workshop_rs/frontend/
parser.rs

1//! Native localized Workshop parser.
2//!
3//! Parses vanilla Workshop text directly into validated, locale-independent
4//! the canonical public [`Program`]. Localized actions, values, events, enums, and structural
5//! keywords resolve through the canonical catalog; malformed input,
6//! unknown spellings, and recognized-but-unsupported constructs are reported
7//! as distinct structured diagnostics with source spans.
8
9pub(crate) use std::collections::HashMap;
10use std::sync::OnceLock;
11
12pub(crate) use crate::core::signatures::ExpectedDomain;
13pub(crate) use crate::core::source::{Position, SourceFile, Span};
14use crate::program::Program;
15pub(crate) use crate::settings::table::KeyKind;
16pub(crate) use crate::settings::{PathPart, table};
17pub(crate) use crate::settings::{Settings, SettingsListElement, SettingsNode};
18pub(crate) use crate::wir::{
19    self, Action, Event, EventTarget, EventTeam, ModifyOp, PlayerEventKind, Value, ValueNode,
20};
21
22pub(crate) use super::lexer::{Token, TokenKind, tokenize};
23pub(crate) use crate::catalog::{Catalog, Kind, Locale};
24pub(crate) use crate::core::error::{Result, WorkshopError};
25
26/// Where action parsing stopped.
27#[derive(Debug, Clone, Copy, PartialEq, Eq)]
28pub(crate) enum Stop {
29    /// The enclosing `}` was consumed.
30    SectionClosed,
31    /// The `End` keyword is next (not consumed).
32    End,
33    /// The `Else If` keyword is next (not consumed).
34    ElseIf,
35    /// The `Else` keyword is next (not consumed).
36    Else,
37}
38
39pub(crate) enum AssignmentOperator {
40    Set,
41    Modify(ModifyOp),
42}
43
44/// Parse localized Workshop text into Workshop IR using the catalog's
45/// canonical call-signature context. Ambiguous bare enum members resolve when
46/// their enclosing call pins one matching domain; unpinned ambiguity remains a
47/// structured canonical value. See [`parse_with_context`] when a
48/// consumer needs to provide additional signature context.
49pub fn parse(input: &str, catalog: &Catalog, locale: &Locale) -> Result<Program> {
50    parse_with_context(input, catalog, locale, catalog)
51}
52
53#[cfg(test)]
54pub(crate) fn parse_wir(input: &str, catalog: &Catalog, locale: &Locale) -> Result<wir::Program> {
55    parse_wir_with_context(input, catalog, locale, catalog)
56}
57
58/// Compatibility entry point for crate-internal storage tests and migration
59/// checks. Ordinary consumers should use [`parse`], which returns [`Program`].
60/// Parse localized Workshop text into Workshop IR, resolving ambiguous bare
61/// enum members from the enclosing call's canonical signature context (#111).
62///
63/// When a bare member spelling matches several enum domains, the parser asks
64/// [`ExpectedDomain::expected_domain`] for the domain the enclosing call's
65/// signature expects at that argument position; the member resolves only when
66/// that expected domain is one of the matching domains (i.e. the signature
67/// pins exactly one). Without a pin all matching candidates are preserved.
68pub fn parse_with_context(
69    input: &str,
70    catalog: &Catalog,
71    locale: &Locale,
72    context: &dyn ExpectedDomain,
73) -> Result<Program> {
74    parse_wir_with_context(input, catalog, locale, context).and_then(Program::from_wir)
75}
76
77pub(crate) fn parse_wir_with_context(
78    input: &str,
79    catalog: &Catalog,
80    locale: &Locale,
81    context: &dyn ExpectedDomain,
82) -> Result<wir::Program> {
83    let tokens = tokenize(input).map_err(|error| {
84        WorkshopError::malformed(error.message, Some(synthetic_span(error.position)))
85    })?;
86    ParseContext {
87        tokens,
88        pos: 0,
89        source: input,
90        catalog,
91        locale: locale.clone(),
92        context,
93        expected_domain: None,
94        condition_pending: false,
95        in_condition: false,
96        call_stack: Vec::new(),
97        target: wir::Program::default(),
98        globals: HashMap::new(),
99        players: HashMap::new(),
100        subroutines: HashMap::new(),
101        next_global_index: 0,
102        next_player_index: 0,
103    }
104    .program()
105}
106
107/// A synthetic single-position span (used before a file registry exists).
108pub(crate) fn synthetic_span(position: Position) -> Span {
109    Span::new(crate::core::ids::Id::from_index(0), position, position)
110}
111
112pub(crate) struct ParseContext<'a> {
113    pub(crate) tokens: Vec<Token>,
114    pub(crate) pos: usize,
115    pub(crate) source: &'a str,
116    pub(crate) catalog: &'a Catalog,
117    pub(crate) locale: Locale,
118    /// Canonical signature context (#111): supplies the expected enum domain
119    /// for the call argument currently being parsed.
120    pub(crate) context: &'a dyn ExpectedDomain,
121    /// The expected enum domain for the value currently being parsed, set by
122    /// [`ParseContext::value_args`] from the enclosing call's signature.
123    pub(crate) expected_domain: Option<&'a str>,
124    /// Marks the next [`ParseContext::value`] call as a rule-level condition
125    /// (`conditions` entries, `If`/`Else If`/`While` conditions) so an action
126    /// call there is diagnosed as an action used as a condition (#340).
127    pub(crate) condition_pending: bool,
128    /// Whether the value currently being parsed is that top-level condition
129    /// expression; nested arguments always parse as ordinary values.
130    pub(crate) in_condition: bool,
131    pub(crate) call_stack: Vec<String>,
132    pub(crate) target: wir::Program,
133    pub(crate) globals: HashMap<String, wir::GlobalVarId>,
134    pub(crate) players: HashMap<String, wir::PlayerVarId>,
135    pub(crate) subroutines: HashMap<String, wir::SubroutineId>,
136    pub(crate) next_global_index: u32,
137    pub(crate) next_player_index: u32,
138}
139
140impl<'a> ParseContext<'a> {
141    /// Run `f` with the signature-pinned expected domain temporarily replaced,
142    /// restoring the enclosing domain afterwards (#111).
143    pub(crate) fn with_domain<T>(
144        &mut self,
145        domain: Option<&'a str>,
146        f: impl FnOnce(&mut Self) -> Result<T>,
147    ) -> Result<T> {
148        let saved = self.expected_domain;
149        self.expected_domain = domain;
150        let result = f(self);
151        self.expected_domain = saved;
152        result
153    }
154
155    /// Parse a value while `domain` is in effect (or `None` for value
156    /// positions a signature must not pin).
157    pub(crate) fn value_in_domain(&mut self, domain: Option<&'a str>) -> Result<wir::ValueId> {
158        self.with_domain(domain, |p| p.value())
159    }
160
161    pub(crate) fn resolve_entry(
162        &self,
163        kind: Kind,
164        spelling: &str,
165    ) -> Option<&'a crate::catalog::CatalogEntry> {
166        self.catalog
167            .resolve(kind, &self.locale, spelling)
168            .or_else(|| {
169                if self.locale != *self.catalog.primary_locale() {
170                    self.catalog
171                        .resolve(kind, self.catalog.primary_locale(), spelling)
172                } else {
173                    None
174                }
175            })
176    }
177
178    pub(crate) fn canonical_keyword(&self, spelling: &str) -> String {
179        self.catalog
180            .resolve(Kind::Structural, &self.locale, spelling)
181            .or_else(|| {
182                (self.locale != *self.catalog.primary_locale()).then(|| {
183                    self.catalog
184                        .resolve(Kind::Structural, self.catalog.primary_locale(), spelling)
185                })?
186            })
187            .map(|entry| entry.id.clone())
188            .unwrap_or_else(|| canonical_keyword(spelling).to_string())
189    }
190
191    pub(crate) fn line_has_assignment(&self) -> bool {
192        let tokens: Vec<_> = self.tokens[self.pos..]
193            .iter()
194            .take_while(|token| !matches!(token.kind, TokenKind::Semi | TokenKind::RBrace))
195            .collect();
196        tokens.iter().any(|token| {
197            matches!(&token.kind, TokenKind::Op(op) if matches!(op.as_str(), "=" | "+=" | "-=" | "*=" | "/=" | "%="))
198        }) || tokens.windows(2).any(|window| {
199            matches!(&window[0].kind, TokenKind::Word(_))
200                && matches!(&window[1].kind, TokenKind::Op(op) if op == "=")
201        })
202    }
203
204    /// Read the maximal phrase of consecutive words (space-joined). Phrases
205    /// may span lines because long Workshop action arguments wrap mid-phrase.
206    pub(crate) fn phrase(&mut self) -> Result<(String, Position, Position)> {
207        let mut words = Vec::new();
208        let (start, mut end) = match self.peek() {
209            Some(Token {
210                kind: TokenKind::Word(word),
211                start,
212                end,
213            }) => {
214                words.push(word.clone());
215                (start, end)
216            }
217            Some(Token {
218                kind: TokenKind::Number { text, .. },
219                start,
220                end,
221            }) => {
222                words.push(text.clone());
223                (start, end)
224            }
225            Some(token) => return Err(self.malformed("expected an identifier", &token)),
226            None => return Err(self.malformed("expected an identifier", self.eof())),
227        };
228        self.pos += 1;
229        while let Some(token) = self.peek() {
230            match token {
231                Token {
232                    kind: TokenKind::Word(word),
233                    end: word_end,
234                    ..
235                } => {
236                    if matches!(
237                        self.peek_at(1).map(|token| token.kind),
238                        Some(TokenKind::Op(equal)) if equal == "="
239                    ) {
240                        break;
241                    }
242                    words.push(word.clone());
243                    end = word_end;
244                    self.pos += 1;
245                }
246                // Enum members embed numbers (`Team 2`, `Ability 2`); the
247                // lexer splits them from the word, so phrases join Number
248                // tokens too (#119 reference evidence).
249                Token {
250                    kind: TokenKind::Number { text, .. },
251                    end: number_end,
252                    ..
253                } => {
254                    words.push(text.clone());
255                    end = number_end;
256                    self.pos += 1;
257                }
258                _ => break,
259            }
260        }
261        Ok((words.join(" "), start, end))
262    }
263
264    /// Read a single-line phrase (stops at a line boundary). Used for names
265    /// that are structurally one per line, such as variable declarations.
266    pub(crate) fn phrase_on_line(&mut self) -> Result<(String, Position, Position)> {
267        self.phrase_on_line_impl(false)
268    }
269
270    /// Read a settings list element name: adjacent tokens stay joined and
271    /// `:` is part of the name, preserving localized spellings such as
272    /// `66号公路` and `Watchpoint: Gibraltar`.
273    pub(crate) fn settings_list_name(&mut self) -> Result<(String, Position, Position)> {
274        self.phrase_on_line_impl(true)
275    }
276
277    fn phrase_on_line_impl(&mut self, list_name: bool) -> Result<(String, Position, Position)> {
278        let mut words = Vec::new();
279        let (start, mut end, line) = match self.peek() {
280            Some(Token {
281                kind: TokenKind::Word(word),
282                start,
283                end,
284            }) => {
285                words.push(word.clone());
286                (start, end, start.line)
287            }
288            Some(Token {
289                kind: TokenKind::Number { text, .. },
290                start,
291                end,
292            }) => {
293                words.push(text.clone());
294                (start, end, start.line)
295            }
296            Some(token) => return Err(self.malformed("expected an identifier", &token)),
297            None => return Err(self.malformed("expected an identifier", self.eof())),
298        };
299        self.pos += 1;
300        while let Some(token) = self.peek() {
301            let (word, word_start, word_end) = match token {
302                Token {
303                    kind: TokenKind::Word(word),
304                    start,
305                    end,
306                } => (word, start, end),
307                Token {
308                    kind: TokenKind::Number { text, .. },
309                    start,
310                    end,
311                } => (text, start, end),
312                Token {
313                    kind: TokenKind::Dot,
314                    start,
315                    end,
316                } => (".".to_string(), start, end),
317                Token {
318                    kind: TokenKind::Op(op),
319                    start,
320                    end,
321                } if matches!(op.as_str(), "-" | "%") => (op.clone(), start, end),
322                Token {
323                    kind: TokenKind::Colon,
324                    start,
325                    end,
326                } if list_name => (":".to_string(), start, end),
327                _ => break,
328            };
329            if word_start.line != line {
330                break;
331            }
332            if list_name && word_start == end {
333                words
334                    .last_mut()
335                    .expect("phrase has a first token")
336                    .push_str(&word);
337            } else {
338                words.push(word);
339            }
340            end = word_end;
341            self.pos += 1;
342        }
343        let joined = words
344            .join(" ")
345            .replace(" .", ".")
346            .replace(". ", ".")
347            .replace(" %", "%");
348        // Canonical spellings keep the spacing around ':' (e.g. fr-FR
349        // "Observatoire : Gibraltar"); list names must not collapse it.
350        let joined = if list_name {
351            joined
352        } else {
353            joined.replace(" : ", ":")
354        };
355        Ok((joined, start, end))
356    }
357
358    pub(crate) fn phrase_on_line_with_colon(&mut self) -> Result<(String, Position, Position)> {
359        let (mut phrase, start, mut end) = self.phrase_on_line()?;
360        if matches!(
361            self.peek(),
362            Some(Token {
363                kind: TokenKind::Colon,
364                ..
365            })
366        ) {
367            let colon_pos = self.pos;
368            self.next();
369            let (rest, _, rest_end) = self.phrase_on_line()?;
370            if matches!(
371                self.peek(),
372                Some(Token {
373                    kind: TokenKind::LBrace,
374                    ..
375                })
376            ) {
377                phrase.push(':');
378                phrase.push(' ');
379                phrase.push_str(&rest);
380                end = rest_end;
381            } else {
382                self.pos = colon_pos;
383            }
384        }
385        Ok((phrase, start, end))
386    }
387
388    pub(crate) fn enum_member_phrase(&mut self) -> Result<(String, Position, Position)> {
389        let first = self
390            .peek()
391            .ok_or_else(|| self.malformed("expected an enum member", self.eof()))?;
392        let start = first.start;
393        let line = first.start.line;
394        let mut end = first.end;
395        let mut parts = Vec::new();
396        while let Some(token) = self.peek() {
397            if token.start.line != line
398                || matches!(token.kind, TokenKind::RParen | TokenKind::Comma)
399            {
400                break;
401            }
402            self.pos += 1;
403            end = token.end;
404            parts.push(raw_token_text(&token.kind));
405        }
406        if parts.is_empty() {
407            return Err(self.malformed("expected an enum member", &first));
408        }
409        Ok((
410            parts
411                .join(" ")
412                .replace(" : ", ":")
413                .replace(" .", ".")
414                .replace(". ", "."),
415            start,
416            end,
417        ))
418    }
419
420    /// Read a text line (tokens until `;`), joining words and dashes into
421    /// the literal text, and consume the terminating `;`. Returns the text
422    /// and the span covering the line's content tokens.
423    pub(crate) fn line_text(&mut self) -> Result<(String, Option<Span>)> {
424        let mut parts = Vec::new();
425        let mut start = None;
426        let mut end = None;
427        loop {
428            match self.peek() {
429                Some(Token {
430                    kind: TokenKind::Semi,
431                    ..
432                }) => {
433                    self.pos += 1;
434                    break;
435                }
436                Some(token) => {
437                    start.get_or_insert(token.start);
438                    end = Some(token.end);
439                    match &token.kind {
440                        TokenKind::Word(word) => parts.push(word.clone()),
441                        TokenKind::Op(op) if op == "-" => parts.push("-".to_string()),
442                        TokenKind::Number { value, .. } => parts.push(value.to_string()),
443                        TokenKind::Dot => parts.push(".".to_string()),
444                        TokenKind::Colon => parts.push(":".to_string()),
445                        _ => return Err(self.malformed("expected a text line", &token)),
446                    }
447                    self.pos += 1;
448                }
449                None => {
450                    return Err(self.malformed("unexpected end of input in line", self.eof()));
451                }
452            }
453        }
454        Ok((
455            parts
456                .join(" ")
457                .replace(" .", ".")
458                .replace(". ", ".")
459                .replace(" : ", ":"),
460            start
461                .zip(end)
462                .map(|(start, end)| Span::new(self.file(), start, end)),
463        ))
464    }
465
466    /// Consume a known keyword phrase, verifying its spelling.
467    pub(crate) fn consume_phrase(&mut self, expected: &str) -> Result<()> {
468        let (phrase, _, _) = self.phrase()?;
469        if phrase != expected {
470            return Err(self.malformed(&format!("expected '{expected}'"), self.previous()));
471        }
472        Ok(())
473    }
474
475    pub(crate) fn expect_keyword(&mut self, expected: &str) -> Result<Position> {
476        match self.next() {
477            Some(Token {
478                kind: TokenKind::Word(word),
479                start,
480                ..
481            }) if self.canonical_keyword(&word) == expected => Ok(start),
482            Some(token) => Err(self.malformed(&format!("expected '{expected}'"), &token)),
483            None => Err(self.malformed(&format!("expected '{expected}'"), self.eof())),
484        }
485    }
486
487    pub(crate) fn expect(&mut self, kind: TokenKind, message: &str) -> Result<()> {
488        match self.next() {
489            Some(token) if token.kind == kind => Ok(()),
490            Some(token) => Err(self.malformed(message, &token)),
491            None => Err(self.malformed(message, self.eof())),
492        }
493    }
494
495    /// Expect the `;` terminating a statement. When the token found instead
496    /// sits on a later line, the missing `;` is reported at the end of the
497    /// statement it should have closed rather than at that unrelated token
498    /// (#340); a same-line offender keeps pointing at the token itself.
499    pub(crate) fn expect_semi(&mut self, message: &str) -> Result<()> {
500        match self.peek() {
501            Some(Token {
502                kind: TokenKind::Semi,
503                ..
504            }) => {
505                self.pos += 1;
506                Ok(())
507            }
508            Some(token) if token.start.line > self.previous().end.line => {
509                let end = self.previous().end;
510                Err(WorkshopError::malformed(
511                    message.to_string(),
512                    Some(Span::new(self.file(), end, end)),
513                ))
514            }
515            Some(token) => Err(self.malformed(message, &token)),
516            None => Err(self.malformed(message, self.eof())),
517        }
518    }
519
520    pub(crate) fn expect_string(&mut self, message: &str) -> Result<String> {
521        match self.next() {
522            Some(Token {
523                kind: TokenKind::String(content),
524                ..
525            }) => Ok(content),
526            Some(token) => Err(self.malformed(message, &token)),
527            None => Err(self.malformed(message, self.eof())),
528        }
529    }
530
531    pub(crate) fn malformed(&self, message: &str, token: &Token) -> WorkshopError {
532        WorkshopError::malformed(
533            message.to_string(),
534            Some(Span::new(self.file(), token.start, token.end)),
535        )
536    }
537
538    pub(crate) fn unknown(&self, kind: &'static str, spelling: &str) -> WorkshopError {
539        WorkshopError::unknown(kind, spelling.to_string(), self.locale.clone(), None)
540    }
541
542    pub(crate) fn peek(&self) -> Option<Token> {
543        self.tokens.get(self.pos).cloned()
544    }
545
546    pub(crate) fn peek_at(&self, offset: usize) -> Option<Token> {
547        self.tokens.get(self.pos + offset).cloned()
548    }
549
550    pub(crate) fn next(&mut self) -> Option<Token> {
551        let token = self.tokens.get(self.pos).cloned();
552        if token.is_some() {
553            self.pos += 1;
554        }
555        token
556    }
557
558    pub(crate) fn previous(&self) -> &Token {
559        self.tokens
560            .get(self.pos.saturating_sub(1))
561            .unwrap_or_else(|| self.tokens.last().unwrap())
562    }
563
564    pub(crate) fn previous_span(&self) -> (Position, Position) {
565        let token = self.previous();
566        (token.start, token.end)
567    }
568
569    pub(crate) fn span_here(&self) -> (Position, Position) {
570        let token = self
571            .peek()
572            .unwrap_or_else(|| self.tokens.last().unwrap().clone());
573        (token.start, token.end)
574    }
575
576    pub(crate) fn eof(&self) -> &Token {
577        self.tokens.last().unwrap()
578    }
579
580    pub(crate) fn file(&self) -> crate::core::ids::Id<SourceFile> {
581        crate::core::ids::Id::from_index(0)
582    }
583}
584
585pub(crate) fn canonical_keyword(keyword: &str) -> &str {
586    static KEYWORDS: OnceLock<HashMap<String, String>> = OnceLock::new();
587    KEYWORDS
588        .get_or_init(|| {
589            serde_json::from_str(include_str!("structural_keywords.json"))
590                .expect("structural keyword data is valid JSON")
591        })
592        .get(keyword)
593        .map(String::as_str)
594        .unwrap_or(keyword)
595}
596
597pub(crate) fn raw_token_text(kind: &TokenKind) -> String {
598    match kind {
599        TokenKind::Word(value) => value.clone(),
600        TokenKind::Number { text, .. } => text.clone(),
601        TokenKind::String(value) => format!("\"{}\"", value.replace('"', "\\\"")),
602        TokenKind::Op(value) => value.clone(),
603        TokenKind::LParen => "(".to_string(),
604        TokenKind::RParen => ")".to_string(),
605        TokenKind::Comma => ",".to_string(),
606        TokenKind::Semi => ";".to_string(),
607        TokenKind::LBrace => "{".to_string(),
608        TokenKind::RBrace => "}".to_string(),
609        TokenKind::Colon => ":".to_string(),
610        TokenKind::Dot => ".".to_string(),
611        TokenKind::LBracket => "[".to_string(),
612        TokenKind::RBracket => "]".to_string(),
613        TokenKind::Eof => String::new(),
614    }
615}
workshop_rs/frontend/parser.rs

workshop_rs/frontend/
parser.rs
