import sys,os,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/review_adjudicated.csv'); d=d[d.label=='F']
def job(v):
    s=sorted(set(d.loc[d.vid==v,'entry_off_s'].astype(int)+5)); return v,len(s),len(grab(v,s))
with cf.ThreadPoolExecutor(4) as ex:
    for r in ex.map(job,list(d.vid.unique())): print(r,flush=True)
rows=[]
for r in d.itertuples():
    f=f"frames/{r.vid}/{int(r.entry_off_s)+5:05d}.png"
    L=labels(f) if os.path.exists(f) else {'top':[],'bot':[]}
    rows.append(dict(cid=r.cid,frame=f,exists=os.path.exists(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/F_ocr.csv',index=False); print('ocr done')