Files
pplus-core/tests/reference.test.mjs
T
alexmandClaude be0db2d74f Widen pseudo for numerical work: sized types, FLOAT32, exact built-ins, USE IEEE
Pseudo could say what a procedure computes only up to INTEGER and FLOAT,
so it could not state what a Mojo or Python program with Int32, UInt8 or
Float32 values will compute, and it had no words for rounding itself.

- lang/parse: INT8..INT64, UINT8..UINT64, FLOAT32, FLOAT64 as types
  (INTEGER or FLOAT with a width, so every "whole number" rule accepts
  them); conversions written as Mojo writes them, INT32(X), FLOAT32(X);
  1_000_000, 0xFF and 0b1010 literals; a type name as an argument
  (MAX_VALUE(INT8)); the USE line (USE IEEE now, library modules next).
- check: widths and precisions never mix silently (mixed-width,
  mixed-precision); a literal takes the type of the sized value beside
  it, as in Mojo, and one that does not fit is an error before any run
  (literal-range); typing for every new built-in.
- interp: a sized result, store, parameter, RETURN or conversion that
  does not fit stops the run (kind overflow), WRAP(X, T) wraps on
  purpose; FLOAT32 results are rounded to single precision; USE IEEE
  lets NaN and infinities flow.
- numeric.js: exact helpers: FMA rounded once, EXACT(X) with every digit
  a FLOAT holds, ULP and NEXT_AFTER for both precisions, bit operations
  in two's complement.
- Built-ins: ASIN ACOS ATAN SINH COSH TANH HYPOT POW CBRT EXPM1 LOG1P E
  TRUNC SIGN COPYSIGN CLAMP FMA IS_NAN IS_INF IS_FINITE INF NAN ULP
  NEXT_AFTER IS_CLOSE EXACT MAX_VALUE MIN_VALUE EPSILON GCD LCM BIT_AND
  BIT_OR BIT_XOR BIT_NOT SHIFT_LEFT SHIFT_RIGHT WRAP. They stay usable
  as names (a variable E, a FUNCTION GCD): the program's own wins.
- vcgen: an overflow obligation on every sized result, then the range
  as a fact (also for sized parameters and loop-cut variables); FLOAT32
  arithmetic as its own uninterpreted functions, constants rounded
  exactly, TOFLOAT32(v) otherwise; argument obligations for SQRT, the
  logarithms, ASIN/ACOS, LOG1P, CLAMP and the shifts; nothing proved
  under USE IEEE.
- testing: the fuzzer tries the edges of a sized type and generates
  FLOAT32 values.
- pseudo-to-mojo: Int8..Float32, conversions, the new built-ins through
  std.math (each name in the harvested index), a note where Mojo wraps.
- Tests: numeric.test.mjs; corpus 49-52; prover fixtures int8-overflow,
  float32-store, ieee-nan; the reference documents every new form.

Gate: 324 pass (core). New Mojo uses std.math names the interpreter has
not been verified on (listed in sync-mojo's KNOWN_GAPS); they need
pixi run mojo-verify on the network before the interpreter runs them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FgvbRdoEheAHqsRmr2WuKZ
2026-10-05 11:43:53 +00:00

131 lines
8.5 KiB
JavaScript

// The reference is only trustworthy if its examples run through the real parsers.
import test from 'node:test';
import assert from 'node:assert/strict';
import { REFERENCE, PSEUDO_REF, LINGO_REF, SPECO_REF, ASM_REF, MATHO_REF, explain, SPECO_SYMBOLS } from '../dist/common/reference.js';
import { parseMatho } from '../dist/services/matho.js';
import { studyOf } from '../dist/services/pseudo.js';
import { readLingo } from '../dist/services/lingo.js';
import { parseSpeco, toAscii, toMath } from '../dist/services/speco.js';
import { parseAsm } from '../dist/services/asm.js';
import { KEYWORDS, BUILTINS, SOFT_BUILTINS } from '../dist/engines/pseudo/lang.js';
import { OPS } from '../dist/engines/asm86/interp.js';
test('every pseudocode statement example parses and checks', () => {
const sec = PSEUDO_REF.sections.find((s) => s.id === 'pseudo-statements');
for (const e of sec.entries) {
const st = studyOf(e.example);
const errs = st.diagnostics.filter((d) => d.severity === 'error');
assert.equal(errs.length, 0, `${e.term}: ${errs.map((d) => `L${d.line} ${d.message}`).join(' | ')}`);
}
});
test('every keyword in the engine has a meaning in the reference, and vice versa', () => {
const sec = PSEUDO_REF.sections.find((s) => s.id === 'pseudo-keywords');
const listed = new Set(sec.entries.map((e) => e.term));
for (const k of KEYWORDS) assert.ok(listed.has(k), `keyword ${k} is missing from the reference`);
for (const e of sec.entries) assert.ok(e.meaning.length > 10, `${e.term} has no meaning`);
for (const b of Object.keys(BUILTINS)) assert.ok(explain('pseudo', b), `built-in ${b} has no hover`);
});
test('every lingo sentence example reads (inside a minimal routine where needed)', () => {
const failures = [];
// kinds for every name the examples mention, so a fragment reads the same way it does on the page
const kinds = 'T is a text. P is a text. S is a text. w is a text. A is a whole number. Score is a whole number.';
const setup = [
'Start with Count at 0.', 'Start with Best at 0.', 'Start with Run at 0.', 'Start with Left at 3.', 'Start with window at "AC".',
'Start with an empty list called Starts.', 'Start with an empty tally called Counts.', 'Start with an empty text called R.',
];
const defs = "The complement of b is 'T' when b is 'A', and 'N' otherwise.";
for (const sec of LINGO_REF.sections.filter((s) => s.id.startsWith('lingo-forms'))) {
for (const e of sec.entries) {
if (!e.example) continue;
const ex = e.example.trim();
const examples = ex.includes(' ') && !/^To /.test(ex) && !/ when /.test(ex) ? ex.split(/\s{2,}/) : [ex];
for (const one of examples) {
let src;
const head = 'To test T, P, S, w, A, B and Score:';
const pre = setup.map((l) => ' ' + l).join('\n');
if (/^The answer is/.test(one)) src = `${head}\n${pre}\n Start with i at 0.\n ${one}\n\n${kinds}`;
else if (/^To /.test(one)) src = `${one}\n${pre}\n The answer is Count.\n\n${kinds}`;
else if (/ when /.test(one) || /^The .* is /.test(one)) src = `To test P, T, S and b:\n Start with i at 0.\n If b is 'A', add 1 to i.\n The answer is the length of T plus the length of P plus the length of S plus i.\n\n${one}${/[.:]$/.test(one) ? '' : '.'}\n\nP is a text. T is a text. S is a text. b is a letter.`;
else if (/^(For|Repeat)/.test(one)) src = `${head}\n${pre}\n${/^For (each|every) .*\bi\b/.test(one) ? '' : ' Start with i at 0.\n'} ${one}\n Add 1 to Count.\n The answer is Count.\n\n${kinds}`;
else if (/^If /.test(one) && /:$/.test(one)) src = `${head}\n${pre}\n ${one}\n Add 1 to Count.\n The answer is Count.\n\n${kinds}`;
else if (sec.title === 'Tests') src = `${head}\n${pre}\n For each position i of T:\n Set b to the letter of T at i.\n ${/^If /.test(one) ? one : `If ${one}, add 1 to Count.`}\n The answer is Count.\n\n${kinds}`;
else if (sec.title === 'Values') src = `${head}\n${pre}\n For each position i of T:\n Set X to ${one}.\n The answer is Count.\n\n${kinds}`;
else src = `${head}\n${pre}\n Start with i at 0.\n${/\bb\b/.test(one) && !/\bB\b/.test(one) ? ' Set b to the letter of T at i.\n' : ''} ${one}\n The answer is Count.\n\n${/complement/.test(one) ? defs + '\n\n' : ''}${kinds}${/^Swap/.test(one) ? ' B is a whole number.' : ''}`;
const r = readLingo(src);
const tolerated = ['pronoun', 'kind-clash', 'unset-read', 'borrowed-input', 'which-letter', 'never-read', 'unreachable'];
const errs = r.diagnostics.filter((d) => d.severity === 'error' && !tolerated.includes(d.id));
if (errs.length) failures.push(`${e.term} → "${one}": ${errs.map((d) => d.id + ': ' + d.message.slice(0, 100)).join(' | ')}`);
}
}
}
assert.deepEqual(failures, []);
});
test('every lingo ambiguity example raises exactly the question it documents', () => {
const sec = LINGO_REF.sections.find((s) => s.id === 'lingo-questions');
for (const e of sec.entries) {
const r = readLingo(e.example);
assert.ok(r.diagnostics.some((d) => d.id === e.term), `${e.term}: got ${r.diagnostics.map((d) => d.id).join(', ')}`);
}
});
test('every speco example parses, and the symbol table round-trips', () => {
const sec = SPECO_REF.sections.find((s) => s.id === 'speco-forms');
for (const e of sec.entries) {
const m = parseSpeco(e.example);
assert.ok(m.ok, `${e.term}: ${m.diagnostics.map((d) => d.message).join(' | ')}`);
}
for (const s of SPECO_SYMBOLS) {
if (s.math.length !== 1) continue;
assert.equal(toAscii(`A ${s.math} B`), `A ${s.ascii} B`);
assert.equal(toMath(`A ${s.ascii} B`), `A ${s.math} B`);
}
});
test('every instruction has a gloss and a hover; setup examples parse', () => {
for (const op of Object.keys(OPS)) assert.ok(explain('asm', op)?.body, `${op} has no gloss`);
for (const e of ASM_REF.sections.find((s) => s.id === 'asm-registers').entries) assert.ok(explain('asm', e.term), e.term);
assert.ok(parseAsm(' mov rax, [rbx]\n ret\n').ok);
assert.ok(REFERENCE.length === 5);
});
test('every pseudo idiom runs and prints what its note says', async () => {
const { runPseudo } = await import('../dist/services/pseudo.js');
const { KEYWORDS: KW, BUILTINS: BI, CONTEXTUAL } = await import('../dist/engines/pseudo/lang.js');
const sec = PSEUDO_REF.sections.find((s) => s.id === 'pseudo-idioms');
assert.ok(sec.entries.length >= 8);
for (const e of sec.entries) {
const st = studyOf(e.example);
const errs = st.diagnostics.filter((d) => d.severity === 'error');
assert.equal(errs.length, 0, `${e.term}: ${errs.map((d) => `L${d.line} ${d.message}`).join(' | ')}`);
const tr = runPseudo(st, { input: [], maxSteps: 200000 });
assert.equal(tr.error, null, `${e.term}: ${tr.error?.message}`);
const out = tr.output.filter((l) => !/^\w+ returned /.test(l));
assert.deepEqual(out, [e.note], e.term);
}
const reserved = PSEUDO_REF.sections.find((s) => s.id === 'pseudo-reserved');
const words = new Set(reserved.entries.map((e) => e.term));
for (const k of KW) assert.ok(words.has(k), `reserved list misses keyword ${k}`);
for (const k of Object.keys(CONTEXTUAL)) assert.ok(words.has(k), `reserved list misses ${k}`);
// the built-ins added for numerical work stay usable as names (a variable E, a FUNCTION GCD), so they are not listed
for (const k of Object.keys(BI)) assert.ok(SOFT_BUILTINS.has(k) ? !words.has(k) : words.has(k), SOFT_BUILTINS.has(k) ? `${k} can be a name, so it is not reserved` : `reserved list misses built-in ${k}`);
});
test('every matho example translates to pseudo that checks, and every refusal is refused', () => {
const forms = MATHO_REF.sections.find((s) => s.id === 'matho-forms');
for (const e of forms.entries) {
const r = parseMatho(e.example);
const errs = r.diagnostics.filter((d) => d.severity === 'error');
assert.ok(r.ok, `${e.term}: ${errs.map((d) => `L${d.line} ${d.id}: ${d.message}`).join(' | ')}\n${r.program.pseudo}`);
const st = studyOf(r.program.pseudo);
assert.ok(st.ok, `${e.term}: the pseudo does not check: ${st.diagnostics.filter((d) => d.severity === 'error').map((d) => d.message).join(' | ')}`);
}
for (const src of ['y = \\int_0^1 x dx', 's = \\sum_{i=0}^{\\infty} x_i', 'y = \\lim_{x \\to 0} x', 'y = \\partial f', 'y = a \\pm b', 'a = b = c']) {
const r = parseMatho(src);
assert.equal(r.ok, false, `${src} should be refused`);
assert.ok(r.diagnostics.some((d) => d.severity === 'error' && /matho-(refused|pm|chain)/.test(d.id)), src);
}
});