app.py

import requests
from paddleocr import PaddleOCR, draw_ocr
from PIL import Image
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

img = "input_data/ocr_input/japan1.jpg"

model_id = "deepseek-ai/deepseek-llm-7b-chat"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype=torch.float16, trust_remote_code=True)

def text_inference(text, language):
    system_prompt = (
        f"Given the following {language} text, extract all words in their base (dictionary) form, including verbs, adjectives, nouns, and particles. "
        "Remove all duplicates. Return the base form words as a comma-separated list, and nothing else."
    )
    user_prompt = f"{system_prompt}\n\nText:\n{text}"

    input_ids = tokenizer.apply_chat_template([{"role": "user", "content": user_prompt}], return_tensors="pt").to(model.device)
    output_ids = model.generate(input_ids, max_new_tokens=256)
    output_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)

    # Parse comma-separated string into list
    words = [word.strip() for word in output_text.split(",") if word.strip()]
    return words

def ocr_inference(img, lang):
	ocr = PaddleOCR(use_angle_cls=True, lang=lang,use_gpu=False)
	img_path = img  
	result = ocr.ocr(img_path, cls=True)[0]
	image = Image.open(img_path).convert('RGB')
	boxes = [line[0] for line in result]
	txts = [line[1][0] for line in result]
	scores = [line[1][1] for line in result]
	return txts

def make_flashcards(words):
    pass;