xmpp_codec.rs

  1//! XML stream parser for XMPP
  2
  3use crate::{ParseError, ParserError};
  4use bytes::{BufMut, BytesMut};
  5use xmpp_parsers::Element;
  6use quick_xml::Writer as EventWriter;
  7use std;
  8use std::cell::RefCell;
  9use std::collections::vec_deque::VecDeque;
 10use std::collections::HashMap;
 11use std::default::Default;
 12use std::fmt::Write;
 13use std::io;
 14use std::iter::FromIterator;
 15use std::rc::Rc;
 16use std::str::from_utf8;
 17use std::borrow::Cow;
 18use tokio_codec::{Decoder, Encoder};
 19use xml5ever::interface::Attribute;
 20use xml5ever::tokenizer::{Tag, TagKind, Token, TokenSink, XmlTokenizer};
 21
 22/// Anything that can be sent or received on an XMPP/XML stream
 23#[derive(Debug, Clone, PartialEq, Eq)]
 24pub enum Packet {
 25    /// `<stream:stream>` start tag
 26    StreamStart(HashMap<String, String>),
 27    /// A complete stanza or nonza
 28    Stanza(Element),
 29    /// Plain text (think whitespace keep-alive)
 30    Text(String),
 31    /// `</stream:stream>` closing tag
 32    StreamEnd,
 33}
 34
 35type QueueItem = Result<Packet, ParserError>;
 36
 37/// Parser state
 38struct ParserSink {
 39    // Ready stanzas, shared with XMPPCodec
 40    queue: Rc<RefCell<VecDeque<QueueItem>>>,
 41    // Parsing stack
 42    stack: Vec<Element>,
 43    ns_stack: Vec<HashMap<Option<String>, String>>,
 44}
 45
 46impl ParserSink {
 47    pub fn new(queue: Rc<RefCell<VecDeque<QueueItem>>>) -> Self {
 48        ParserSink {
 49            queue,
 50            stack: vec![],
 51            ns_stack: vec![],
 52        }
 53    }
 54
 55    fn push_queue(&self, pkt: Packet) {
 56        self.queue.borrow_mut().push_back(Ok(pkt));
 57    }
 58
 59    fn push_queue_error(&self, e: ParserError) {
 60        self.queue.borrow_mut().push_back(Err(e));
 61    }
 62
 63    /// Lookup XML namespace declaration for given prefix (or no prefix)
 64    fn lookup_ns(&self, prefix: &Option<String>) -> Option<&str> {
 65        for nss in self.ns_stack.iter().rev() {
 66            if let Some(ns) = nss.get(prefix) {
 67                return Some(ns);
 68            }
 69        }
 70
 71        None
 72    }
 73
 74    fn handle_start_tag(&mut self, tag: Tag) {
 75        let mut nss = HashMap::new();
 76        let is_prefix_xmlns = |attr: &Attribute| {
 77            attr.name
 78                .prefix
 79                .as_ref()
 80                .map(|prefix| prefix.eq_str_ignore_ascii_case("xmlns"))
 81                .unwrap_or(false)
 82        };
 83        for attr in &tag.attrs {
 84            match attr.name.local.as_ref() {
 85                "xmlns" => {
 86                    nss.insert(None, attr.value.as_ref().to_owned());
 87                }
 88                prefix if is_prefix_xmlns(attr) => {
 89                    nss.insert(Some(prefix.to_owned()), attr.value.as_ref().to_owned());
 90                }
 91                _ => (),
 92            }
 93        }
 94        self.ns_stack.push(nss);
 95
 96        let el = {
 97            let mut el_builder = Element::builder(tag.name.local.as_ref());
 98            if let Some(el_ns) =
 99                self.lookup_ns(&tag.name.prefix.map(|prefix| prefix.as_ref().to_owned()))
100            {
101                el_builder = el_builder.ns(el_ns);
102            }
103            for attr in &tag.attrs {
104                match attr.name.local.as_ref() {
105                    "xmlns" => (),
106                    _ if is_prefix_xmlns(attr) => (),
107                    _ => {
108                        let attr_name = if let Some(ref prefix) = attr.name.prefix {
109                            Cow::Owned(format!("{}:{}", prefix, attr.name.local))
110                        } else {
111                            Cow::Borrowed(attr.name.local.as_ref())
112                        };
113                        el_builder = el_builder.attr(attr_name, attr.value.as_ref());
114                    }
115                }
116            }
117            el_builder.build()
118        };
119
120        if self.stack.is_empty() {
121            let attrs = HashMap::from_iter(tag.attrs.iter().map(|attr| {
122                (
123                    attr.name.local.as_ref().to_owned(),
124                    attr.value.as_ref().to_owned(),
125                )
126            }));
127            self.push_queue(Packet::StreamStart(attrs));
128        }
129
130        self.stack.push(el);
131    }
132
133    fn handle_end_tag(&mut self) {
134        let el = self.stack.pop().unwrap();
135        self.ns_stack.pop();
136
137        match self.stack.len() {
138            // </stream:stream>
139            0 => self.push_queue(Packet::StreamEnd),
140            // </stanza>
141            1 => self.push_queue(Packet::Stanza(el)),
142            len => {
143                let parent = &mut self.stack[len - 1];
144                parent.append_child(el);
145            }
146        }
147    }
148}
149
150impl TokenSink for ParserSink {
151    fn process_token(&mut self, token: Token) {
152        match token {
153            Token::TagToken(tag) => match tag.kind {
154                TagKind::StartTag => self.handle_start_tag(tag),
155                TagKind::EndTag => self.handle_end_tag(),
156                TagKind::EmptyTag => {
157                    self.handle_start_tag(tag);
158                    self.handle_end_tag();
159                }
160                TagKind::ShortTag => self.push_queue_error(ParserError::ShortTag),
161            },
162            Token::CharacterTokens(tendril) => match self.stack.len() {
163                0 | 1 => self.push_queue(Packet::Text(tendril.into())),
164                len => {
165                    let el = &mut self.stack[len - 1];
166                    el.append_text_node(tendril);
167                }
168            },
169            Token::EOFToken => self.push_queue(Packet::StreamEnd),
170            Token::ParseError(s) => {
171                // println!("ParseError: {:?}", s);
172                self.push_queue_error(ParserError::Parse(ParseError(s)));
173            }
174            _ => (),
175        }
176    }
177
178    // fn end(&mut self) {
179    // }
180}
181
182/// Stateful encoder/decoder for a bytestream from/to XMPP `Packet`
183pub struct XMPPCodec {
184    /// Outgoing
185    ns: Option<String>,
186    /// Incoming
187    parser: XmlTokenizer<ParserSink>,
188    /// For handling incoming truncated utf8
189    // TODO: optimize using  tendrils?
190    buf: Vec<u8>,
191    /// Shared with ParserSink
192    queue: Rc<RefCell<VecDeque<QueueItem>>>,
193}
194
195impl XMPPCodec {
196    /// Constructor
197    pub fn new() -> Self {
198        let queue = Rc::new(RefCell::new(VecDeque::new()));
199        let sink = ParserSink::new(queue.clone());
200        // TODO: configure parser?
201        let parser = XmlTokenizer::new(sink, Default::default());
202        XMPPCodec {
203            ns: None,
204            parser,
205            queue,
206            buf: vec![],
207        }
208    }
209}
210
211impl Default for XMPPCodec {
212    fn default() -> Self {
213        Self::new()
214    }
215}
216
217impl Decoder for XMPPCodec {
218    type Item = Packet;
219    type Error = ParserError;
220
221    fn decode(&mut self, buf: &mut BytesMut) -> Result<Option<Self::Item>, Self::Error> {
222        let buf1: Box<AsRef<[u8]>> = if !self.buf.is_empty() && !buf.is_empty() {
223            let mut prefix = std::mem::replace(&mut self.buf, vec![]);
224            prefix.extend_from_slice(buf.take().as_ref());
225            Box::new(prefix)
226        } else {
227            Box::new(buf.take())
228        };
229        let buf1 = buf1.as_ref().as_ref();
230        match from_utf8(buf1) {
231            Ok(s) => {
232                if !s.is_empty() {
233                    // println!("<< {}", s);
234                    let tendril = FromIterator::from_iter(s.chars());
235                    self.parser.feed(tendril);
236                }
237            }
238            // Remedies for truncated utf8
239            Err(e) if e.valid_up_to() >= buf1.len() - 3 => {
240                // Prepare all the valid data
241                let mut b = BytesMut::with_capacity(e.valid_up_to());
242                b.put(&buf1[0..e.valid_up_to()]);
243
244                // Retry
245                let result = self.decode(&mut b);
246
247                // Keep the tail back in
248                self.buf.extend_from_slice(&buf1[e.valid_up_to()..]);
249
250                return result;
251            }
252            Err(e) => {
253                // println!("error {} at {}/{} in {:?}", e, e.valid_up_to(), buf1.len(), buf1);
254                return Err(ParserError::Utf8(e));
255            }
256        }
257
258        match self.queue.borrow_mut().pop_front() {
259            None => Ok(None),
260            Some(result) => result.map(|pkt| Some(pkt)),
261        }
262    }
263
264    fn decode_eof(&mut self, buf: &mut BytesMut) -> Result<Option<Self::Item>, Self::Error> {
265        self.decode(buf)
266    }
267}
268
269impl Encoder for XMPPCodec {
270    type Item = Packet;
271    type Error = io::Error;
272
273    fn encode(&mut self, item: Self::Item, dst: &mut BytesMut) -> Result<(), Self::Error> {
274        let remaining = dst.capacity() - dst.len();
275        let max_stanza_size: usize = 2usize.pow(16);
276        if remaining < max_stanza_size {
277            dst.reserve(max_stanza_size - remaining);
278        }
279
280        fn to_io_err<E: Into<Box<dyn std::error::Error + Send + Sync>>>(e: E) -> io::Error {
281            io::Error::new(io::ErrorKind::InvalidInput, e)
282        }
283
284        match item {
285            Packet::StreamStart(start_attrs) => {
286                let mut buf = String::new();
287                write!(buf, "<stream:stream")
288                    .map_err(to_io_err)?;
289                for (name, value) in start_attrs {
290                    write!(buf, " {}=\"{}\"", escape(&name), escape(&value))
291                        .map_err(to_io_err)?;
292                    if name == "xmlns" {
293                        self.ns = Some(value);
294                    }
295                }
296                write!(buf, ">\n")
297                    .map_err(to_io_err)?;
298
299                // print!(">> {}", buf);
300                write!(dst, "{}", buf)
301                    .map_err(to_io_err)
302            }
303            Packet::Stanza(stanza) => {
304                stanza
305                    .write_to_inner(&mut EventWriter::new(WriteBytes::new(dst)))
306                    .and_then(|_| {
307                        // println!(">> {:?}", dst);
308                        Ok(())
309                    })
310                    .map_err(|e| to_io_err(format!("{}", e)))
311            }
312            Packet::Text(text) => {
313                write_text(&text, dst)
314                    .and_then(|_| {
315                        // println!(">> {:?}", dst);
316                        Ok(())
317                    })
318                    .map_err(to_io_err)
319            }
320            Packet::StreamEnd => {
321                write!(dst, "</stream:stream>\n")
322                    .map_err(to_io_err)
323            }
324        }
325    }
326}
327
328/// Write XML-escaped text string
329pub fn write_text<W: Write>(text: &str, writer: &mut W) -> Result<(), std::fmt::Error> {
330    write!(writer, "{}", escape(text))
331}
332
333/// Copied from `RustyXML` for now
334pub fn escape(input: &str) -> String {
335    let mut result = String::with_capacity(input.len());
336
337    for c in input.chars() {
338        match c {
339            '&' => result.push_str("&amp;"),
340            '<' => result.push_str("&lt;"),
341            '>' => result.push_str("&gt;"),
342            '\'' => result.push_str("&apos;"),
343            '"' => result.push_str("&quot;"),
344            o => result.push(o),
345        }
346    }
347    result
348}
349
350/// BytesMut impl only std::fmt::Write but not std::io::Write. The
351/// latter trait is required for minidom's
352/// `Element::write_to_inner()`.
353struct WriteBytes<'a> {
354    dst: &'a mut BytesMut,
355}
356
357impl<'a> WriteBytes<'a> {
358    fn new(dst: &'a mut BytesMut) -> Self {
359        WriteBytes { dst }
360    }
361}
362
363impl<'a> std::io::Write for WriteBytes<'a> {
364    fn write(&mut self, buf: &[u8]) -> std::result::Result<usize, std::io::Error> {
365        self.dst.put_slice(buf);
366        Ok(buf.len())
367    }
368
369    fn flush(&mut self) -> std::result::Result<(), std::io::Error> {
370        Ok(())
371    }
372}
373
374#[cfg(test)]
375mod tests {
376    use super::*;
377    use bytes::BytesMut;
378
379    #[test]
380    fn test_stream_start() {
381        let mut c = XMPPCodec::new();
382        let mut b = BytesMut::with_capacity(1024);
383        b.put(r"<?xml version='1.0'?><stream:stream xmlns:stream='http://etherx.jabber.org/streams' version='1.0' xmlns='jabber:client'>");
384        let r = c.decode(&mut b);
385        assert!(match r {
386            Ok(Some(Packet::StreamStart(_))) => true,
387            _ => false,
388        });
389    }
390
391    #[test]
392    fn test_stream_end() {
393        let mut c = XMPPCodec::new();
394        let mut b = BytesMut::with_capacity(1024);
395        b.put(r"<?xml version='1.0'?><stream:stream xmlns:stream='http://etherx.jabber.org/streams' version='1.0' xmlns='jabber:client'>");
396        let r = c.decode(&mut b);
397        assert!(match r {
398            Ok(Some(Packet::StreamStart(_))) => true,
399            _ => false,
400        });
401        b.clear();
402        b.put(r"</stream:stream>");
403        let r = c.decode(&mut b);
404        assert!(match r {
405            Ok(Some(Packet::StreamEnd)) => true,
406            _ => false,
407        });
408    }
409
410    #[test]
411    fn test_truncated_stanza() {
412        let mut c = XMPPCodec::new();
413        let mut b = BytesMut::with_capacity(1024);
414        b.put(r"<?xml version='1.0'?><stream:stream xmlns:stream='http://etherx.jabber.org/streams' version='1.0' xmlns='jabber:client'>");
415        let r = c.decode(&mut b);
416        assert!(match r {
417            Ok(Some(Packet::StreamStart(_))) => true,
418            _ => false,
419        });
420
421        b.clear();
422        b.put(r"<test>ß</test");
423        let r = c.decode(&mut b);
424        assert!(match r {
425            Ok(None) => true,
426            _ => false,
427        });
428
429        b.clear();
430        b.put(r">");
431        let r = c.decode(&mut b);
432        assert!(match r {
433            Ok(Some(Packet::Stanza(ref el))) if el.name() == "test" && el.text() == "ß" => true,
434            _ => false,
435        });
436    }
437
438    #[test]
439    fn test_truncated_utf8() {
440        let mut c = XMPPCodec::new();
441        let mut b = BytesMut::with_capacity(1024);
442        b.put(r"<?xml version='1.0'?><stream:stream xmlns:stream='http://etherx.jabber.org/streams' version='1.0' xmlns='jabber:client'>");
443        let r = c.decode(&mut b);
444        assert!(match r {
445            Ok(Some(Packet::StreamStart(_))) => true,
446            _ => false,
447        });
448
449        b.clear();
450        b.put(&b"<test>\xc3"[..]);
451        let r = c.decode(&mut b);
452        assert!(match r {
453            Ok(None) => true,
454            _ => false,
455        });
456
457        b.clear();
458        b.put(&b"\x9f</test>"[..]);
459        let r = c.decode(&mut b);
460        assert!(match r {
461            Ok(Some(Packet::Stanza(ref el))) if el.name() == "test" && el.text() == "ß" => true,
462            _ => false,
463        });
464    }
465
466    /// test case for https://gitlab.com/xmpp-rs/tokio-xmpp/issues/3
467    #[test]
468    fn test_atrribute_prefix() {
469        let mut c = XMPPCodec::new();
470        let mut b = BytesMut::with_capacity(1024);
471        b.put(r"<?xml version='1.0'?><stream:stream xmlns:stream='http://etherx.jabber.org/streams' version='1.0' xmlns='jabber:client'>");
472        let r = c.decode(&mut b);
473        assert!(match r {
474            Ok(Some(Packet::StreamStart(_))) => true,
475            _ => false,
476        });
477
478        b.clear();
479        b.put(r"<status xml:lang='en'>Test status</status>");
480        let r = c.decode(&mut b);
481        assert!(match r {
482            Ok(Some(Packet::Stanza(ref el))) if el.name() == "status" && el.text() == "Test status" && el.attr("xml:lang").map_or(false, |a| a == "en") => true,
483            _ => false,
484        });
485
486    }
487
488    /// By default, encode() only get's a BytesMut that has 8kb space reserved.
489    #[test]
490    fn test_large_stanza() {
491        use futures::{Future, Sink};
492        use std::io::Cursor;
493        use tokio_codec::FramedWrite;
494        let framed = FramedWrite::new(Cursor::new(vec![]), XMPPCodec::new());
495        let mut text = "".to_owned();
496        for _ in 0..2usize.pow(15) {
497            text = text + "A";
498        }
499        let stanza = Element::builder("message")
500            .append(Element::builder("body").append(&text).build())
501            .build();
502        let framed = framed.send(Packet::Stanza(stanza)).wait().expect("send");
503        assert_eq!(
504            framed.get_ref().get_ref(),
505            &("<message><body>".to_owned() + &text + "</body></message>").as_bytes()
506        );
507    }
508}