Add reliable execution reporting and coordinated dashboard views

scripts/report.py

@@ -0,0 +1,115 @@
+#!/usr/bin/env python3
+"""Persist and deliver an execution report; never mark integration accepted."""
+import argparse
+import hashlib
+import json
+from pathlib import Path
+import re
+import sys
+import time
+from control import Client,APIError
+from worker import atomic_json,session_lock
+
+
+def deliver_file(client,path,now=None):
+    now=time.time() if now is None else now
+    with session_lock('task-report:'+str(path.resolve())):
+        record=json.loads(path.read_text(encoding='utf-8'))
+        if record['status']!='pending' or record.get('next_retry',0)>now:
+            return record
+        try:
+            receipt=client.call('/api/task-reports',record['payload'])
+        except Exception as exc:
+            record['attempts']=record.get('attempts',0)+1
+            terminal=isinstance(exc,APIError) and exc.status in (400,403,409)
+            record['status']='blocked' if terminal or record['attempts']>=10 else 'pending'
+            record['next_retry']=now+min(60,2**record['attempts'])
+            record['error']='Report delivery requires inspection' if record['status']=='blocked' else 'Temporary delivery interruption'
+            atomic_json(path,record)
+            return record
+        record.update(status='accepted',receipt=receipt,accepted_at=now)
+        record.pop('error',None)
+        atomic_json(path,record)
+        return record
+
+
+def submit(cfg,payload):
+    root=Path(cfg['outbox_dir']).expanduser().resolve()
+    root.mkdir(parents=True,exist_ok=True,mode=0o700)
+    canonical=json.dumps(payload,sort_keys=True,ensure_ascii=True)
+    report_id=payload.get('report_id') or 'report-'+hashlib.sha256(canonical.encode()).hexdigest()[:32]
+    if not isinstance(report_id,str) or not re.fullmatch(r'[A-Za-z0-9_.:-]{1,128}',report_id):
+        raise ValueError('Invalid report id')
+    payload={**payload,'report_id':report_id}
+    path=root/(hashlib.sha256(report_id.encode()).hexdigest()+'.json')
+    if path.is_symlink():raise ValueError('Report file cannot be a symlink')
+    with session_lock('task-report:'+str(path)):
+        if path.exists():
+            record=json.loads(path.read_text(encoding='utf-8'))
+            if record['payload']!=payload:raise ValueError('Report id already used for different content')
+        else:
+            atomic_json(path,{'status':'pending','created':time.time(),'payload':payload,'attempts':0})
+    return deliver_file(Client(cfg['control']),path)
+
+
+def flush(cfg):
+    root=Path(cfg['outbox_dir']).expanduser().resolve()
+    root.mkdir(parents=True,exist_ok=True,mode=0o700)
+    client=Client(cfg['control'])
+    counts={'pending':0,'accepted':0,'blocked':0,'superseded':0}
+    records=[]
+    for path in root.glob('*.json'):
+        if not re.fullmatch(r'[a-f0-9]{64}\.json',path.name):continue
+        if path.is_symlink():raise ValueError('Unexpected report symlink')
+        records.append((path,json.loads(path.read_text(encoding='utf-8'))))
+    completed={}
+    for _,record in records:
+        payload=record['payload']
+        if payload['stage']=='execution_completed' and record['status'] in ('pending','accepted'):
+            key=(payload['task_id'],payload['session_id'])
+            completed[key]=max(completed.get(key,0),record['created'])
+    for path,record in records:
+        payload=record['payload'];key=(payload['task_id'],payload['session_id'])
+        if record['status']=='pending' and payload['stage']!='execution_completed' and record['created']<completed.get(key,0):
+            with session_lock('task-report:'+str(path.resolve())):
+                current=json.loads(path.read_text(encoding='utf-8'))
+                if current['status']=='pending':
+                    current.update(status='superseded',superseded_at=time.time())
+                    atomic_json(path,current)
+                    record.update(current)
+    pending=sorted(((path,r) for path,r in records if r['status']=='pending'),
+                   key=lambda item:(item[1]['payload']['stage']!='execution_completed',item[1]['created']))
+    for path,record in pending[:100]:
+        try:record.update(deliver_file(client,path))
+        except (BlockingIOError,PermissionError):continue
+    for _,record in records:
+        counts[record['status']]=counts.get(record['status'],0)+1
+    return counts
+
+
+def main():
+    p=argparse.ArgumentParser(description=__doc__)
+    p.add_argument('--config',required=True)
+    p.add_argument('--task-id',required=True)
+    p.add_argument('--session-id',required=True)
+    p.add_argument('--stage',choices=['execution_completed','blocked','progress','failed'],required=True)
+    p.add_argument('--progress',required=True)
+    p.add_argument('--summary',required=True)
+    p.add_argument('--evidence',action='append',default=[])
+    p.add_argument('--version')
+    p.add_argument('--report-id')
+    p.add_argument('--needs-attention',action='store_true')
+    a=p.parse_args()
+    cfg=json.loads(Path(a.config).read_text(encoding='utf-8-sig'))
+    payload={'task_id':a.task_id,'session_id':a.session_id,'stage':a.stage,
+             'current_progress':a.progress,'summary':a.summary,'evidence':a.evidence,
+             'needs_attention':a.needs_attention}
+    if a.version:payload['version']=a.version
+    if a.report_id:payload['report_id']=a.report_id
+    result=submit(cfg,payload)
+    print(json.dumps({'report_id':result['payload']['report_id'],'delivery':result['status'],
+                      'execution_report_only':True,'integration_accepted':False},ensure_ascii=True))
+    return 0 if result['status']=='accepted' else 75
+
+
+if __name__=='__main__':sys.exit(main())