Merge pull request #325 from BlueGreenMagick/doctype-whitespace

Event::DocType skip initial whitespaces

src/events/mod.rs

@@ -885,7 +885,7 @@ pub enum Event<'a> {
  Decl(BytesDecl<'a>),
  /// Processing instruction `<?...?>`.
  PI(BytesText<'a>),
- /// Doctype `<!DOCTYPE...>`.
+ /// Doctype `<!DOCTYPE ...>`.
  DocType(BytesText<'a>),
  /// End of XML document.
  Eof,

src/reader.rs

@@ -371,7 +371,12 @@ impl<R: BufRead> Reader<R> {
  Ok(Event::CData(BytesText::from_plain(&buf[8..buf.len() - 2])))
  } else if uncased_starts_with(buf, b"!DOCTYPE") {
  debug_assert!(len >= 8, "Minimum length guaranteed by read_bang_elem");
- Ok(Event::DocType(BytesText::from_escaped(&buf[8..])))
+ let start = buf[8..]
+ .iter()
+ .position(|b| !is_whitespace(*b))
+ .unwrap_or_else(|| buf.len() - 8);
+ debug_assert!(start < buf.len() - 8, "DocType must have a name");
+ Ok(Event::DocType(BytesText::from_escaped(&buf[8 + start..])))
  } else {
  unreachable!("Proper bang start guaranteed by read_bang_elem");
  }

src/writer.rs

@@ -119,7 +119,7 @@ impl<W: Write> Writer<W> {
  }
  Event::Decl(ref e) => self.write_wrapped(b"<?", e, b"?>"),
  Event::PI(ref e) => self.write_wrapped(b"<?", e, b"?>"),
- Event::DocType(ref e) => self.write_wrapped(b"<!DOCTYPE", e, b">"),
+ Event::DocType(ref e) => self.write_wrapped(b"<!DOCTYPE ", e, b">"),
  Event::Eof => Ok(()),
  };
  if let Some(i) = self.indent.as_mut() {

tests/documents/html5.txt

@@ -1,4 +1,4 @@
-DocType( html)
+DocType(html)
 Characters(
 )
 StartElement(a, attr-error: error while parsing attribute at position 7: Attribute value must start with a quote.)

tests/documents/sample_4_full.txt

@@ -1,4 +1,5 @@
 StartDocument(1.0, utf-8)
+DocType(data SYSTEM "abcd.dtd")
 StartElement({urn:x}p:data [z=">"])
 Whitespace("\n ")
 Comment(" abcd < > & ")

tests/documents/sample_4_short.txt

@@ -1,15 +1,16 @@
 StartDocument(1.0, utf-8)
-DocType(DOCTYPE data SYSTEM "abcd.dtd")
+DocType(data SYSTEM "abcd.dtd")
 StartElement({urn:x}p:data [z=">"])
+Comment( abcd < > & )
 StartElement(a)
-Characters("test")
+Characters(test)
 EndElement(a)
 StartElement(b)
-Characters("kkss\" = ddd\' >")
+Characters(kkss" = ddd' >)
 EndElement(b)
-Characters("<a>ddddd</b>!e3--><!-- ddckx")
+CData(<a>ddddd</b>!e3--><!-- ddckx)
 StartElement(c)
 EndElement(c)
-Characters("<![CDATA[zzzz]]>")
+CData(<![CDATA[zzzz]]>)
 EndElement({urn:x}p:data)
 EndDocument

tests/serde-migrated.rs

@@ -98,7 +98,6 @@ fn test_namespaces() {
 }
 
 #[test]
-#[ignore] // FIXME
 fn test_doctype() {
  #[derive(PartialEq, Serialize, Deserialize, Debug)]
  struct Envelope {