Files

26 lines
1.4 KiB
Python

import xlrd, json, pathlib, decimal, hashlib, re
out=pathlib.Path(__file__).parent
rows=[]; summary=[]; seen={}
for p in sorted(pathlib.Path('/Users/lingniu/Documents').glob('里程统计报表*.xls')):
s=xlrd.open_workbook(str(p)).sheet_by_index(0)
meta={'file':p.name,'sha256':hashlib.sha256(p.read_bytes()).hexdigest(),'rows':s.nrows-1}
if s.cell_value(0,2)!='日期':
meta.update(excluded='No daily date column',nonzero=sum(decimal.Decimal(str(s.cell_value(i,3)))!=0 for i in range(1,s.nrows)))
summary.append(meta); continue
dates=[]
for i in range(1,s.nrows):
plate,phone,day,km=map(lambda x:str(x).strip(),s.row_values(i))
import datetime
datetime.date.fromisoformat(day)
assert re.fullmatch(r'(?:\d+|\d{1,3}(?:,\d{3})+)(?:\.\d+)?',km), km
amount=decimal.Decimal(km.replace(',','')); assert amount.is_finite() and amount>=0
key=(plate,day)
assert key not in seen, ('duplicate',key)
seen[key]=amount; dates.append(day)
rows.append(dict(plate=plate,phone=phone,date=day,km=str(amount),file=p.name,row=i+1))
meta.update(date_min=min(dates),date_max=max(dates),vehicles=len({r['plate'] for r in rows if r['file']==p.name}))
summary.append(meta)
(out/'rows.json').write_text(json.dumps(rows,ensure_ascii=False))
(out/'input-summary.json').write_text(json.dumps(summary,ensure_ascii=False,indent=2))
print(json.dumps(summary,ensure_ascii=False,indent=2))