02_SIR_namecard_process.py

file_identifier='2026-EROLLGEN-S25-161-SIR-FinalRoll-Revision1-ENG-136-WI'
base_folder=f'/Users/ashhadulislam/projects/social/SIR/namecard_sets/{file_identifier}'


answer_list=[]
counter=0
for f in os.listdir(base_folder):
    #print(f)
    image_file_location=f'{base_folder}/{f}'

    
    with open(image_file_location, "rb") as img_file:
        base64_image = base64.b64encode(img_file.read()).decode("utf-8")

    data_url = f"data:image/png;base64,{base64_image}"
    answer=namecard_info_extract(openrouter_api_key,namecard_info_extractor_prompt,data_url)
    answer['file_identifier']=file_identifier
    answer['file_name']=f
    answer_list.append(answer)
    counter+=1

    if counter%50==0:
        print(counter,'/',len(os.listdir(base_folder)),time.time())



import pandas as pd
cleaned_answer_list = clean_answer_list(answer_list)

df=pd.DataFrame(cleaned_answer_list)
df=df.sort_values(by=['file_name'])
save_folder='/Users/ashhadulislam/projects/social/SIR/output_sheets'
df.to_csv(f'{save_folder}/{file_identifier}.csv',index=False)