Skip to main content

substrate_data/forms/text/
parser.rs

1//! The parser: text to a value, with the spans of its parts when asked.
2
3use super::{Doc, LineCol, Span, Spans, SyntaxError, scan};
4use crate::model::value::{Map, Value};
5
6/// `\\u` escapes are hexadecimal.
7const HEX_RADIX: u32 = 16;
8
9/// Deeper than this is an error, not a stack overflow.
10pub const MAX_DEPTH: usize = 256;
11
12/// Parses text.
13///
14/// # Errors
15///
16/// [`SyntaxError`] when the text doesn't parse.
17pub fn parse(text: &str) -> Result<Doc, SyntaxError> {
18    let (value, spans) = run::<true>(text)?;
19    Ok(Doc { value, spans })
20}
21
22/// Parses text, without the positions (faster).
23///
24/// # Errors
25///
26/// [`SyntaxError`] when the text doesn't parse.
27pub fn parse_value(text: &str) -> Result<Value, SyntaxError> {
28    run::<false>(text).map(|(v, _)| v)
29}
30
31fn run<const SPANS: bool>(text: &str) -> Result<(Value, Spans), SyntaxError> {
32    let mut p = Parser::<SPANS> {
33        src: text.as_bytes(),
34        text,
35        i: 0,
36        depth: 0,
37    };
38    if text.starts_with('\u{feff}') {
39        p.i = 3;
40    }
41    p.ws()?;
42    let (value, spans) = p.value()?;
43    p.ws()?;
44    if p.i < p.src.len() {
45        return Err(p.error("unexpected text after the value"));
46    }
47    Ok((value, spans))
48}
49
50/// `SPANS`: whether to record where each value is.
51#[derive(Debug)]
52struct Parser<'a, const SPANS: bool> {
53    src: &'a [u8],
54    text: &'a str,
55    i: usize,
56    depth: usize,
57}
58
59impl<const SPANS: bool> Parser<'_, SPANS> {
60    fn error(&self, message: impl Into<String>) -> SyntaxError {
61        self.error_at(self.i, message)
62    }
63
64    fn error_at(&self, offset: usize, message: impl Into<String>) -> SyntaxError {
65        let mut message = message.into();
66        if offset >= self.src.len() {
67            message.push_str(", found the end of the file");
68        }
69        SyntaxError {
70            message,
71            at: LineCol::of(self.text, offset),
72            offset,
73        }
74    }
75
76    fn peek(&self) -> Option<u8> {
77        self.src.get(self.i).copied()
78    }
79
80    /// Skips whitespace and comments.
81    fn ws(&mut self) -> Result<(), SyntaxError> {
82        loop {
83            let rest = &self.src[self.i..];
84            self.i += rest
85                .iter()
86                .position(|&b| !matches!(b, b' ' | b'\t' | b'\n' | b'\r'))
87                .unwrap_or(rest.len());
88            if self.peek() != Some(b'/') {
89                return Ok(());
90            }
91            match self.src.get(self.i + 1) {
92                Some(b'/') => {
93                    self.i += scan::line_end(&self.src[self.i..]);
94                }
95                Some(b'*') => {
96                    let start = self.i;
97                    match scan::block_comment_end(&self.src[self.i + 2..]) {
98                        Some(end) => self.i += 2 + end + 2,
99                        None => return Err(self.error_at(start, "unclosed `/*` comment")),
100                    }
101                }
102                _ => return Ok(()),
103            }
104        }
105    }
106
107    fn value(&mut self) -> Result<(Value, Spans), SyntaxError> {
108        let start = self.i;
109        let (value, children) = match self.peek() {
110            Some(b'{') => self.nested(Parser::map)?,
111            Some(b'[') => self.nested(Parser::list)?,
112            Some(b'"') => (Value::Str(self.string()?), Vec::new()),
113            Some(b'-' | b'0'..=b'9') => (self.number()?, Vec::new()),
114            Some(b't') if self.src[self.i..].starts_with(b"true") => {
115                self.i += 4;
116                (Value::Bool(true), Vec::new())
117            }
118            Some(b'f') if self.src[self.i..].starts_with(b"false") => {
119                self.i += 5;
120                (Value::Bool(false), Vec::new())
121            }
122            Some(b'n') if self.src[self.i..].starts_with(b"null") => {
123                self.i += 4;
124                (Value::Null, Vec::new())
125            }
126            Some(b'\'') => return Err(self.error("strings use double quotes")),
127            _ => return Err(self.error("expected a value")),
128        };
129        let span = Span::new(start, self.i);
130        Ok((
131            value,
132            Spans {
133                span,
134                key: None,
135                children,
136            },
137        ))
138    }
139
140    fn nested(
141        &mut self,
142        f: impl FnOnce(&mut Self) -> Result<(Value, Vec<Spans>), SyntaxError>,
143    ) -> Result<(Value, Vec<Spans>), SyntaxError> {
144        self.depth += 1;
145        if self.depth > MAX_DEPTH {
146            return Err(self.error(format!("nested deeper than {MAX_DEPTH} levels")));
147        }
148        let r = f(self);
149        self.depth -= 1;
150        r
151    }
152
153    fn list(&mut self) -> Result<(Value, Vec<Spans>), SyntaxError> {
154        self.i += 1;
155        let mut items = Vec::new();
156        let mut spans = Vec::new();
157        loop {
158            self.ws()?;
159            if self.peek() == Some(b']') {
160                self.i += 1;
161                break;
162            }
163            let (v, s) = self.value()?;
164            items.push(v);
165            if SPANS {
166                spans.push(s);
167            }
168            self.ws()?;
169            match self.peek() {
170                Some(b',') => self.i += 1,
171                Some(b']') => {
172                    self.i += 1;
173                    break;
174                }
175                _ => return Err(self.error("expected `,` or `]`")),
176            }
177        }
178        Ok((Value::List(items), spans))
179    }
180
181    fn map(&mut self) -> Result<(Value, Vec<Spans>), SyntaxError> {
182        self.i += 1;
183        let mut map = Map::new();
184        let mut spans = Vec::new();
185        // Duplicate keys: a scan for small objects, a set once they grow.
186        let mut seen: Option<std::collections::HashSet<String>> = None;
187        loop {
188            self.ws()?;
189            if self.peek() == Some(b'}') {
190                self.i += 1;
191                break;
192            }
193            let key_start = self.i;
194            let key = self.key()?;
195            let key_span = Span::new(key_start, self.i);
196            if is_duplicate(&map, &mut seen, &key) {
197                return Err(
198                    self.error_at(key_start, format!("`{key}` is already set in this object"))
199                );
200            }
201            self.ws()?;
202            if self.peek() != Some(b':') {
203                return Err(self.error("expected `:` after the key"));
204            }
205            self.i += 1;
206            self.ws()?;
207            let (v, mut s) = self.value()?;
208            s.key = Some(key_span);
209            map.push_unchecked(key.into(), v);
210            if SPANS {
211                spans.push(s);
212            }
213            self.ws()?;
214            match self.peek() {
215                Some(b',') => self.i += 1,
216                Some(b'}') => {
217                    self.i += 1;
218                    break;
219                }
220                _ => return Err(self.error("expected `,` or `}`")),
221            }
222        }
223        Ok((Value::Map(map), spans))
224    }
225
226    /// An object's key: a string, or a bare name.
227    fn key(&mut self) -> Result<String, SyntaxError> {
228        let start = self.i;
229        match self.peek() {
230            Some(b'"') => self.string(),
231            Some(first) if first.is_ascii_alphabetic() || first == b'_' || first == b'$' => {
232                while self
233                    .peek()
234                    .is_some_and(|c| c.is_ascii_alphanumeric() || c == b'_' || c == b'$')
235                {
236                    self.i += 1;
237                }
238                Ok(self.text[start..self.i].to_string())
239            }
240            _ => Err(self.error("expected a key")),
241        }
242    }
243
244    fn string(&mut self) -> Result<String, SyntaxError> {
245        let open = self.i;
246        self.i += 1;
247        let mut out = String::new();
248        loop {
249            let Some(n) = scan::string_stop(&self.src[self.i..]) else {
250                return Err(self.error_at(open, "unclosed string"));
251            };
252            out.push_str(&self.text[self.i..self.i + n]);
253            self.i += n;
254            match self.src[self.i] {
255                b'"' => {
256                    self.i += 1;
257                    return Ok(out);
258                }
259                b'\\' => self.escape(&mut out)?,
260                b'\n' => return Err(self.error("line break in a string (write `\\n`)")),
261                _ => return Err(self.error("control character in a string")),
262            }
263        }
264    }
265
266    fn escape(&mut self, out: &mut String) -> Result<(), SyntaxError> {
267        let at = self.i;
268        self.i += 1;
269        let c = self.peek().ok_or_else(|| self.error("unclosed string"))?;
270        self.i += 1;
271        match c {
272            b'"' => out.push('"'),
273            b'\\' => out.push('\\'),
274            b'/' => out.push('/'),
275            b'b' => out.push('\u{8}'),
276            b'f' => out.push('\u{c}'),
277            b'n' => out.push('\n'),
278            b'r' => out.push('\r'),
279            b't' => out.push('\t'),
280            b'u' => {
281                let hi = self.hex4(at)?;
282                let ch = if (0xD800..0xDC00).contains(&hi) {
283                    if !self.src[self.i..].starts_with(b"\\u") {
284                        return Err(self.error_at(at, "unpaired surrogate in `\\u` escape"));
285                    }
286                    self.i += 2;
287                    let lo = self.hex4(at)?;
288                    if !(0xDC00..0xE000).contains(&lo) {
289                        return Err(self.error_at(at, "unpaired surrogate in `\\u` escape"));
290                    }
291                    char::from_u32(0x10000 + ((hi - 0xD800) << 10) + (lo - 0xDC00))
292                } else {
293                    char::from_u32(hi)
294                };
295                out.push(
296                    ch.ok_or_else(|| self.error_at(at, "unpaired surrogate in `\\u` escape"))?,
297                );
298            }
299            _ => {
300                return Err(self.error_at(
301                    at,
302                    "unknown escape (`\\n`, `\\t`, `\\\"`, `\\\\`, `\\uXXXX`, ...)",
303                ));
304            }
305        }
306        Ok(())
307    }
308
309    fn hex4(&mut self, at: usize) -> Result<u32, SyntaxError> {
310        let digits = self
311            .text
312            .get(self.i..self.i + 4)
313            .filter(|d| d.bytes().all(|b| b.is_ascii_hexdigit()));
314        let d = digits.ok_or_else(|| self.error_at(at, "`\\u` needs four hex digits"))?;
315        let code = u32::from_str_radix(d, HEX_RADIX)
316            .map_err(|_| self.error_at(at, "`\\u` needs four hex digits"))?;
317        self.i += 4;
318        Ok(code)
319    }
320
321    fn number(&mut self) -> Result<Value, SyntaxError> {
322        let start = self.i;
323        if self.peek() == Some(b'-') {
324            self.i += 1;
325        }
326        let digits = |p: &mut Self| {
327            let s = p.i;
328            while p.peek().is_some_and(|c| c.is_ascii_digit()) {
329                p.i += 1;
330            }
331            p.i - s
332        };
333        let int_start = self.i;
334        if digits(self) == 0 {
335            return Err(self.error("expected digits"));
336        }
337        if self.src[int_start] == b'0' && self.i - int_start > 1 {
338            return Err(self.error_at(int_start, "numbers don't start with 0"));
339        }
340        let mut float = false;
341        if self.peek() == Some(b'.') {
342            self.i += 1;
343            float = true;
344            if digits(self) == 0 {
345                return Err(self.error("expected digits after `.`"));
346            }
347        }
348        if matches!(self.peek(), Some(b'e' | b'E')) {
349            self.i += 1;
350            float = true;
351            if matches!(self.peek(), Some(b'+' | b'-')) {
352                self.i += 1;
353            }
354            if digits(self) == 0 {
355                return Err(self.error("expected digits in the exponent"));
356            }
357        }
358        let s = &self.text[start..self.i];
359        if !float && let Ok(n) = s.parse::<i64>() {
360            return Ok(Value::Int(n));
361        }
362        Ok(Value::Float(
363            s.parse::<f64>()
364                .map_err(|_| self.error_at(start, "invalid number"))?,
365        ))
366    }
367}
368
369/// Whether `key` is already in `map`: a scan for small objects, a set
370/// (built in `seen` the first time) once they grow.
371fn is_duplicate(
372    map: &Map,
373    seen: &mut Option<std::collections::HashSet<String>>,
374    key: &str,
375) -> bool {
376    match seen {
377        Some(set) => !set.insert(key.to_string()),
378        None if map.len() < 16 => map.get(key).is_some(),
379        None => {
380            let mut set: std::collections::HashSet<String> =
381                map.keys().map(str::to_string).collect();
382            let dup = !set.insert(key.to_string());
383            *seen = Some(set);
384            dup
385        }
386    }
387}