mileage-bonus/replace_gps_jun.py

#!/usr/bin/env python3
"""6月测试车辆：用GPS日里程替换仪表盘里程，写入源数据文件"""
import openpyxl, pymysql, os
from datetime import datetime, date
from collections import defaultdict

os.chdir('/Users/kkfluous/Downloads')

# 1. 读取测试车辆清单
fp = '/Users/kkfluous/Library/Containers/com.tencent.xinWeChat/Data/Documents/xwechat_files/wxid_1704407055112_3af8/temp/RWTemp/2026-07/65b41c10aa83cabb7a365b2b8437c1c7/6月测试车辆.xlsx'
wb = openpyxl.load_workbook(fp, data_only=True)
ws = wb.active
rows = list(ws.iter_rows(min_row=1, values_only=True))
test_plates = set()
for row in rows[1:]:
    for ci in [0, 3, 6]:
        p = str(row[ci]).strip() if row[ci] else ''
        if p and p != 'None':
            test_plates.add(p)
wb.close()
print(f'测试车辆: {len(test_plates)}辆')

# 2. 备份
src = '租赁任务考核_2026年6月.xlsx'
bak = '租赁任务考核_2026年6月_备份.xlsx'
if not os.path.exists(bak):
    import shutil
    shutil.copy2(src, bak)
    print(f'已备份: {bak}')

# 3. 读源文件
wb = openpyxl.load_workbook(src)
ws = wb['业务考核视图']
h = [c.value for c in next(ws.iter_rows(min_row=1, max_row=1))]
recs = []
for row in ws.iter_rows(min_row=2, values_only=True):
    r = dict(zip(h, row))
    r['row_num'] = len(recs) + 2
    recs.append(r)

# 测试车辆在源数据中的记录
test_recs = [r for r in recs if r['车牌号'] in test_plates]
print(f'测试车辆在6月源数据中的记录: {len(test_recs)}条 ({len(set(r["车牌号"] for r in test_recs))}辆)')

# 4. 连数据库取GPS
conn = pymysql.connect(host='101.133.130.65', port=3306, user='root', password='ln123!@#',
                       database='hydrogen_energy', connect_timeout=30)
cur = conn.cursor()

def get_gps_km(plate, start_str, end_str):
    """按考核起止日期汇总GPS日里程(km)"""
    if not start_str or not end_str:
        return None
    try:
        sd = datetime.strptime(str(start_str)[:10], '%Y-%m-%d').date()
        ed = datetime.strptime(str(end_str)[:10], '%Y-%m-%d').date()
    except:
        return None
    
    cur.execute("""
        SELECT dates, total_mileage FROM ln_vehicle_g7_mileage
        WHERE plate_number = %s AND dates >= %s AND dates <= %s
        ORDER BY dates
    """, (plate, sd, ed))
    rows = cur.fetchall()
    if not rows:
        return None
    
    # total_mileage is daily cumulative (meters). Get daily delta then sum.
    # Actually check: is it daily km or cumulative? From last time: total_mileage/100000 = km
    # Let's check if values are ascending (cumulative) or fluctuating (daily)
    vals = [r[1] for r in rows]
    is_cumulative = all(vals[i] <= vals[i+1] for i in range(len(vals)-1)) if len(vals) > 1 else False
    
    if is_cumulative:
        # Daily delta: today - yesterday
        deltas = []
        prev = None
        for v in vals:
            if prev is not None and v > prev:
                deltas.append(v - prev)
            elif prev is None:
                deltas.append(v)  # first day
            prev = v
        total_km = sum(deltas) / 100000
    else:
        total_km = sum(vals) / 100000
    
    return round(total_km, 1)

# 批量查询: build plate→[(start,end,row_num)] map
queries = defaultdict(list)
for r in test_recs:
    plate = r['车牌号']
    start = r.get('考核开始日期')
    end = r.get('考核结束日期')
    queries[plate].append((start, end, r['row_num']))

# Process each unique plate
gps_cache = {}
replaced = 0
no_data = 0
no_data_plates = set()

for plate in sorted(queries.keys()):
    # Get all GPS data for this plate in June for caching
    cur.execute("""
        SELECT dates, total_mileage FROM ln_vehicle_g7_mileage
        WHERE plate_number = %s AND dates >= '2026-06-01' AND dates <= '2026-06-30'
        ORDER BY dates
    """, (plate,))
    all_rows = cur.fetchall()
    if not all_rows:
        no_data += len(queries[plate])
        no_data_plates.add(plate)
        continue
    
    for start, end, row_num in queries[plate]:
        km = get_gps_km(plate, start, end)
        if km is None:
            no_data += 1
            no_data_plates.add(plate)
        else:
            gps_cache[row_num] = km

conn.close()

# 5. 更新源文件
# Need to add '原TBOX里程' and '测试里程' columns if not exist
# Check existing columns
last_col_letter = openpyxl.utils.get_column_letter(ws.max_column)
need_new_cols = '原TBOX里程(km)' not in h and '测试里程(km)' not in h

if need_new_cols:
    # Add two new columns
    tbox_col = ws.max_column + 1
    test_col = ws.max_column + 2
    ws.cell(row=1, column=tbox_col, value='原TBOX里程(km)')
    ws.cell(row=1, column=test_col, value='测试里程(km)')
    # Find the actual行驶里程 column index
    actual_idx = h.index('实际行驶里程(km)') + 1  # 1-indexed
else:
    tbox_idx = h.index('原TBOX里程(km)') + 1 if '原TBOX里程(km)' in h else None
    test_idx = h.index('测试里程(km)') + 1 if '测试里程(km)' in h else None

# Apply replacements
for row_num, gps_km in gps_cache.items():
    # Find column indices in current sheet
    row_data = [c.value for c in ws[row_num]]
    h_current = [c.value for c in ws[1]]
    
    actual_col = h_current.index('实际行驶里程(km)') + 1
    old_val = ws.cell(row=row_num, column=actual_col).value
    
    if need_new_cols:
        tbox_col = h_current.index('原TBOX里程(km)') + 1 if '原TBOX里程(km)' in h_current else ws.max_column - 1
        test_col = h_current.index('测试里程(km)') + 1 if '测试里程(km)' in h_current else ws.max_column
    else:
        tbox_col = h_current.index('原TBOX里程(km)') + 1
        test_col = h_current.index('测试里程(km)') + 1
    
    # Write original TBOX mileage
    ws.cell(row=row_num, column=tbox_col, value=float(old_val) if old_val else 0)
    # Write test mileage (TBOX - GPS)
    tbox_val = float(old_val or 0)
    test_km = round(tbox_val - gps_km, 1)
    ws.cell(row=row_num, column=test_col, value=test_km)
    # Replace actual mileage with GPS
    ws.cell(row=row_num, column=actual_col, value=gps_km)
    replaced += 1

wb.save(src)
wb.close()

print(f'\n替换结果:')
print(f'  成功替换: {replaced}条')
print(f'  无GPS数据: {no_data}条 ({len(no_data_plates)}辆)')
if no_data_plates:
    print(f'  无数据车牌: {sorted(no_data_plates)}')
print(f'\n已保存: {src}')