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

workshop_rs/frontend/
parser.rs
