Add textbook formalization pipeline (extract -> LLM formalize -> elaboration check -> library assembly)

- 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.

textbook/README.md

@@ -0,0 +1,133 @@
+# Textbook formalization pipeline
+
+把一本教科书（目前是 Goodman, *Algebra: Abstract and Concrete*）里正文的
+Theorem/Proposition/Lemma/Corollary 转成 lean-bench 能跑的任务 JSONL，配套
+Definition 清单（优先复用 Mathlib，Mathlib 没有的才手写），再由 lean-agent 在
+一个真正带 Mathlib 的 Lean 项目里验证/证明。这条流水线本身不跑 Lean，也不联网
+以外没有第三方依赖。
+
+## 这条流水线不做什么
+
+- 不产出证明。生成的只是**待证命题**（`statement` 字段），证明留给
+  `lean-agent`/`lean-bench` 去跑。
+- 不在 Cowork 沙盒里验证 Lean 语句能不能通过类型检查——沙盒里没有装
+  Lean/Lake/Mathlib（架构也不兼容：本机 `~/.elan` 下是 macOS 二进制，沙盒是
+  Linux VM，跑不了），所以 `check_statements.py` 必须在你本机、且是一个已经
+  `lake exe cache get` 过的 Mathlib 项目里跑，见下面"本机环境准备"。
+- 不会把教科书 PDF 或者大段原文提取结果提交进 git——`.gitignore` 里已经排除
+  了 `*.pdf` 和 `textbook/*.txt`，这些是版权内容，只应该留在你本机。只有代码、
+  生成的最终数据集（短的 Lean 语句 + 自己改写的简短英文释义）会进仓库。
+
+## 用法
+
+### 1. 从 PDF 提取一章的版式文本
+
+```sh
+pdftotext -layout -f <起始pdf页码> -l <结束pdf页码> uiowa-algebra.pdf textbook/chapterNN_layout.txt
+```
+
+页码用 `pdfinfo`/`pypdf` 先核对——书里印的页码和 PDF 物理页码不是一回事（前
+面有目录、序言）。
+
+### 2. 切出一个 section，列出带编号的条目清单
+
+```sh
+python3 textbook/extract_section.py textbook/chapterNN_layout.txt \
+  --start-heading "1.6. Divisibility in the Integers" \
+  --end-heading "1.7. Modular Arithmetic" \
+  --section-prefix "1.6." \
+  --out-text textbook/section_1_6.txt \
+  --out-index textbook/section_1_6.index.json
+```
+
+`--start-heading`/`--end-heading` 必须和版式文本里独占一行的章节标题**逐字**
+一致（提取时打印出的书本身的标题行）；`--end-heading` 留空就切到文件末尾。
+输出的 index JSON 是这个 section 里所有 `Theorem/Proposition/Lemma/
+Corollary/Definition N.N.N` 开头的条目清单（自动去重了跨页重复的标签），用来
+让下一步的 LLM 不漏项、不编项目。
+
+### 3. 用 LLM 把 section 形式化成候选 Lean 语句
+
+```sh
+export DASHSCOPE_API_KEY=...            # 或者写进 lean-bench/.env（已 gitignore）
+python3 textbook/formalize.py \
+  textbook/section_1_6.txt textbook/section_1_6.index.json \
+  --model qwen-max --base-url https://dashscope.aliyuncs.com/compatible-mode/v1 \
+  --out-dataset datasets/goodman-1.6.jsonl \
+  --out-definitions textbook/goodman-1.6.definitions.json \
+  --out-excluded textbook/goodman-1.6.excluded.jsonl
+```
+
+模型会为每条 Theorem/Proposition/Lemma/Corollary 产出一个 Lean 4 命题
+（`statement`）、需要的 `prelude`（通常就是 `import Mathlib`）、一句自己改写
+的英文释义，以及用到的 Mathlib 标识符列表；为每条相关的 Definition 产出一个
+清单条目——如果 Mathlib 已经有对应概念（`mathlib_ref` 非空）就不写 Lean 代
+码，只有 Mathlib 没有的概念才会给出手写的 `def`。判断这条目根本没法形式化
+（纯计算题、内容太模糊/被 PDF 提取污染到读不出原意）的条目，会带着
+`include_in_bench: false` 和原因单独写进 `--out-excluded`，不会进真正的数据
+集，但也不会被静默丢弃——都留痕方便你复核。
+
+源 PDF 的数学符号提取质量很差（≡、∈、上下标经常丢失或乱码），这一步依赖模
+型自己对标准教材内容的知识去猜测/还原真实含义，**不是逐字翻译提取出来的文
+本**。所以生成的候选语句必须经过下一步验证，不能直接信。
+
+### 4. 在你本机验证候选语句能不能通过类型检查
+
+见下面"本机环境准备"先搭好一个带 Mathlib 的 Lake 项目，然后：
+
+```sh
+python3 textbook/check_statements.py datasets/goodman-1.6.jsonl \
+  --lean-project /path/to/your/mathlib-project \
+  --lean-command '["lake","env","lean"]' \
+  --report textbook/goodman-1.6.elab-report.jsonl
+```
+
+这一步只检查"这句 Lean 能不能通过类型检查"（用 `sorry` 占位），**不检查证明
+是否存在、也不检查是否忠实于原书**——分别对应两件不同的事：语法/引用是否正
+确，跟数学内容是否正确。失败的条目（`FAIL`）要么是模型编造了不存在的
+Mathlib 标识符，要么是语句本身写错了，需要你告诉我具体哪条、Lean 报的什么
+错，我再改。
+
+### 5. 用 lean-agent/lean-bench 真正证明
+
+```sh
+lean-bench datasets/goodman-1.6.jsonl --output results/goodman-1.6-001 \
+  --lean-project /path/to/your/mathlib-project --lean-command '["lake","env","lean"]' \
+  --model ... --base-url ...
+```
+
+### 6. 把证出来的定理组装成一份可复用的库文件
+
+```sh
+python3 textbook/build_library.py datasets/goodman-1.6.jsonl \
+  results/goodman-1.6-001/results.jsonl --out lib/Chapter01Section06.lean
+```
+
+只有这次跑真的证出来的定理（`results.jsonl` 里 `success: true`）才会被收进
+去，直接复用验证时确认过的那个证明，不会引入新的证明风险。后续章节要用这些
+已证定理时，`import` 这份文件——注意这只是把已证结果攒成一个文件，不是一个真
+正的 Lake 包/模块边界。
+
+## 本机环境准备：一个带 Mathlib 的 Lake 项目
+
+Cowork 这边的沙盒没有 Lean，`check_statements.py` 和真正跑 `lean-agent`/
+`lean-bench` 都需要在你本机（或者别的有 Lean 的环境）完成。如果还没有这样一
+个项目：
+
+```sh
+cd ~/ProjectResearch   # 随便一个地方
+lake +leanprover/lean4:v4.33.1 new mathlib-bench math
+cd mathlib-bench
+lake exe cache get     # 下载 Mathlib 预编译缓存，比从源码编译快得多
+lake build
+```
+
+之后 `--lean-project` 就指到 `mathlib-bench` 这个目录，`--lean-command` 用
+`'["lake","env","lean"]'`。
+
+## 反馈循环
+
+我（claude-bot）在 ProjectResearch 里生成/修改数据集和代码 → 提交并 push 到
+`Zayden-Jung/lean-bench`（用 claude-bot 的 token） → 你在本机 `git pull` →
+跑 `check_statements.py` / `lean-bench` → 把失败的条目 id 和 Lean 报的错发
+给我 → 我改完再 push。不需要额外工具，就是正常的 git 往返。

textbook/build_library.py

@@ -0,0 +1,82 @@
+"""Assemble a reusable Lean library file from a completed lean-bench run.
+
+Only theorems that actually solved (results.jsonl success == true) go in; each one's
+exact winning proof (its last, successful attempt) is reused verbatim, re-wrapped in the
+same `theorem <id> : <statement> := by ...` shape lean-bench's verifier already checked,
+so the assembled file carries no new proof risk beyond what that run already confirmed --
+this script does not itself call Lean.
+
+Plain `import ...` preludes are deduplicated and hoisted to the top of the file; any
+other prelude content (a local definition, say) is kept attached to its own theorem
+instead of being collapsed, since it isn't necessarily shared with the rest.
+
+A library assembled this way is only as trustworthy as the run it came from: rerun
+check_statements.py / lean-bench yourself before importing it into something else you
+care about, and treat --module-name as documentation, not a real Lean module boundary
+(this script emits one flat .lean file, not a package).
+"""
+import argparse
+import json
+from pathlib import Path
+
+
+def load_jsonl(path):
+    return [json.loads(line) for line in path.read_text(encoding='utf-8').splitlines() if line.strip()]
+
+
+def build(dataset_path, results_path):
+    dataset = {t['id']: t for t in load_jsonl(dataset_path)}
+    results = {r['id']: r for r in load_jsonl(results_path)}
+
+    shared_imports = set()
+    per_theorem_prelude = {}
+    for task_id, task in dataset.items():
+        prelude = task.get('prelude', '').strip()
+        if not prelude:
+            continue
+        if all(line.strip().startswith('import ') for line in prelude.splitlines()):
+            shared_imports.update(line.strip() for line in prelude.splitlines())
+        else:
+            per_theorem_prelude[task_id] = prelude
+
+    lines = ['-- Generated by textbook/build_library.py -- do not hand-edit proofs;',
+             '-- regenerate from a fresh lean-bench run instead.', '']
+    lines.extend(sorted(shared_imports))
+    lines.append('')
+
+    solved, skipped = [], []
+    for task_id, task in dataset.items():
+        result = results.get(task_id)
+        if not result or not result.get('success') or not result.get('attempts'):
+            skipped.append(task_id)
+            continue
+        proof = result['attempts'][-1]['proof']
+        comment = f'-- {task.get("source_label", task_id)}: {task.get("natural_language", "")}'.rstrip()
+        lines.append(comment)
+        if task_id in per_theorem_prelude:
+            lines.append(per_theorem_prelude[task_id])
+        lean_name = task_id.replace('-', '_')
+        lines.append(f'theorem {lean_name} : {task["statement"]} := by')
+        lines.extend('  ' + line for line in proof.splitlines())
+        lines.append('')
+        solved.append(task_id)
+
+    return '\n'.join(lines), solved, skipped
+
+
+def main():
+    parser = argparse.ArgumentParser(description='Assemble a Lean library file from a solved lean-bench dataset')
+    parser.add_argument('dataset', type=Path)
+    parser.add_argument('results', type=Path, help="a lean-bench run's results.jsonl")
+    parser.add_argument('--out', type=Path, required=True)
+    args = parser.parse_args()
+
+    text, solved, skipped = build(args.dataset, args.results)
+    args.out.write_text(text + '\n', encoding='utf-8')
+    print(f'{len(solved)} theorems assembled -> {args.out}')
+    if skipped:
+        print(f'{len(skipped)} skipped (not solved in this run): {", ".join(skipped)}')
+
+
+if __name__ == '__main__':
+    main()

textbook/check_statements.py

@@ -0,0 +1,75 @@
+"""Elaboration-only sanity check for candidate Lean statements from formalize.py.
+
+Checks whether `theorem _ : statement := by sorry` type-checks under the task's
+prelude -- i.e. whether the statement is well-formed Lean using identifiers that
+actually exist. This does NOT verify the statement is provable, or even that it
+faithfully represents the book (that's still on you to spot-check); it only catches
+hallucinated Mathlib names and syntax errors for free, before spending any lean-agent
+proof attempts on them.
+
+Must run on a machine with `lean`/`lake` installed and, for anything using Mathlib
+(almost everything here), a Lake project that already has Mathlib built -- point
+--lean-project at it and pass --lean-command '["lake","env","lean"]'. This script
+cannot run inside a Cowork sandbox: see lean-bench/textbook/README.md.
+"""
+import argparse
+import json
+import subprocess
+import tempfile
+from dataclasses import dataclass
+from pathlib import Path
+
+
+@dataclass
+class ElabResult:
+    ok: bool
+    diagnostics: str
+
+
+def elaborates(prelude, statement, command=None, cwd=None, timeout=30):
+    command = command or ['lean']
+    source = (prelude + '\n' if prelude else '') + 'theorem lean_bench_sanity_check : ' + statement + ' := by sorry\n'
+    with tempfile.TemporaryDirectory(prefix='lean-sanity-', dir=cwd) as directory:
+        path = Path(directory) / 'SanityCheck.lean'
+        path.write_text(source, encoding='utf-8')
+        try:
+            result = subprocess.run([*command, str(path)], cwd=cwd, capture_output=True,
+                                     text=True, timeout=timeout)
+        except subprocess.TimeoutExpired:
+            return ElabResult(False, 'Lean timeout')
+    diagnostics = result.stdout + result.stderr
+    ok = result.returncode == 0 and 'error:' not in diagnostics
+    return ElabResult(ok, diagnostics[-8000:])
+
+
+def main():
+    parser = argparse.ArgumentParser(description='Check that candidate statements elaborate (not that they are provable)')
+    parser.add_argument('dataset', type=Path, help='JSONL with id, statement, optional prelude')
+    parser.add_argument('--lean-command', default='["lean"]', help='JSON argv array; e.g. ["lake","env","lean"]')
+    parser.add_argument('--lean-project', help='Lean/Lake project working directory (needs Mathlib built)')
+    parser.add_argument('--timeout', type=float, default=60)
+    parser.add_argument('--report', type=Path, help='Write JSONL of {id, ok, diagnostics} here')
+    args = parser.parse_args()
+
+    command = json.loads(args.lean_command)
+    tasks = [json.loads(line) for line in args.dataset.read_text(encoding='utf-8').splitlines() if line.strip()]
+    reports = []
+    failed = 0
+    for task in tasks:
+        result = elaborates(task.get('prelude', ''), task['statement'], command, args.lean_project, args.timeout)
+        reports.append({"id": task['id'], "ok": result.ok, "diagnostics": result.diagnostics})
+        print(('OK  ' if result.ok else 'FAIL') + ' ' + task['id'])
+        if not result.ok:
+            failed += 1
+
+    if args.report:
+        with args.report.open('w', encoding='utf-8') as stream:
+            for r in reports:
+                stream.write(json.dumps(r, ensure_ascii=False) + '\n')
+
+    print(f'{len(tasks) - failed}/{len(tasks)} elaborate cleanly')
+    raise SystemExit(1 if failed else 0)
+
+
+if __name__ == '__main__':
+    main()

textbook/extract_section.py

@@ -0,0 +1,73 @@
+"""Slice one book section out of a `pdftotext -layout` dump and list its labeled
+Theorem/Proposition/Lemma/Corollary/Definition statements, as input for formalize.py.
+
+Produces two files: the section's raw text, and a JSON checklist of labeled items
+(kind, label, first line) in book order. The checklist exists so formalize.py's LLM
+step has an explicit inventory to cover and cannot silently skip or invent items.
+"""
+import argparse
+import json
+import re
+from pathlib import Path
+
+LABEL_RE = re.compile(r'^(Definition|Theorem|Proposition|Lemma|Corollary)\s+(\d+\.\d+(?:\.\d+)?)\.?\s*(.*)$')
+
+
+def find_heading_line(lines, heading):
+    for i, line in enumerate(lines):
+        if line.strip() == heading:
+            return i
+    raise ValueError(f'Heading not found: {heading!r}')
+
+
+def extract_section(full_text, start_heading, end_heading):
+    lines = full_text.split('\n')
+    start = find_heading_line(lines, start_heading)
+    end = find_heading_line(lines, end_heading) if end_heading else len(lines)
+    if end <= start:
+        raise ValueError('end_heading must come after start_heading')
+    return '\n'.join(lines[start:end])
+
+
+def labeled_statements(section_text, section_prefix):
+    """Unique labeled items in section_text whose label starts with section_prefix
+    (e.g. '1.6.'), keeping only the first occurrence of each label: later matches are
+    almost always proof continuations across a page break or cross-references, not new
+    statements. Returned sorted in book order (numeric, not lexical)."""
+    seen = {}
+    for line in section_text.split('\n'):
+        line = line.strip()
+        m = LABEL_RE.match(line)
+        if not m:
+            continue
+        kind, label, rest = m.group(1), m.group(2), m.group(3)
+        if not label.startswith(section_prefix) or label in seen:
+            continue
+        seen[label] = {"kind": kind, "label": label, "first_line": line[:120]}
+
+    def sort_key(item):
+        return tuple(int(p) for p in item['label'].split('.'))
+
+    return sorted(seen.values(), key=sort_key)
+
+
+def main():
+    parser = argparse.ArgumentParser(description='Slice one section out of a pdftotext -layout dump')
+    parser.add_argument('layout_txt', type=Path)
+    parser.add_argument('--start-heading', required=True, help="e.g. '1.6. Divisibility in the Integers'")
+    parser.add_argument('--end-heading', help="e.g. '1.7. Modular Arithmetic'; omit for end of file")
+    parser.add_argument('--section-prefix', required=True, help="e.g. '1.6.'")
+    parser.add_argument('--out-text', type=Path, required=True)
+    parser.add_argument('--out-index', type=Path, required=True)
+    args = parser.parse_args()
+
+    full_text = args.layout_txt.read_text(encoding='utf-8')
+    section_text = extract_section(full_text, args.start_heading, args.end_heading)
+    args.out_text.write_text(section_text, encoding='utf-8')
+    index = labeled_statements(section_text, args.section_prefix)
+    args.out_index.write_text(json.dumps(index, ensure_ascii=False, indent=2) + '\n', encoding='utf-8')
+    print(f'{len(index)} labeled statements -> {args.out_index}')
+
+
+if __name__ == '__main__':
+    main()

textbook/formalize.py

@@ -0,0 +1,165 @@
+"""LLM-assisted formalization of a textbook section into Lean 4 candidate statements.
+
+Reads a section's extracted text and labeled-statement checklist (from
+extract_section.py) and asks an LLM to produce:
+  (a) a Lean 4 theorem statement for every genuine Theorem/Proposition/Lemma/Corollary,
+      preferring Mathlib's existing vocabulary over reinventing definitions, and
+  (b) a manifest of every Definition the theorems depend on, pointing at the matching
+      Mathlib name when one exists and only asking for hand-written Lean `def`s when it
+      doesn't.
+
+This step never runs Lean. The extracted PDF text can have corrupted math symbols
+(missing =/in, dropped superscripts) from the source PDF's font encoding; the model is
+told to lean on its own knowledge of standard textbook mathematics plus the section
+context, and to flag low-confidence items rather than guess silently. Always run
+check_statements.py against a real Mathlib-backed Lean project before trusting the
+output, and before spending lean-agent proof attempts on it.
+"""
+import argparse
+import json
+import re
+from pathlib import Path
+
+try:
+    from .llm import ChatClient, load_dotenv  # imported as textbook.formalize
+except ImportError:
+    from llm import ChatClient, load_dotenv  # run directly: python3 textbook/formalize.py
+
+SYSTEM_PROMPT = """You are formalizing a section of an undergraduate abstract algebra \
+textbook (Goodman, "Algebra: Abstract and Concrete") into Lean 4 using Mathlib.
+
+You will receive:
+1. The raw text of one book section, extracted from a PDF. WARNING: the source PDF's \
+math font encoding is broken in this extraction -- symbols like congruence, set \
+membership, comparisons, superscripts and subscripts are frequently dropped or garbled. \
+Use your own knowledge of standard undergraduate number theory / algebra to reconstruct \
+the intended mathematics from context; do not invent content that is not actually in the \
+section.
+2. A checklist of every labeled Theorem/Proposition/Lemma/Corollary/Definition in the \
+section (by number), so you know exactly what to cover -- do not skip any, and do not \
+invent ones that are not listed.
+
+Produce exactly two JSON arrays, "definitions" and "theorems", as a single JSON object \
+inside one ```json fenced code block, and nothing else outside that block.
+
+Each element of "definitions" (only for Definition-labeled entries the theorems below \
+actually depend on):
+{
+  "label": "Definition 1.6.3",
+  "concept": "short name, e.g. 'prime number'",
+  "mathlib_ref": "Mathlib identifier that already means this, e.g. 'Nat.Prime', or null \
+if Mathlib has no equivalent",
+  "lean_code": "a standalone Lean 4 `def ...` ONLY if mathlib_ref is null, else null",
+  "notes": "anything a formalizer should know, e.g. where the book's definition differs \
+from Mathlib's convention"
+}
+
+Each element of "theorems" (one per Theorem/Proposition/Lemma/Corollary in the \
+checklist):
+{
+  "id": "goodman-1-6-6",
+  "source_label": "Theorem 1.6.6",
+  "kind": "theorem",
+  "statement": "the Lean 4 proposition only, i.e. what goes after `theorem name : `, \
+using Mathlib vocabulary (Nat.Prime, Nat.gcd, Int.gcd, etc.) wherever possible instead \
+of reinventing definitions",
+  "prelude": "Lean import lines this statement needs, usually just \\"import Mathlib\\"",
+  "natural_language": "a short original English restatement in your own words (do not \
+quote the source text verbatim)",
+  "requires": ["list of Mathlib identifiers the statement uses"],
+  "include_in_bench": true,
+  "exclude_reason": null
+}
+
+Set "include_in_bench" to false (with a short "exclude_reason") instead of omitting an \
+entry when a checklist item turns out to not be a genuine formalizable claim (e.g. it is \
+actually a definition misfiled under a theorem-style label, a purely numerical one-off \
+computation with no general statement, or the extracted text is too corrupted to recover \
+its meaning confidently) -- still emit the entry so nothing silently disappears, just \
+mark it excluded.
+
+id must be the label with dots and spaces replaced by hyphens, lowercased, prefixed \
+"goodman-". Do not include a proof, "sorry", or "by" -- statement is only the \
+proposition, no theorem keyword or name. Do not wrap the JSON in anything other than a \
+single ```json fenced block."""
+
+
+def user_prompt(section_text, checklist):
+    checklist_lines = '\n'.join(f"- {item['kind']} {item['label']}: {item['first_line']}"
+                                 for item in checklist)
+    return (f"Checklist of labeled items in this section:\n{checklist_lines}\n\n"
+            f"Section text:\n{section_text}")
+
+
+def extract_json_block(text):
+    match = re.search(r'```json\s*\n(.*?)```', text, re.S)
+    payload = match.group(1) if match else text
+    return json.loads(payload)
+
+
+def validate(data):
+    if not isinstance(data, dict) or 'theorems' not in data or 'definitions' not in data:
+        raise ValueError('Response must be a JSON object with "theorems" and "definitions" arrays')
+    theorems, definitions = data['theorems'], data['definitions']
+    if not isinstance(theorems, list) or not isinstance(definitions, list):
+        raise ValueError('"theorems" and "definitions" must be arrays')
+    seen_ids = set()
+    for item in theorems:
+        for field in ('id', 'source_label', 'statement', 'prelude', 'include_in_bench'):
+            if field not in item:
+                raise ValueError(f'theorem entry missing field {field!r}: {item}')
+        if item['id'] in seen_ids:
+            raise ValueError(f'Duplicate theorem id: {item["id"]}')
+        seen_ids.add(item['id'])
+    for item in definitions:
+        for field in ('label', 'concept', 'mathlib_ref'):
+            if field not in item:
+                raise ValueError(f'definition entry missing field {field!r}: {item}')
+    return theorems, definitions
+
+
+def main():
+    parser = argparse.ArgumentParser(description='LLM-formalize a book section into Lean 4 candidate statements')
+    parser.add_argument('section_text', type=Path)
+    parser.add_argument('statement_index', type=Path)
+    parser.add_argument('--out-dataset', type=Path, required=True, help='JSONL of theorems with include_in_bench=true')
+    parser.add_argument('--out-definitions', type=Path, required=True, help='JSON manifest of definitions')
+    parser.add_argument('--out-excluded', type=Path, help='JSONL of theorem entries the model excluded, for review')
+    parser.add_argument('--model', required=True)
+    parser.add_argument('--base-url', default='https://dashscope.aliyuncs.com/compatible-mode/v1')
+    parser.add_argument('--api-key-env', default='DASHSCOPE_API_KEY')
+    parser.add_argument('--env-file', type=Path, default=Path(__file__).resolve().parents[1] / '.env')
+    parser.add_argument('--timeout', type=float, default=300)
+    args = parser.parse_args()
+
+    load_dotenv(args.env_file)
+    client = ChatClient(args.model, args.base_url, args.api_key_env, args.timeout)
+    checklist = json.loads(args.statement_index.read_text(encoding='utf-8'))
+    section_text = args.section_text.read_text(encoding='utf-8')
+    messages = [{"role": "system", "content": SYSTEM_PROMPT},
+                {"role": "user", "content": user_prompt(section_text, checklist)}]
+    reply = client.complete(messages)
+    data = extract_json_block(reply)
+    theorems, definitions = validate(data)
+
+    included = [t for t in theorems if t.get('include_in_bench')]
+    excluded = [t for t in theorems if not t.get('include_in_bench')]
+
+    with args.out_dataset.open('w', encoding='utf-8') as stream:
+        for item in included:
+            stream.write(json.dumps(item, ensure_ascii=False) + '\n')
+    args.out_definitions.write_text(json.dumps(definitions, ensure_ascii=False, indent=2) + '\n', encoding='utf-8')
+    if args.out_excluded:
+        with args.out_excluded.open('w', encoding='utf-8') as stream:
+            for item in excluded:
+                stream.write(json.dumps(item, ensure_ascii=False) + '\n')
+
+    print(f'{len(included)} theorems -> {args.out_dataset}')
+    if excluded:
+        note = f' (see {args.out_excluded})' if args.out_excluded else ' (pass --out-excluded to keep them)'
+        print(f'{len(excluded)} excluded{note}')
+    print(f'{len(definitions)} definitions -> {args.out_definitions}')
+
+
+if __name__ == '__main__':
+    main()

textbook/llm.py

@@ -0,0 +1,46 @@
+"""Minimal OpenAI-compatible chat client, standalone (no third-party deps, no import of
+lean_agent) so this pipeline works regardless of whether lean-agent is pip-installed.
+Mirrors lean_agent.core.HTTPProvider on purpose; kept separate deliberately."""
+import json
+import os
+import urllib.request
+
+
+def load_dotenv(path):
+    """Load KEY=VALUE lines from a .env file into os.environ (does not overwrite vars
+    already set). Silently does nothing if the file does not exist."""
+    if not path.exists():
+        return
+    for line in path.read_text(encoding='utf-8').splitlines():
+        line = line.strip()
+        if not line or line.startswith('#') or '=' not in line:
+            continue
+        key, _, value = line.partition('=')
+        key = key.strip()
+        value = value.strip().strip('"').strip("'")
+        os.environ.setdefault(key, value)
+
+
+class ChatClient:
+    """OpenAI-compatible /chat/completions caller."""
+
+    def __init__(self, model, base_url, api_key_env='DASHSCOPE_API_KEY', timeout=300):
+        self.model = model
+        self.base_url = base_url.rstrip('/')
+        self.api_key_env = api_key_env
+        self.timeout = timeout
+
+    def complete(self, messages, temperature=0.2):
+        key = os.environ.get(self.api_key_env)
+        if not key:
+            raise RuntimeError(f'Missing API key in environment variable {self.api_key_env}')
+        headers = {"Content-Type": "application/json", "Authorization": "Bearer " + key}
+        payload = {"model": self.model, "messages": messages, "temperature": temperature}
+        request = urllib.request.Request(self.base_url + '/chat/completions',
+                                          data=json.dumps(payload).encode(), headers=headers)
+        with urllib.request.urlopen(request, timeout=self.timeout) as response:
+            data = json.load(response)
+        content = data["choices"][0]["message"]["content"]
+        if not isinstance(content, str):
+            raise ValueError("Provider returned non-text content")
+        return content

textbook/section_1_6.index.json

@@ -0,0 +1,117 @@
+[
+  {
+    "kind": "Proposition",
+    "label": "1.6.1",
+    "first_line": "Proposition 1.6.1."
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.2",
+    "first_line": "Proposition 1.6.2. Let a, b, c, u, and v denote integers."
+  },
+  {
+    "kind": "Definition",
+    "label": "1.6.3",
+    "first_line": "Definition 1.6.3. A natural number is prime if it is greater than 1 and not"
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.4",
+    "first_line": "Proposition 1.6.4. Any natural number other than 1 can be written as a"
+  },
+  {
+    "kind": "Theorem",
+    "label": "1.6.6",
+    "first_line": "Theorem 1.6.6. There are infinitely many prime numbers."
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.7",
+    "first_line": "Proposition 1.6.7. Given integers a and d , with d \u0015 1, there exist unique"
+  },
+  {
+    "kind": "Definition",
+    "label": "1.6.8",
+    "first_line": "Definition 1.6.8. A natural number d is the greatest common divisor of"
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.9",
+    "first_line": "Proposition 1.6.9. For integers n and m, let"
+  },
+  {
+    "kind": "Lemma",
+    "label": "1.6.10",
+    "first_line": "Lemma 1.6.10. Let m and n be nonzero integers. If a natural number d"
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.11",
+    "first_line": "Proposition 1.6.11. The natural number nr is the greatest common divisor"
+  },
+  {
+    "kind": "Corollary",
+    "label": "1.6.13",
+    "first_line": "Corollary 1.6.13. Let m and n be nonzero integers, and write d D"
+  },
+  {
+    "kind": "Definition",
+    "label": "1.6.14",
+    "first_line": "Definition 1.6.14. Nonzero integers m and n are relatively prime if"
+  },
+  {
+    "kind": "Corollary",
+    "label": "1.6.15",
+    "first_line": "Corollary 1.6.15. Two nonzero integers m and n are relatively prime if"
+  },
+  {
+    "kind": "Corollary",
+    "label": "1.6.17",
+    "first_line": "Corollary 1.6.17. Suppose that a and b are relatively prime natural num-"
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.18",
+    "first_line": "Proposition 1.6.18. If p is a prime number and a is any nonzero integer,"
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.19",
+    "first_line": "Proposition 1.6.19. Let p be a prime number, and a and b nonzero inte-"
+  },
+  {
+    "kind": "Corollary",
+    "label": "1.6.20",
+    "first_line": "Corollary 1.6.20. Suppose that a prime number p divides a product"
+  },
+  {
+    "kind": "Theorem",
+    "label": "1.6.21",
+    "first_line": "Theorem 1.6.21. The prime factorization of a natural number is unique."
+  },
+  {
+    "kind": "Definition",
+    "label": "1.6.22",
+    "first_line": "Definition 1.6.22. A natural number d is the greatest common divisor of"
+  },
+  {
+    "kind": "Lemma",
+    "label": "1.6.23",
+    "first_line": "Lemma 1.6.23. Given nonzero integers a1 ; a2 ; : : : ; an (n \u0015 2), there is a"
+  },
+  {
+    "kind": "Proposition",
+    "label": "1.6.24",
+    "first_line": "Proposition 1.6.24. The greatest common divisor of nonzero inte-"
+  },
+  {
+    "kind": "Definition",
+    "label": "1.6.25",
+    "first_line": "Definition 1.6.25. We say that non-zero integers a1 ; : : : ; an are relatively"
+  },
+  {
+    "kind": "Lemma",
+    "label": "1.6.26",
+    "first_line": "Lemma 1.6.26. Let a1 ; : : : ; an be pairwise relatively prime integers, and"
+  }
+]