1use std::collections::VecDeque;
2use std::io::{BufReader, Read};
3
4use quick_xml::encoding::Decoder;
5use quick_xml::escape::unescape;
6use quick_xml::events::{BytesEnd, BytesRef, BytesStart, BytesText, Event};
7use quick_xml::Reader;
8use quick_xml::XmlVersion;
9
10#[derive(Clone, Debug, PartialEq, Eq)]
11pub struct OwnedName {
12 pub local_name: String,
13 pub namespace: Option<String>,
14 pub prefix: Option<String>,
15}
16
17#[derive(Clone, Debug, PartialEq, Eq)]
18pub struct OwnedAttribute {
19 pub name: OwnedName,
20 pub value: String,
21}
22
23#[derive(Clone, Debug, Default, PartialEq, Eq)]
24pub struct Namespace {
25 mappings: Vec<(String, String)>,
26}
27
28impl Namespace {
29 pub fn empty() -> Self {
30 Self {
31 mappings: Vec::new(),
32 }
33 }
34}
35
36impl IntoIterator for Namespace {
37 type Item = (String, String);
38 type IntoIter = std::vec::IntoIter<(String, String)>;
39
40 fn into_iter(self) -> Self::IntoIter {
41 self.mappings.into_iter()
42 }
43}
44
45#[derive(Clone, Debug, PartialEq)]
46pub enum XmlEvent {
47 StartElement {
48 name: OwnedName,
49 attributes: Vec<OwnedAttribute>,
50 namespace: Namespace,
51 },
52 EndElement {
53 name: OwnedName,
54 },
55 Characters(String),
56 Whitespace(String),
57 EndDocument,
58}
59
60pub struct EventReader<R: Read> {
61 reader: Reader<BufReader<R>>,
62 buf: Vec<u8>,
63 pending: VecDeque<XmlEvent>,
64 finished: bool,
65}
66
67impl<R: Read> EventReader<R> {
68 pub fn new(reader: R) -> Self {
69 let mut reader = Reader::from_reader(BufReader::new(reader));
70 {
71 let config = reader.config_mut();
72 config.trim_text(false);
73 config.check_end_names = true;
74 config.expand_empty_elements = false;
75 }
76 Self {
77 reader,
78 buf: Vec::new(),
79 pending: VecDeque::new(),
80 finished: false,
81 }
82 }
83
84 pub fn next_event(&mut self) -> Result<XmlEvent, quick_xml::Error> {
85 self.read_next()
86 }
87
88 fn read_next(&mut self) -> Result<XmlEvent, quick_xml::Error> {
89 if let Some(event) = self.pending.pop_front() {
90 return Ok(event);
91 }
92
93 loop {
94 self.buf.clear();
95 match self.reader.read_event_into(&mut self.buf)? {
96 Event::Start(element) => {
97 let decoder = self.reader.decoder();
98 let event = Self::build_start_event(element, decoder)?;
99 return Ok(event);
100 }
101 Event::Empty(element) => {
102 let decoder = self.reader.decoder();
103 let event = Self::build_start_event(element, decoder)?;
104 if let XmlEvent::StartElement { name, .. } = &event {
105 self.pending
106 .push_back(XmlEvent::EndElement { name: name.clone() });
107 }
108 return Ok(event);
109 }
110 Event::End(element) => {
111 let name = build_name_from_end(&element)?;
112 return Ok(XmlEvent::EndElement { name });
113 }
114 Event::Text(text) => {
115 let text = decode_text(text)?;
116 return self.read_text_event(text);
117 }
118 Event::GeneralRef(reference) => {
119 let text = decode_reference(reference)?;
120 return self.read_text_event(text);
121 }
122 Event::CData(text) => {
123 let decoded = self.reader.decoder().decode(text.as_ref())?.into_owned();
124 return Ok(XmlEvent::Characters(decoded));
125 }
126 Event::Eof => {
127 self.finished = true;
128 return Ok(XmlEvent::EndDocument);
129 }
130 Event::Decl(_) | Event::PI(_) | Event::Comment(_) | Event::DocType(_) => {
131 }
133 }
134 }
135 }
136
137 fn read_text_event(&mut self, mut text: String) -> Result<XmlEvent, quick_xml::Error> {
138 loop {
139 self.buf.clear();
140 match self.reader.read_event_into(&mut self.buf)? {
141 Event::Text(next_text) => {
142 text.push_str(&decode_text(next_text)?);
143 }
144 Event::GeneralRef(reference) => {
145 text.push_str(&decode_reference(reference)?);
146 }
147 Event::Start(element) => {
148 let decoder = self.reader.decoder();
149 let event = Self::build_start_event(element, decoder)?;
150 self.pending.push_back(event);
151 break;
152 }
153 Event::Empty(element) => {
154 let decoder = self.reader.decoder();
155 let event = Self::build_start_event(element, decoder)?;
156 let end_event = if let XmlEvent::StartElement { name, .. } = &event {
157 Some(XmlEvent::EndElement { name: name.clone() })
158 } else {
159 None
160 };
161 self.pending.push_back(event);
162 if let Some(end_event) = end_event {
163 self.pending.push_back(end_event);
164 }
165 break;
166 }
167 Event::End(element) => {
168 let name = build_name_from_end(&element)?;
169 self.pending.push_back(XmlEvent::EndElement { name });
170 break;
171 }
172 Event::CData(text) => {
173 let decoded = self.reader.decoder().decode(text.as_ref())?.into_owned();
174 self.pending.push_back(XmlEvent::Characters(decoded));
175 break;
176 }
177 Event::Eof => {
178 self.pending.push_back(XmlEvent::EndDocument);
179 break;
180 }
181 Event::Decl(_) | Event::PI(_) | Event::Comment(_) | Event::DocType(_) => {
182 break;
183 }
184 }
185 }
186
187 if text.chars().all(char::is_whitespace) {
188 Ok(XmlEvent::Whitespace(text))
189 } else {
190 Ok(XmlEvent::Characters(text))
191 }
192 }
193
194 fn build_start_event(
195 element: BytesStart<'_>,
196 decoder: Decoder,
197 ) -> Result<XmlEvent, quick_xml::Error> {
198 let name = build_name_from_start(&element)?;
199 let attributes = build_attributes(&element, decoder)?;
200 Ok(XmlEvent::StartElement {
201 name,
202 attributes,
203 namespace: Namespace::empty(),
204 })
205 }
206}
207
208impl<R: Read> Iterator for EventReader<R> {
209 type Item = Result<XmlEvent, quick_xml::Error>;
210
211 fn next(&mut self) -> Option<Self::Item> {
212 if self.finished {
213 return None;
214 }
215 match self.read_next() {
216 Ok(XmlEvent::EndDocument) => {
217 self.finished = true;
218 Some(Ok(XmlEvent::EndDocument))
219 }
220 Ok(event) => Some(Ok(event)),
221 Err(e) => {
222 self.finished = true;
223 Some(Err(e))
224 }
225 }
226 }
227}
228
229fn build_name_from_start(element: &BytesStart<'_>) -> Result<OwnedName, quick_xml::Error> {
230 let name = element.name();
231 Ok(split_qname(name.as_ref()))
232}
233
234fn build_name_from_end(element: &BytesEnd<'_>) -> Result<OwnedName, quick_xml::Error> {
235 let name = element.name();
236 Ok(split_qname(name.as_ref()))
237}
238
239fn decode_text(text: BytesText<'_>) -> Result<String, quick_xml::Error> {
240 Ok(text.xml_content(XmlVersion::Implicit1_0)?.into_owned())
241}
242
243fn decode_reference(reference: BytesRef<'_>) -> Result<String, quick_xml::Error> {
244 let reference = reference.xml_content(XmlVersion::Implicit1_0)?;
245 let escaped = format!("&{reference};");
246 Ok(unescape(&escaped)?.into_owned())
247}
248
249fn split_qname(raw: &[u8]) -> OwnedName {
250 let text = String::from_utf8_lossy(raw).into_owned();
251 if let Some(idx) = text.find(':') {
252 let prefix = text[..idx].to_string();
253 let local = text[idx + 1..].to_string();
254 OwnedName {
255 local_name: local,
256 namespace: None,
257 prefix: Some(prefix),
258 }
259 } else {
260 OwnedName {
261 local_name: text,
262 namespace: None,
263 prefix: None,
264 }
265 }
266}
267
268fn build_attributes(
269 element: &BytesStart<'_>,
270 decoder: Decoder,
271) -> Result<Vec<OwnedAttribute>, quick_xml::Error> {
272 let mut attributes = Vec::new();
273 for attr_result in element.attributes().with_checks(false) {
274 let attr = attr_result.map_err(quick_xml::Error::from)?;
275 let value = attr
276 .decoded_and_normalized_value(XmlVersion::Implicit1_0, decoder)?
277 .into_owned();
278 let name = split_qname(attr.key.as_ref());
279 attributes.push(OwnedAttribute { name, value });
280 }
281 Ok(attributes)
282}
283
284#[cfg(test)]
285mod tests {
286 use super::*;
287
288 #[test]
289 fn decodes_xml_entities_in_text_and_attributes() {
290 let xml = r#"<w:t w:val="A & B">A & B</w:t>"#;
291 let mut reader = EventReader::new(xml.as_bytes());
292
293 let start = reader.next_event().unwrap();
294 match start {
295 XmlEvent::StartElement {
296 name, attributes, ..
297 } => {
298 assert_eq!(name.prefix.as_deref(), Some("w"));
299 assert_eq!(name.local_name, "t");
300 assert_eq!(attributes[0].name.prefix.as_deref(), Some("w"));
301 assert_eq!(attributes[0].name.local_name, "val");
302 assert_eq!(attributes[0].value, "A & B");
303 }
304 event => panic!("expected start element, got {event:?}"),
305 }
306
307 assert_eq!(
308 reader.next_event().unwrap(),
309 XmlEvent::Characters("A & B".to_string())
310 );
311 assert!(matches!(
312 reader.next_event().unwrap(),
313 XmlEvent::EndElement { name } if name.local_name == "t"
314 ));
315 assert_eq!(reader.next_event().unwrap(), XmlEvent::EndDocument);
316 }
317}