Upload sync.py with huggingface_hub

sync.py

@@ -0,0 +1,264 @@
+# -*- coding: utf-8 -*-
+"""
+雙向同步 Local ↔ HF Space 資料庫
+智慧合併：兩邊都留最多最新的，重複的以最新為主
+用法: python sync.py
+"""
+
+import configparser
+import sqlite3
+import shutil
+import tempfile
+import os
+from pathlib import Path
+from datetime import datetime
+
+cfg = configparser.ConfigParser()
+cfg.read('Myconfig.ini')
+TOKEN = cfg['Hugging Face']['Token'].strip()
+USERNAME = cfg['Hugging Face']['Username'].strip()
+REPO_ID = f'{USERNAME}/pradsa'
+
+LOCAL_DB = Path('pra_data.db')
+BACKUP_DIR = LOCAL_DB.parent / 'backups'
+
+
+def get_db_info(db_path):
+    if not Path(db_path).exists():
+        return {'patients': 0, 'reports': 0, 'users': 0, 'latest': ''}
+    conn = sqlite3.connect(str(db_path))
+    conn.row_factory = sqlite3.Row
+    try:
+        patients = conn.execute('SELECT COUNT(*) as c FROM patients').fetchone()['c']
+        reports = conn.execute('SELECT COUNT(*) as c FROM reports').fetchone()['c']
+        users = conn.execute('SELECT COUNT(*) as c FROM users').fetchone()['c']
+        latest = conn.execute(
+            "SELECT MAX(COALESCE(updated_at, created_at)) as t FROM reports"
+        ).fetchone()['t'] or ''
+    except Exception:
+        patients, reports, users, latest = 0, 0, 0, ''
+    conn.close()
+    return {'patients': patients, 'reports': reports, 'users': users, 'latest': latest}
+
+
+def download_cloud_db():
+    from huggingface_hub import hf_hub_download
+    try:
+        path = hf_hub_download(
+            repo_id=REPO_ID, filename='pra_data.db',
+            repo_type='space', token=TOKEN,
+            local_dir=tempfile.gettempdir(),
+            force_download=True,
+        )
+        return path
+    except Exception as e:
+        print(f'  雲端 DB 下載失敗: {e}')
+        return None
+
+
+def upload_db(db_path):
+    from huggingface_hub import HfApi
+    api = HfApi(token=TOKEN)
+    api.upload_file(
+        path_or_fileobj=str(db_path),
+        path_in_repo='pra_data.db',
+        repo_id=REPO_ID,
+        repo_type='space',
+    )
+
+
+def backup_local():
+    if LOCAL_DB.exists():
+        BACKUP_DIR.mkdir(exist_ok=True)
+        ts = datetime.now().strftime('%Y%m%d_%H%M%S')
+        dst = BACKUP_DIR / f'pra_data_before_sync_{ts}.db'
+        shutil.copy2(str(LOCAL_DB), str(dst))
+        return dst
+    return None
+
+
+def merge_dbs(local_path, cloud_path):
+    """
+    智慧合併兩個 DB。
+    - patients: 以 chart_no 為 key，兩邊都留
+    - reports: 以 (chart_no, report_date, pra_class) 為 key
+      重複的保留 updated_at/created_at 較新的
+    - antibody_strength: 跟著 report 走
+    - users: 以 username 為 key，兩邊都留
+    回傳合併後的 DB 路徑
+    """
+    merged_path = str(Path(tempfile.gettempdir()) / 'pra_merged.db')
+    shutil.copy2(str(local_path), merged_path)
+
+    local = sqlite3.connect(merged_path)
+    local.row_factory = sqlite3.Row
+    cloud = sqlite3.connect(str(cloud_path))
+    cloud.row_factory = sqlite3.Row
+
+    stats = {'patients_added': 0, 'reports_added': 0, 'reports_updated': 0, 'users_added': 0}
+
+    # --- Merge patients ---
+    cloud_patients = cloud.execute('SELECT * FROM patients').fetchall()
+    for cp in cloud_patients:
+        existing = local.execute('SELECT id FROM patients WHERE chart_no=?', (cp['chart_no'],)).fetchone()
+        if not existing:
+            local.execute(
+                'INSERT INTO patients (patient_name, chart_no, donor_hla, created_at) VALUES (?,?,?,?)',
+                (cp['patient_name'], cp['chart_no'], cp['donor_hla'] or '', cp['created_at']))
+            stats['patients_added'] += 1
+        else:
+            # Update donor_hla if cloud has it and local doesn't
+            local_donor = local.execute('SELECT donor_hla FROM patients WHERE id=?', (existing['id'],)).fetchone()
+            if cp['donor_hla'] and not (local_donor and local_donor['donor_hla']):
+                local.execute('UPDATE patients SET donor_hla=? WHERE id=?', (cp['donor_hla'], existing['id']))
+
+    # --- Merge reports ---
+    # Build chart_no -> local patient_id mapping
+    local_patients = {r['chart_no']: r['id'] for r in local.execute('SELECT id, chart_no FROM patients').fetchall()}
+
+    cloud_reports = cloud.execute('''
+        SELECT r.*, p.chart_no FROM reports r JOIN patients p ON p.id = r.patient_id
+    ''').fetchall()
+
+    for cr in cloud_reports:
+        chart_no = cr['chart_no']
+        local_pid = local_patients.get(chart_no)
+        if not local_pid:
+            continue
+
+        # Check if same report exists locally
+        existing = local.execute(
+            'SELECT id, COALESCE(updated_at, created_at) as ts FROM reports WHERE patient_id=? AND report_date=? AND pra_class=?',
+            (local_pid, cr['report_date'], cr['pra_class'])
+        ).fetchone()
+
+        cloud_ts = cr['updated_at'] or cr['created_at'] or ''
+
+        if not existing:
+            # Cloud has a report local doesn't -> add
+            cur = local.execute(
+                '''INSERT INTO reports (patient_id, report_date, pra_class, pra_percent, overall,
+                   specificity, comment, status, submitted_by, created_at, updated_at)
+                   VALUES (?,?,?,?,?,?,?,?,?,?,?)''',
+                (local_pid, cr['report_date'], cr['pra_class'], cr['pra_percent'],
+                 cr['overall'], cr['specificity'], cr['comment'], cr['status'] or 'draft',
+                 cr['submitted_by'] or '', cr['created_at'], cr['updated_at']))
+            new_rid = cur.lastrowid
+            # Copy antibody_strength
+            cloud_abs = cloud.execute('SELECT * FROM antibody_strength WHERE report_id=?', (cr['id'],)).fetchall()
+            for ab in cloud_abs:
+                local.execute(
+                    '''INSERT INTO antibody_strength (report_id, antigen, allele, max_mfi, mean_mfi, no_of_beads, bead_ids)
+                       VALUES (?,?,?,?,?,?,?)''',
+                    (new_rid, ab['antigen'], ab['allele'], ab['max_mfi'], ab['mean_mfi'], ab['no_of_beads'], ab['bead_ids']))
+            stats['reports_added'] += 1
+
+        elif cloud_ts > (existing['ts'] or ''):
+            # Cloud is newer -> update
+            rid = existing['id']
+            local.execute(
+                '''UPDATE reports SET pra_percent=?, overall=?, specificity=?, comment=?,
+                   status=?, submitted_by=?, updated_at=? WHERE id=?''',
+                (cr['pra_percent'], cr['overall'], cr['specificity'], cr['comment'],
+                 cr['status'] or 'draft', cr['submitted_by'] or '', cr['updated_at'], rid))
+            # Replace antibody_strength
+            local.execute('DELETE FROM antibody_strength WHERE report_id=?', (rid,))
+            cloud_abs = cloud.execute('SELECT * FROM antibody_strength WHERE report_id=?', (cr['id'],)).fetchall()
+            for ab in cloud_abs:
+                local.execute(
+                    '''INSERT INTO antibody_strength (report_id, antigen, allele, max_mfi, mean_mfi, no_of_beads, bead_ids)
+                       VALUES (?,?,?,?,?,?,?)''',
+                    (rid, ab['antigen'], ab['allele'], ab['max_mfi'], ab['mean_mfi'], ab['no_of_beads'], ab['bead_ids']))
+            stats['reports_updated'] += 1
+
+    # --- Merge users ---
+    stats['users_updated'] = 0
+    cloud_users = cloud.execute('SELECT * FROM users').fetchall()
+    for cu in cloud_users:
+        existing = local.execute('SELECT id, COALESCE(edit_time, created_at) as ts FROM users WHERE username=?',
+                                 (cu['username'],)).fetchone()
+        cloud_ts = ''
+        try:
+            cloud_ts = cu['edit_time'] or cu['created_at'] or ''
+        except (IndexError, KeyError):
+            cloud_ts = cu['created_at'] if 'created_at' in cu.keys() else ''
+
+        if not existing:
+            # Cloud only -> add
+            cols = ['username', 'display_name', 'password_hash', 'password_plain', 'role', 'created_at', 'edit_time']
+            vals = []
+            for c in cols:
+                try:
+                    vals.append(cu[c] or '')
+                except (IndexError, KeyError):
+                    vals.append('')
+            local.execute(
+                f'INSERT INTO users ({",".join(cols)}) VALUES ({",".join(["?"]*len(cols))})', vals)
+            stats['users_added'] += 1
+        elif cloud_ts > (existing['ts'] or ''):
+            # Cloud is newer -> update
+            local.execute(
+                '''UPDATE users SET display_name=?, password_hash=?, password_plain=?, role=?, edit_time=?
+                   WHERE id=?''',
+                (cu['display_name'] or '', cu['password_hash'], cu['password_plain'] or '',
+                 cu['role'] or 'user', cloud_ts, existing['id']))
+            stats['users_updated'] += 1
+
+    local.commit()
+    local.close()
+    cloud.close()
+
+    return merged_path, stats
+
+
+def sync():
+    print('=== PRA DB Sync ===\n')
+
+    local_info = get_db_info(LOCAL_DB)
+    print(f'Local:  {local_info["patients"]} patients, {local_info["reports"]} reports, {local_info["users"]} users, latest: {local_info["latest"] or "(none)"}')
+
+    print('Downloading cloud DB...')
+    cloud_path = download_cloud_db()
+    if not cloud_path:
+        print('無法取得雲端 DB')
+        return
+
+    cloud_info = get_db_info(cloud_path)
+    print(f'Cloud:  {cloud_info["patients"]} patients, {cloud_info["reports"]} reports, {cloud_info["users"]} users, latest: {cloud_info["latest"] or "(none)"}')
+    print()
+
+    if (local_info['latest'] == cloud_info['latest']
+            and local_info['reports'] == cloud_info['reports']
+            and local_info['patients'] == cloud_info['patients']
+            and local_info['users'] == cloud_info['users']):
+        print('兩邊一致，不需要同步')
+        return
+
+    # Merge
+    print('合併中...')
+    bk = backup_local()
+    if bk:
+        print(f'  已備份 Local → {bk.name}')
+
+    merged_path, stats = merge_dbs(LOCAL_DB, cloud_path)
+
+    print(f'  新增 patients: {stats["patients_added"]}')
+    print(f'  新增 reports:  {stats["reports_added"]}')
+    print(f'  更新 reports:  {stats["reports_updated"]}')
+    print(f'  新增 users:    {stats["users_added"]}')
+    print(f'  更新 users:    {stats["users_updated"]}')
+
+    # Replace local with merged
+    shutil.copy2(merged_path, str(LOCAL_DB))
+    merged_info = get_db_info(LOCAL_DB)
+    print(f'\nMerged: {merged_info["patients"]} patients, {merged_info["reports"]} reports, {merged_info["users"]} users')
+
+    # Upload merged to cloud
+    print('上傳合併結果到雲端...')
+    upload_db(LOCAL_DB)
+
+    print('\n同步完成')
+
+
+if __name__ == '__main__':
+    sync()