| """Convert reviewed JSONL examples to explicit Ares/Xiphos chat-section training text. |
| Input is local, human-curated JSONL. This module never downloads data or auto-labels it. |
| """ |
| import argparse,json |
| from pathlib import Path |
| REQUIRED={'ares':('user','assistant'),'xiphos':('request','plan')} |
| def format_row(role,row): |
| if role=='ares': |
| user,assistant=(str(row[x]).strip() for x in REQUIRED[role]) |
| if not user or not assistant:raise ValueError('empty user or assistant') |
| system=str(row.get('system','You are Ares, a helpful and honest assistant.')).strip() |
| return f'<system>{system}</system><user>{user}</user><assistant>{assistant}</assistant>' |
| request,plan=(str(row[x]).strip() for x in REQUIRED[role]) |
| if not request or not plan:raise ValueError('empty request or plan') |
| system='You are Xiphos. Draft reviewable plans; do not execute tools or code.' |
| return f'<system>{system}</system><user>{request}</user><assistant><plan>{plan}</plan></assistant>' |
| def main(): |
| p=argparse.ArgumentParser();p.add_argument('--role',choices=('ares','xiphos'),required=True);p.add_argument('--input',required=True);p.add_argument('--out',required=True);a=p.parse_args();written=0 |
| Path(a.out).parent.mkdir(parents=True,exist_ok=True) |
| with open(a.out,'w',encoding='utf8') as out: |
| for line_no,line in enumerate(open(a.input,encoding='utf8'),1): |
| try:out.write(format_row(a.role,json.loads(line))+'\n');written+=1 |
| except (json.JSONDecodeError,KeyError,ValueError) as e:raise SystemExit(f'Invalid reviewed example at line {line_no}: {e}') |
| print({'role':a.role,'examples':written,'out':a.out}) |
| if __name__=='__main__':main() |
|
|
