models/aya_vision.py

import os
import json
import re
from PIL import Image
import torch
from transformers import AutoProcessor, AutoModelForImageTextToText
from transformers import Pix2StructForConditionalGeneration, Pix2StructProcessor

# Set Hugging Face Token from env
hf_token = os.getenv("HF_TOKEN")

# Lazy-load model objects
aya_model = None
aya_processor = None

ocr_model = None
ocr_processor = None

# Load prompt
def load_prompt():
    #with open("prompts/prompt.txt", "r", encoding="utf-8") as f:
    #    return f.read()
    return os.getenv("PROMPT_TEXT", "⚠️ PROMPT_TEXT not found in secrets.")


# Try extracting JSON from text
def try_extract_json(text):
    if not text or not text.strip():
        return None
    try:
        return json.loads(text)
    except json.JSONDecodeError:
        start = text.find('{')
        if start == -1:
            return None
        brace_count = 0
        json_candidate = ''
        for i in range(start, len(text)):
            char = text[i]
            if char == '{':
                brace_count += 1
            elif char == '}':
                brace_count -= 1
            json_candidate += char
            if brace_count == 0:
                break
        try:
            return json.loads(json_candidate)
        except json.JSONDecodeError:
            return None


# OCR text from Pix2Struct
def extract_all_text_pix2struct(image: Image.Image):
    global ocr_processor, ocr_model

    if ocr_processor is None or ocr_model is None:
        ocr_processor = Pix2StructProcessor.from_pretrained("google/pix2struct-textcaps-base")
        ocr_model = Pix2StructForConditionalGeneration.from_pretrained("google/pix2struct-textcaps-base")
        device = "cuda" if torch.cuda.is_available() else "cpu"
        ocr_model = ocr_model.to(device)

    inputs = ocr_processor(images=image, return_tensors="pt").to(ocr_model.device)
    predictions = ocr_model.generate(**inputs, max_new_tokens=512)
    output_text = ocr_processor.decode(predictions[0], skip_special_tokens=True)
    return output_text.strip()


# Add fallback names if missing
def assign_event_gateway_names_from_ocr(json_data: dict, ocr_text: str):
    if not ocr_text or not json_data:
        return json_data

    def assign_best_guess(obj):
        if not obj.get("name") or obj["name"].strip() == "":
            obj["name"] = "(label unknown)"

    for evt in json_data.get("events", []):
        assign_best_guess(evt)

    for gw in json_data.get("gateways", []):
        assign_best_guess(gw)

    return json_data


# Main inference function
def run_model(image: Image.Image):
    global aya_model, aya_processor

    if aya_model is None or aya_processor is None:
        model_id = "CohereForAI/aya-vision-8b"
        aya_processor = AutoProcessor.from_pretrained(model_id)
        aya_model = AutoModelForImageTextToText.from_pretrained(
            model_id, device_map="auto", torch_dtype=torch.float16
        )

    prompt = load_prompt()

    messages = [
        {
            "role": "user",
            "content": [
                {"type": "image", "image": image},
                {"type": "text", "text": prompt}
            ]
        }
    ]

    inputs = aya_processor.apply_chat_template(
        messages,
        padding=True,
        add_generation_prompt=True,
        tokenize=True,
        return_dict=True,
        return_tensors="pt"
    ).to(aya_model.device)

    gen_tokens = aya_model.generate(
        **inputs,
        max_new_tokens=5000,
        do_sample=True,
        temperature=0.3,
    )

    output_text = aya_processor.tokenizer.decode(
        gen_tokens[0][inputs.input_ids.shape[1]:],
        skip_special_tokens=True
    )

    parsed_json = try_extract_json(output_text)

    # OCR enhancement
    ocr_text = extract_all_text_pix2struct(image)
    parsed_json = assign_event_gateway_names_from_ocr(parsed_json, ocr_text)

    return {
        "json": parsed_json,
        "raw": output_text
    }