Skip to main content

blue_lang_runtime/
pipeline.rs

1//! The blue pipeline: **parse → check → erase → run**, in that order, once.
2//!
3//! The order is the whole reason this module exists. Each stage is available
4//! separately for tools that want one, but the *default* path is a single
5//! function, because two of the four orderings are silently wrong:
6//!
7//! - **Erase before check** discards every annotation, so a program with type
8//!   errors passes. The checker sees `(define …)` and has nothing to check.
9//! - **Run before check** reports a type error after the side effects.
10//!
11//! Neither fails loudly. Both produce a green run on a program that should
12//! have been rejected. Leaving the order to each caller means every caller
13//! is one reordering away from turning the type checker off — so the order
14//! lives here, and callers ask for a *result*, not a sequence of steps.
15
16use tatara_lisp::Sexp;
17use tatara_lisp_eval::Value;
18
19use blue_lang_check::{NameTable, Namespace};
20
21use crate::erase::erase_types;
22use crate::inputs::Inputs;
23use crate::uses::Entry;
24
25/// Why a run stopped short.
26#[derive(Debug, thiserror::Error)]
27pub enum RunError {
28    #[error("parse error: {0}")]
29    Parse(String),
30    /// The check stage rejected the program — a type error, an unbound name,
31    /// or any other error-severity rule in `blue_lang_check::RULES`. Carries
32    /// every error, not just the first: a caller fixing one wants to see the
33    /// rest. (The variant keeps its first name because it is public API on a
34    /// released crate.)
35    ///
36    /// Each one is already rendered `file:line:col: message` against the file
37    /// the offending form came from — which, in a program with imports, is
38    /// frequently not the file the user named. The rendering happens here
39    /// rather than at the consumer because here is where the file table exists;
40    /// see `uses::ResolvedProgram::locate`.
41    #[error("{} error(s) in the check stage:\n{}", .0.len(), .0.join("\n"))]
42    Types(Vec<String>),
43    /// **No longer reachable, and that is the point.** This reported "blue
44    /// emitted a tree the reader could not read back" — a failure only a
45    /// print-then-reparse hop could have. [`crate::lower_to_spanned`] deleted
46    /// the hop, so there is nothing left to fail: the tree the evaluator gets
47    /// IS the tree erasure produced, not a re-reading of its text.
48    ///
49    /// Kept rather than removed because it is public API on a released crate
50    /// and a consumer may still match on it, per ★★ MODULARIZE, DON'T DELETE.
51    /// It is retired, not orphaned — if a future stage ever serialises again
52    /// it has a typed home. **Nothing constructs it today**; do not read its
53    /// presence as evidence the pipeline can still fail this way.
54    #[error("the emitted tatara-lisp could not be read back: {0}")]
55    Lower(String),
56    /// The program ran and raised.
57    ///
58    /// Rendered `file:line:col: message`, then the blue call stack the error
59    /// unwound through, innermost first (`crate::observe::Fault`). The
60    /// position is resolved against the file of the innermost frame's
61    /// definition, so a raise inside a callee from another file names the
62    /// callee's file and line; before the stack existed that case could only
63    /// name the caller's file, with no position.
64    #[error("runtime error: {0}")]
65    Eval(Box<crate::observe::Fault>),
66    /// A `use("name")` could not be resolved.
67    ///
68    /// Its own variant rather than folded into `Parse`, because the reader's
69    /// next action is different: a parse error is in the source in front of
70    /// them, an import error is in their packaging — a missing bidama, a
71    /// BLUE_PATH that does not contain it, or no loader at all.
72    #[error("import error: {0}")]
73    Import(String),
74}
75
76/// What a run produced, plus what the checker did on the way.
77#[derive(Debug)]
78pub struct Run {
79    pub value: Value,
80    /// Nodes the type walk visited. Zero for a fully untyped program — this
81    /// is what makes "no annotations, no analysis" a *measurement* rather
82    /// than a claim.
83    pub visited: usize,
84    /// Declarations that carried an annotation.
85    pub typed_decls: usize,
86    /// Boundaries where typed code meets untyped code.
87    pub seams: usize,
88}
89
90/// Parse blue source to tatara-lisp forms.
91pub fn parse(src: &str) -> Result<Vec<Sexp>, RunError> {
92    parse_with_depth(src, blue_lang_syntax::MAX_EXPR_DEPTH)
93}
94
95/// [`parse`] with the parser's nesting bound supplied by the caller.
96///
97/// The bound exists so a stack overflow — which `catch_unwind` cannot catch —
98/// arrives as a typed `Err` instead. It is a *limit*, not a dialect: raising
99/// it changes no program's meaning, which is exactly why it is safe to expose
100/// as configuration (`blue-lang-cli`'s `config` module holds the rule).
101pub fn parse_with_depth(src: &str, max_depth: usize) -> Result<Vec<Sexp>, RunError> {
102    blue_lang_syntax::parse_program_with_depth(src, max_depth)
103        .map_err(|e| RunError::Parse(e.to_string()))
104}
105
106/// [`parse_with_depth`] keeping **every node's** source span.
107///
108/// The door for anything that will report a position to a human. It exists here,
109/// beside the spanless one, so a caller that wants spans still parses under the
110/// CONFIGURED nesting bound — a separate `blue_lang_syntax` call would be the
111/// second door `parse_with_depth`'s own docs exist to prevent, with
112/// `max_expr_depth` true of some subcommands and not others.
113pub fn parse_tree_with_depth(
114    src: &str,
115    max_depth: usize,
116) -> Result<Vec<blue_lang_syntax::Spanned>, RunError> {
117    blue_lang_syntax::parse_program_tree_with_depth(src, max_depth)
118        .map_err(|e| RunError::Parse(e.to_string()))
119}
120
121/// [`parse`] keeping **every node's** source span.
122///
123/// The spanned twin of [`parse`], at the same default bound — the door for
124/// anything downstream of a parse that will report a position, which since
125/// `use` learned to carry file identity is every path through
126/// [`run_in_surface`].
127pub fn parse_tree(src: &str) -> Result<Vec<blue_lang_syntax::Spanned>, RunError> {
128    parse_tree_with_depth(src, blue_lang_syntax::MAX_EXPR_DEPTH)
129}
130
131/// Run blue source with no build inputs.
132pub fn run(src: &str) -> Result<Run, RunError> {
133    run_with_inputs(src, Inputs::new())
134}
135
136/// Run blue source, giving the macro phase access to verified build inputs.
137///
138/// `inputs` is already verified — [`Inputs`] cannot hold bytes that do not match
139/// their declared hash — so nothing here re-checks. The capability a macro gains
140/// is exactly "these hashed bytes", never a path.
141pub fn run_with_inputs(src: &str, inputs: Inputs) -> Result<Run, RunError> {
142    run_with_loader(src, inputs, &crate::uses::NoLoader)
143}
144
145/// Run blue source with a loader, so `use("name")` can resolve.
146///
147/// Split from [`run_with_inputs`] rather than folded into it because loading a
148/// package reads a filesystem, and this crate has a `wasm32-unknown-unknown`
149/// consumer with zero host imports. The capability is injected by callers that
150/// have it — `blue_lang_pkg::LoadPath` is the real one — and absent by default,
151/// where a `use` is a typed error naming the package.
152pub fn run_with_loader(
153    src: &str,
154    inputs: Inputs,
155    loader: &dyn crate::uses::Loader,
156) -> Result<Run, RunError> {
157    run_in_surface(Entry::anonymous(src), inputs, loader, None)
158}
159
160/// Run blue source written in a `yakugo` surface.
161///
162/// The pack applies at PARSE time and nowhere else — by the time the checker
163/// sees the program it is canonical, so every stage below is identical whatever
164/// surface the author wrote in. That is what makes a surface a surface: it
165/// changes how a program is spelled and nothing about how it runs.
166///
167/// `entry` carries the source AND the file it was read from, because a type
168/// error has to be reported somewhere: a caller with a path should pass it, and
169/// one without ([`Entry::anonymous`]) gets diagnostics that say so rather than
170/// diagnostics that guess. Imported packages name themselves through the
171/// loader either way.
172///
173/// # Errors
174///
175/// As [`run_with_loader`].
176pub fn run_in_surface(
177    entry: Entry<'_>,
178    inputs: Inputs,
179    loader: &dyn crate::uses::Loader,
180    surface: Option<&blue_lang_syntax::yakugo::Yakugo>,
181) -> Result<Run, RunError> {
182    // The interpreter is built BEFORE the check stage, because it is the check
183    // stage's name oracle: every builtin, macro and special form a program may
184    // name is read off the interpreter that will run it, never off a list.
185    // Building it is not a stage; nothing is evaluated until `eval_prepared`.
186    let mut interp = crate::interpreter_hostless();
187    crate::inputs::install_input_primitives(&mut interp, inputs);
188    let prepared = prepare(entry, loader, surface, &builtin_names(&interp))?;
189    let mut value = eval_prepared(&mut interp, &prepared, &mut ())?;
190    if let Some(v) = call_main(&mut interp, &prepared, &mut ())? {
191        value = v;
192    }
193    Ok(Run {
194        value,
195        visited: prepared.visited,
196        typed_decls: prepared.typed_decls,
197        seams: prepared.seams,
198    })
199}
200
201/// A run watched by a call observer, and the definitions the observer was
202/// built over.
203pub struct Observed {
204    pub definitions: std::sync::Arc<crate::observe::Definitions>,
205    /// The run itself: its value, or the error it stopped on.
206    pub outcome: Result<Run, RunError>,
207}
208
209/// [`run_in_surface`] with a call observer installed for the evaluation.
210/// `watch` builds the observer from the program's definitions once every stage
211/// before evaluation has passed; the profiler, the tracer and coverage are
212/// all built this way (`crate::observe::Recorder`).
213///
214/// # Errors
215///
216/// A parse, import or check error, which leaves nothing to run. A runtime
217/// error is the [`Observed::outcome`], so what ran before it is still read.
218pub fn run_observed(
219    entry: Entry<'_>,
220    inputs: Inputs,
221    loader: &dyn crate::uses::Loader,
222    surface: Option<&blue_lang_syntax::yakugo::Yakugo>,
223    watch: impl FnOnce(
224        &std::sync::Arc<crate::observe::Definitions>,
225    ) -> std::sync::Arc<dyn crate::observe::CallObserver>,
226) -> Result<Observed, RunError> {
227    let mut interp = crate::interpreter_hostless();
228    crate::inputs::install_input_primitives(&mut interp, inputs);
229    let prepared = prepare(entry, loader, surface, &builtin_names(&interp))?;
230    let definitions = std::sync::Arc::new(crate::observe::Definitions::of(
231        &prepared.program,
232        &prepared.erased,
233    ));
234    interp.set_call_observer(Some(watch(&definitions)));
235    let outcome = eval_prepared(&mut interp, &prepared, &mut ())
236        .and_then(|value| Ok(call_main(&mut interp, &prepared, &mut ())?.unwrap_or(value)))
237        .map(|value| Run {
238            value,
239            visited: prepared.visited,
240            typed_decls: prepared.typed_decls,
241            seams: prepared.seams,
242        });
243    interp.set_call_observer(None);
244    Ok(Observed {
245        definitions,
246        outcome,
247    })
248}
249
250/// A program past every stage before evaluation: parsed, its imports
251/// resolved, type-checked and erased. What [`run_in_surface`] and
252/// [`crate::hosted::load_hosted`] both evaluate, so the stage order this
253/// module exists to own is written once for both.
254pub(crate) struct Prepared {
255    program: crate::uses::ResolvedProgram,
256    erased: Vec<tatara_lisp::Spanned>,
257    visited: usize,
258    typed_decls: usize,
259    seams: usize,
260}
261
262impl Prepared {
263    /// The forms the evaluator will run: every file's, resolved and erased.
264    pub(crate) fn erased(&self) -> &[tatara_lisp::Spanned] {
265        &self.erased
266    }
267
268    /// The entry file's namespace.
269    pub(crate) fn entry_namespace(&self) -> Namespace {
270        let first = (0..self.program.forms().len())
271            .find(|i| self.program.owner_of(*i) == Some(crate::uses::ResolvedProgram::ENTRY));
272        match first {
273            Some(i) => namespace_of(&self.program, i),
274            None => match self
275                .program
276                .file(crate::uses::ResolvedProgram::ENTRY)
277                .and_then(|f| f.package.clone())
278            {
279                Some(p) => Namespace::Bidama(p),
280                None => Namespace::File(String::new()),
281            },
282        }
283    }
284}
285
286/// Parse, resolve, check and erase, in that order.
287///
288/// `builtins` is the name table of the interpreter that will evaluate the
289/// result ([`builtin_names`]); the check stage resolves the program's names
290/// against it.
291pub(crate) fn prepare(
292    entry: Entry<'_>,
293    loader: &dyn crate::uses::Loader,
294    surface: Option<&blue_lang_syntax::yakugo::Yakugo>,
295    builtins: &NameTable,
296) -> Result<Prepared, RunError> {
297    let mut program = parse_and_resolve(entry, loader, surface)?;
298
299    // `test` blocks are declarations for the harness, not code to run.
300    //
301    // Dropped here rather than in `resolve_uses`, because `blue test` calls
302    // the resolver and then NEEDS the entry file's blocks — so the two
303    // callers want different things and the split has to live at this level.
304    //
305    // Without this, `blue run` on a file containing its own tests fails with
306    // `unbound symbol: deftest`: every package in the bidama distribution
307    // carries tests, so every one of them was unrunnable.
308    //
309    // Through `retain`, which drops each form's owner with it. A plain filter
310    // over the forms alone would slide every later form onto the wrong file.
311    program.retain(|f| !crate::uses::is_test_form(f));
312
313    // CHECK, on the annotated tree — the only tree that has annotations.
314    //
315    // **On the REAL spanned tree, including every imported package's.** This
316    // was the one caller that checked a spanless lift, because `resolve_uses`
317    // flattened the entry file and its imports into one list and `Span` is a
318    // byte range with no file identity — so a real span here would have
319    // reported an imported package's error at that offset in the ENTRY file, a
320    // precise-looking answer pointing at unrelated code.
321    //
322    // The fix is not a wider `Span` (that type is upstream, and its own docs
323    // put file identity on the caller: spans "are meaningful only relative to
324    // the string that produced them, which the caller is responsible for
325    // holding onto"). blue holds onto it BESIDE the span, per top-level form —
326    // see `uses::ResolvedProgram`.
327    let (outcome, names) = check_stage(&program, builtins, false);
328    if !outcome.ok() {
329        return Err(RunError::Types(
330            outcome
331                .errors()
332                // `file:line:col: error[CODE]: message`, resolved against the
333                // file the form actually came from, then its help line.
334                .map(|d| render(&program, d))
335                .collect(),
336        ));
337    }
338
339    // ERASE, so the interpreter never sees a type — **on the spanned tree,
340    // and back out as one.**
341    //
342    // This used to be `erase_types(&program.sexps())` followed by a
343    // `lower_to_spanned` that stamped `Span::synthetic()` over every node, and
344    // the comment here said carrying real positions through was "a larger
345    // piece and is NOT built". It was not larger: erasure only ever DELETES
346    // nodes — the single node it invents is the `define` replacing
347    // `define-typed` — so there was never anything for a synthetic span to
348    // stand in for. See `crate::erase`.
349    //
350    // No lowering step follows. The tree the evaluator receives IS the tree the
351    // parser built, minus annotations, with the author's byte offsets intact.
352    let erased = erase_types(&lower(&program, &names));
353
354    Ok(Prepared {
355        program,
356        erased,
357        visited: outcome.stats.visited,
358        typed_decls: outcome.stats.typed_decls,
359        seams: outcome.seams.len(),
360    })
361}
362
363/// Parse the entry (in `surface`, if one is given) and splice in its imports.
364pub(crate) fn parse_and_resolve(
365    entry: Entry<'_>,
366    loader: &dyn crate::uses::Loader,
367    surface: Option<&blue_lang_syntax::yakugo::Yakugo>,
368) -> Result<crate::uses::ResolvedProgram, RunError> {
369    let forms = match surface {
370        Some(pack) => blue_lang_syntax::parse_program_tree_in(entry.text, pack)
371            .map_err(|e| RunError::Parse(e.to_string()))?,
372        None => loader
373            .parse(entry.text)
374            .map_err(|e| RunError::Parse(e.to_string()))?,
375    };
376
377    // RESOLVE imports first, so everything below sees ONE program.
378    //
379    // Before the check on purpose: imported code is type-checked at the point
380    // its consumer imports it, rather than at whatever later moment its code
381    // first runs. A package that does not typecheck should break its importer's
382    // build, not their production run.
383    //
384    // One program, but not one FILE: the result records which file each
385    // top-level form came from, which is what lets a diagnostic below name a
386    // place instead of only a problem.
387    crate::uses::resolve_uses(forms, entry, loader).map_err(RunError::Import)
388}
389
390/// **The check stage**: every rule in `blue_lang_check::RULES`, over a
391/// resolved program, with waivers applied.
392///
393/// One function, called by every door — `run`, the hosted loader, `blue test`
394/// and `blue check` — so a rule added to the registry is enforced by all of
395/// them at once. That is what makes an architecture rule a pipeline error
396/// rather than an optional report: there is no door that skips it.
397///
398/// Warnings (unused bindings, unused waivers) are reported for the ENTRY
399/// file only; an imported package's warnings belong to its author. Errors
400/// are reported for every file, because an imported package with one is
401/// broken for its importer.
402pub(crate) fn check_stage(
403    program: &crate::uses::ResolvedProgram,
404    builtins: &NameTable,
405    with_tests: bool,
406) -> (blue_lang_check::Outcome, NameTable) {
407    use crate::uses::ResolvedProgram;
408    let forms = program.forms();
409    let is_entry = |i: usize| program.owner_of(i) == Some(ResolvedProgram::ENTRY);
410
411    // Rules B0003–B0005: the typing ladder.
412    let mut outcome = blue_lang_check::check_program(forms);
413
414    // Rules B0001–B0002: names. The program's own scopes come first (the
415    // entry file, then each bidama in import order), then the interpreter's.
416    let table = program_names(program, builtins);
417    let whole = |i: usize| with_tests && is_entry(i);
418    let (names, resolved) = blue_lang_check::names::check_names_whole(
419        forms,
420        &table,
421        &|i| namespace_of(program, i),
422        &is_entry,
423        &whole,
424    );
425    let stored = program
426        .file(ResolvedProgram::ENTRY)
427        .is_some_and(|f| f.path.is_some());
428    outcome.diagnostics.extend(names);
429    outcome.stats.names_resolved = resolved;
430    if stored {
431        let script = program
432            .file(ResolvedProgram::ENTRY)
433            .is_some_and(|f| f.package.is_none());
434        let found = blue_lang_check::form_rules::check(forms, &is_entry, script);
435        let silenced: Vec<tatara_lisp::Span> = found
436            .iter()
437            .filter(|d| d.code == blue_lang_check::Code::B0038)
438            .map(|d| d.span)
439            .collect();
440        outcome
441            .diagnostics
442            .retain(|d| !(d.code == blue_lang_check::Code::B0002 && silenced.contains(&d.span)));
443        outcome.diagnostics.extend(found);
444    }
445
446    // Waivers, per file: B0007 for a malformed one, B0008 for one that
447    // suppresses nothing.
448    let mut waivers = Vec::new();
449    for file in program.files() {
450        let spans: Vec<(usize, tatara_lisp::Span)> = forms
451            .iter()
452            .enumerate()
453            .filter(|(i, _)| program.owner_of(*i) == Some(file.id))
454            .map(|(i, f)| (i, f.span))
455            .collect();
456        let (found, malformed) = blue_lang_check::waiver::collect(&file.text, &spans);
457        waivers.extend(found);
458        outcome.diagnostics.extend(malformed);
459    }
460    outcome.waived = blue_lang_check::waiver::apply(&mut outcome.diagnostics, waivers, &is_entry);
461    if !stored {
462        outcome.diagnostics.retain(|d| {
463            !matches!(
464                d.code,
465                blue_lang_check::Code::B0023
466                    | blue_lang_check::Code::B0036
467                    | blue_lang_check::Code::B0037
468            )
469        });
470    }
471    // A rule whose census has not reached zero is computed, and counted, but
472    // not enforced: its findings move to `census` (see `RULES`' `ratchet`).
473    let (enforced, census): (Vec<_>, Vec<_>) = std::mem::take(&mut outcome.diagnostics)
474        .into_iter()
475        .partition(|d| d.code.rule().active());
476    outcome.diagnostics = enforced;
477    outcome.census = census;
478    outcome
479        .diagnostics
480        .sort_by_key(|d| (d.top_level, d.span.start, d.code));
481    (outcome, table)
482}
483
484/// The whole name table the check stage resolves against: the program's
485/// own scopes (each file and bidama, by namespace) followed by `builtins`.
486///
487/// **A top-level call to a builtin macro can define names**: `defflow(slug,
488/// trim, downcase)` expands to `(define slug …)`, and nothing in the source
489/// spells `define`. Such a call is expanded here, on a hostless fork, and what
490/// its expansion defines is bound in the calling file's namespace. Only
491/// BUILTIN macros are expanded: their bodies are the runtime's own
492/// templates. A program's own `defmacro` is not run at check time — checking a
493/// file (and an editor checking on every keystroke) must not execute the code
494/// it checks — so names a program macro generates are not seen; none of the
495/// corpus's macros generates one (measured 2026-09-29).
496#[must_use]
497pub fn program_names(program: &crate::uses::ResolvedProgram, builtins: &NameTable) -> NameTable {
498    use blue_lang_check::Binding;
499    let mut table = NameTable::new();
500    table.add_program(program.forms(), |i| namespace_of(program, i));
501    let builtin_macro = |head: &str| {
502        builtins
503            .scopes()
504            .iter()
505            .any(|s| s.namespace == Namespace::Macro && s.get(head).is_some())
506    };
507    let mut expander: Option<tatara_lisp_eval::Interpreter<()>> = None;
508    for (i, form) in program.forms().iter().enumerate() {
509        let Some(head) = form
510            .as_list()
511            .and_then(|items| items.first())
512            .and_then(tatara_lisp::Spanned::as_symbol)
513        else {
514            continue;
515        };
516        if !builtin_macro(head) {
517            continue;
518        }
519        let interp = expander.get_or_insert_with(crate::interpreter_hostless);
520        // A failed expansion defines nothing here; the evaluator will report
521        // it, with its position, when the form runs.
522        let Ok(expanded) = interp.fully_expand(form, &mut ()) else {
523            continue;
524        };
525        let nullary = blue_lang_syntax::scope::nullary_definition(&expanded);
526        for (name, span, kind) in blue_lang_check::names::definitions_of(&expanded) {
527            table.define(
528                namespace_of(program, i),
529                Binding {
530                    nullary: nullary == Some(name.as_str()),
531                    name,
532                    kind,
533                    span: Some(span),
534                    top_level: Some(i),
535                },
536            );
537        }
538    }
539    for scope in builtins.scopes() {
540        table.push(scope.clone());
541    }
542    table.spell_bare_like(builtins);
543    let form_file: Vec<usize> = (0..program.forms().len())
544        .map(|i| program.owner_of(i).map_or(usize::MAX, |f| f.index()))
545        .collect();
546    let mut imports: std::collections::BTreeMap<usize, blue_lang_check::names::FileImports> =
547        std::collections::BTreeMap::new();
548    for (file, import) in program.imports() {
549        let entry = imports.entry(file.index()).or_default();
550        entry.add(&import.package, import.names.iter().map(|(n, _)| n.clone()));
551        let at = import.span.start;
552        entry.use_at = Some(entry.use_at.map_or(at, |a| a.min(at)));
553    }
554    // A FACADE — a bidama that defines nothing and `use`s others (zenbu) —
555    // exists so a consumer can depend on "everything": `use("zenbu")` makes
556    // every member's QUALIFIER reachable (`toukei::median`), never a bare
557    // name. Followed to a fixpoint, so a facade of facades works too.
558    let facade_uses =
559        |pkg: &str,
560         imports: &std::collections::BTreeMap<usize, blue_lang_check::names::FileImports>| {
561            let defines = table.bidama(pkg).is_some_and(|s| !s.is_empty());
562            if defines {
563                return std::collections::BTreeSet::new();
564            }
565            program
566                .files()
567                .iter()
568                .filter(|f| f.package.as_deref() == Some(pkg))
569                .filter_map(|f| imports.get(&f.id.index()))
570                .flat_map(|i| i.uses.iter().cloned())
571                .collect::<std::collections::BTreeSet<String>>()
572        };
573    loop {
574        let mut grew = false;
575        let files: Vec<usize> = imports.keys().copied().collect();
576        for f in files {
577            let used: Vec<String> = imports[&f].uses.iter().cloned().collect();
578            for p in used {
579                for member in facade_uses(&p, &imports) {
580                    if let Some(i) = imports.get_mut(&f) {
581                        if i.uses.insert(member.clone()) {
582                            i.via.insert(member, p.clone());
583                            grew = true;
584                        }
585                    }
586                }
587            }
588        }
589        if !grew {
590            break;
591        }
592    }
593    let mut pkg_uses: std::collections::BTreeMap<String, std::collections::BTreeSet<String>> =
594        std::collections::BTreeMap::new();
595    for (file, import) in program.imports() {
596        if let Some(p) = program.file(*file).and_then(|f| f.package.clone()) {
597            pkg_uses
598                .entry(p)
599                .or_default()
600                .insert(import.package.clone());
601        }
602    }
603    table.attach_files(form_file, imports);
604    table.attach_package_uses(pkg_uses);
605    table.attach_needs(program.needs().clone());
606    // Where each file lives and which distribution each bidama ships in: a
607    // bridge reference (B0021) is an error inside the bidama's own
608    // distribution and a warning outside it.
609    let canonical = |p: &std::path::Path| {
610        std::fs::canonicalize(p)
611            .unwrap_or_else(|_| p.to_path_buf())
612            .display()
613            .to_string()
614    };
615    let mut file_paths = std::collections::BTreeMap::new();
616    let mut pkg_roots = std::collections::BTreeMap::new();
617    for f in program.files() {
618        if let Some(path) = &f.path {
619            file_paths.insert(f.id.index(), canonical(path));
620            if let (Some(pkg), Some(root)) =
621                (&f.package, path.parent().and_then(std::path::Path::parent))
622            {
623                pkg_roots
624                    .entry(pkg.clone())
625                    .or_insert_with(|| canonical(root));
626            }
627        }
628    }
629    table.attach_paths(file_paths, pkg_roots);
630    let versions = program.versions().clone();
631    table.apply_legacy(program.forms(), &|i| namespace_of(program, i), &|p| {
632        versions.get(p).cloned()
633    });
634    table
635}
636
637/// The forms the evaluator runs, from a checked program: every `use`
638/// declaration inert (`nil`, so each index still names its top-level form)
639/// and every name at the key it runs under — the RESOLVED tree under
640/// per-bidama namespaces (`blue_lang_check::names::Resolved::resolved_tree`).
641///
642/// A definition runs as `retsu/first` (a script's as `%root/f`), a
643/// reference to it as the same key, and a builtin as its bare name. Keys
644/// cannot collide across namespaces, so a bidama's `first` no longer
645/// replaces the builtin for anyone who did not import it, and an importer's
646/// `helper` no longer replaces a bidama's own: the runtime binds exactly what
647/// the check stage resolved.
648#[must_use]
649pub fn lower(
650    program: &crate::uses::ResolvedProgram,
651    names: &NameTable,
652) -> Vec<tatara_lisp::Spanned> {
653    lower_with(program, names, true)
654}
655
656#[must_use]
657pub fn lower_with(
658    program: &crate::uses::ResolvedProgram,
659    names: &NameTable,
660    simplify_chains: bool,
661) -> Vec<tatara_lisp::Spanned> {
662    let mut resolved = blue_lang_check::names::resolve_program(program.forms(), names, &|i| {
663        namespace_of(program, i)
664    });
665    let simplified = if simplify_chains {
666        chain_simplified(program.forms(), &resolved)
667    } else {
668        None
669    };
670    let forms: &[tatara_lisp::Spanned] = simplified.as_deref().unwrap_or(program.forms());
671    if simplified.is_some() {
672        resolved =
673            blue_lang_check::names::resolve_program(forms, names, &|i| namespace_of(program, i));
674    }
675    resolved
676        .resolved_tree(forms, blue_lang_check::names::Rule::Namespaced)
677        .into_iter()
678        .zip(forms)
679        .enumerate()
680        .map(|(i, (resolved, written))| {
681            if crate::uses::use_target(written).is_some()
682                || blue_lang_syntax::scope::legacy_target(written).is_some()
683            {
684                tatara_lisp::Spanned::new(written.span, tatara_lisp::SpannedForm::Nil)
685            } else {
686                let file = program.owner_of(i).and_then(|id| program.file(id));
687                match file {
688                    Some(f) => locate_raises(&resolved, &raise_site(f)),
689                    None => resolved,
690                }
691            }
692        })
693        .collect()
694}
695
696fn chain_simplified(
697    forms: &[tatara_lisp::Spanned],
698    resolved: &blue_lang_check::names::Resolved,
699) -> Option<Vec<tatara_lisp::Spanned>> {
700    let respell: std::collections::HashMap<usize, blue_lang_check::respell::Respelling> = resolved
701        .respellings
702        .iter()
703        .filter(|(_, r)| matches!(r.respell, blue_lang_check::respell::Respell::Chain(_)))
704        .cloned()
705        .collect();
706    if respell.is_empty() {
707        return None;
708    }
709    let bare = std::collections::HashSet::new();
710    let canon = Canon {
711        bare: &bare,
712        respell: &respell,
713    };
714    Some(forms.iter().map(|f| canon.form(f)).collect())
715}
716
717fn raise_site(file: &crate::uses::SourceFile) -> impl Fn(tatara_lisp::Span) -> String + '_ {
718    move |span| {
719        let (line, _) =
720            blue_lang_syntax::Span::line_col(&file.text, span.start.min(file.text.len()));
721        let shown = file
722            .path
723            .as_ref()
724            .map_or_else(|| "<source>".to_string(), |p| p.display().to_string());
725        format!("{shown}:{line}")
726    }
727}
728
729fn locate_raises(
730    form: &tatara_lisp::Spanned,
731    site: &dyn Fn(tatara_lisp::Span) -> String,
732) -> tatara_lisp::Spanned {
733    use tatara_lisp::{Atom, Spanned, SpannedForm};
734    let inner = match &form.form {
735        SpannedForm::List(items) => {
736            let mut out: Vec<Spanned> = items.iter().map(|i| locate_raises(i, site)).collect();
737            let raising = items.first().and_then(Spanned::as_symbol) == Some("raise")
738                && (3..=4).contains(&items.len());
739            if raising {
740                out[0] = Spanned::new(
741                    items[0].span,
742                    SpannedForm::Atom(Atom::Symbol(crate::errors::RAISE_AT.to_string())),
743                );
744                out.insert(
745                    1,
746                    Spanned::new(items[0].span, SpannedForm::Atom(Atom::Str(site(form.span)))),
747                );
748            }
749            SpannedForm::List(out)
750        }
751        SpannedForm::Quote(_) | SpannedForm::Quasiquote(_) => return form.clone(),
752        other => other.clone(),
753    };
754    Spanned::new(form.span, inner)
755}
756
757/// The namespace top-level form `i` of `program` defines into.
758#[must_use]
759pub fn namespace_of(program: &crate::uses::ResolvedProgram, i: usize) -> Namespace {
760    let file = program.owner_of(i).and_then(|id| program.file(id));
761    match file {
762        Some(f) => match &f.package {
763            Some(p) => Namespace::Bidama(p.clone()),
764            None => Namespace::File(
765                f.path
766                    .as_ref()
767                    .map_or_else(|| "<anonymous>".to_string(), |p| p.display().to_string()),
768            ),
769        },
770        None => Namespace::File("<unknown file>".to_string()),
771    }
772}
773
774/// Names only the test harness binds.
775///
776/// In scope for EVERY check, not only for test blocks. A bidama may carry a
777/// test helper — a plain `def` that asserts, called only from its `test`
778/// blocks (`heni`'s `hn_end_to_end` is one) — and every program importing that
779/// bidama would otherwise be refused for a function it never calls. The
780/// refusal would be scoped to the importer's whole program, not to the fault.
781/// The cost, stated: an `assert` outside any test that `blue run` actually
782/// reaches is caught at runtime (`blue-assert` is unbound there), not here.
783///
784/// `blue-lang-test` registers these on its own interpreter, and this crate
785/// cannot depend on it (it depends on this one), so the list lives beside the
786/// only name in it: the lowered `assert`, which `blue-lang-syntax` owns.
787pub const HARNESS_NAMES: &[&str] = &[blue_lang_syntax::LOWERED_ASSERT];
788
789/// The check stage's builtin scopes, read off the interpreter that will run
790/// the program: the test harness's names ([`HARNESS_NAMES`]), then its special
791/// forms, its macros, and every value it binds — the last three in the order
792/// the evaluator consults for a head.
793///
794/// Read from a live interpreter rather than a list, for the reason
795/// `Interpreter::resolve_head` gives: a list is a second copy of the truth and
796/// goes stale on the next installer that lands. An embedder's own
797/// primitives are in scope here because the embedder installs them first.
798#[must_use]
799pub fn builtin_names<H: 'static>(interp: &tatara_lisp_eval::Interpreter<H>) -> NameTable {
800    use blue_lang_check::{Binding, Scope, ScopeKind};
801    use tatara_lisp_eval::HeadBinding;
802    let mut special = Scope::new(Namespace::SpecialForm);
803    let mut macros = Scope::new(Namespace::Macro);
804    let mut values = Scope::new(Namespace::Builtin);
805    // `reserved_head_names` reads the ROOT frame only, so a name installed on
806    // a fork (the input primitives, on every `run`) is invisible to it; those
807    // are asked about by name instead, which `resolve_head` answers through
808    // every frame.
809    let mut names: Vec<std::sync::Arc<str>> = interp.reserved_head_names().into_iter().collect();
810    names.extend(
811        crate::inputs::INPUT_PRIMITIVES
812            .iter()
813            .filter(|n| interp.resolve_head(n).is_some())
814            .map(|n| std::sync::Arc::from(*n)),
815    );
816    for name in names {
817        let (scope, kind) = match interp.resolve_head(&name) {
818            Some(HeadBinding::SpecialForm) => (&mut special, ScopeKind::SpecialForm),
819            Some(HeadBinding::Macro) => (&mut macros, ScopeKind::Macro),
820            _ => (&mut values, ScopeKind::Value),
821        };
822        let nullary = kind == ScopeKind::Value
823            && interp.lookup_global(&name).is_some_and(|v| {
824                nullary_value(&v)
825                    || (crate::docs::doc_of(&name).is_some_and(|d| d.signature == &*name)
826                        && matches!(
827                            v,
828                            tatara_lisp_eval::Value::NativeFn(_)
829                                | tatara_lisp_eval::Value::Closure(_)
830                        ))
831            });
832        scope.bind(Binding {
833            name: name.to_string(),
834            kind,
835            span: None,
836            top_level: None,
837            nullary,
838        });
839    }
840    let mut table = NameTable::new();
841    let mut harness = Scope::new(Namespace::Harness);
842    for n in HARNESS_NAMES {
843        harness.bind(Binding {
844            name: (*n).to_string(),
845            kind: ScopeKind::Value,
846            span: None,
847            top_level: None,
848            nullary: false,
849        });
850    }
851    table.push(harness);
852    table.push(special);
853    table.push(macros);
854    table.push(values);
855    table.spell_bare_only(
856        crate::docs::NAMES
857            .iter()
858            .filter(|d| d.signature == d.name)
859            .map(|d| d.name.to_string())
860            .collect(),
861    );
862    table
863}
864
865#[must_use]
866pub fn nullary_value(value: &tatara_lisp_eval::Value) -> bool {
867    match value {
868        tatara_lisp_eval::Value::NativeFn(f) => {
869            matches!(f.arity, tatara_lisp_eval::ffi::Arity::Exact(0))
870        }
871        tatara_lisp_eval::Value::Closure(c) => c.params.is_empty() && c.rest.is_none(),
872        _ => false,
873    }
874}
875
876/// A diagnostic as a person reads it: `file:line:col: error[B0001]: message`,
877/// then an indented `help:` line when it has one.
878#[must_use]
879pub fn render(program: &crate::uses::ResolvedProgram, d: &blue_lang_check::Diagnostic) -> String {
880    let head = d.to_string();
881    let mut out = program.locate(d.top_level, d.span, &head).to_string();
882    if let Some(help) = &d.help {
883        out.push_str("\n  help: ");
884        out.push_str(help);
885    }
886    out
887}
888
889/// What [`check_entry`] checks.
890#[derive(Clone, Copy, Debug, PartialEq, Eq)]
891pub enum Checking {
892    /// The program `blue run` would evaluate: test blocks dropped.
893    Program,
894    /// The program plus the entry file's test blocks, as `blue test` and
895    /// `blue check` see it.
896    WithTests,
897}
898
899/// A program through the check stage, with everything the stage found.
900pub struct Checked {
901    pub program: crate::uses::ResolvedProgram,
902    pub outcome: blue_lang_check::Outcome,
903    /// The name table the stage resolved against: every program scope, then
904    /// the builtins. What an editor completes from.
905    pub names: NameTable,
906}
907
908impl Checked {
909    /// The forms an evaluator runs: [`lower`]ed, then erased — the tree
910    /// `run` evaluates, for a door that brings its own evaluator (the VM).
911    #[must_use]
912    pub fn erased(&self) -> Vec<tatara_lisp::Spanned> {
913        erase_types(&lower(&self.program, &self.names))
914    }
915
916    /// What `blue test` hands the harness: the program [`lower`]ed, spans
917    /// projected away.
918    #[must_use]
919    pub fn evaluable(&self) -> Vec<tatara_lisp::Sexp> {
920        lower(&self.program, &self.names)
921            .iter()
922            .map(tatara_lisp::Spanned::to_sexp)
923            .collect()
924    }
925
926    /// Every non-local reference, resolved under both rules — today's flat
927    /// one and per-bidama namespaces — and the resolved trees built from them
928    /// (`blue ast --resolved`).
929    #[must_use]
930    pub fn resolve(&self) -> blue_lang_check::names::Resolved {
931        blue_lang_check::names::resolve_program(self.program.forms(), &self.names, &|i| {
932            namespace_of(&self.program, i)
933        })
934    }
935}
936
937impl Checked {
938    #[must_use]
939    pub fn call_explicit_forms(&self) -> Vec<tatara_lisp::Spanned> {
940        self.resolve().call_explicit_tree(self.program.forms())
941    }
942
943    #[must_use]
944    pub fn canonical_entry(&self, src: &str) -> Option<Vec<tatara_lisp::Spanned>> {
945        use crate::uses::ResolvedProgram;
946        let program = &self.program;
947        let entry: Vec<&tatara_lisp::Spanned> = program
948            .forms()
949            .iter()
950            .enumerate()
951            .filter(|(i, _)| program.owner_of(*i) == Some(ResolvedProgram::ENTRY))
952            .map(|(_, f)| f)
953            .collect();
954        let written = blue_lang_syntax::parse_program_tree(src).ok()?;
955        let same = written.len() == entry.len()
956            && written
957                .iter()
958                .zip(&entry)
959                .all(|(w, e)| w.to_sexp() == e.to_sexp() && w.span == e.span);
960        if !same {
961            return None;
962        }
963        let resolved = self.resolve();
964        let bare: std::collections::HashSet<usize> = resolved
965            .parenthesized_nullary_calls()
966            .into_iter()
967            .filter(|r| program.owner_of(r.top_level) == Some(ResolvedProgram::ENTRY))
968            .map(blue_lang_check::names::Reference::node_address)
969            .collect();
970        let waived: Vec<tatara_lisp::Span> = self
971            .outcome
972            .waived
973            .iter()
974            .map(|w| w.diagnostic.span)
975            .collect();
976        let respell: std::collections::HashMap<usize, blue_lang_check::respell::Respelling> =
977            resolved
978                .respellings
979                .iter()
980                .filter(|(_, r)| !waived.contains(&r.span))
981                .cloned()
982                .collect();
983        let canon = Canon {
984            bare: &bare,
985            respell: &respell,
986        };
987        let forms: Vec<tatara_lisp::Spanned> = entry.into_iter().map(|f| canon.form(f)).collect();
988        let script = program
989            .file(ResolvedProgram::ENTRY)
990            .is_some_and(|f| f.package.is_none());
991        let forms = if script {
992            blue_lang_check::form_rules::without_prefix(&forms).unwrap_or(forms)
993        } else {
994            forms
995        };
996        let forms = blue_lang_check::form_rules::without_main_guard(&forms).unwrap_or(forms);
997        Some(blue_lang_check::form_rules::eta_reduced(&forms).unwrap_or(forms))
998    }
999}
1000
1001struct Canon<'a> {
1002    bare: &'a std::collections::HashSet<usize>,
1003    respell: &'a std::collections::HashMap<usize, blue_lang_check::respell::Respelling>,
1004}
1005
1006impl Canon<'_> {
1007    fn form(&self, form: &tatara_lisp::Spanned) -> tatara_lisp::Spanned {
1008        use tatara_lisp::{Spanned, SpannedForm};
1009        let inner = match &form.form {
1010            SpannedForm::List(items) => match items.as_slice() {
1011                [head] if self.bare.contains(&(std::ptr::from_ref(head) as usize)) => {
1012                    return Spanned::new(form.span, head.form.clone());
1013                }
1014                [head, ..]
1015                    if self
1016                        .respell
1017                        .contains_key(&(std::ptr::from_ref(head) as usize)) =>
1018                {
1019                    let r = &self.respell[&(std::ptr::from_ref(head) as usize)];
1020                    let walked: Vec<Spanned> = items.iter().map(|i| self.form(i)).collect();
1021                    return blue_lang_check::respell::apply(form, &walked, r);
1022                }
1023                _ => SpannedForm::List(items.iter().map(|i| self.form(i)).collect()),
1024            },
1025            SpannedForm::Quote(i) => SpannedForm::Quote(Box::new(self.form(i))),
1026            SpannedForm::Quasiquote(i) => SpannedForm::Quasiquote(Box::new(self.form(i))),
1027            SpannedForm::Unquote(i) => SpannedForm::Unquote(Box::new(self.form(i))),
1028            SpannedForm::UnquoteSplice(i) => SpannedForm::UnquoteSplice(Box::new(self.form(i))),
1029            other => other.clone(),
1030        };
1031        Spanned::new(form.span, inner)
1032    }
1033}
1034
1035/// Parse, resolve and CHECK, without running: the door for `blue check`,
1036/// `blue test` and the language server.
1037///
1038/// Returns every diagnostic, errors and warnings alike, rather than failing on
1039/// the first error: a reader fixing a file wants the whole list, and an agent
1040/// repairing one wants it in one pass. Whether errors stop anything is the
1041/// caller's decision — `blue test` refuses to run tests over a program with
1042/// one, exactly as `run` refuses to evaluate it.
1043///
1044/// The name oracle is the interpreter `blue run` would build: hostless plus
1045/// input primitives.
1046///
1047/// # Errors
1048///
1049/// A parse or import error, which leaves no program to check.
1050pub fn check_entry(
1051    entry: Entry<'_>,
1052    loader: &dyn crate::uses::Loader,
1053    surface: Option<&blue_lang_syntax::yakugo::Yakugo>,
1054    checking: Checking,
1055) -> Result<Checked, RunError> {
1056    let mut program = parse_and_resolve(entry, loader, surface)?;
1057    if checking == Checking::Program {
1058        program.retain(|f| !crate::uses::is_test_form(f));
1059    }
1060    let mut interp = crate::interpreter_hostless();
1061    crate::inputs::install_input_primitives(&mut interp, Inputs::new());
1062    let builtins = builtin_names(&interp);
1063    let (outcome, names) = check_stage(&program, &builtins, checking == Checking::WithTests);
1064    Ok(Checked {
1065        program,
1066        outcome,
1067        names,
1068    })
1069}
1070
1071#[must_use]
1072pub fn spelling_code(spelling: blue_lang_syntax::Spelling) -> blue_lang_check::Code {
1073    use blue_lang_check::Code;
1074    use blue_lang_syntax::Spelling;
1075    match spelling {
1076        Spelling::LogicalSymbol => Code::B0025,
1077        Spelling::Unless => Code::B0026,
1078        Spelling::ElseIf => Code::B0027,
1079        Spelling::CallForm => Code::B0028,
1080        Spelling::KeywordRocket => Code::B0029,
1081        Spelling::ParameterList => Code::B0030,
1082        Spelling::OperatorSpelling => Code::B0031,
1083        Spelling::TrailingComma => Code::B0032,
1084        Spelling::NegatedComparison => Code::B0034,
1085        Spelling::ElseNil => Code::B0035,
1086        Spelling::TryCatch => Code::B0037,
1087        Spelling::Apply => Code::B0042,
1088    }
1089}
1090
1091#[must_use]
1092pub fn form_diagnostics(text: &str) -> Vec<blue_lang_check::Diagnostic> {
1093    match blue_lang_fmt::format_source_lossless(text) {
1094        Ok(formatted) => form_diagnostics_against(text, &formatted),
1095        Err(_) => Vec::new(),
1096    }
1097}
1098
1099#[must_use]
1100pub fn form_diagnostics_against(text: &str, formatted: &str) -> Vec<blue_lang_check::Diagnostic> {
1101    use blue_lang_check::{Code, Diagnostic};
1102    if formatted.trim_end() == text.trim_end() {
1103        return Vec::new();
1104    }
1105    let help = "`blue fmt --write` writes it; `blue run`, `blue test` and `blue check` rewrite a writable file before compiling it";
1106    let mut out: Vec<Diagnostic> = blue_lang_syntax::spelling_notes(text)
1107        .unwrap_or_default()
1108        .into_iter()
1109        .map(|n| {
1110            let message = if n.canonical.is_empty() {
1111                format!("`{}` is not written", n.written)
1112            } else {
1113                format!("`{}` is written `{}`", n.written, n.canonical)
1114            };
1115            Diagnostic::new(spelling_code(n.spelling), message, n.span)
1116                .at_top_level(0)
1117                .with_help(help)
1118        })
1119        .collect();
1120    out.extend(
1121        literal_diagnostics(text)
1122            .into_iter()
1123            .map(|d| d.with_help(help)),
1124    );
1125    if out.is_empty() {
1126        let at = text
1127            .lines()
1128            .zip(formatted.lines())
1129            .position(|(a, b)| a != b)
1130            .unwrap_or_else(|| text.lines().count().min(formatted.lines().count()));
1131        let start: usize = text.lines().take(at).map(|l| l.len() + 1).sum();
1132        let start = start.min(text.len());
1133        let end = text[start..].find('\n').map_or(text.len(), |e| start + e);
1134        out.push(
1135            Diagnostic::new(
1136                Code::B0024,
1137                format!("line {} is not laid out as blue writes it", at + 1),
1138                tatara_lisp::Span::new(start, end),
1139            )
1140            .at_top_level(0)
1141            .with_help(help),
1142        );
1143    }
1144    out.sort_by_key(|d| (d.span.start, d.code));
1145    out
1146}
1147
1148fn literal_diagnostics(text: &str) -> Vec<blue_lang_check::Diagnostic> {
1149    use blue_lang_syntax::{Atom, TokenKind};
1150    let Ok(toks) = blue_lang_syntax::lex(text) else {
1151        return Vec::new();
1152    };
1153    toks.iter()
1154        .filter_map(|t| {
1155            let atom = match &t.kind {
1156                TokenKind::Int(v) => Atom::Int(*v),
1157                TokenKind::Float(v) => Atom::Float(*v),
1158                TokenKind::Str(v) => Atom::Str(v.clone()),
1159                _ => return None,
1160            };
1161            let written = text.get(t.span.start..t.span.end)?;
1162            let canonical = blue_lang_fmt::render_literal(&atom);
1163            (written != canonical).then(|| {
1164                blue_lang_check::Diagnostic::new(
1165                    blue_lang_check::Code::B0033,
1166                    format!("`{written}` is written `{canonical}`"),
1167                    t.span,
1168                )
1169                .at_top_level(0)
1170            })
1171        })
1172        .collect()
1173}
1174
1175/// A syntax error in `text` as a [`B0006`](blue_lang_check::Code::B0006)
1176/// diagnostic, so a tool reading `--format json` gets a parse failure in the
1177/// same shape as every other finding. `None` when `text` parses.
1178#[must_use]
1179pub fn syntax_diagnostic(text: &str) -> Option<blue_lang_check::Diagnostic> {
1180    let e = blue_lang_syntax::parse_program_tree(text).err()?;
1181    Some(
1182        blue_lang_check::Diagnostic::new(blue_lang_check::Code::B0006, e.message, e.span)
1183            .at_top_level(0),
1184    )
1185}
1186
1187/// A runtime message as the author wrote the names in it: a definition runs
1188/// under its key (`retsu/first`, `%root/f`), and a message naming one in
1189/// backticks names it `retsu::first`, or `f` for the author's own script.
1190#[must_use]
1191pub fn display_keys(message: &str) -> String {
1192    let mut out = String::with_capacity(message.len());
1193    let mut parts = message.split('`');
1194    if let Some(first) = parts.next() {
1195        out.push_str(first);
1196    }
1197    for (i, part) in parts.enumerate() {
1198        out.push('`');
1199        // Odd pieces (0-based here) are inside backticks.
1200        if i % 2 == 0 {
1201            out.push_str(&crate::observe::display_key(part));
1202        } else {
1203            out.push_str(part);
1204        }
1205    }
1206    out
1207}
1208
1209/// Evaluate a prepared program's top-level forms into `interp`, with `host`.
1210pub(crate) fn eval_prepared<H: 'static>(
1211    interp: &mut tatara_lisp_eval::Interpreter<H>,
1212    prepared: &Prepared,
1213    host: &mut H,
1214) -> Result<Value, RunError> {
1215    let Prepared {
1216        program, erased, ..
1217    } = prepared;
1218
1219    // RUN, one top-level form at a time, **counting them**.
1220    //
1221    // This is literally the loop `Interpreter::eval_program` runs internally
1222    // (tatara-lisp-eval `eval.rs`) — unrolled here for exactly one reason: the
1223    // INDEX. `resolve_uses` flattens N files into one form list and a `Span` is
1224    // a byte range with no file identity, so an offset alone is ambiguous
1225    // across files — 66 means something in every one of them. The index is the
1226    // join key back to `ResolvedProgram`'s file table, and `eval_program`
1227    // consumes it internally and hands back only the error.
1228    //
1229    // The same key, the same renderer and the same file table the type errors
1230    // above already use. A runtime error was the one diagnostic in this
1231    // function still reporting a bare message.
1232    let mut value = Value::Nil;
1233    for (top_level, form) in erased.iter().enumerate() {
1234        value = interp.eval_top_form(form, host).map_err(|e| {
1235            // The message, its position and the blue call stack, read off the
1236            // frames the error unwound through (`observe::Fault`). The message
1237            // is `short_message` rather than upstream's `Display`, whose
1238            // trailing ` at {span}` is a bare byte range with no file, through
1239            // `messages::describe`, which names the function a wrong argument
1240            // count was called on.
1241            RunError::Eval(Box::new(crate::observe::Fault::of(
1242                program,
1243                erased,
1244                top_level,
1245                &e,
1246                &interp.take_trace(),
1247            )))
1248        })?;
1249    }
1250    Ok(value)
1251}
1252
1253pub const MAIN: &str = "main";
1254
1255#[must_use]
1256pub fn entry_main(program: &crate::uses::ResolvedProgram) -> Option<(usize, tatara_lisp::Span)> {
1257    use crate::uses::ResolvedProgram;
1258    let stored = program
1259        .file(ResolvedProgram::ENTRY)
1260        .is_some_and(|f| f.path.is_some());
1261    if !stored {
1262        return None;
1263    }
1264    program.forms().iter().enumerate().find_map(|(i, f)| {
1265        if program.owner_of(i) != Some(ResolvedProgram::ENTRY) {
1266            return None;
1267        }
1268        let items = f.as_list()?;
1269        let sig = items.get(1)?.as_list()?;
1270        let named_main = matches!(
1271            items.first().and_then(tatara_lisp::Spanned::as_symbol),
1272            Some("define" | "define-typed")
1273        ) && sig.len() == 1
1274            && sig[0].as_symbol() == Some(MAIN);
1275        named_main.then_some((i, sig[0].span))
1276    })
1277}
1278
1279pub(crate) fn call_main<H: 'static>(
1280    interp: &mut tatara_lisp_eval::Interpreter<H>,
1281    prepared: &Prepared,
1282    host: &mut H,
1283) -> Result<Option<Value>, RunError> {
1284    let Some((top_level, span)) = entry_main(&prepared.program) else {
1285        return Ok(None);
1286    };
1287    let ns = namespace_of(&prepared.program, top_level);
1288    let call = tatara_lisp::Spanned::new(
1289        span,
1290        tatara_lisp::SpannedForm::List(vec![tatara_lisp::Spanned::new(
1291            span,
1292            tatara_lisp::SpannedForm::Atom(tatara_lisp::Atom::Symbol(blue_lang_check::names::key(
1293                &ns, MAIN,
1294            ))),
1295        )]),
1296    );
1297    interp.eval_top_form(&call, host).map(Some).map_err(|e| {
1298        RunError::Eval(Box::new(crate::observe::Fault::of(
1299            &prepared.program,
1300            &prepared.erased,
1301            top_level,
1302            &e,
1303            &interp.take_trace(),
1304        )))
1305    })
1306}
1307
1308/// **A runtime error names a place.** The gates for the last stage that
1309/// reported a bare message.
1310///
1311/// Each fixture is sized so that resolving the failing span against ANY file
1312/// but the right one lands somewhere else — a different path and a different
1313/// line — so a green run here is evidence about the join and not about the
1314/// renderer being called at all.
1315///
1316/// **Each raise is an unbound name, WAIVED.** Since the check stage resolves
1317/// names (B0001), an unbound call no longer reaches the evaluator; these
1318/// fixtures need exactly that runtime raise, so each carries
1319/// `# waive B0001: …` above the definition holding it — the one escape hatch,
1320/// used for what it is for. The waiver adds a line above each failing call,
1321/// which is why the positions below are one line later than their red runs
1322/// (recorded against the unwaived fixtures) show.
1323#[cfg(test)]
1324mod position_tests {
1325    use super::*;
1326    use crate::uses::{Entry, Loader};
1327    use std::collections::BTreeMap;
1328    use std::path::Path;
1329
1330    struct MemLoader(BTreeMap<&'static str, &'static str>);
1331
1332    impl Loader for MemLoader {
1333        fn load(&self, name: &str) -> Result<Vec<(String, String)>, String> {
1334            self.0
1335                .get(name)
1336                .map(|s| vec![(format!("{name}.b"), (*s).to_owned())])
1337                .ok_or_else(|| format!("no bidama named \"{name}\""))
1338        }
1339    }
1340
1341    /// A package whose OWN top-level code raises, deep inside a file that is
1342    /// deliberately taller than either of the other two.
1343    ///
1344    /// Byte 66 — where `kore_wa_sonzai_shinai` starts — is past the end of the
1345    /// entry file (30 bytes) and past the end of `kotae.b` (21 bytes). So the
1346    /// only file this span can honestly resolve against is this one, and the
1347    /// other two now refuse to answer rather than walking off their ends.
1348    const BAKUHATSU: &str = "\
1349def tsukawanai_a()
1350  1
1351end
1352
1353def tsukawanai_b()
1354  2
1355end
1356
1357# waive B0001: the fixture needs a raise at runtime
1358def bakuhatsu()
1359  kore_wa_sonzai_shinai()
1360end
1361
1362bakuhatsu";
1363
1364    const KOTAE: &str = "def kotae()\n  42\nend";
1365
1366    /// The same failing call, but only ever reached by a CALLER in another
1367    /// file — the shape the call stack resolved (the last position test).
1368    const BAKUHATSU2: &str =
1369        "# waive B0001: a raise at runtime\ndef yobu()\n  kore_wa_sonzai_shinai()\nend";
1370
1371    fn loader() -> MemLoader {
1372        MemLoader(BTreeMap::from([
1373            ("kotae", KOTAE),
1374            ("bakuhatsu", BAKUHATSU),
1375            ("bakuhatsu2", BAKUHATSU2),
1376        ]))
1377    }
1378
1379    fn run_named(path: &str, src: &str) -> RunError {
1380        run_in_surface(
1381            Entry {
1382                path: Some(Path::new(path)),
1383                text: src,
1384            },
1385            Inputs::new(),
1386            &loader(),
1387            None,
1388        )
1389        .expect_err("the fixture must raise")
1390    }
1391
1392    /// **The load-bearing gate: a raise inside an IMPORTED package reports the
1393    /// IMPORTED file's path and line.**
1394    ///
1395    /// Hand-computed. In `bakuhatsu.b` the failing call sits on line 11 at
1396    /// column 3, and the whole point of the fixture's height is that no other
1397    /// file in the program can produce that pair: the entry file has 2 lines
1398    /// and `kotae.b` has 3, so *any* mis-attribution is visible as a different
1399    /// path AND a different position, never as a coincidence.
1400    ///
1401    /// **Three red runs, one per edit this change makes** — each isolates a
1402    /// different half of the mechanism, which is what stops the gate being a
1403    /// tautology over "some renderer ran".
1404    ///
1405    /// 1. `eval_program` + `.map_err(|e| RunError::Eval(e.to_string()))`
1406    ///    restored (the pre-change code):
1407    ///    ```text
1408    ///    left:  "runtime error: unbound symbol: kore_wa_sonzai_shinai at 74..95"
1409    ///    right: "runtime error: bakuhatsu.b:10:3: unbound symbol `kore_wa_sonzai_shinai`"
1410    ///    ```
1411    ///    No place at all — the state this change is fixing — and the trailing
1412    ///    `at 74..95` is upstream's `Display`, an offset with no file, which is
1413    ///    why the loop renders `short_message` instead.
1414    ///
1415    /// 2. The loop kept, but `0` passed to `locate` in place of `top_level`:
1416    ///    ```text
1417    ///    left:  "runtime error: kotae.b: unbound symbol `kore_wa_sonzai_shinai`"
1418    ///    right: "runtime error: bakuhatsu.b:10:3: unbound symbol `kore_wa_sonzai_shinai`"
1419    ///    ```
1420    ///    **This is the one that proves the INDEX is load-bearing** rather than
1421    ///    the renderer: form 0 belongs to `kotae.b`, so the wrong file is named
1422    ///    — and named without a position, because the guard in `locate` sees
1423    ///    that byte 74 is not a range in a 21-byte file and declines to invent
1424    ///    one. A fixture with one imported package could not tell these apart.
1425    ///
1426    /// 3. `crate::lower_to_spanned(&crate::to_sexps(&erased))` reinstated
1427    ///    between erasure and the loop:
1428    ///    ```text
1429    ///    left:  "runtime error: bakuhatsu.b: unbound symbol `kore_wa_sonzai_shinai`"
1430    ///    right: "runtime error: bakuhatsu.b:10:3: unbound symbol `kore_wa_sonzai_shinai`"
1431    ///    ```
1432    ///    The right FILE with no position — which is exactly the shape of the
1433    ///    bug: the index survived the lift and every span did not.
1434    /// kotae loads first so bakuhatsu's forms do not start at index 0: the
1435    /// order is the fixture, and the waiver says so (B0015 wants them sorted).
1436    const TWO_IMPORTS: &str =
1437        "use(\"kotae\")\n# waive B0015: kotae loads first on purpose\nuse(\"bakuhatsu\")\n";
1438
1439    #[test]
1440    fn a_raise_inside_an_imported_package_names_that_package() {
1441        let err = run_named("entry.b", TWO_IMPORTS);
1442        assert_eq!(
1443            err.to_string(),
1444            "runtime error: bakuhatsu.b:11:3: unbound symbol `kore_wa_sonzai_shinai`\n  \
1445             at bakuhatsu::bakuhatsu (bakuhatsu.b:11:3)\n  \
1446             at <top level> (bakuhatsu.b:14:1)"
1447        );
1448    }
1449
1450    /// Anti-vacuity for the fixture itself: the position asserted above is a
1451    /// FACT ABOUT `bakuhatsu.b`, checked against the source text rather than
1452    /// against the thing that produced it.
1453    ///
1454    /// Without this, `11:3` is just a number that happened to come out of the
1455    /// code under test, and a change that moved every reported line by one
1456    /// would move this assertion with it.
1457    #[test]
1458    fn line_11_column_3_of_the_fixture_is_the_failing_call() {
1459        let line = BAKUHATSU.lines().nth(10).expect("line 11 exists");
1460        assert_eq!(line, "  kore_wa_sonzai_shinai()");
1461        assert_eq!(
1462            &line[2..],
1463            "kore_wa_sonzai_shinai()",
1464            "column 3 (1-indexed) is where the call starts"
1465        );
1466        // And the other two files cannot reach that far, which is what makes a
1467        // mis-attribution loud instead of plausible.
1468        let offset = BAKUHATSU.find("kore_wa_sonzai_shinai").expect("in fixture");
1469        assert!(
1470            offset > KOTAE.len(),
1471            "kotae.b could answer for byte {offset}"
1472        );
1473        assert!(
1474            offset > TWO_IMPORTS.len(),
1475            "the entry file could answer for byte {offset}"
1476        );
1477    }
1478
1479    /// **The `FileId(0)` path: a raise in the ENTRY file reports the entry
1480    /// file.** The single-file case, which is every program that imports
1481    /// nothing — and the one where the index and the span cannot disagree.
1482    ///
1483    /// **Red run** (2026-08-12), `eval_program` +
1484    /// `.map_err(|e| RunError::Eval(e.to_string()))` restored:
1485    /// ```text
1486    /// left:  "runtime error: unbound symbol: nani_mo_nai at 17..28"
1487    /// right: "runtime error: honmono.b:5:1: unbound symbol `nani_mo_nai`"
1488    /// ```
1489    ///
1490    /// **Second red run**, `lower_to_spanned` reinstated after erasure — the
1491    /// one that isolates the SPAN half from the index half, since a
1492    /// single-file program cannot get its file wrong:
1493    /// ```text
1494    /// left:  "runtime error: honmono.b: unbound symbol `nani_mo_nai`"
1495    /// right: "runtime error: honmono.b:5:1: unbound symbol `nani_mo_nai`"
1496    /// ```
1497    #[test]
1498    fn a_raise_in_the_entry_file_names_the_entry_file() {
1499        let err = run_named(
1500            "honmono.b",
1501            "def f()\n  1\nend\n\n# waive B0001: a raise at runtime\nnani_mo_nai()\n",
1502        );
1503        assert_eq!(
1504            err.to_string(),
1505            "runtime error: honmono.b:6:1: unbound symbol `nani_mo_nai`"
1506        );
1507    }
1508
1509    /// A raise from a form that came through the SURFACE-level erasure still
1510    /// reports a real position — the annotated path, where erasure rewrites
1511    /// the node rather than passing it through.
1512    ///
1513    /// Separate from the two above because the erasure rewrite is the only
1514    /// place a node is built rather than kept, and a synthetic span there
1515    /// would be invisible to a fixture whose failing form is untouched.
1516    ///
1517    /// **Red run** (2026-08-12), `lower_to_spanned` reinstated after erasure:
1518    /// ```text
1519    /// left:  "runtime error: chuu.b: unbound symbol `mada_nai`"
1520    /// right: "runtime error: chuu.b:2:3: unbound symbol `mada_nai`"
1521    /// ```
1522    #[test]
1523    fn a_raise_inside_an_annotated_def_still_names_its_line() {
1524        let err = run_named(
1525            "chuu.b",
1526            "# waive B0001: a raise at runtime\ndef f(a: Int) -> Int\n  mada_nai(a)\nend\n\nf(1)\n",
1527        );
1528        assert_eq!(
1529            err.to_string(),
1530            "runtime error: chuu.b:3:3: unbound symbol `mada_nai`\n  \
1531             at f (chuu.b:3:3)\n  \
1532             at <top level> (chuu.b:6:1)"
1533        );
1534    }
1535
1536    /// **The limit this module used to state, closed by the call stack.**
1537    ///
1538    /// A top-level form in the entry file calls a function defined in an
1539    /// imported one, and the raise happens inside the callee. The executing
1540    /// form is the entry's and the span is the callee's bytes, so until the
1541    /// evaluator reported the frames an error unwound through, the best blue
1542    /// could do was name the caller's file with no position (the guard in
1543    /// `uses::ResolvedProgram::locate` refusing an offset it could not
1544    /// justify). The innermost frame's definition now names the file the
1545    /// offsets belong to, and the stack shows the call that led there.
1546    ///
1547    /// **Red run** (2026-10-10): `Fault::of` locating against the executing
1548    /// top-level form instead of the innermost frame's definition:
1549    /// ```text
1550    /// left:  "runtime error: yobidashi.b: unbound symbol `kore_wa_sonzai_shinai`\n  at bakuhatsu2::yobu (bakuhatsu2.b:3:3)\n  at <top level> (yobidashi.b:3:1)"
1551    /// ```
1552    /// The old answer, kept by the guard: the caller's file and no position.
1553    #[test]
1554    fn a_raise_in_a_callee_from_another_file_names_the_callees_file() {
1555        // The comment is padding, and load-bearing: `bakuhatsu2.b`'s failing
1556        // call starts at byte 47 (after its waiver line) and ends at 70, and
1557        // this file is 57 bytes, so the start is in range here and the end is
1558        // not — a mis-attribution would still be refused rather than placed.
1559        let err = run_named(
1560            "yobidashi.b",
1561            "use(\"bakuhatsu2\")\n# pad: 47's in range\nbakuhatsu2::yobu\n",
1562        );
1563        assert_eq!(
1564            err.to_string(),
1565            "runtime error: bakuhatsu2.b:3:3: unbound symbol `kore_wa_sonzai_shinai`\n  \
1566             at bakuhatsu2::yobu (bakuhatsu2.b:3:3)\n  \
1567             at <top level> (yobidashi.b:3:1)"
1568        );
1569    }
1570}
1571
1572#[cfg(test)]
1573mod tests {
1574    use super::*;
1575
1576    fn int(src: &str) -> i64 {
1577        match run(src).unwrap_or_else(|e| panic!("{src:?}: {e}")).value {
1578            Value::Int(v) => v,
1579            other => panic!("{src:?} produced {other:?}"),
1580        }
1581    }
1582
1583    /// **The sliding scale, as one assertion.** Annotating changes the
1584    /// analysis and nothing else.
1585    #[test]
1586    fn annotating_buys_analysis_and_changes_nothing_else() {
1587        let plain = run("def add(a, b)\n  a + b\nend\nadd(2, 3)").expect("plain");
1588        let typed = run("def add(a: Int, b: Int) -> Int\n  a + b\nend\nadd(2, 3)").expect("typed");
1589
1590        assert!(matches!(plain.value, Value::Int(5)));
1591        assert!(
1592            matches!(typed.value, Value::Int(5)),
1593            "the annotated program must compute the same answer"
1594        );
1595        assert_eq!(plain.visited, 0, "no annotations means no analysis");
1596        assert!(
1597            typed.visited > 0,
1598            "an annotation must actually buy analysis, not just decorate"
1599        );
1600        assert_eq!(plain.typed_decls, 0);
1601        assert_eq!(typed.typed_decls, 1);
1602    }
1603
1604    /// **Checking happens before erasure.** This is the test that catches the
1605    /// reordering: a program with a declared-type violation must be rejected,
1606    /// and it can only be rejected while the annotations still exist.
1607    #[test]
1608    fn a_type_error_is_reported_and_the_program_does_not_run() {
1609        let err = run("def add(a: Int, b: Int) -> Str\n  a + b\nend\nadd(1, 2)")
1610            .expect_err("a declared Str return from an Int body must be rejected");
1611        assert!(
1612            matches!(err, RunError::Types(ref d) if !d.is_empty()),
1613            "expected type diagnostics, got {err}"
1614        );
1615    }
1616
1617    /// And the untyped version of the same program runs, so the rejection
1618    /// above is the annotation's doing rather than a parse failure.
1619    #[test]
1620    fn the_same_program_without_annotations_runs() {
1621        assert_eq!(int("def add(a, b)\n  a + b\nend\nadd(1, 2)"), 3);
1622    }
1623
1624    #[test]
1625    fn a_parse_error_is_reported_as_one() {
1626        assert!(matches!(run("def (").unwrap_err(), RunError::Parse(_)));
1627    }
1628
1629    /// Every stage reports in its own vocabulary, so a failure names which
1630    /// stage failed rather than surfacing as a generic error.
1631    #[test]
1632    fn a_runtime_error_is_reported_as_one() {
1633        let err = run("throw(error(\"boom\"))").expect_err("a raise");
1634        assert!(matches!(err, RunError::Eval(_)), "got {err}");
1635    }
1636
1637    /// Stdlib and primitives are both reachable through the pipeline — the
1638    /// gap that made `6 % 3` fail.
1639    #[test]
1640    fn the_pipeline_reaches_both_runtime_layers() {
1641        assert_eq!(int("6 % 3"), 0);
1642        assert_eq!(int("7 % 3"), 1);
1643        assert_eq!(int("2 + 3 * 4"), 14);
1644    }
1645
1646    /// **The deleted hop was a no-op on everything blue emits — so removing it
1647    /// is a swap, not a behaviour change.**
1648    ///
1649    /// The old lowering printed the erased tree and read it back through
1650    /// `tatara_lisp::read_spanned`. This walks a corpus and asserts the two
1651    /// paths land on the same tree, which is the equivalence the swap rests on.
1652    /// It is stated as a *measurement over this corpus*, not as a theorem:
1653    /// the round trip is not identity in general (that is precisely why it had
1654    /// to go), it merely happened to be identity for the bytes blue emits.
1655    #[test]
1656    fn the_deleted_round_trip_agreed_with_the_direct_lowering() {
1657        let corpus = [
1658            "def add(a, b)\n  a + b\nend\nadd(2, 3)",
1659            "def fact(n)\n  if n < 2\n    1\n  else\n    n * fact(n - 1)\n  end\nend\nfact(5)",
1660            "def f(a, b)\n  c = a + b\n  c * 2\nend\nf(1, 2)",
1661            "defmacro sq(e)\n  quote\n    unquote(e) * unquote(e)\n  end\nend\nsq(2 + 3)",
1662            "\"a string with spaces, a ( and a )\"",
1663            "def g(a: Int) -> Int\n  a + 1\nend\ng(1)",
1664            "6 % 3",
1665            "1.5 + 2.25",
1666        ];
1667        for src in corpus {
1668            // Projected to `Sexp` because that is what the deleted hop
1669            // operated on. Erasure itself no longer goes near it.
1670            let erased = crate::to_sexps(&erase_types(&parse_tree(src).expect("parse")));
1671
1672            let direct: Vec<Sexp> = crate::lower_to_spanned(&erased)
1673                .iter()
1674                .map(tatara_lisp::Spanned::to_sexp)
1675                .collect();
1676            assert_eq!(direct, erased, "the direct lowering must be the identity");
1677
1678            let text = erased
1679                .iter()
1680                .map(ToString::to_string)
1681                .collect::<Vec<_>>()
1682                .join("\n");
1683            let round_tripped: Vec<Sexp> = tatara_lisp::read_spanned(&text)
1684                .unwrap_or_else(|e| panic!("{src:?}: the old path could not read back: {e:?}"))
1685                .iter()
1686                .map(tatara_lisp::Spanned::to_sexp)
1687                .collect();
1688            assert_eq!(
1689                round_tripped, erased,
1690                "{src:?}: the old print-and-reparse path changed the tree"
1691            );
1692        }
1693    }
1694
1695    /// Anti-vacuity for the test above: the round trip really is *not* the
1696    /// identity in general, so agreeing on the corpus was a property of what
1697    /// blue happens to emit rather than a property of the reader.
1698    ///
1699    /// **`Atom::Symbol`'s `Display` writes the name raw, with no escaping.**
1700    /// `Atom::Str` escapes and its docs explain at length why; the symbol arm
1701    /// is `f.write_str(s)`. So print-then-read is not inverse over the symbol
1702    /// domain, and the failure is *silent*: a symbol containing a space prints
1703    /// as two tokens, reads back as two symbols, and the result is a perfectly
1704    /// well-formed tree with a different meaning. No error, nothing to catch.
1705    ///
1706    /// Measured 2026-08-02 across the separators: `a b` and `x'y` come back
1707    /// `Ok` with a different tree; `x)y`, `x"y` and `x;y` come back `Err`;
1708    /// `x{y` and `x[y` DO round-trip at this level — those two are one symbol
1709    /// in and one symbol out, so the brace-fusion reported in tatara *source*
1710    /// is not what bites a printed tree. The silent pair is what makes this a
1711    /// miscompile class rather than a noisy one.
1712    #[test]
1713    fn the_round_trip_is_not_the_identity_in_general() {
1714        let tree = Sexp::List(vec![
1715            Sexp::Atom(tatara_lisp::Atom::Symbol("f".into())),
1716            Sexp::Atom(tatara_lisp::Atom::Symbol("a b".into())),
1717        ]);
1718        let text = tree.to_string();
1719        let back: Vec<Sexp> = tatara_lisp::read_spanned(&text)
1720            .expect("it reads back cleanly — that IS the problem")
1721            .iter()
1722            .map(tatara_lisp::Spanned::to_sexp)
1723            .collect();
1724        assert_ne!(
1725            back,
1726            vec![tree.clone()],
1727            "if print-then-read became inverse over symbols, the class would be \
1728             closed upstream and this test should be deleted rather than relaxed"
1729        );
1730        // …and the direct lowering is unaffected by any of it.
1731        let direct: Vec<Sexp> = crate::lower_to_spanned(std::slice::from_ref(&tree))
1732            .iter()
1733            .map(tatara_lisp::Spanned::to_sexp)
1734            .collect();
1735        assert_eq!(direct, vec![tree]);
1736    }
1737}
1738
1739#[cfg(test)]
1740mod macro_tests {
1741    use super::*;
1742
1743    fn int(src: &str) -> i64 {
1744        match run(src).unwrap_or_else(|e| panic!("{src:?}: {e}")).value {
1745            Value::Int(v) => v,
1746            other => panic!("{src:?} produced {other:?}"),
1747        }
1748    }
1749
1750    /// **A blue macro expands and runs.** Tenet 2's surface, end to end.
1751    #[test]
1752    fn a_macro_expands_and_runs() {
1753        assert_eq!(
1754            int("defmacro double(x)\n  quote\n    unquote(x) + unquote(x)\n  end\nend\ndouble(21)"),
1755            42
1756        );
1757    }
1758
1759    /// A macro receives *source forms*, not values — so it can duplicate its
1760    /// argument, which a function cannot do without re-evaluating it.
1761    #[test]
1762    fn a_macro_operates_on_syntax_not_values() {
1763        assert_eq!(
1764            int("defmacro sq(e)\n  quote\n    unquote(e) * unquote(e)\n  end\nend\nsq(2 + 3)"),
1765            25,
1766            "the argument form `2 + 3` must be substituted twice"
1767        );
1768    }
1769
1770    /// **A runaway macro is a typed error, not a dead compiler.** This is the
1771    /// property that makes the metaprogramming surface safe to hand to a user.
1772    ///
1773    /// The assertion was `contains("expansion limit")` and now reads
1774    /// `contains("expansion")`, because the pipeline renders `short_message`
1775    /// rather than `Display`: upstream words the same fact as "exceeded 256
1776    /// expansion steps" instead of "exceeded the expansion limit of 256
1777    /// rewrites". The PROPERTY is unchanged and still asserted — the macro is
1778    /// named, and the failure is attributed to expansion rather than to
1779    /// evaluation. **This reword was the whole measured blast radius of that
1780    /// switch**, one test in the workspace.
1781    ///
1782    /// The number is deliberately NOT pinned. It is upstream's constant, this
1783    /// test owns none of it, and matching it would turn an upstream bump into
1784    /// a red run here that proves nothing about blue.
1785    ///
1786    /// The third clause is new capability rather than repair: the message now
1787    /// carries `<anonymous>:6:1`, so a runaway macro says WHERE it ran away.
1788    #[test]
1789    fn a_runaway_macro_fails_the_compilation_rather_than_the_process() {
1790        let err =
1791            run("defmacro forever(x)\n  quote\n    forever(unquote(x))\n  end\nend\nforever(1)")
1792                .expect_err("a self-referential macro must be rejected");
1793        let msg = err.to_string();
1794        assert!(
1795            msg.contains("forever") && msg.contains("expansion"),
1796            "the error must name the macro and the limit it hit: {msg}"
1797        );
1798        assert!(
1799            msg.contains("<anonymous>:6:1"),
1800            "and must place the call that ran away — line 6 is `forever(1)`: {msg}"
1801        );
1802    }
1803}
1804
1805#[cfg(test)]
1806mod input_tests {
1807    use super::*;
1808    use crate::inputs::{Declaration, Inputs};
1809
1810    /// A schema a macro will generate code from.
1811    const SCHEMA: &[u8] = b"3";
1812
1813    fn with_schema(src: &str) -> Result<Run, RunError> {
1814        let hash = Inputs::hash_of(SCHEMA);
1815        let mut inputs = Inputs::new();
1816        inputs
1817            .bind(
1818                &Declaration {
1819                    name: "schema".to_string(),
1820                    hash,
1821                },
1822                SCHEMA.to_vec(),
1823            )
1824            .expect("bind");
1825        run_with_inputs(src, inputs)
1826    }
1827
1828    fn decl_line() -> String {
1829        let mut s = String::from("definput(\"schema\", \"");
1830        s.push_str(&Inputs::hash_of(SCHEMA));
1831        s.push_str("\")\n");
1832        s
1833    }
1834
1835    /// **A macro reads a declared build input.** This is §VI OPEN #6 closed —
1836    /// the spec names it as gating blue's whole "stronger than Ruby's
1837    /// metaprogramming" claim, because a macro that cannot read a schema cannot
1838    /// generate code from one.
1839    #[test]
1840    fn a_macro_can_read_a_declared_build_input() {
1841        let src = decl_line() + "input(\"schema\")";
1842        let out = with_schema(&src).expect("run");
1843        assert!(
1844            matches!(out.value, Value::Str(ref s) if &**s == "3"),
1845            "got {:?}",
1846            out.value
1847        );
1848    }
1849
1850    /// **An undeclared input is an error, not a file read and not nil.**
1851    /// Returning nil is how a macro generates an empty table and nobody notices
1852    /// until runtime.
1853    #[test]
1854    fn an_undeclared_input_is_an_error() {
1855        let err = with_schema("input(\"not_declared\")").expect_err("must fail");
1856        let msg = err.to_string();
1857        assert!(msg.contains("not_declared"), "must name it: {msg}");
1858        assert!(msg.contains("definput"), "and say how to declare it: {msg}");
1859    }
1860
1861    /// **There is no path-based read at all.** The capability is the absence of
1862    /// the primitive, not a check inside one — so this is an unbound symbol.
1863    ///
1864    /// Holds for the DEFAULT surface — the one every embedder gets. The `sys`
1865    /// cargo feature (CLI only) is the one declared exception: it is the
1866    /// operator's own trusted host surface, and is asserted in
1867    /// `sys_read_file_is_the_trusted_cli_only_exception` below.
1868    #[cfg(not(feature = "sys"))]
1869    #[test]
1870    fn there_is_no_ambient_file_read() {
1871        for attempt in [
1872            "read_file(\"/etc/passwd\")",
1873            "File(\"/etc/passwd\")",
1874            "slurp(\"/etc/passwd\")",
1875            "open(\"/etc/passwd\")",
1876        ] {
1877            let err = with_schema(attempt).expect_err("must not resolve");
1878            assert!(
1879                err.to_string().contains("unbound"),
1880                "{attempt} must be UNBOUND — a capability removed by absence, \
1881                 not guarded by a check: {err}"
1882            );
1883        }
1884    }
1885
1886    /// With the `sys` feature compiled in, `read_file` IS bound — that is the
1887    /// point of the feature. The doctrine does not move: this is the operator's
1888    /// own machine (the CLI), not an embedder's sandbox. Pin the boundary so a
1889    /// future default-build change is heard, and assert that `input()` still
1890    /// works beside it.
1891    #[cfg(feature = "sys")]
1892    #[test]
1893    fn sys_read_file_is_the_trusted_cli_only_exception() {
1894        let err = with_schema("definitely_not_a_primitive(\"x\")").expect_err("must not resolve");
1895        assert!(err.to_string().contains("unbound"), "{err}");
1896        assert!(
1897            with_schema("read_file(\"/etc/passwd\")").is_ok(),
1898            "with `sys` on, read_file is the trusted CLI surface"
1899        );
1900        let out = with_schema("input(\"schema\")").expect("run");
1901        assert!(
1902            matches!(out.value, Value::Str(ref s) if &**s == "3"),
1903            "input() still binds beside the sys surface: {:?}",
1904            out.value
1905        );
1906    }
1907
1908    /// Anti-vacuity: with no inputs supplied at all, even a declared name fails
1909    /// — so the success above is the binding's doing.
1910    #[test]
1911    fn a_declared_input_with_no_bytes_supplied_fails() {
1912        let src = decl_line() + "input(\"schema\")";
1913        assert!(run(&src).is_err(), "no bytes were supplied");
1914    }
1915}
1916
1917#[cfg(test)]
1918mod tier2_tests {
1919    use super::*;
1920    use crate::inputs::{Declaration, Inputs};
1921
1922    /// **The Tier-2 conversion §V.6.3 said was gated: a macro that emits real
1923    /// declarations FROM A SCHEMA.**
1924    ///
1925    /// `theory/BLUE.md` §VI OPEN #6 states the blocker plainly — "tenet 2
1926    /// installs a `NoLoader`, so a macro cannot read a schema — which gates
1927    /// every Tier-2 conversion in §V.6 and therefore blue's whole 'stronger than
1928    /// Ruby's metaprogramming' claim."
1929    ///
1930    /// Here the schema supplies a *value the generated code depends on*, read at
1931    /// expansion time. Ruby and Elixir can both do this — with the whole
1932    /// filesystem open. blue does it through a name bound to a content hash.
1933    #[test]
1934    fn a_macro_generates_code_from_a_schema() {
1935        let schema = b"7";
1936        let mut inputs = Inputs::new();
1937        inputs
1938            .bind(
1939                &Declaration {
1940                    name: "arity".to_string(),
1941                    hash: Inputs::hash_of(schema),
1942                },
1943                schema.to_vec(),
1944            )
1945            .expect("bind");
1946
1947        // The macro reads the input at EXPANSION time and splices the value it
1948        // found into the code it emits.
1949        let mut src = String::from("definput(\"arity\", \"");
1950        src.push_str(&Inputs::hash_of(schema));
1951        src.push_str("\")\n");
1952        src.push_str(
1953            "defmacro from_schema()\n  quote\n    unquote(to_int(input(\"arity\")))\n  end\nend\n\
1954             from_schema() * 6",
1955        );
1956
1957        let out = run_with_inputs(&src, inputs).expect("run");
1958        assert!(
1959            matches!(out.value, Value::Int(42)),
1960            "the schema's 7 must reach the generated code: got {:?}",
1961            out.value
1962        );
1963    }
1964}