kbvm/xkb/compose/
parser.rs

1pub(crate) mod error;
2#[cfg(test)]
3mod tests;
4
5use {
6    crate::{
7        Keysym, ModifierMask,
8        xkb::{
9            code::Code,
10            code_map::CodeMap,
11            code_slice::CodeSlice,
12            compose::{
13                parser::error::{Expected, LHS, ParserError},
14                token::Token,
15            },
16            context::Environment,
17            diagnostic::{DiagnosticKind, DiagnosticSink},
18            interner::Interner,
19            meaning::{Meaning, MeaningCache},
20            span::{SpanExt, SpanUnit, Spanned},
21        },
22    },
23    isnt::std_1::primitive::IsntSliceExt,
24    kbvm_proc::ad_hoc_display,
25    std::sync::Arc,
26};
27
28struct Parser<'a, 'b, 'c> {
29    map: &'a mut CodeMap,
30    diagnostics: &'a mut DiagnosticSink<'b, 'c>,
31    tokens: &'a [Spanned<Token>],
32    interner: &'a mut Interner,
33    meaning_cache: &'a mut MeaningCache,
34    pos: usize,
35    env: &'a Environment,
36    locale_file: Option<&'a str>,
37}
38
39#[derive(Debug)]
40pub(crate) enum Line {
41    Include(CodeSlice<'static>),
42    Production(Production),
43}
44
45#[derive(Debug)]
46pub(crate) struct Production {
47    pub(crate) steps: Vec<Step>,
48    pub(crate) string: Option<CodeSlice<'static>>,
49    pub(crate) keysym: Option<Keysym>,
50}
51
52#[derive(Copy, Clone, Debug, PartialEq, Eq, PartialOrd, Ord)]
53pub(crate) struct Step {
54    pub(crate) keysym: Keysym,
55}
56
57pub(crate) fn parse_line(
58    map: &mut CodeMap,
59    diagnostics: &mut DiagnosticSink,
60    interner: &mut Interner,
61    meaning_cache: &mut MeaningCache,
62    tokens: &[Spanned<Token>],
63    env: &Environment,
64    locale_file: Option<&str>,
65) -> Result<Option<Spanned<Line>>, Spanned<ParserError>> {
66    let line = Parser {
67        map,
68        diagnostics,
69        tokens,
70        interner,
71        meaning_cache,
72        pos: 0,
73        env,
74        locale_file,
75    }
76    .parse_line()?;
77    let lo = tokens.first().map(|t| t.span.lo).unwrap_or_default();
78    let hi = tokens.last().map(|t| t.span.hi).unwrap_or_default();
79    Ok(line.map(|l| l.spanned(lo, hi)))
80}
81
82impl Parser<'_, '_, '_> {
83    fn next(
84        &mut self,
85        expected: &'static [Expected],
86    ) -> Result<Spanned<Token>, Spanned<ParserError>> {
87        if self.pos >= self.tokens.len() {
88            return Err(self.expected_but_eof(self.tokens.last().unwrap().span, expected));
89        }
90        let token = self.tokens[self.pos];
91        self.pos += 1;
92        Ok(token)
93    }
94
95    fn try_peek(&mut self) -> Option<Spanned<Token>> {
96        self.tokens.get(self.pos).copied()
97    }
98
99    fn peek(
100        &mut self,
101        expected: &'static [Expected],
102    ) -> Result<Spanned<Token>, Spanned<ParserError>> {
103        if self.pos >= self.tokens.len() {
104            return Err(self.expected_but_eof(self.tokens.last().unwrap().span, expected));
105        }
106        Ok(self.tokens[self.pos])
107    }
108
109    fn parse_line(&mut self) -> Result<Option<Line>, Spanned<ParserError>> {
110        let t = self.peek(LHS)?;
111        if let Token::Ident(i) = t.val {
112            let meaning = self.meaning_cache.get_case_insensitive(self.interner, i);
113            if meaning == Meaning::Include {
114                self.pos += 1;
115                return self.parse_include();
116            }
117        }
118        self.parse_production()
119    }
120
121    fn parse_include(&mut self) -> Result<Option<Line>, Spanned<ParserError>> {
122        let t = self.next(&[Expected::AnyString])?;
123        let Token::String(s) = t.val else {
124            return Err(self.unexpected_token(&[Expected::AnyString], t));
125        };
126        let mut bytes = self.interner.get(s).to_owned();
127        if bytes.contains(&b'%') {
128            let mut out = vec![];
129            let mut last_was_percent = false;
130            for (offset, &b) in bytes.as_bytes().iter().enumerate() {
131                if last_was_percent {
132                    last_was_percent = false;
133                    let lo = t.span.lo + offset as SpanUnit;
134                    let hi = lo + 2;
135                    match b {
136                        b'%' => out.push(b'%'),
137                        b'H' => {
138                            if let Some(home) = &self.env.home {
139                                out.extend_from_slice(home.as_bytes());
140                            } else {
141                                self.diagnostics.push(
142                                    self.map,
143                                    DiagnosticKind::HomeNotSet,
144                                    ad_hoc_display!("$HOME is not set").spanned(lo, hi),
145                                );
146                                return Ok(None);
147                            }
148                        }
149                        b'L' => {
150                            if let Some(dir) = &self.locale_file {
151                                out.extend_from_slice(dir.as_bytes());
152                            } else {
153                                self.diagnostics.push(
154                                    self.map,
155                                    DiagnosticKind::LocaleComposeFileNotResolved,
156                                    ad_hoc_display!("%L could not be resolved").spanned(lo, hi),
157                                );
158                                return Ok(None);
159                            }
160                        }
161                        b'S' => {
162                            out.extend_from_slice(self.env.xlocaledir.as_bytes());
163                        }
164                        _ => {
165                            self.diagnostics.push(
166                                self.map,
167                                DiagnosticKind::UnknownComposeIncludeEscape,
168                                ad_hoc_display!("unknown escape sequence %{}", b as char => char)
169                                    .spanned(lo, hi),
170                            );
171                            return Ok(None);
172                        }
173                    }
174                } else if b == b'%' {
175                    last_was_percent = true;
176                } else {
177                    out.push(b);
178                }
179            }
180            bytes = Code::new(&Arc::new(out)).to_slice().to_owned();
181        }
182        if let Some(next) = self.try_peek() {
183            return Err(self.expected_eol(next));
184        }
185        Ok(Some(Line::Include(bytes)))
186    }
187
188    fn parse_production(&mut self) -> Result<Option<Line>, Spanned<ParserError>> {
189        let mut steps = vec![];
190        loop {
191            let (mask, _mods) = self.parse_modifiers()?;
192            let token = self.next(LHS)?;
193            if token.val == token![:] && mask.0 == 0 && steps.is_not_empty() {
194                break;
195            }
196            let Token::Keysym(ks) = token.val else {
197                return Err(self.unexpected_token(&[Expected::AnyKeysym], token));
198            };
199            let ks = self.interner.get(ks);
200            let Some(ks) = Keysym::from_str(ks) else {
201                return Err(self.unknown_keysym(ks, token.span));
202            };
203            steps.push(Step { keysym: ks });
204        }
205        let mut string = None;
206        let mut keysym = None;
207        let mut t = self.next(&[Expected::AnyString, Expected::AnyIdent])?;
208        'rhs: {
209            if let Token::String(s) = t.val {
210                string = Some(self.interner.get(s).to_owned());
211                match self.try_peek() {
212                    Some(n) => {
213                        self.pos += 1;
214                        t = n
215                    }
216                    _ => break 'rhs,
217                }
218            }
219            let Token::Ident(ks) = t.val else {
220                return Err(self.unexpected_token(&[Expected::AnyIdent], t));
221            };
222            let ks = self.interner.get(ks);
223            let Some(ks) = Keysym::from_str(ks) else {
224                return Err(self.unknown_keysym(ks, t.span));
225            };
226            keysym = Some(ks);
227        }
228        if let Some(next) = self.try_peek() {
229            return Err(self.expected_eol(next));
230        }
231        Ok(Some(Line::Production(Production {
232            steps,
233            string,
234            keysym,
235        })))
236    }
237
238    fn parse_modifiers(&mut self) -> Result<(ModifierMask, ModifierMask), Spanned<ParserError>> {
239        let Some(t) = self.try_peek() else {
240            return Ok((ModifierMask(0), ModifierMask(0)));
241        };
242        let mut mask = ModifierMask(0);
243        if let Token::Ident(i) = t.val {
244            let meaning = self.meaning_cache.get_case_insensitive(self.interner, i);
245            if meaning == Meaning::None {
246                self.pos += 1;
247                return Ok((ModifierMask(!0), ModifierMask(0)));
248            }
249        }
250        if t.val == token![!] {
251            mask.0 = !0;
252            self.pos += 1;
253        }
254        let mut mods = ModifierMask(0);
255        while let Some(mut t) = self.try_peek() {
256            let mut tilde = false;
257            if t.val == token![~] {
258                self.pos += 1;
259                tilde = true;
260                t = self.peek(&[Expected::AnyModifier])?;
261            }
262            if let Token::Ident(i) = t.val {
263                let meaning = self.meaning_cache.get_case_insensitive(self.interner, i);
264                let mm = match meaning {
265                    Meaning::Ctrl => ModifierMask::CONTROL,
266                    Meaning::Caps | Meaning::Lock => ModifierMask::LOCK,
267                    Meaning::Shift => ModifierMask::SHIFT,
268                    Meaning::Alt | Meaning::Meta => ModifierMask::MOD1,
269                    _ => return Err(self.unexpected_token(&[Expected::AnyModifier], t)),
270                };
271                mask |= mm;
272                if tilde {
273                    mods &= !mm;
274                } else {
275                    mods |= mm;
276                }
277                self.pos += 1;
278            } else if tilde {
279                return Err(self.unexpected_token(&[Expected::AnyModifier], t));
280            } else {
281                break;
282            }
283        }
284        Ok((mask, mods))
285    }
286}