1#[derive(Debug, Clone)]
7pub struct Element {
8 pub name: String,
9 pub attrs: Vec<(String, String)>,
10 pub children: Vec<Element>,
11 pub line: usize,
13}
14
15impl Element {
16 pub fn attr(&self, key: &str) -> Option<&str> {
17 self.attrs
18 .iter()
19 .find(|(k, _)| k == key)
20 .map(|(_, v)| v.as_str())
21 }
22}
23
24struct Parser<'a> {
25 src: &'a str,
26 pos: usize,
27 line_at: std::cell::Cell<(usize, usize)>,
30}
31
32impl<'a> Parser<'a> {
33 fn new(src: &'a str) -> Self {
34 Parser {
35 src,
36 pos: 0,
37 line_at: std::cell::Cell::new((0, 1)),
38 }
39 }
40
41 fn line(&self) -> usize {
42 let (at, line) = self.line_at.get();
43 let (from, line) = if self.pos >= at { (at, line) } else { (0, 1) };
44 let line = line
45 + self.src.as_bytes()[from..self.pos]
46 .iter()
47 .filter(|&&b| b == b'\n')
48 .count();
49 self.line_at.set((self.pos, line));
50 line
51 }
52
53 fn err<T>(&self, msg: impl std::fmt::Display) -> Result<T, String> {
54 Err(format!("line {}: {msg}", self.line()))
55 }
56
57 fn rest(&self) -> &'a str {
58 &self.src[self.pos..]
59 }
60
61 fn skip_ws(&mut self) {
62 let trimmed = self.rest().trim_start();
63 self.pos = self.src.len() - trimmed.len();
64 }
65
66 fn skip_misc(&mut self) -> Result<(), String> {
68 loop {
69 self.skip_ws();
70 let rest = self.rest();
71 let end = if rest.starts_with("<!--") {
72 rest.find("-->").map(|i| i + 3)
73 } else if rest.starts_with("<?") {
74 rest.find("?>").map(|i| i + 2)
75 } else if rest.starts_with("<!") {
76 rest.find('>').map(|i| i + 1)
77 } else {
78 return Ok(());
79 };
80 match end {
81 Some(n) => self.pos += n,
82 None => return self.err("unterminated comment or declaration"),
83 }
84 }
85 }
86
87 fn name(&mut self) -> Result<String, String> {
88 let rest = self.rest();
89 let len = rest
90 .find(|c: char| !(c.is_alphanumeric() || matches!(c, '_' | '-' | ':' | '.')))
91 .unwrap_or(rest.len());
92 if len == 0 {
93 return self.err("expected a name");
94 }
95 self.pos += len;
96 Ok(rest[..len].to_string())
97 }
98
99 fn element(&mut self) -> Result<Element, String> {
100 let line = self.line();
101 if !self.rest().starts_with('<') {
102 return self.err("expected an element");
103 }
104 self.pos += 1;
105 let name = self.name()?;
106 let mut attrs: Vec<(String, String)> = Vec::new();
107 loop {
108 self.skip_ws();
109 let rest = self.rest();
110 if rest.starts_with("/>") {
111 self.pos += 2;
112 return Ok(Element {
113 name,
114 attrs,
115 children: Vec::new(),
116 line,
117 });
118 }
119 if rest.starts_with('>') {
120 self.pos += 1;
121 break;
122 }
123 if rest.is_empty() {
124 return self.err(format!("unterminated <{name}>"));
125 }
126 let key = self.name()?;
127 self.skip_ws();
128 if !self.rest().starts_with('=') {
129 return self.err(format!("expected `=` after attribute `{key}`"));
130 }
131 self.pos += 1;
132 self.skip_ws();
133 let quote = match self.rest().chars().next() {
134 Some(q @ ('"' | '\'')) => q,
135 _ => return self.err(format!("attribute `{key}`: expected a quoted value")),
136 };
137 self.pos += 1;
138 let Some(end) = self.rest().find(quote) else {
139 return self.err(format!("attribute `{key}`: unterminated value"));
140 };
141 let raw = &self.rest()[..end];
142 let value = unescape(raw).map_err(|e| format!("line {}: {e}", self.line()))?;
143 self.pos += end + 1;
144 if attrs.iter().any(|(k, _)| *k == key) {
145 return self.err(format!("<{name}>: duplicate attribute `{key}`"));
146 }
147 attrs.push((key, value));
148 }
149 let mut children = Vec::new();
150 loop {
151 self.skip_misc()?;
152 let rest = self.rest();
153 if rest.starts_with("</") {
154 self.pos += 2;
155 let close = self.name()?;
156 if close != name {
157 return self.err(format!("</{close}> closes <{name}>"));
158 }
159 self.skip_ws();
160 if !self.rest().starts_with('>') {
161 return self.err("expected `>`");
162 }
163 self.pos += 1;
164 return Ok(Element {
165 name,
166 attrs,
167 children,
168 line,
169 });
170 }
171 if rest.is_empty() {
172 return self.err(format!("<{name}> is never closed"));
173 }
174 if !rest.starts_with('<') {
175 return self.err(format!(
176 "<{name}>: text content isn't allowed — use a `text` attribute"
177 ));
178 }
179 children.push(self.element()?);
180 }
181 }
182}
183
184fn unescape(raw: &str) -> Result<String, String> {
185 let mut out = String::with_capacity(raw.len());
186 let mut rest = raw;
187 while let Some(i) = rest.find('&') {
188 out.push_str(&rest[..i]);
189 let Some(semi) = rest[i..].find(';') else {
190 return Err("`&` without `;` — write `&`".into());
191 };
192 let entity = &rest[i + 1..i + semi];
193 match entity {
194 "amp" => out.push('&'),
195 "lt" => out.push('<'),
196 "gt" => out.push('>'),
197 "quot" => out.push('"'),
198 "apos" => out.push('\''),
199 _ => {
200 let code = if let Some(hex) = entity.strip_prefix("#x") {
201 u32::from_str_radix(hex, 16).ok()
202 } else if let Some(dec) = entity.strip_prefix('#') {
203 dec.parse().ok()
204 } else {
205 None
206 };
207 match code.and_then(char::from_u32) {
208 Some(c) => out.push(c),
209 None => return Err(format!("unknown entity `&{entity};`")),
210 }
211 }
212 }
213 rest = &rest[i + semi + 1..];
214 }
215 out.push_str(rest);
216 Ok(out)
217}
218
219pub fn parse(src: &str) -> Result<Element, String> {
221 let src = src.strip_prefix('\u{feff}').unwrap_or(src);
222 let mut p = Parser::new(src);
223 p.skip_misc()?;
224 let root = p.element()?;
225 p.skip_misc()?;
226 if !p.rest().is_empty() {
227 return p.err("content after the root element");
228 }
229 Ok(root)
230}
231