1 | //! This is an implementation of [`Reader`] for reading from a `&[u8]` as
|
2 | //! underlying byte stream. This implementation supports not using an
|
3 | //! intermediate buffer as the byte slice itself can be used to borrow from.
|
4 |
|
5 | use std::borrow::Cow;
|
6 |
|
7 | #[cfg (feature = "encoding" )]
|
8 | use crate::reader::EncodingRef;
|
9 | #[cfg (feature = "encoding" )]
|
10 | use encoding_rs::{Encoding, UTF_8};
|
11 |
|
12 | use crate::errors::{Error, Result};
|
13 | use crate::events::Event;
|
14 | use crate::name::QName;
|
15 | use crate::reader::{is_whitespace, BangType, ReadElementState, Reader, Span, XmlSource};
|
16 |
|
17 | use memchr;
|
18 |
|
19 | /// This is an implementation for reading from a `&[u8]` as underlying byte stream.
|
20 | /// This implementation supports not using an intermediate buffer as the byte slice
|
21 | /// itself can be used to borrow from.
|
22 | impl<'a> Reader<&'a [u8]> {
|
23 | /// Creates an XML reader from a string slice.
|
24 | #[allow (clippy::should_implement_trait)]
|
25 | pub fn from_str(s: &'a str) -> Self {
|
26 | // Rust strings are guaranteed to be UTF-8, so lock the encoding
|
27 | #[cfg (feature = "encoding" )]
|
28 | {
|
29 | let mut reader = Self::from_reader(s.as_bytes());
|
30 | reader.state.encoding = EncodingRef::Explicit(UTF_8);
|
31 | reader
|
32 | }
|
33 |
|
34 | #[cfg (not(feature = "encoding" ))]
|
35 | Self::from_reader(s.as_bytes())
|
36 | }
|
37 |
|
38 | /// Read an event that borrows from the input rather than a buffer.
|
39 | ///
|
40 | /// There is no asynchronous `read_event_async()` version of this function,
|
41 | /// because it is not necessary -- the contents are already in memory and no IO
|
42 | /// is needed, therefore there is no potential for blocking.
|
43 | ///
|
44 | /// # Examples
|
45 | ///
|
46 | /// ```
|
47 | /// # use pretty_assertions::assert_eq;
|
48 | /// use quick_xml::events::Event;
|
49 | /// use quick_xml::reader::Reader;
|
50 | ///
|
51 | /// let mut reader = Reader::from_str(r#"
|
52 | /// <tag1 att1 = "test">
|
53 | /// <tag2><!--Test comment-->Test</tag2>
|
54 | /// <tag2>Test 2</tag2>
|
55 | /// </tag1>
|
56 | /// "# );
|
57 | /// reader.trim_text(true);
|
58 | ///
|
59 | /// let mut count = 0;
|
60 | /// let mut txt = Vec::new();
|
61 | /// loop {
|
62 | /// match reader.read_event().unwrap() {
|
63 | /// Event::Start(e) => count += 1,
|
64 | /// Event::Text(e) => txt.push(e.unescape().unwrap().into_owned()),
|
65 | /// Event::Eof => break,
|
66 | /// _ => (),
|
67 | /// }
|
68 | /// }
|
69 | /// assert_eq!(count, 3);
|
70 | /// assert_eq!(txt, vec!["Test" .to_string(), "Test 2" .to_string()]);
|
71 | /// ```
|
72 | #[inline ]
|
73 | pub fn read_event(&mut self) -> Result<Event<'a>> {
|
74 | self.read_event_impl(())
|
75 | }
|
76 |
|
77 | /// Reads until end element is found. This function is supposed to be called
|
78 | /// after you already read a [`Start`] event.
|
79 | ///
|
80 | /// Returns a span that cover content between `>` of an opening tag and `<` of
|
81 | /// a closing tag or an empty slice, if [`expand_empty_elements`] is set and
|
82 | /// this method was called after reading expanded [`Start`] event.
|
83 | ///
|
84 | /// Manages nested cases where parent and child elements have the _literally_
|
85 | /// same name.
|
86 | ///
|
87 | /// If corresponding [`End`] event will not be found, the [`Error::UnexpectedEof`]
|
88 | /// will be returned. In particularly, that error will be returned if you call
|
89 | /// this method without consuming the corresponding [`Start`] event first.
|
90 | ///
|
91 | /// The `end` parameter should contain name of the end element _in the reader
|
92 | /// encoding_. It is good practice to always get that parameter using
|
93 | /// [`BytesStart::to_end()`] method.
|
94 | ///
|
95 | /// The correctness of the skipped events does not checked, if you disabled
|
96 | /// the [`check_end_names`] option.
|
97 | ///
|
98 | /// There is no asynchronous `read_to_end_async()` version of this function,
|
99 | /// because it is not necessary -- the contents are already in memory and no IO
|
100 | /// is needed, therefore there is no potential for blocking.
|
101 | ///
|
102 | /// # Namespaces
|
103 | ///
|
104 | /// While the `Reader` does not support namespace resolution, namespaces
|
105 | /// does not change the algorithm for comparing names. Although the names
|
106 | /// `a:name` and `b:name` where both prefixes `a` and `b` resolves to the
|
107 | /// same namespace, are semantically equivalent, `</b:name>` cannot close
|
108 | /// `<a:name>`, because according to [the specification]
|
109 | ///
|
110 | /// > The end of every element that begins with a **start-tag** MUST be marked
|
111 | /// > by an **end-tag** containing a name that echoes the element's type as
|
112 | /// > given in the **start-tag**
|
113 | ///
|
114 | /// # Examples
|
115 | ///
|
116 | /// This example shows, how you can skip XML content after you read the
|
117 | /// start event.
|
118 | ///
|
119 | /// ```
|
120 | /// # use pretty_assertions::assert_eq;
|
121 | /// use quick_xml::events::{BytesStart, Event};
|
122 | /// use quick_xml::reader::Reader;
|
123 | ///
|
124 | /// let mut reader = Reader::from_str(r#"
|
125 | /// <outer>
|
126 | /// <inner>
|
127 | /// <inner></inner>
|
128 | /// <inner/>
|
129 | /// <outer></outer>
|
130 | /// <outer/>
|
131 | /// </inner>
|
132 | /// </outer>
|
133 | /// "# );
|
134 | /// reader.trim_text(true);
|
135 | ///
|
136 | /// let start = BytesStart::new("outer" );
|
137 | /// let end = start.to_end().into_owned();
|
138 | ///
|
139 | /// // First, we read a start event...
|
140 | /// assert_eq!(reader.read_event().unwrap(), Event::Start(start));
|
141 | ///
|
142 | /// // ...then, we could skip all events to the corresponding end event.
|
143 | /// // This call will correctly handle nested <outer> elements.
|
144 | /// // Note, however, that this method does not handle namespaces.
|
145 | /// reader.read_to_end(end.name()).unwrap();
|
146 | ///
|
147 | /// // At the end we should get an Eof event, because we ate the whole XML
|
148 | /// assert_eq!(reader.read_event().unwrap(), Event::Eof);
|
149 | /// ```
|
150 | ///
|
151 | /// [`Start`]: Event::Start
|
152 | /// [`End`]: Event::End
|
153 | /// [`BytesStart::to_end()`]: crate::events::BytesStart::to_end
|
154 | /// [`expand_empty_elements`]: Self::expand_empty_elements
|
155 | /// [`check_end_names`]: Self::check_end_names
|
156 | /// [the specification]: https://www.w3.org/TR/xml11/#dt-etag
|
157 | pub fn read_to_end(&mut self, end: QName) -> Result<Span> {
|
158 | Ok(read_to_end!(self, end, (), read_event_impl, {}))
|
159 | }
|
160 |
|
161 | /// Reads content between start and end tags, including any markup. This
|
162 | /// function is supposed to be called after you already read a [`Start`] event.
|
163 | ///
|
164 | /// Manages nested cases where parent and child elements have the _literally_
|
165 | /// same name.
|
166 | ///
|
167 | /// This method does not unescape read data, instead it returns content
|
168 | /// "as is" of the XML document. This is because it has no idea what text
|
169 | /// it reads, and if, for example, it contains CDATA section, attempt to
|
170 | /// unescape it content will spoil data.
|
171 | ///
|
172 | /// Any text will be decoded using the XML current [`decoder()`].
|
173 | ///
|
174 | /// Actually, this method perform the following code:
|
175 | ///
|
176 | /// ```ignore
|
177 | /// let span = reader.read_to_end(end)?;
|
178 | /// let text = reader.decoder().decode(&reader.inner_slice[span]);
|
179 | /// ```
|
180 | ///
|
181 | /// # Examples
|
182 | ///
|
183 | /// This example shows, how you can read a HTML content from your XML document.
|
184 | ///
|
185 | /// ```
|
186 | /// # use pretty_assertions::assert_eq;
|
187 | /// # use std::borrow::Cow;
|
188 | /// use quick_xml::events::{BytesStart, Event};
|
189 | /// use quick_xml::reader::Reader;
|
190 | ///
|
191 | /// let mut reader = Reader::from_str("
|
192 | /// <html>
|
193 | /// <title>This is a HTML text</title>
|
194 | /// <p>Usual XML rules does not apply inside it
|
195 | /// <p>For example, elements not needed to be "closed"
|
196 | /// </html>
|
197 | /// " );
|
198 | /// reader.trim_text(true);
|
199 | ///
|
200 | /// let start = BytesStart::new("html" );
|
201 | /// let end = start.to_end().into_owned();
|
202 | ///
|
203 | /// // First, we read a start event...
|
204 | /// assert_eq!(reader.read_event().unwrap(), Event::Start(start));
|
205 | /// // ...and disable checking of end names because we expect HTML further...
|
206 | /// reader.check_end_names(false);
|
207 | ///
|
208 | /// // ...then, we could read text content until close tag.
|
209 | /// // This call will correctly handle nested <html> elements.
|
210 | /// let text = reader.read_text(end.name()).unwrap();
|
211 | /// assert_eq!(text, Cow::Borrowed(r#"
|
212 | /// <title>This is a HTML text</title>
|
213 | /// <p>Usual XML rules does not apply inside it
|
214 | /// <p>For example, elements not needed to be "closed"
|
215 | /// "# ));
|
216 | /// assert!(matches!(text, Cow::Borrowed(_)));
|
217 | ///
|
218 | /// // Now we can enable checks again
|
219 | /// reader.check_end_names(true);
|
220 | ///
|
221 | /// // At the end we should get an Eof event, because we ate the whole XML
|
222 | /// assert_eq!(reader.read_event().unwrap(), Event::Eof);
|
223 | /// ```
|
224 | ///
|
225 | /// [`Start`]: Event::Start
|
226 | /// [`decoder()`]: Self::decoder()
|
227 | pub fn read_text(&mut self, end: QName) -> Result<Cow<'a, str>> {
|
228 | // self.reader will be changed, so store original reference
|
229 | let buffer = self.reader;
|
230 | let span = self.read_to_end(end)?;
|
231 |
|
232 | self.decoder().decode(&buffer[0..span.len()])
|
233 | }
|
234 | }
|
235 |
|
236 | ////////////////////////////////////////////////////////////////////////////////////////////////////
|
237 |
|
238 | /// Implementation of `XmlSource` for `&[u8]` reader using a `Self` as buffer
|
239 | /// that will be borrowed by events. This implementation provides a zero-copy deserialization
|
240 | impl<'a> XmlSource<'a, ()> for &'a [u8] {
|
241 | #[cfg (not(feature = "encoding" ))]
|
242 | fn remove_utf8_bom(&mut self) -> Result<()> {
|
243 | if self.starts_with(crate::encoding::UTF8_BOM) {
|
244 | *self = &self[crate::encoding::UTF8_BOM.len()..];
|
245 | }
|
246 | Ok(())
|
247 | }
|
248 |
|
249 | #[cfg (feature = "encoding" )]
|
250 | fn detect_encoding(&mut self) -> Result<Option<&'static Encoding>> {
|
251 | if let Some((enc, bom_len)) = crate::encoding::detect_encoding(self) {
|
252 | *self = &self[bom_len..];
|
253 | return Ok(Some(enc));
|
254 | }
|
255 | Ok(None)
|
256 | }
|
257 |
|
258 | fn read_bytes_until(
|
259 | &mut self,
|
260 | byte: u8,
|
261 | _buf: (),
|
262 | position: &mut usize,
|
263 | ) -> Result<Option<&'a [u8]>> {
|
264 | // search byte must be within the ascii range
|
265 | debug_assert!(byte.is_ascii());
|
266 | if self.is_empty() {
|
267 | return Ok(None);
|
268 | }
|
269 |
|
270 | Ok(Some(if let Some(i) = memchr::memchr(byte, self) {
|
271 | *position += i + 1;
|
272 | let bytes = &self[..i];
|
273 | *self = &self[i + 1..];
|
274 | bytes
|
275 | } else {
|
276 | *position += self.len();
|
277 | let bytes = &self[..];
|
278 | *self = &[];
|
279 | bytes
|
280 | }))
|
281 | }
|
282 |
|
283 | fn read_bang_element(
|
284 | &mut self,
|
285 | _buf: (),
|
286 | position: &mut usize,
|
287 | ) -> Result<Option<(BangType, &'a [u8])>> {
|
288 | // Peeked one bang ('!') before being called, so it's guaranteed to
|
289 | // start with it.
|
290 | debug_assert_eq!(self[0], b'!' );
|
291 |
|
292 | let bang_type = BangType::new(self[1..].first().copied())?;
|
293 |
|
294 | if let Some((bytes, i)) = bang_type.parse(&[], self) {
|
295 | *position += i;
|
296 | *self = &self[i..];
|
297 | return Ok(Some((bang_type, bytes)));
|
298 | }
|
299 |
|
300 | // Note: Do not update position, so the error points to
|
301 | // somewhere sane rather than at the EOF
|
302 | Err(bang_type.to_err())
|
303 | }
|
304 |
|
305 | fn read_element(&mut self, _buf: (), position: &mut usize) -> Result<Option<&'a [u8]>> {
|
306 | if self.is_empty() {
|
307 | return Ok(None);
|
308 | }
|
309 |
|
310 | let mut state = ReadElementState::Elem;
|
311 |
|
312 | if let Some((bytes, i)) = state.change(self) {
|
313 | // Position now just after the `>` symbol
|
314 | *position += i;
|
315 | *self = &self[i..];
|
316 | return Ok(Some(bytes));
|
317 | }
|
318 |
|
319 | // Note: Do not update position, so the error points to a sane place
|
320 | // rather than at the EOF.
|
321 | Err(Error::UnexpectedEof("Element" .to_string()))
|
322 |
|
323 | // FIXME: Figure out why the other one works without UnexpectedEof
|
324 | }
|
325 |
|
326 | fn skip_whitespace(&mut self, position: &mut usize) -> Result<()> {
|
327 | let whitespaces = self
|
328 | .iter()
|
329 | .position(|b| !is_whitespace(*b))
|
330 | .unwrap_or(self.len());
|
331 | *position += whitespaces;
|
332 | *self = &self[whitespaces..];
|
333 | Ok(())
|
334 | }
|
335 |
|
336 | fn skip_one(&mut self, byte: u8, position: &mut usize) -> Result<bool> {
|
337 | // search byte must be within the ascii range
|
338 | debug_assert!(byte.is_ascii());
|
339 | if self.first() == Some(&byte) {
|
340 | *self = &self[1..];
|
341 | *position += 1;
|
342 | Ok(true)
|
343 | } else {
|
344 | Ok(false)
|
345 | }
|
346 | }
|
347 |
|
348 | fn peek_one(&mut self) -> Result<Option<u8>> {
|
349 | Ok(self.first().copied())
|
350 | }
|
351 | }
|
352 |
|
353 | #[cfg (test)]
|
354 | mod test {
|
355 | use crate::reader::test::check;
|
356 | use crate::reader::XmlSource;
|
357 |
|
358 | /// Default buffer constructor just pass the byte array from the test
|
359 | fn identity<T>(input: T) -> T {
|
360 | input
|
361 | }
|
362 |
|
363 | check!(
|
364 | #[test]
|
365 | read_event_impl,
|
366 | read_until_close,
|
367 | identity,
|
368 | ()
|
369 | );
|
370 |
|
371 | #[cfg (feature = "encoding" )]
|
372 | mod encoding {
|
373 | use crate::events::Event;
|
374 | use crate::reader::Reader;
|
375 | use encoding_rs::UTF_8;
|
376 | use pretty_assertions::assert_eq;
|
377 |
|
378 | /// Checks that XML declaration cannot change the encoding from UTF-8 if
|
379 | /// a `Reader` was created using `from_str` method
|
380 | #[test ]
|
381 | fn str_always_has_utf8() {
|
382 | let mut reader = Reader::from_str("<?xml encoding='UTF-16'?>" );
|
383 |
|
384 | assert_eq!(reader.decoder().encoding(), UTF_8);
|
385 | reader.read_event().unwrap();
|
386 | assert_eq!(reader.decoder().encoding(), UTF_8);
|
387 |
|
388 | assert_eq!(reader.read_event().unwrap(), Event::Eof);
|
389 | }
|
390 | }
|
391 | }
|
392 | |