#!/usr/bin/env python3
"""S4: End-to-end pipeline verification."""
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("=== S4: End-to-End Verification ===\n")

# 1. Find a scored article in learn_queue
conn = sqlite3.connect(DB)
rows = conn.execute("""
    SELECT lq.id, lq.article_url, lq.title, lq.learn_type, lq.learn_target, 
           lq.score, lq.confidence, lq.status,
           a.url as article_url_full, dr.content as summary
    FROM learn_queue lq
    LEFT JOIN articles a ON lq.dedup_key = a.dedup_key
    LEFT JOIN daily_reports dr ON dr.article_url = lq.article_url AND dr.report_date = 'summary'
    WHERE lq.score >= 6 AND lq.status = 'pending'
    ORDER BY lq.score DESC LIMIT 1
""").fetchone()

if not rows:
    # Try without status filter
    rows = conn.execute("""
        SELECT lq.id, lq.article_url, lq.title, lq.learn_type, lq.learn_target,
               lq.score, lq.confidence, lq.status
        FROM learn_queue lq
        WHERE lq.score >= 6
        ORDER BY lq.score DESC LIMIT 1
    """).fetchone()

if rows:
    lid = rows[0]
    ltype = rows[3]
    ltarget = rows[4]
    score = rows[5]
    status = rows[7] if len(rows) > 7 else "?"
    title = (rows[2] or "?")[:50]
    
    check("Find classified article", True, f"id={lid} type={ltype} score={score} status={status} title={title}")
    print(f"    target={ltarget}")
    
    # 2. Check summary exists for this article
    summary_row = conn.execute(
        "SELECT content FROM daily_reports WHERE article_url = ? AND report_date = 'summary' LIMIT 1",
        (rows[1],)
    ).fetchone()
    has_summary = summary_row and summary_row[0] and len(summary_row[0]) > 200
    check("Summary available", has_summary,
          f"len={len(summary_row[0]) if summary_row else 0}")
    
    # 3. Check dedup logic - is this already in memory/skill?
    if ltype == "skill":
        has_dedup = conn.execute(
            "SELECT COUNT(*) FROM learn_queue WHERE dedup_key = (SELECT dedup_key FROM learn_queue WHERE id=?) AND status IN ('completed','processing')", (lid,)
        ).fetchone()[0]
        check("Dedup check", has_dedup == 0, f"duplicates={has_dedup}")
    
    # 4. Check learn-and-build skill exists
    from pathlib import Path
    lb_skill = Path("/data/data/com.termux/files/home/.dsh/skills/wechat/learn-and-build/SKILL.md")
    check("learn-and-build skill exists", lb_skill.exists())
    
    # 5. Verify cron pipeline can self-heal
    scan_path = Path("/data/data/com.termux/files/home/.dsh/skills/wechat-group-digest/scripts/scan_articles.py")
    check("scan_articles.py exists", scan_path.exists())
    
    # 6. Overall pipeline check
    conn2 = sqlite3.connect(DB)
    total_arts = conn2.execute("SELECT COUNT(*) FROM articles").fetchone()[0]
    total_reports = conn2.execute("SELECT COUNT(*) FROM daily_reports WHERE report_date='summary'").fetchone()[0]
    total_learn = conn2.execute("SELECT COUNT(*) FROM learn_queue WHERE score >= 6").fetchone()[0]
    total_pending_classify = conn2.execute("""
        SELECT COUNT(*) FROM learn_queue WHERE status = 'pending' AND score >= 6
    """).fetchone()[0]
    conn2.close()
    
    check("Pipeline: articles → summaries", total_reports > 0,
          f"{total_arts} arts → {total_reports} summaries")
    check("Pipeline: summaries → classified", total_learn > 0,
          f"{total_reports} summaries → {total_learn} classified (≥6)")
    check("Pipeline: pending review", total_pending_classify >= 0,
          f"{total_pending_classify} pending")

conn.close()

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