# Upstox HISTORICAL candles only (v2 historical-candle). No live feeds. Token never printed.
import requests,pandas as pd,gzip,json,time,os,urllib.parse
B='/workspace/research/flow-forensics-2026-10/data'; os.makedirs(f'{B}/1m',exist_ok=True)
tok=open('/home/box/.secrets/upstox_access_token').read().strip()
H={'User-Agent':'Mozilla/5.0','Accept':'application/json','Authorization':'Bearer '+tok}
ins=pd.DataFrame(json.load(gzip.open(f'{B}/NSE.json.gz')))
cons=pd.read_csv('/workspace/market-live-stack/bots/upstox/nifty50_constituents.csv')
eq=ins[(ins.segment=='NSE_EQ')&(ins.instrument_type=='EQ')].set_index('isin')
keys={r.Symbol:eq.loc[r['ISIN Code'],'instrument_key'] for _,r in cons.iterrows()}
keys.update({'NIFTY':'NSE_INDEX|Nifty 50','BANKNIFTY':'NSE_INDEX|Nifty Bank','NIFTYFUT':'NSE_FO|48704','BANKNIFTYFUT':'NSE_FO|48699','NIFTYFUT_NOV':'NSE_FO|61471','INDIAVIX':'NSE_INDEX|India VIX'})
json.dump(keys,open(f'{B}/keys.json','w'),indent=1)
def get(k,iv,to,fr):
    u=f'https://api.upstox.com/v2/historical-candle/{urllib.parse.quote(k,safe="")}/{iv}/{to}/{fr}'
    for i in range(4):
        r=requests.get(u,headers=H,timeout=30)
        if r.status_code==200: return r.json()['data']['candles']
        print('retry',k,iv,r.status_code,r.text[:120]); time.sleep(2+i*3)
    return []
cols=['ts','open','high','low','close','volume','oi']
for s,k in keys.items():
    d=pd.DataFrame(get(k,'day','2026-10-23','2026-08-01'),columns=cols); d.to_csv(f'{B}/1m/{s}_day.csv',index=False)
    rows=[]
    for to,fr in [('2026-10-23','2026-10-01'),('2026-09-30','2026-09-08')]:
        rows+=get(k,'1minute',to,fr); time.sleep(0.25)
    m=pd.DataFrame(rows,columns=cols).drop_duplicates('ts').sort_values('ts'); m.to_csv(f'{B}/1m/{s}_1m.csv',index=False)
    print(s,k,'day',len(d),'1m',len(m),m.ts.min() if len(m) else None,m.ts.max() if len(m) else None,flush=True)