ai_member_xiaoyan/scripts/fix_p4_dialogue.py

#!/usr/bin/env python3
"""
Fix P4 题目1/题目2 fields: re-add dialogue text
The original script's pop("dialogue") corrupted module-level data during dry run.
This script extracts dialogues from the rewrite script source and writes correct 题目1/题目2.
"""
import json, subprocess, os, re

APP_TOKEN = "CMHSbUUjka3TrUsaxxEc297ongf"
TABLE_ID = "tblVmeDtBDKsAEfz"

def get_token():
    CRED_FILE = "/root/.openclaw/credentials/xiaoyan/config.json"
    with open(CRED_FILE) as f:
        cred = json.load(f)
    app_id = cred['apps'][0]['appId']
    app_secret = cred['apps'][0]['appSecret']
    result = subprocess.run([
        "curl", "-s", "-X", "POST",
        "https://open.feishu.cn/open-apis/auth/v3/tenant_access_token/internal",
        "-H", "Content-Type: application/json",
        "-d", json.dumps({"app_id": app_id, "app_secret": app_secret})
    ], capture_output=True, text=True)
    return json.loads(result.stdout)['tenant_access_token']

def get_records(token):
    result = subprocess.run([
        "curl", "-s", "-X", "GET",
        f"https://open.feishu.cn/open-apis/bitable/v1/apps/{APP_TOKEN}/tables/{TABLE_ID}/records?page_size=50",
        "-H", f"Authorization: Bearer {token}",
    ], capture_output=True, text=True)
    return json.loads(result.stdout).get('data', {}).get('items', [])

def update_record(token, record_id, fields):
    body = json.dumps({"fields": fields}, ensure_ascii=False)
    result = subprocess.run([
        "curl", "-s", "-X", "PUT",
        f"https://open.feishu.cn/open-apis/bitable/v1/apps/{APP_TOKEN}/tables/{TABLE_ID}/records/{record_id}",
        "-H", f"Authorization: Bearer {token}",
        "-H", "Content-Type: application/json; charset=utf-8",
        "-d", body
    ], capture_output=True, text=True)
    return json.loads(result.stdout)

def make_text_with_dialogue(questions):
    """Generate 题目 text from question list, with dialogue"""
    lines = []
    for i, q in enumerate(questions, 1):
        lines.append(f"{i}.")
        lines.append(f"【描述】{q['questionDesc']}")
        lines.append(f"【听力文本】")
        if 'dialogue' in q:
            lines.append(q['dialogue'])
        lines.append(f"【题目】")
        lines.append(q['question'])
        for j, opt in enumerate(q['options']):
            marker = "（正确）" if j == q['answer'][0] else ""
            lines.append(f"{chr(65+j)}. {opt}{marker}")
        lines.append("")
    return "\n".join(lines)

# Step 1: Extract all dialogues from rewrite script by re-executing the data definitions
# We exec the script data without running main()
script_path = os.path.dirname(os.path.abspath(__file__))
rewrite_script = os.path.join(script_path, 'rewrite_p4_all.py')

with open(rewrite_script) as f:
    script_content = f.read()

# Remove the main() call at the end and replace with our own
# Find and extract dialogue data for each QSID
# Pattern: QSID_XXXXXX_FIRST_DIALOGUES = [...] and QSID_XXXXXX_SECOND_DIALOGUES = [...]

# Parse the script content to get all dialogue arrays
qsid_dialogues = {}

# Find all DIALOGUES blocks
pattern = r"QSID_(\d{6})_(FIRST|SECOND)_DIALOGUES\s*=\s*\[(.*?)\]"
matches = list(re.finditer(pattern, script_content, re.DOTALL))

for m in matches:
    qsid = m.group(1)
    block_type = m.group(2).lower()
    dialogues_raw = m.group(3)
    
    # Extract individual dialogue strings (triple-quoted)
    dialogue_pattern = r'"""(.*?)"""'
    dialogues = re.findall(dialogue_pattern, dialogues_raw, re.DOTALL)
    dialogues = [d.strip() for d in dialogues]
    
    key = f"{qsid}_{block_type}"
    qsid_dialogues[key] = dialogues
    print(f"Extracted {len(dialogues)} dialogues for {qsid} {block_type}")

# Now assign dialogues and regenerate text
token = get_token()
records = get_records(token)

for rec in records:
    f = rec.get('fields', {})
    qsid = f.get('题目集合 ID')
    if not qsid:
        continue
    
    jd = f.get('jsonData')
    if not jd:
        continue
    
    try:
        parsed = json.loads(jd)
    except:
        continue
    
    first_qs = parsed.get('first', {}).get('questionSet', [])
    second_qs = parsed.get('second', {}).get('questionSet', [])
    
    # Assign dialogues
    first_dials = qsid_dialogues.get(f"{qsid}_first", [])
    second_dials = qsid_dialogues.get(f"{qsid}_second", [])
    
    for i, d in enumerate(first_dials):
        if i < len(first_qs):
            first_qs[i]['dialogue'] = d
    
    for i, d in enumerate(second_dials):
        if i < len(second_qs):
            second_qs[i]['dialogue'] = d
    
    # Generate text
    q1_text = make_text_with_dialogue(first_qs) if first_qs else ""
    q2_text = make_text_with_dialogue(second_qs) if second_qs else ""
    
    # Check if dialogue is actually included
    has_dialogue = 'Ben' in q1_text or 'Lucy' in q1_text or 'Daisy' in q1_text
    if not has_dialogue:
        print(f"⚠️ {qsid}: 题目1 still missing dialogue!")
        print(f"  first_dials count: {len(first_dials)}, first_qs count: {len(first_qs)}")
        continue
    
    # Write
    result = update_record(token, rec['record_id'], {
        "题目1": q1_text,
        "题目2": q2_text,
    })
    
    code = result.get('code', -1)
    if code == 0:
        print(f"✅ {qsid}: 题目1={len(q1_text)}bytes 题目2={len(q2_text)}bytes")
    else:
        print(f"❌ {qsid}: {result.get('msg', 'unknown')}")

print("\nDone!")