Spaces:
Running
Running
File size: 8,095 Bytes
f492127 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 | """
HaramGuard — Database Layer
============================
SQLite persistence for risk events, operational decisions, and coordinator plans.
All agents write/read through this single interface.
"""
import os
import json
import sqlite3
from datetime import datetime
from typing import Optional
from core.models import RiskResult, Decision
class HajjFlowDB:
def __init__(self, path: str = 'outputs/hajjflow_rt.db'):
os.makedirs(os.path.dirname(path), exist_ok=True)
self.conn = sqlite3.connect(path, check_same_thread=False)
self._init_tables()
print(f'🗄️ [DB] Connected -> {path}')
def _init_tables(self):
self.conn.executescript("""
CREATE TABLE IF NOT EXISTS risk_events (
id INTEGER PRIMARY KEY AUTOINCREMENT,
frame_id INTEGER,
timestamp TEXT,
risk_score REAL,
risk_level TEXT,
trend TEXT,
window_avg REAL,
window_max INTEGER
);
CREATE TABLE IF NOT EXISTS op_decisions (
id INTEGER PRIMARY KEY AUTOINCREMENT,
frame_id INTEGER,
timestamp TEXT,
context TEXT,
priority TEXT,
actions TEXT,
risk_score REAL,
risk_level TEXT
);
CREATE TABLE IF NOT EXISTS coordinator_plans (
id INTEGER PRIMARY KEY AUTOINCREMENT,
frame_id INTEGER,
timestamp TEXT,
threat_level TEXT,
summary TEXT,
arabic_alert TEXT,
confidence REAL,
raw_json TEXT
);
CREATE TABLE IF NOT EXISTS reflection_log (
id INTEGER PRIMARY KEY AUTOINCREMENT,
frame_id INTEGER,
timestamp TEXT,
original_level TEXT,
original_score REAL,
corrected_level TEXT,
corrected_score REAL,
bias_detected INTEGER,
critique TEXT,
person_count INTEGER
);
""")
self.conn.commit()
# ── Writes ───────────────────────────────────────────────────────────
def save_risk_event(self, rr: RiskResult):
self.conn.execute(
'INSERT INTO risk_events '
'(frame_id,timestamp,risk_score,risk_level,trend,window_avg,window_max) '
'VALUES (?,?,?,?,?,?,?)',
(rr.frame_id,
datetime.fromtimestamp(rr.timestamp).isoformat(),
rr.risk_score, rr.risk_level, rr.trend,
rr.window_avg, rr.window_max))
self.conn.commit()
def save_decision(self, d: Decision):
self.conn.execute(
'INSERT INTO op_decisions '
'(frame_id,timestamp,context,priority,actions,risk_score,risk_level) '
'VALUES (?,?,?,?,?,?,?)',
(d.frame_id, d.timestamp, d.context, d.priority,
json.dumps(d.actions), d.risk_score, d.risk_level))
self.conn.commit()
def save_coordinator_plan(self, frame_id: int, plan: dict):
self.conn.execute(
'INSERT INTO coordinator_plans '
'(frame_id,timestamp,threat_level,summary,arabic_alert,confidence,raw_json) '
'VALUES (?,?,?,?,?,?,?)',
(frame_id, datetime.now().isoformat(),
plan.get('threat_level', 'UNKNOWN'),
plan.get('executive_summary', ''),
plan.get('arabic_alert', ''),
plan.get('confidence_score', 0),
json.dumps(plan, ensure_ascii=False)))
self.conn.commit()
def save_reflection(self, reflection: dict):
self.conn.execute(
'INSERT INTO reflection_log '
'(frame_id,timestamp,original_level,original_score,'
'corrected_level,corrected_score,bias_detected,critique,person_count) '
'VALUES (?,?,?,?,?,?,?,?,?)',
(reflection['frame_id'],
datetime.now().isoformat(),
reflection['original_level'],
reflection['original_score'],
reflection['corrected_level'],
reflection['corrected_score'],
int(reflection['bias_detected']),
reflection['critique'],
reflection['person_count']))
self.conn.commit()
# ── Reads ────────────────────────────────────────────────────────────
def get_last_p0_time(self, context: str) -> Optional[datetime]:
cur = self.conn.execute(
'SELECT timestamp FROM op_decisions '
'WHERE context=? AND priority="P0" '
'ORDER BY timestamp DESC LIMIT 1', (context,))
row = cur.fetchone()
return datetime.fromisoformat(row[0]) if row else None
def get_recent_decisions(self, limit: int = 10) -> list:
"""
Return recent decisions enriched with coordinator plan data
(selected_gates, justification) via LEFT JOIN on frame_id.
"""
cur = self.conn.execute(
'SELECT d.frame_id, d.timestamp, d.context, d.priority, '
' d.actions, d.risk_score, d.risk_level, '
' cp.raw_json AS plan_json '
'FROM op_decisions d '
'LEFT JOIN coordinator_plans cp ON cp.frame_id = d.frame_id '
'ORDER BY d.id DESC LIMIT ?', (limit,))
rows = []
for row in cur.fetchall():
record = {
'frame_id': row[0],
'timestamp': row[1],
'context': row[2],
'priority': row[3],
'actions': row[4],
'risk_score': row[5],
'risk_level': row[6],
}
# Enrich with coordinator plan fields if available
if row[7]:
try:
plan = json.loads(row[7])
record['selected_gates'] = plan.get('selected_gates', [])
record['justification'] = plan.get('actions_justification', '')
record['arabic_alert'] = plan.get('arabic_alert', '')
record['threat_level'] = plan.get('threat_level', '')
record['confidence'] = plan.get('confidence_score', 0)
record['immediate_actions'] = plan.get('immediate_actions', [])
except Exception:
pass
rows.append(record)
return rows
def get_risk_history(self, limit: int = 60) -> list:
cur = self.conn.execute(
'SELECT frame_id,risk_score,risk_level,trend,window_avg '
'FROM risk_events ORDER BY id DESC LIMIT ?', (limit,))
cols = [c[0] for c in cur.description]
return list(reversed([dict(zip(cols, row)) for row in cur.fetchall()]))
def get_reflection_summary(self) -> dict:
cur = self.conn.execute(
'SELECT bias_detected, original_level, corrected_level FROM reflection_log')
rows = cur.fetchall()
if not rows:
return {'total': 0, 'bias_events': 0, 'bias_rate_pct': 0}
total = len(rows)
biased = sum(1 for r in rows if r[0])
return {
'total': total,
'bias_events': biased,
'bias_rate_pct': round(biased / total * 100, 1),
}
def print_summary(self):
for tbl in ['risk_events', 'op_decisions', 'coordinator_plans', 'reflection_log']:
n = self.conn.execute(f'SELECT COUNT(*) FROM {tbl}').fetchone()[0]
print(f' {tbl:<25} -> {n:4d} rows')
|