# --- START OF FILE app.py --- import sys import os import re import gradio as gr import torch from transformers import AutoModelForCausalLM, AutoTokenizer from huggingface_hub import login from dotenv import load_dotenv # --- FIX: Add project root to Python's path --- project_root = os.path.dirname(os.path.abspath(__file__)) sys.path.insert(0, project_root) # --- Updated Spaces import for Zero-GPU compatibility --- try: import spaces print("'spaces' module imported successfully.") except ImportError: print("Warning: 'spaces' module not found. Using dummy decorator for local execution.") class DummySpaces: def GPU(self, *args, **kwargs): def decorator(func): print(f"Note: Dummy @GPU decorator used for function '{func.__name__}'.") return func return decorator spaces = DummySpaces() # --- Step 1: Hugging Face Authentication --- load_dotenv() HF_TOKEN = os.getenv("HF_TOKEN") if not HF_TOKEN: raise ValueError("FATAL: Hugging Face token not found. Please set the HF_TOKEN environment variable.") print("--- Logging in to Hugging Face Hub ---") login(token=HF_TOKEN) # --- Step 2: Initialize Model and Tokenizer --- MODEL_NAME = "Gregniuki/ERNIE-4.5-0.3B-PT-Translator-EN-PL-EN" print(f"--- Loading model from Hugging Face Hub: {MODEL_NAME} ---") device = torch.device("cuda" if torch.cuda.is_available() else "cpu") dtype = torch.bfloat16 if device.type == "cuda" else torch.float32 print(f"--- Using device: {device}, dtype: {dtype} ---") try: tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True) model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, torch_dtype=dtype, trust_remote_code=True).to(device) model.eval() print("--- Model and Tokenizer Loaded Successfully ---") except Exception as e: raise RuntimeError(f"FATAL: Could not load components. Error: {e}") # --- Helper Functions --- def chunk_text(text: str, max_size: int) -> list[str]: if not text: return [] chunks, start_index = [], 0 while start_index < len(text): end_index = start_index + max_size if end_index >= len(text): chunks.append(text[start_index:]); break split_pos = text.rfind('.', start_index, end_index) if split_pos != -1: chunk, start_index = text[start_index : split_pos + 1], split_pos + 1 else: chunk, start_index = text[start_index:end_index], end_index chunks.append(chunk.strip()) return [c for c in chunks if c] def do_translation_get_ids(text_to_translate: str) -> list[int]: """Runs a single translation and returns ONLY the raw output token IDs.""" if not text_to_translate.strip(): return [] messages = [{"role": "user", "content": text_to_translate}] prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) model_inputs = tokenizer([prompt], add_special_tokens=False, return_tensors="pt").to(device) generated_ids_tensor = model.generate(**model_inputs, max_new_tokens=2048, do_sample=True, temperature=0.7, top_p=0.95, top_k=50) input_token_len = model_inputs.input_ids.shape[1] return generated_ids_tensor[0][input_token_len:].tolist() def preprocess_text(text: str) -> str: """Intelligently cleans text by handling newlines.""" if not text: return "" text = re.sub(r'\n{2,}', ' ', text) text = text.replace('\n', ' ') text = re.sub(r'\s{2,}', ' ', text) return text.strip() # --- Step 3: Core Translation Function (DEFINITIVE TOKEN-LEVEL DIFF) --- @spaces.GPU @torch.no_grad() def translate_with_chunks(input_text: str, chunk_size: int, context_words: int, progress=gr.Progress()) -> str: """ Processes chunks using a precise token-level diff to remove overlap. This is the most robust method for this model. """ progress(0, desc="Starting...") processed_text = preprocess_text(input_text) if not processed_text: return "Input text is empty. Please enter some text to translate." text_chunks = chunk_text(processed_text, chunk_size) if len(processed_text) > chunk_size else [processed_text] num_chunks = len(text_chunks) print(f"Processing {num_chunks} chunk(s).") all_results = [] english_context = "" for i, chunk in enumerate(text_chunks): progress(0.2 + (i / num_chunks) * 0.7, desc=f"Translating chunk {i+1}/{num_chunks}") if not english_context or context_words == 0: # First chunk or context disabled: Translate directly and decode output_ids = do_translation_get_ids(chunk) final_translation_for_chunk = tokenizer.decode(output_ids, skip_special_tokens=True).strip() else: # --- The Token-Level Diff Logic --- prompt_with_context = (english_context + " " + chunk).strip() # 1. Get token IDs for the context translation context_ids = do_translation_get_ids(english_context) # 2. Get token IDs for the full translation full_ids = do_translation_get_ids(prompt_with_context) # 3. Find the first point of difference at the token level diff_index = 0 for j in range(min(len(context_ids), len(full_ids))): if context_ids[j] != full_ids[j]: break diff_index += 1 # 4. The clean translation starts from the point of difference clean_ids = full_ids[diff_index:] final_translation_for_chunk = tokenizer.decode(clean_ids, skip_special_tokens=True).strip() all_results.append(final_translation_for_chunk) print(f"Chunk {i+1} processed successfully.") if context_words > 0: # Update context with words from the *source* English chunk words = chunk.split() english_context = " ".join(words[-context_words:]) full_output = " ".join(all_results) progress(1.0, desc="Done!") return full_output # --- Step 4: Create and Launch the Gradio App --- print("\n--- Initializing Gradio Interface ---") app = gr.Interface( fn=translate_with_chunks, inputs=[ gr.Textbox(lines=15, label="Input Text", placeholder="Enter long text to process here..."), gr.Slider(minimum=256, maximum=2048, value=512, step=64, label="Character Chunk Size"), gr.Slider( minimum=0, maximum=50, value=20, step=5, label="Context Overlap (English Words)", info="Number of English words from the previous chunk to use as context. A token-level comparison is used to reliably remove the overlap." ) ], outputs=gr.Textbox(lines=15, label="Model Output", interactive=False), title="ERNIE 4.5 Context-Aware Translator", description="Processes long text using a precise, token-level diffing algorithm to ensure high-quality, consistent translations.", allow_flagging="never" ) if __name__ == "__main__": app.queue().launch()