||||
Err(crate::error::Error::InvalidElement) => (),
|
||||
Err(crate::error::Error::ParserError(rxml::Error::NotWellFormed(
|
||||
rxml::error::WFError::InvalidEof(_),
|
||||
))) => (),
|
||||
err => panic!("No or wrong error: {:?}", err),
|
||||
}
|
||||
}
|
||||
|
|
|
|||
154
minidom/src/tree_builder.rs
Normal file
154
minidom/src/tree_builder.rs
Normal file
|
|
@ -0,0 +1,154 @@
|
|||
// Copyright (c) 2022 Astro <astro@spaceboyz.net>
|
||||
|
||||
//! SAX events to DOM tree conversion
|
||||
|
||||
use crate::prefixes::{Prefix, Prefixes};
|
||||
use crate::{Element, Error};
|
||||
use rxml::RawEvent;
|
||||
use std::collections::BTreeMap;
|
||||
|
||||
/// Tree-building parser state
|
||||
pub struct TreeBuilder {
|
||||
next_tag: Option<(Prefix, String, Prefixes, BTreeMap<String, String>)>,
|
||||
/// Parsing stack
|
||||
stack: Vec<Element>,
|
||||
/// Namespace set stack by prefix
|
||||
prefixes_stack: Vec<Prefixes>,
|
||||
/// Document root element if finished
|
||||
pub root: Option<Element>,
|
||||
}
|
||||
|
||||
impl TreeBuilder {
|
||||
/// Create a new one
|
||||
pub fn new() -> Self {
|
||||
TreeBuilder {
|
||||
next_tag: None,
|
||||
stack: vec![],
|
||||
prefixes_stack: vec![],
|
||||
root: None,
|
||||
}
|
||||
}
|
||||
|
||||
/// Stack depth
|
||||
pub fn depth(&self) -> usize {
|
||||
self.stack.len()
|
||||
}
|
||||
|
||||
/// Get the top-most element from the stack but don't remove it
|
||||
pub fn top(&mut self) -> Option<&Element> {
|
||||
self.stack.last()
|
||||
}
|
||||
|
||||
/// Pop the top-most element from the stack
|
||||
fn pop(&mut self) -> Option<Element> {
|
||||
self.prefixes_stack.pop();
|
||||
self.stack.pop()
|
||||
}
|
||||
|
||||
/// Unshift the first child of the top element
|
||||
pub fn unshift_child(&mut self) -> Option<Element> {
|
||||
let depth = self.stack.len();
|
||||
if depth > 0 {
|
||||
self.stack[depth - 1].unshift_child()
|
||||
} else {
|
||||
None
|
||||
}
|
||||
}
|
||||
|
||||
/// Lookup XML namespace declaration for given prefix (or no prefix)
|
||||
fn lookup_prefix(&self, prefix: &Option<String>) -> Option<&str> {
|
||||
for nss in self.prefixes_stack.iter().rev() {
|
||||
if let Some(ns) = nss.get(prefix) {
|
||||
return Some(ns);
|
||||
}
|
||||
}
|
||||
|
||||
None
|
||||
}
|
||||
|
||||
fn process_end_tag(&mut self) -> Result<(), Error> {
|
||||
if let Some(el) = self.pop() {
|
||||
if self.depth() > 0 {
|
||||
let top = self.stack.len() - 1;
|
||||
self.stack[top].append_child(el);
|
||||
} else {
|
||||
self.root = Some(el);
|
||||
}
|
||||
}
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn process_text(&mut self, text: String) {
|
||||
if self.depth() > 0 {
|
||||
let top = self.stack.len() - 1;
|
||||
self.stack[top].append_text_node(text);
|
||||
}
|
||||
}
|
||||
|
||||
/// Process a Event that you got out of a RawParser
|
||||
pub fn process_event(&mut self, event: RawEvent) -> Result<(), Error> {
|
||||
match event {
|
||||
RawEvent::XMLDeclaration(_, _) => {}
|
||||
|
||||
RawEvent::ElementHeadOpen(_, (prefix, name)) => {
|
||||
self.next_tag = Some((
|
||||
prefix.map(|prefix| prefix.as_str().to_owned()),
|
||||
name.as_str().to_owned(),
|
||||
Prefixes::default(),
|
||||
BTreeMap::new(),
|
||||
))
|
||||
}
|
||||
|
||||
RawEvent::Attribute(_, (prefix, name), value) => {
|
||||
self.next_tag
|
||||
.as_mut()
|
||||
.map(
|
||||
|(_, _, ref mut prefixes, ref mut attrs)| match (prefix, name) {
|
||||
(None, xmlns) if xmlns == "xmlns" => {
|
||||
prefixes.insert(None, value);
|
||||
}
|
||||
(Some(xmlns), prefix) if xmlns.as_str() == "xmlns" => {
|
||||
prefixes.insert(Some(prefix.as_str().to_owned()), value);
|
||||
}
|
||||
(Some(prefix), name) => {
|
||||
attrs.insert(
|
||||
format!("{}:{}", prefix, name),
|
||||
value.as_str().to_owned(),
|
||||
);
|
||||
}
|
||||
(None, name) => {
|
||||
attrs.insert(name.as_str().to_owned(), value.as_str().to_owned());
|
||||
}
|
||||
},
|
||||
);
|
||||
}
|
||||
|
||||
RawEvent::ElementHeadClose(_) => {
|
||||
if let Some((prefix, name, prefixes, attrs)) = self.next_tag.take() {
|
||||
self.prefixes_stack.push(prefixes.clone());
|
||||
|
||||
let namespace = self
|
||||
.lookup_prefix(&prefix.clone().map(|prefix| prefix.as_str().to_owned()))
|
||||
.ok_or(Error::MissingNamespace)?
|
||||
.to_owned();
|
||||
let el = Element::new(
|
||||
name.as_str().to_owned(),
|
||||
namespace,
|
||||
Some(prefix.map(|prefix| prefix.as_str().to_owned())),
|
||||
prefixes,
|
||||
attrs,
|
||||
vec![],
|
||||
);
|
||||
self.stack.push(el);
|
||||
}
|
||||
}
|
||||
|
||||
RawEvent::ElementFoot(_) => self.process_end_tag()?,
|
||||
|
||||
RawEvent::Text(_, text) => self.process_text(text.as_str().to_owned()),
|
||||
}
|
||||
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
Loading…
Reference in a new issue