import sys,pandas as pd,concurrent.futures as cf; sys.path.insert(0,'scripts'); from grab import grab; from ocr import labels
d=pd.read_csv('data/candidates.csv'); vids=sys.argv[1:] or list(d.vid.unique())
def job(v):
    s=sorted(set(d.loc[d.vid==v,'entry_off_s'].astype(int)+5)); fs=grab(v,s); return v,len(s),len(fs)
with cf.ThreadPoolExecutor(3) as ex:
    for r in ex.map(job,vids): print(r,flush=True)
rows=[]
for _,r in d.iterrows():
    f=f"frames/{r.vid}/{int(r.entry_off_s)+5:05d}.png"
    try: L=labels(f)
    except Exception as e: L={'top':[],'bot':[]}
    rows.append(dict(cid=r.cid,frame=f,top=';'.join(a+b for a,b in L['top']),bot=';'.join(a+b for a,b in L['bot'])))
pd.DataFrame(rows).to_csv('data/candidates_ocr.csv',index=False); print('ocr done')