"""Build a two-sheet human checklist from agent-curated, source-mapped JSON.""" import argparse import json import math from pathlib import Path LAYOUT = { 'procurement': ('采购内容总览', ('category', 'item', 'summary', 'action'), ('类别', '采购内容（用途）', '核心要求', '需准备 / 确认'), (14, 34, 59, 48)), 'commercial': ('商务履约清单', ('topic', 'agreement', 'numbers', 'action'), ('事项', '核心约定', '关键数字', '需准备 / 确认'), (18, 55, 23, 55)), } def validate(data): if not isinstance(data, dict): raise ValueError('input must be an object') if not isinstance(data.get('title'), str) or not data['title'].strip(): raise ValueError('title is required') sources = data.get('sources', []) if not isinstance(sources, list) or not sources: raise ValueError('sources are required') source_ids = set() for source in sources: sid = source.get('id') if not isinstance(sid, str) or not sid.strip() or sid in source_ids or not source.get('name'): raise ValueError('invalid or duplicate source id/name') source_ids.add(sid) requirements = data.get('requirements', []) if not isinstance(requirements, list) or not requirements: raise ValueError('requirements are required') reqs = {} for req in requirements: rid = req.get('id') if not isinstance(rid, str) or not rid.strip() or rid in reqs: raise ValueError('invalid or duplicate requirement id') if req.get('source_id') not in source_ids or not req.get('locator'): raise ValueError(f'{rid}: unknown source or missing locator') if not isinstance(req.get('critical'), bool): raise ValueError(f'{rid}: critical must be boolean') tokens = req.get('must_keep', []) if not isinstance(tokens, list) or any(not isinstance(t, str) or not t.strip() for t in tokens): raise ValueError(f'{rid}: invalid must_keep') if req['critical'] and not tokens: raise ValueError(f'{rid}: critical requirement needs explicit preserved terms') reqs[rid] = req mapped = {rid: [] for rid in reqs} warnings = [] for key, (title, fields, _, _) in LAYOUT.items(): rows = data.get(key) if not isinstance(rows, list) or not rows: raise ValueError(f'{key}: nonempty list required') seen = set() for index, row in enumerate(rows, 1): if any(not isinstance(row.get(f), str) for f in fields): raise ValueError(f'{key} row {index}: visible fields must be text') if not row[fields[0]].strip() or not row[fields[1]].strip(): raise ValueError(f'{key} row {index}: empty item') identity = tuple(row[f].strip() for f in fields) if identity in seen: raise ValueError(f'{key}: duplicate visible row') seen.add(identity) refs = row.get('refs') if not isinstance(refs, list) or not refs or any(not isinstance(r, str) or r not in reqs for r in refs): raise ValueError(f'{key} row {index}: missing or unknown refs') text = '\n'.join(row[f] for f in fields) for rid in set(refs): mapped[rid].append(text) if any(len(row[f]) > 80 for f in fields): warnings.append(f'{title} row {index}: long cell; review before publication') for rid, req in reqs.items(): if not mapped[rid]: raise ValueError(f'{rid}: requirement omitted from summary') joined = '\n'.join(mapped[rid]) for token in req.get('must_keep', []): if token not in joined: raise ValueError(f'{rid}: preserved term missing: {token}') return {'source_count': len(sources), 'requirement_count': len(reqs), 'critical_count': sum(r['critical'] for r in reqs.values()), 'sheets': {v[0]: len(data[k]) for k, v in LAYOUT.items()}, 'warnings': warnings, 'coverage': 'all mapped; semantic review still required'} def build(data, output): report = validate(data) output = Path(output) trace = output.with_suffix('.trace.json') if output.suffix.lower() != '.xlsx': raise ValueError('output must end in .xlsx') if output.exists() or trace.exists(): raise ValueError('output or trace exists; choose a new version path') from openpyxl import Workbook, load_workbook from openpyxl.styles import Alignment, Font, PatternFill from openpyxl.utils import get_column_letter wb = Workbook() wb.remove(wb.active) for key, (title, fields, headers, widths) in LAYOUT.items(): ws = wb.create_sheet(title) for rno, values in enumerate([headers] + [[r[f] for f in fields] for r in data[key]], 1): lines = 1 for cno, (text, width) in enumerate(zip(values, widths), 1): cell = ws.cell(rno, cno, text) cell.data_type = 's' # User text beginning with '=' stays text. cell.font = Font(name='微软雅黑', size=12, bold=rno == 1, color='FFFFFF' if rno == 1 else '263445') cell.alignment = Alignment(vertical='top', wrap_text=True) cell.fill = PatternFill('solid', fgColor='174B6B' if rno == 1 else ('F1F6FA' if rno % 2 == 0 else 'FFFFFF')) estimate = sum(max(1, math.ceil(sum(2 if ord(c) > 127 else 1 for c in line) / (width - 2))) for line in text.split('\n')) lines = max(lines, estimate) ws.column_dimensions[get_column_letter(cno)].width = width ws.row_dimensions[rno].height = 32 if rno == 1 else min(409, max(54, 20 * lines + 10)) ws.freeze_panes = 'C2' ws.auto_filter.ref = ws.dimensions output.parent.mkdir(parents=True, exist_ok=True) wb.save(output) check = load_workbook(output) for key, (title, fields, headers, _) in LAYOUT.items(): assert check[title].max_column == 4 and check[title].max_row == len(data[key]) + 1 expected = [list(headers)] + [[r[f] for f in fields] for r in data[key]] actual = [[c.value or '' for c in row] for row in check[title]] assert actual == expected, title assert all(c.data_type not in ('f', 'e') for row in check[title] for c in row) trace.write_text(json.dumps({'report': report, 'sources': data['sources'], 'requirements': data['requirements'], 'row_mapping': {k: [r['refs'] for r in data[k]] for k in LAYOUT}}, ensure_ascii=False, indent=2)) return report def main(): parser = argparse.ArgumentParser(description=__doc__) parser.add_argument('--input', required=True) parser.add_argument('--output') parser.add_argument('--validate-only', action='store_true') args = parser.parse_args() data = json.loads(Path(args.input).read_text()) try: if args.validate_only: report = validate(data) else: if not args.output: parser.error('--output is required unless --validate-only') report = build(data, args.output) print(json.dumps(report, ensure_ascii=False, indent=2)) except ValueError as exc: parser.exit(1, f'{exc}\n') if __name__ == '__main__': main()