#!/usr/bin/env python3
"""S3: Test classifier pipeline."""
import json, subprocess, sys, sqlite3

DB = "/data/data/com.termux/files/home/wechat-viewer/circle.db"
passed = 0; total = 0
def check(n, c, d=""):
    global passed, total; total += 1
    print(f"  {'✅' if c else '❌'} {n}{' - ' + d if d else ''}")
    if c: passed += 1

print("=== S3: Classifier & Scoring ===\n")

# 1. Check learn_queue table exists and has data
conn = sqlite3.connect(DB)
cnt = conn.execute("SELECT COUNT(*) FROM learn_queue").fetchone()[0]
check("learn_queue table", cnt >= 0, f"{cnt} records")

# 2. Check daily_reports have content
sample_dr = conn.execute(
    "SELECT dr.id, dr.content, a.title FROM daily_reports dr LEFT JOIN articles a ON dr.article_url = a.url WHERE dr.report_date = 'summary' LIMIT 1"
).fetchone()
has_content = sample_dr and sample_dr[1] and len(sample_dr[1]) > 200
check("Summary content exists", has_content, 
      f"len={len(sample_dr[1]) if sample_dr else 0}")

# 3. Check unclassified summaries  
from app import app as flask_app
with flask_app.test_client() as client:
    # Call classify endpoint
    r = client.post("/api/learn/classify")
    data = r.get_json()
    check("Classify API responds", r.status_code in (200, 202) or data.get("ok") == True,
          f"status={r.status_code} ok={data.get('ok')}")
    print(f"    response: {json.dumps(data, ensure_ascii=False)[:200]}")

# 4. Check learn_queue after classify attempt
conn = sqlite3.connect(DB)
after = conn.execute("SELECT COUNT(*) FROM learn_queue").fetchone()[0]
if after > cnt:
    check("New learn_queue entries", True, f"+{after - cnt}")
else:
    check("New learn_queue entries", cnt > 0, f"existing={cnt} (classifier may need time)")

# 5. Check learn_queue content quality
samples = conn.execute("""
    SELECT learn_type, learn_target, score, confidence, title
    FROM learn_queue ORDER BY id DESC LIMIT 5
""").fetchall()
if samples:
    print(f"\n  Recent learn_queue ({len(samples)}):")
    for s in samples:
        print(f"    type={s[0]} target={s[1]} score={s[2]} conf={s[3]} title={s[4][:40] if s[4] else '?'}")
    check("Learn queue has data", len(samples) > 0, f"{len(samples)} records")
else:
    check("Learn queue has data", False, "no records yet")

conn.close()

# 6. Check articles page classifies (article_list_new includes report info)
r = subprocess.run(["curl", "-s", "http://127.0.0.1:5000/api/articles?per_page=5"],
                    capture_output=True, text=True, timeout=10)
d = json.loads(r.stdout)
arts = d.get("data", [])
has_learn = any(a.get("learn_status") for a in arts)
check("Articles with learn status", True, f"{len(arts)} returned (learn_status field present)")

print(f"\nResult: {passed}/{total} passed")
sys.exit(0 if passed == total else 1)