code/data_creation/diff_label_text_creation_bangla.py

from openai import OpenAI
import json, os

source_language = "Bengali"
if source_language == "English":
    source_lang_code = "en"
elif source_language == "Spanish":
    source_lang_code = "es"
elif source_language == "French":
    source_lang_code = "fr"
elif source_language == "Portuguese":
    source_lang_code = "pt"
elif source_language == "Bengali":
    source_lang_code = "bn"
else:
    assert False, "Unsupported language"
print(f"{source_language}")
with open(
    "/home/mshahidul/readctrl/prompts/syn_data_generation/syn_data_gen_diff_label_Bangla.txt",
    "r",
    encoding="utf-8",
) as f:
    prompt_template = f.read()


api_file = "/home/mshahidul/api_new.json"
with open(api_file, "r", encoding="utf-8") as f:
    api_keys = json.load(f)
openai_api_key = api_keys["openai"]

client = OpenAI(api_key=openai_api_key)


def openai_return(prompt, model="gpt-5"):
    """Send a prompt to GPT and parse JSON."""
    response = client.chat.completions.create(
        model=model,
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt}
        ]
    )
    content = response.choices[0].message.content.strip()
    cleaned = content.replace("```json", "").replace("```", "").strip()
    try:
        return json.loads(cleaned)
    except json.JSONDecodeError:
        print("⚠️ JSON parse failed — storing raw text.")
        return cleaned

input_path = "/home/mshahidul/readctrl/data/translated_data/translation_wo_judge/multiclinsum_gs_train_en2bn_gemma(0_200).json"
save_path = f"/home/mshahidul/readctrl/data/synthetic_dataset_diff_labels/syn_data_diff_labels_{source_lang_code}_0_80.json"
res=[]
if os.path.exists(save_path):
    with open(save_path, "r", encoding="utf-8") as f:
        res = json.load(f)
import tqdm
with open(input_path, "r", encoding="utf-8") as f:
    data = json.load(f)
for idx, item in tqdm.tqdm(enumerate(data)):
    fulltext_bn = item.get("translated_fulltext")
    summary_bn = item.get("translated_summary")
    if not fulltext_bn or not summary_bn:
        print(f"Skipping idx={idx}, id={item.get('id')} due to missing translated fields.")
        continue

    prompt = (
        prompt_template
        .replace("<<<FULL_TEXT>>>", fulltext_bn)
        .replace("<<<SOURCE_LANGUAGE>>>", source_language)
        .replace("<<<GOLD_SUMMARY>>>", summary_bn)
    )
    # import ipdb; ipdb.set_trace()
    sample = openai_return(prompt, model="gpt-5")

    res.append({
        "id": item["id"],
        "index": idx ,
        "fulltext": fulltext_bn,
        "diff_label_texts": sample
    })

    if len(res) % 2 == 0:
        with open(save_path, "w", encoding="utf-8") as f:
            json.dump(res, f, indent=2, ensure_ascii=False)
        print(f"Saved {len(res)} samples so far.")

with open(save_path, "w", encoding="utf-8") as f:
    json.dump(res, f, indent=2, ensure_ascii=False)