1 | //! Parsing library for [`defmt`] format strings. |
2 | //! |
3 | //! This is an implementation detail of [`defmt`] and not meant to be consumed by other tools at the |
4 | //! moment so all the API is unstable. |
5 | //! |
6 | //! [`defmt`]: https://github.com/knurling-rs/defmt |
7 | |
8 | #![cfg_attr (docsrs, feature(doc_cfg))] |
9 | #![doc (html_logo_url = "https://knurling.ferrous-systems.com/knurling_logo_light_text.svg" )] |
10 | |
11 | mod display_hint; |
12 | #[cfg (test)] |
13 | mod tests; |
14 | mod types; |
15 | |
16 | use std::{borrow::Cow, ops::Range}; |
17 | |
18 | pub use crate::{ |
19 | display_hint::{DisplayHint, TimePrecision}, |
20 | types::Type, |
21 | }; |
22 | |
23 | /// The kinds of error this library can return |
24 | #[derive (thiserror::Error, Debug, PartialEq, Eq, Clone)] |
25 | pub enum Error { |
26 | #[error("invalid type specifier `{0:?}`" )] |
27 | InvalidTypeSpecifier(String), |
28 | #[error("unable to parse given integer" )] |
29 | InvalidInteger(#[from] std::num::ParseIntError), |
30 | #[error("invalid array specifier (missing length)" )] |
31 | InvalidArraySpecifierMissingLength, |
32 | #[error("invalid array specifier (missing `]`" )] |
33 | InvalidArraySpecifierMissingBracket, |
34 | #[error("trailing data after bitfield range" )] |
35 | TrailingDataAfterBitfieldRange, |
36 | #[error("malformed format string (missing display hint after ':')" )] |
37 | MalformedFormatString, |
38 | #[error("unknown display hint: {0:?}" )] |
39 | UnknownDisplayHint(String), |
40 | #[error("unexpected content `{0:?}` in format string" )] |
41 | UnexpectedContentInFormatString(String), |
42 | #[error("unmatched `{{` in format string" )] |
43 | UnmatchedOpenBracket, |
44 | #[error("unmatched `}}` in format string" )] |
45 | UnmatchedCloseBracket, |
46 | #[error("conflicting types for argument {0}: used as {1:?} and {2:?}" )] |
47 | ConflictingTypes(usize, Type, Type), |
48 | #[error("argument {0} is not used in this format string" )] |
49 | UnusedArgument(usize), |
50 | } |
51 | |
52 | /// A parameter of the form `{{0=Type:hint}}` in a format string. |
53 | #[derive (Clone, Debug, Eq, PartialEq)] |
54 | pub struct Parameter { |
55 | /// The argument index to display at this position. |
56 | pub index: usize, |
57 | /// The type of the argument to display, e.g. '=u8', '=bool'. |
58 | pub ty: Type, |
59 | /// The display hint, e.g. ':x', ':b', ':a'. |
60 | pub hint: Option<DisplayHint>, |
61 | } |
62 | |
63 | /// A part of a format string. |
64 | #[derive (Clone, Debug, Eq, PartialEq)] |
65 | pub enum Fragment<'f> { |
66 | /// A literal string (eg. `"literal "` in `"literal {:?}"`). |
67 | Literal(Cow<'f, str>), |
68 | |
69 | /// A format parameter. |
70 | Parameter(Parameter), |
71 | } |
72 | |
73 | /// A parsed formatting parameter (contents of `{` `}` block). |
74 | /// |
75 | /// # Syntax |
76 | /// |
77 | /// ```notrust |
78 | /// param := '{' [ argument ] [ '=' argtype ] [ ':' format_spec ] '}' |
79 | /// argument := integer |
80 | /// |
81 | /// argtype := bitfield | '?' | format-array | '[?]' | byte-array | '[u8]' | 'istr' | 'str' | |
82 | /// 'bool' | 'char' | 'u8' | 'u16' | 'u32' | 'u64' | 'u128' | 'usize' | 'i8' | 'i16' | 'i32' | |
83 | /// 'i64' | 'i128 | 'isize' | 'f32' | 'f64' |
84 | /// bitfield := integer '..' integer |
85 | /// format-array := '[?;' spaces integer ']' |
86 | /// byte-array := '[u8;' spaces integer ']' |
87 | /// spaces := ' '* |
88 | /// |
89 | /// format_spec := [ zero_pad ] type |
90 | /// zero_pad := '0' integer |
91 | /// type := 'a' | 'b' | 'o' | 'x' | 'X' | '?' | 'us' |
92 | /// ``` |
93 | #[derive (Debug, PartialEq)] |
94 | struct Param { |
95 | index: Option<usize>, |
96 | ty: Type, |
97 | hint: Option<DisplayHint>, |
98 | } |
99 | |
100 | /// The log level |
101 | #[derive (Clone, Copy, Debug, Eq, PartialEq, PartialOrd)] |
102 | pub enum Level { |
103 | Trace, |
104 | Debug, |
105 | Info, |
106 | Warn, |
107 | Error, |
108 | } |
109 | |
110 | impl Level { |
111 | pub fn as_str(self) -> &'static str { |
112 | match self { |
113 | Level::Trace => "trace" , |
114 | Level::Debug => "debug" , |
115 | Level::Info => "info" , |
116 | Level::Warn => "warn" , |
117 | Level::Error => "error" , |
118 | } |
119 | } |
120 | } |
121 | |
122 | fn parse_range(mut s: &str) -> Option<(Range<u8>, usize /* consumed */)> { |
123 | // consume first number |
124 | let start_digits = s |
125 | .as_bytes() |
126 | .iter() |
127 | .take_while(|b| (**b as char).is_ascii_digit()) |
128 | .count(); |
129 | let start = s[..start_digits].parse().ok()?; |
130 | |
131 | // next two `char`s should be `..` |
132 | if &s[start_digits..start_digits + 2] != ".." { |
133 | return None; |
134 | } |
135 | s = &s[start_digits + 2..]; |
136 | |
137 | // consume second number |
138 | let end_digits = s |
139 | .as_bytes() |
140 | .iter() |
141 | .take_while(|b| (**b as char).is_ascii_digit()) |
142 | .count(); |
143 | let end = s[..end_digits].parse().ok()?; |
144 | |
145 | // check for faulty state |
146 | if end <= start || start >= 128 || end > 128 { |
147 | return None; |
148 | } |
149 | |
150 | Some((start..end, start_digits + end_digits + 2)) |
151 | } |
152 | |
153 | /// Parse and consume an array at the beginning of `s`. |
154 | /// |
155 | /// Return the length of the array. |
156 | fn parse_array(mut s: &str) -> Result<usize, Error> { |
157 | // skip spaces |
158 | let len_pos: usize = s |
159 | .find(|c: char| c != ' ' ) |
160 | .ok_or(err:Error::InvalidArraySpecifierMissingLength)?; |
161 | s = &s[len_pos..]; |
162 | |
163 | // consume length |
164 | let after_len: usize = s |
165 | .find(|c: char| !c.is_ascii_digit()) |
166 | .ok_or(err:Error::InvalidArraySpecifierMissingBracket)?; |
167 | let len: usize = s[..after_len].parse::<usize>()?; |
168 | s = &s[after_len..]; |
169 | |
170 | // consume final `]` |
171 | if s != "]" { |
172 | return Err(Error::InvalidArraySpecifierMissingBracket); |
173 | } |
174 | |
175 | Ok(len) |
176 | } |
177 | |
178 | /// Parser mode |
179 | #[derive (Clone, Copy, Debug, Eq, PartialEq)] |
180 | pub enum ParserMode { |
181 | /// Rejects unknown display hints |
182 | Strict, |
183 | /// Accepts unknown display hints |
184 | ForwardsCompatible, |
185 | } |
186 | |
187 | /// Parse `Param` from `&str` |
188 | /// |
189 | /// * example `input`: `0=Type:hint` (note: no curly braces) |
190 | fn parse_param(mut input: &str, mode: ParserMode) -> Result<Param, Error> { |
191 | const TYPE_PREFIX: &str = "=" ; |
192 | const HINT_PREFIX: &str = ":" ; |
193 | |
194 | // First, optional argument index. |
195 | let mut index = None; |
196 | let index_end = input |
197 | .find(|c: char| !c.is_ascii_digit()) |
198 | .unwrap_or(input.len()); |
199 | |
200 | if index_end != 0 { |
201 | index = Some(input[..index_end].parse::<usize>()?); |
202 | } |
203 | |
204 | // Then, optional type |
205 | let mut ty = Type::default(); // when no explicit type; use the default one |
206 | input = &input[index_end..]; |
207 | |
208 | if input.starts_with(TYPE_PREFIX) { |
209 | // skip the prefix |
210 | input = &input[TYPE_PREFIX.len()..]; |
211 | |
212 | // type is delimited by `HINT_PREFIX` or end-of-string |
213 | let type_end = input.find(HINT_PREFIX).unwrap_or(input.len()); |
214 | let type_fragment = &input[..type_end]; |
215 | |
216 | const FORMAT_ARRAY_START: &str = "[?;" ; |
217 | const U8_ARRAY_START: &str = "[u8;" ; |
218 | |
219 | // what comes next is the type |
220 | ty = if let Ok(ty) = type_fragment.parse() { |
221 | Ok(ty) |
222 | } else if let Some(s) = type_fragment.strip_prefix(U8_ARRAY_START) { |
223 | Ok(Type::U8Array(parse_array(s)?)) |
224 | } else if let Some(s) = type_fragment.strip_prefix(FORMAT_ARRAY_START) { |
225 | Ok(Type::FormatArray(parse_array(s)?)) |
226 | } else if let Some((range, used)) = parse_range(type_fragment) { |
227 | // Check for bitfield syntax. |
228 | match used != type_fragment.len() { |
229 | true => Err(Error::TrailingDataAfterBitfieldRange), |
230 | false => Ok(Type::BitField(range)), |
231 | } |
232 | } else { |
233 | Err(Error::InvalidTypeSpecifier(input.to_owned())) |
234 | }?; |
235 | |
236 | input = &input[type_end..]; |
237 | } |
238 | |
239 | // Then, optional hint |
240 | let mut hint = None; |
241 | |
242 | if input.starts_with(HINT_PREFIX) { |
243 | // skip the prefix |
244 | input = &input[HINT_PREFIX.len()..]; |
245 | if input.is_empty() { |
246 | return Err(Error::MalformedFormatString); |
247 | } |
248 | |
249 | hint = match (DisplayHint::parse(input), mode) { |
250 | (Some(a), _) => Some(a), |
251 | (None, ParserMode::Strict) => return Err(Error::UnknownDisplayHint(input.to_owned())), |
252 | (None, ParserMode::ForwardsCompatible) => Some(DisplayHint::Unknown(input.to_owned())), |
253 | }; |
254 | } else if !input.is_empty() { |
255 | return Err(Error::UnexpectedContentInFormatString(input.to_owned())); |
256 | } |
257 | |
258 | Ok(Param { index, ty, hint }) |
259 | } |
260 | |
261 | fn push_literal<'f>(frag: &mut Vec<Fragment<'f>>, unescaped_literal: &'f str) -> Result<(), Error> { |
262 | // Replace `{{` with `{` and `}}` with `}`. Single braces are errors. |
263 | |
264 | // Scan for single braces first. The rest is trivial. |
265 | let mut last_open = false; |
266 | let mut last_close = false; |
267 | for c in unescaped_literal.chars() { |
268 | match c { |
269 | '{' => last_open = !last_open, |
270 | '}' => last_close = !last_close, |
271 | _ if last_open => return Err(Error::UnmatchedOpenBracket), |
272 | _ if last_close => return Err(Error::UnmatchedCloseBracket), |
273 | _ => {} |
274 | } |
275 | } |
276 | |
277 | // Handle trailing unescaped `{` or `}`. |
278 | if last_open { |
279 | return Err(Error::UnmatchedOpenBracket); |
280 | } else if last_close { |
281 | return Err(Error::UnmatchedCloseBracket); |
282 | } |
283 | |
284 | // FIXME: This always allocates a `String`, so the `Cow` is useless. |
285 | let literal = unescaped_literal.replace("{{" , "{" ).replace("}}" , "}" ); |
286 | frag.push(Fragment::Literal(literal.into())); |
287 | Ok(()) |
288 | } |
289 | |
290 | /// Returns `Some(smallest_bit_index, largest_bit_index)` contained in `params` if |
291 | /// `params` contains any bitfields. Otherwise `None`. |
292 | pub fn get_max_bitfield_range<'a, I>(params: I) -> Option<(u8, u8)> |
293 | where |
294 | I: Iterator<Item = &'a Parameter> + Clone, |
295 | { |
296 | let largest_bit_index: Option = paramsimpl Iterator |
297 | .clone() |
298 | .map(|param: &'a Parameter| match ¶m.ty { |
299 | Type::BitField(range: &Range) => range.end, |
300 | _ => unreachable!(), |
301 | }) |
302 | .max(); |
303 | |
304 | let smallest_bit_index: Option = paramsimpl Iterator |
305 | .map(|param: &'a Parameter| match ¶m.ty { |
306 | Type::BitField(range: &Range) => range.start, |
307 | _ => unreachable!(), |
308 | }) |
309 | .min(); |
310 | |
311 | match (smallest_bit_index, largest_bit_index) { |
312 | (Some(smallest: u8), Some(largest: u8)) => Some((smallest, largest)), |
313 | (None, None) => None, |
314 | _ => unreachable!(), |
315 | } |
316 | } |
317 | |
318 | pub fn parse(format_string: &str, mode: ParserMode) -> Result<Vec<Fragment<'_>>, Error> { |
319 | let mut fragments = Vec::new(); |
320 | |
321 | // Index after the `}` of the last format specifier. |
322 | let mut end_pos = 0; |
323 | |
324 | // Next argument index assigned to a parameter without an explicit one. |
325 | let mut next_arg_index = 0; |
326 | |
327 | let mut chars = format_string.char_indices(); |
328 | while let Some((brace_pos, ch)) = chars.next() { |
329 | if ch != '{' { |
330 | // Part of a literal fragment. |
331 | continue; |
332 | } |
333 | |
334 | // Peek at the next char. |
335 | if chars.as_str().starts_with('{' ) { |
336 | // Escaped `{{`, also part of a literal fragment. |
337 | chars.next(); // Move after both `{`s. |
338 | continue; |
339 | } |
340 | |
341 | if brace_pos > end_pos { |
342 | // There's a literal fragment with at least 1 character before this parameter fragment. |
343 | let unescaped_literal = &format_string[end_pos..brace_pos]; |
344 | push_literal(&mut fragments, unescaped_literal)?; |
345 | } |
346 | |
347 | // Else, this is a format specifier. It ends at the next `}`. |
348 | let len = chars |
349 | .as_str() |
350 | .find('}' ) |
351 | .ok_or(Error::UnmatchedOpenBracket)?; |
352 | end_pos = brace_pos + 1 + len + 1; |
353 | |
354 | // Parse the contents inside the braces. |
355 | let param_str = &format_string[brace_pos + 1..][..len]; |
356 | let param = parse_param(param_str, mode)?; |
357 | fragments.push(Fragment::Parameter(Parameter { |
358 | index: param.index.unwrap_or_else(|| { |
359 | // If there is no explicit index, assign the next one. |
360 | let idx = next_arg_index; |
361 | next_arg_index += 1; |
362 | idx |
363 | }), |
364 | ty: param.ty, |
365 | hint: param.hint, |
366 | })); |
367 | } |
368 | |
369 | // Trailing literal. |
370 | if end_pos != format_string.len() { |
371 | push_literal(&mut fragments, &format_string[end_pos..])?; |
372 | } |
373 | |
374 | // Check for argument type conflicts. |
375 | let mut args = Vec::new(); |
376 | for frag in &fragments { |
377 | if let Fragment::Parameter(Parameter { index, ty, .. }) = frag { |
378 | if args.len() <= *index { |
379 | args.resize(*index + 1, None); |
380 | } |
381 | |
382 | match &args[*index] { |
383 | None => args[*index] = Some(ty.clone()), |
384 | Some(other_ty) => match (other_ty, ty) { |
385 | (Type::BitField(_), Type::BitField(_)) => {} // FIXME: Bitfield range shouldn't be part of the type. |
386 | (a, b) if a != b => { |
387 | return Err(Error::ConflictingTypes(*index, a.clone(), b.clone())) |
388 | } |
389 | _ => {} |
390 | }, |
391 | } |
392 | } |
393 | } |
394 | |
395 | // Check that argument indices are dense (all arguments must be used). |
396 | for (index, arg) in args.iter().enumerate() { |
397 | if arg.is_none() { |
398 | return Err(Error::UnusedArgument(index)); |
399 | } |
400 | } |
401 | |
402 | Ok(fragments) |
403 | } |
404 | |