claude-bot
caedf2e6f4
- extract_section.py: slice a book section out of a pdftotext -layout dump and list its labeled Theorem/Proposition/Lemma/Corollary/Definition items. - formalize.py: LLM-formalize a section into Lean 4 candidate statements (preferring Mathlib vocabulary) plus a definitions manifest. - check_statements.py: elaboration-only sanity check against a real Mathlib-backed Lean project (must run outside the Cowork sandbox, which has no Lean toolchain and can't run macOS binaries anyway). - build_library.py: assemble solved lean-bench results into a reusable .lean file, reusing each winning proof verbatim. - llm.py: standalone OpenAI-compatible chat client + .env loader, kept independent of lean_agent so this pipeline works without it installed. - Source PDFs and raw/layout text dumps are gitignored (copyrighted material); only code and the final short Lean statement datasets are tracked.
59 lines
2.1 KiB
Python
59 lines
2.1 KiB
Python
import json
|
|
import unittest
|
|
|
|
from textbook.formalize import extract_json_block, validate
|
|
|
|
VALID_REPLY = """Here is the result:
|
|
```json
|
|
{
|
|
"definitions": [
|
|
{"label": "Definition 1.6.3", "concept": "prime number", "mathlib_ref": "Nat.Prime",
|
|
"lean_code": null, "notes": ""}
|
|
],
|
|
"theorems": [
|
|
{"id": "goodman-1-6-6", "source_label": "Theorem 1.6.6", "kind": "theorem",
|
|
"statement": "\\u2200 n : \\u2115, \\u2203 p, p > n \\u2227 p.Prime",
|
|
"prelude": "import Mathlib", "natural_language": "There are infinitely many primes.",
|
|
"requires": ["Nat.Prime"], "include_in_bench": true, "exclude_reason": null}
|
|
]
|
|
}
|
|
```
|
|
"""
|
|
|
|
|
|
class FormalizeParsingTests(unittest.TestCase):
|
|
def test_extract_json_block_from_fenced_reply(self):
|
|
data = extract_json_block(VALID_REPLY)
|
|
self.assertEqual(set(data), {"definitions", "theorems"})
|
|
|
|
def test_extract_json_block_without_fence_falls_back_to_whole_text(self):
|
|
payload = '{"theorems": [], "definitions": []}'
|
|
self.assertEqual(extract_json_block(payload), json.loads(payload))
|
|
|
|
def test_validate_accepts_well_formed_response(self):
|
|
data = extract_json_block(VALID_REPLY)
|
|
theorems, definitions = validate(data)
|
|
self.assertEqual(len(theorems), 1)
|
|
self.assertEqual(len(definitions), 1)
|
|
|
|
def test_validate_rejects_missing_field(self):
|
|
data = {"theorems": [{"id": "x", "source_label": "Theorem 1", "statement": "True"}],
|
|
"definitions": []}
|
|
with self.assertRaises(ValueError):
|
|
validate(data)
|
|
|
|
def test_validate_rejects_duplicate_ids(self):
|
|
entry = {"id": "dup", "source_label": "Theorem 1", "kind": "theorem", "statement": "True",
|
|
"prelude": "import Mathlib", "include_in_bench": True}
|
|
data = {"theorems": [entry, dict(entry)], "definitions": []}
|
|
with self.assertRaises(ValueError):
|
|
validate(data)
|
|
|
|
def test_validate_rejects_non_object_response(self):
|
|
with self.assertRaises(ValueError):
|
|
validate([])
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|