harn_parser/
analysis.rs

1use std::collections::{HashMap, HashSet};
2use std::hash::{Hash, Hasher};
3use std::path::Path;
4
5use harn_lexer::{Lexer, LexerError, Token};
6
7use crate::InlayHintInfo;
8use crate::{Parser, ParserError, SNode, TypeChecker, TypeDiagnostic};
9
10/// Stable source identity used by the incremental analysis cache.
11#[derive(Debug, Clone, PartialEq, Eq, Hash)]
12pub struct SourceId(String);
13
14impl SourceId {
15    pub fn new(value: impl Into<String>) -> Self {
16        Self(value.into())
17    }
18
19    pub fn path(path: &Path) -> Self {
20        Self(path.to_string_lossy().into_owned())
21    }
22
23    pub fn as_str(&self) -> &str {
24        &self.0
25    }
26}
27
28/// Monotonic caller-owned version for a source input.
29#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash)]
30pub struct SourceVersion(pub u64);
31
32/// Deterministic content digest for a source input.
33#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
34pub struct SourceDigest(u64);
35
36impl SourceDigest {
37    pub fn from_source(source: &str) -> Self {
38        let mut hash = 0xcbf29ce484222325u64;
39        for byte in source.as_bytes() {
40            hash ^= u64::from(*byte);
41            hash = hash.wrapping_mul(0x100000001b3);
42        }
43        Self(hash)
44    }
45}
46
47#[derive(Debug, Clone, Copy, PartialEq, Eq)]
48pub enum SourceUpdate {
49    Inserted,
50    Changed,
51    Unchanged,
52}
53
54#[derive(Debug, Clone, Default, PartialEq, Eq)]
55pub struct AnalysisStats {
56    pub lex_runs: usize,
57    pub parse_runs: usize,
58    pub typecheck_runs: usize,
59}
60
61#[derive(Debug, Clone)]
62pub struct ParseOutput {
63    pub source: String,
64    pub program: Vec<SNode>,
65}
66
67#[derive(Debug, Clone)]
68pub struct TypeCheckOutput {
69    pub source: String,
70    pub program: Vec<SNode>,
71    pub diagnostics: Vec<TypeDiagnostic>,
72    pub inlay_hints: Vec<InlayHintInfo>,
73}
74
75#[derive(Debug, Clone)]
76pub enum AnalysisError {
77    MissingSource(SourceId),
78    Lex {
79        source: String,
80        error: LexerError,
81    },
82    Parse {
83        source: String,
84        errors: Vec<ParserError>,
85    },
86}
87
88impl AnalysisError {
89    pub fn source(&self) -> Option<&str> {
90        match self {
91            AnalysisError::MissingSource(_) => None,
92            AnalysisError::Lex { source, .. } | AnalysisError::Parse { source, .. } => Some(source),
93        }
94    }
95}
96
97#[derive(Debug, Clone, Default)]
98pub struct TypeCheckConfig {
99    pub strict_types: bool,
100    pub privileged_wire_builtins: bool,
101    pub imported_names: Option<HashSet<String>>,
102    pub imported_type_decls: Vec<SNode>,
103    pub imported_callable_decls: Vec<SNode>,
104    pub namespace_imports: Vec<(String, crate::NamespaceImportBinding)>,
105}
106
107impl TypeCheckConfig {
108    pub fn new() -> Self {
109        Self::default()
110    }
111
112    pub fn with_strict_types(mut self, strict_types: bool) -> Self {
113        self.strict_types = strict_types;
114        self
115    }
116
117    pub fn with_privileged_wire_builtins(mut self, enabled: bool) -> Self {
118        self.privileged_wire_builtins = enabled;
119        self
120    }
121
122    pub fn with_imported_names(mut self, imported_names: Option<HashSet<String>>) -> Self {
123        self.imported_names = imported_names;
124        self
125    }
126
127    pub fn with_imported_type_decls(mut self, imported_type_decls: Vec<SNode>) -> Self {
128        self.imported_type_decls = imported_type_decls;
129        self
130    }
131
132    pub fn with_imported_callable_decls(mut self, imported_callable_decls: Vec<SNode>) -> Self {
133        self.imported_callable_decls = imported_callable_decls;
134        self
135    }
136
137    pub fn with_namespace_imports(
138        mut self,
139        namespace_imports: Vec<(String, crate::NamespaceImportBinding)>,
140    ) -> Self {
141        self.namespace_imports = namespace_imports;
142        self
143    }
144
145    fn cache_key(&self) -> TypeCheckCacheKey {
146        let mut imported_names = self
147            .imported_names
148            .as_ref()
149            .map(|names| names.iter().cloned().collect::<Vec<_>>());
150        if let Some(names) = &mut imported_names {
151            names.sort();
152        }
153        let mut namespace_aliases: Vec<String> = self
154            .namespace_imports
155            .iter()
156            .map(|(alias, _)| alias.clone())
157            .collect();
158        namespace_aliases.sort();
159        TypeCheckCacheKey {
160            strict_types: self.strict_types,
161            privileged_wire_builtins: self.privileged_wire_builtins,
162            imported_names,
163            imported_type_decls_digest: debug_digest(&self.imported_type_decls),
164            imported_callable_decls_digest: debug_digest(&self.imported_callable_decls),
165            namespace_imports_digest: debug_digest(&namespace_aliases),
166        }
167    }
168
169    fn build_checker(&self) -> TypeChecker {
170        let mut checker = TypeChecker::with_strict_types(self.strict_types);
171        checker = checker.with_privileged_wire_builtins(self.privileged_wire_builtins);
172        if let Some(imported) = self.imported_names.clone() {
173            checker = checker.with_imported_names(imported);
174        }
175        if !self.imported_type_decls.is_empty() {
176            checker = checker.with_imported_type_decls(self.imported_type_decls.clone());
177        }
178        if !self.imported_callable_decls.is_empty() {
179            checker = checker.with_imported_callable_decls(self.imported_callable_decls.clone());
180        }
181        if !self.namespace_imports.is_empty() {
182            checker = checker.with_namespace_imports(self.namespace_imports.clone());
183        }
184        checker
185    }
186}
187
188#[derive(Debug, Clone, PartialEq, Eq, Hash)]
189struct TypeCheckCacheKey {
190    strict_types: bool,
191    privileged_wire_builtins: bool,
192    imported_names: Option<Vec<String>>,
193    imported_type_decls_digest: u64,
194    imported_callable_decls_digest: u64,
195    namespace_imports_digest: u64,
196}
197
198#[derive(Debug, Clone)]
199struct CachedTypeCheck {
200    diagnostics: Vec<TypeDiagnostic>,
201    inlay_hints: Vec<InlayHintInfo>,
202}
203
204#[derive(Debug, Clone)]
205struct SourceEntry {
206    source: String,
207    version: SourceVersion,
208    digest: SourceDigest,
209    tokens: Option<Result<Vec<Token>, LexerError>>,
210    program: Option<Result<Vec<SNode>, Vec<ParserError>>>,
211    typechecks: HashMap<TypeCheckCacheKey, CachedTypeCheck>,
212}
213
214impl SourceEntry {
215    fn new(source: String, version: SourceVersion, digest: SourceDigest) -> Self {
216        Self {
217            source,
218            version,
219            digest,
220            tokens: None,
221            program: None,
222            typechecks: HashMap::new(),
223        }
224    }
225
226    fn replace_source(&mut self, source: String, version: SourceVersion, digest: SourceDigest) {
227        self.source = source;
228        self.version = version;
229        self.digest = digest;
230        self.tokens = None;
231        self.program = None;
232        self.typechecks.clear();
233    }
234}
235
236/// Persistent query cache for lexing, parsing, and type checking Harn sources.
237#[derive(Debug, Default)]
238pub struct AnalysisDatabase {
239    entries: HashMap<SourceId, SourceEntry>,
240    stats: AnalysisStats,
241}
242
243impl AnalysisDatabase {
244    pub fn new() -> Self {
245        Self::default()
246    }
247
248    pub fn stats(&self) -> AnalysisStats {
249        self.stats.clone()
250    }
251
252    pub fn set_source(
253        &mut self,
254        id: SourceId,
255        source: String,
256        version: SourceVersion,
257    ) -> SourceUpdate {
258        let digest = SourceDigest::from_source(&source);
259        match self.entries.get_mut(&id) {
260            None => {
261                self.entries
262                    .insert(id, SourceEntry::new(source, version, digest));
263                SourceUpdate::Inserted
264            }
265            Some(entry) if entry.digest == digest => {
266                entry.version = version;
267                SourceUpdate::Unchanged
268            }
269            Some(entry) => {
270                entry.replace_source(source, version, digest);
271                SourceUpdate::Changed
272            }
273        }
274    }
275
276    pub fn set_parsed_source(
277        &mut self,
278        id: SourceId,
279        source: String,
280        version: SourceVersion,
281        program: Vec<SNode>,
282    ) -> SourceUpdate {
283        let digest = SourceDigest::from_source(&source);
284        match self.entries.get_mut(&id) {
285            None => {
286                let mut entry = SourceEntry::new(source, version, digest);
287                entry.program = Some(Ok(program));
288                self.entries.insert(id, entry);
289                SourceUpdate::Inserted
290            }
291            Some(entry) if entry.digest == digest => {
292                entry.version = version;
293                entry.program = Some(Ok(program));
294                SourceUpdate::Unchanged
295            }
296            Some(entry) => {
297                entry.replace_source(source, version, digest);
298                entry.program = Some(Ok(program));
299                SourceUpdate::Changed
300            }
301        }
302    }
303
304    pub fn parse(&mut self, id: &SourceId) -> Result<ParseOutput, AnalysisError> {
305        if let Some(entry) = self.entries.get(id) {
306            if let Some(program) = &entry.program {
307                return match program {
308                    Ok(program) => Ok(ParseOutput {
309                        source: entry.source.clone(),
310                        program: program.clone(),
311                    }),
312                    Err(errors) => Err(AnalysisError::Parse {
313                        source: entry.source.clone(),
314                        errors: errors.clone(),
315                    }),
316                };
317            }
318        }
319
320        let mut lexed = false;
321        let mut parsed_now = false;
322        let entry = self.entry_mut(id)?;
323        if entry.tokens.is_none() {
324            lexed = true;
325            let mut lexer = Lexer::new(&entry.source);
326            entry.tokens = Some(lexer.tokenize());
327        }
328        let tokens = match entry.tokens.as_ref().expect("tokens initialized") {
329            Ok(tokens) => tokens.clone(),
330            Err(error) => {
331                let source = entry.source.clone();
332                let error = error.clone();
333                if lexed {
334                    self.stats.lex_runs += 1;
335                }
336                return Err(AnalysisError::Lex { source, error });
337            }
338        };
339
340        if entry.program.is_none() {
341            parsed_now = true;
342            let mut parser = Parser::new(tokens);
343            entry.program = Some(match parser.parse() {
344                Ok(program) => Ok(program),
345                Err(error) => {
346                    let mut errors = parser.all_errors().to_vec();
347                    if errors.is_empty() {
348                        errors.push(error);
349                    }
350                    Err(errors)
351                }
352            });
353        }
354
355        let result = match entry.program.as_ref().expect("program initialized") {
356            Ok(program) => Ok(ParseOutput {
357                source: entry.source.clone(),
358                program: program.clone(),
359            }),
360            Err(errors) => Err(AnalysisError::Parse {
361                source: entry.source.clone(),
362                errors: errors.clone(),
363            }),
364        };
365        if lexed {
366            self.stats.lex_runs += 1;
367        }
368        if parsed_now {
369            self.stats.parse_runs += 1;
370        }
371        result
372    }
373
374    pub fn typecheck(
375        &mut self,
376        id: &SourceId,
377        config: TypeCheckConfig,
378    ) -> Result<TypeCheckOutput, AnalysisError> {
379        let parsed = self.parse(id)?;
380        let key = config.cache_key();
381        if let Some(cached) = self
382            .entries
383            .get(id)
384            .expect("parse verified source entry")
385            .typechecks
386            .get(&key)
387        {
388            return Ok(TypeCheckOutput {
389                source: parsed.source,
390                program: parsed.program,
391                diagnostics: cached.diagnostics.clone(),
392                inlay_hints: cached.inlay_hints.clone(),
393            });
394        }
395
396        self.stats.typecheck_runs += 1;
397        let (diagnostics, inlay_hints) = config
398            .build_checker()
399            .check_with_hints(&parsed.program, &parsed.source);
400        let cached = CachedTypeCheck {
401            diagnostics: diagnostics.clone(),
402            inlay_hints: inlay_hints.clone(),
403        };
404        self.entries
405            .get_mut(id)
406            .expect("parse verified source entry")
407            .typechecks
408            .insert(key, cached);
409        Ok(TypeCheckOutput {
410            source: parsed.source,
411            program: parsed.program,
412            diagnostics,
413            inlay_hints,
414        })
415    }
416
417    fn entry_mut(&mut self, id: &SourceId) -> Result<&mut SourceEntry, AnalysisError> {
418        self.entries
419            .get_mut(id)
420            .ok_or_else(|| AnalysisError::MissingSource(id.clone()))
421    }
422}
423
424fn debug_digest<T: std::fmt::Debug>(value: &T) -> u64 {
425    let mut hasher = StableHasher::default();
426    format!("{value:?}").hash(&mut hasher);
427    hasher.finish()
428}
429
430#[derive(Default)]
431struct StableHasher(u64);
432
433impl Hasher for StableHasher {
434    fn finish(&self) -> u64 {
435        self.0
436    }
437
438    fn write(&mut self, bytes: &[u8]) {
439        let mut hash = if self.0 == 0 {
440            0xcbf29ce484222325u64
441        } else {
442            self.0
443        };
444        for byte in bytes {
445            hash ^= u64::from(*byte);
446            hash = hash.wrapping_mul(0x100000001b3);
447        }
448        self.0 = hash;
449    }
450}
451
452#[cfg(test)]
453mod tests {
454    use super::*;
455    use crate::DiagnosticSeverity;
456
457    fn source_id() -> SourceId {
458        SourceId::new("test.harn")
459    }
460
461    #[test]
462    fn parse_reuses_cached_program_for_unchanged_source() {
463        let mut db = AnalysisDatabase::new();
464        let id = source_id();
465        assert_eq!(
466            db.set_source(id.clone(), "const x = 1\n".to_string(), SourceVersion(1)),
467            SourceUpdate::Inserted
468        );
469        db.parse(&id).expect("initial parse");
470        db.parse(&id).expect("cached parse");
471        assert_eq!(db.stats().lex_runs, 1);
472        assert_eq!(db.stats().parse_runs, 1);
473
474        assert_eq!(
475            db.set_source(id.clone(), "const x = 1\n".to_string(), SourceVersion(2)),
476            SourceUpdate::Unchanged
477        );
478        db.parse(&id).expect("same digest parse");
479        assert_eq!(db.stats().lex_runs, 1);
480        assert_eq!(db.stats().parse_runs, 1);
481    }
482
483    #[test]
484    fn source_change_invalidates_parse_and_typecheck_outputs() {
485        let mut db = AnalysisDatabase::new();
486        let id = source_id();
487        db.set_source(id.clone(), "const x = 1\n".to_string(), SourceVersion(1));
488        db.typecheck(&id, TypeCheckConfig::new())
489            .expect("initial check");
490        assert_eq!(
491            db.set_source(id.clone(), "const x = 2\n".to_string(), SourceVersion(2)),
492            SourceUpdate::Changed
493        );
494        db.typecheck(&id, TypeCheckConfig::new())
495            .expect("changed check");
496        assert_eq!(db.stats().lex_runs, 2);
497        assert_eq!(db.stats().parse_runs, 2);
498        assert_eq!(db.stats().typecheck_runs, 2);
499    }
500
501    #[test]
502    fn parsed_source_seed_skips_lex_and_parse() {
503        let mut db = AnalysisDatabase::new();
504        let id = source_id();
505        let source = "const x = 1\n".to_string();
506        let mut lexer = Lexer::new(&source);
507        let tokens = lexer.tokenize().expect("tokenize");
508        let mut parser = Parser::new(tokens);
509        let program = parser.parse().expect("parse");
510
511        assert_eq!(
512            db.set_parsed_source(id.clone(), source, SourceVersion(1), program),
513            SourceUpdate::Inserted
514        );
515        db.typecheck(&id, TypeCheckConfig::new())
516            .expect("seeded check");
517        assert_eq!(db.stats().lex_runs, 0);
518        assert_eq!(db.stats().parse_runs, 0);
519        assert_eq!(db.stats().typecheck_runs, 1);
520    }
521
522    #[test]
523    fn typecheck_cache_is_keyed_by_options() {
524        let mut db = AnalysisDatabase::new();
525        let id = source_id();
526        db.set_source(
527            id.clone(),
528            "pipeline main() {\n  const x = read_file(\"a\")\n  log(x.foo)\n}\n".to_string(),
529            SourceVersion(1),
530        );
531        db.typecheck(&id, TypeCheckConfig::new())
532            .expect("default check");
533        db.typecheck(&id, TypeCheckConfig::new())
534            .expect("cached default check");
535        db.typecheck(&id, TypeCheckConfig::new().with_strict_types(true))
536            .expect("strict check");
537        assert_eq!(db.stats().typecheck_runs, 2);
538    }
539
540    #[test]
541    fn typecheck_diagnostics_are_cached_with_hints() {
542        let mut db = AnalysisDatabase::new();
543        let id = source_id();
544        db.set_source(
545            id.clone(),
546            "pipeline main() {\n  const x: int = \"nope\"\n}\n".to_string(),
547            SourceVersion(1),
548        );
549        let first = db.typecheck(&id, TypeCheckConfig::new()).expect("check");
550        let second = db.typecheck(&id, TypeCheckConfig::new()).expect("cached");
551        assert!(first
552            .diagnostics
553            .iter()
554            .any(|diag| diag.severity == DiagnosticSeverity::Error));
555        assert_eq!(first.diagnostics.len(), second.diagnostics.len());
556        assert_eq!(db.stats().typecheck_runs, 1);
557    }
558}
harn_parser/analysis.rs

harn_parser/
analysis.rs
