app.py

import argparse
import copy
import os
import re
import subprocess
import tempfile
import base64
from pathlib import Path
import fitz
import gradio as gr
import time
import html
from openai import OpenAI
from s3_uploads import upload_to_s3
from environs import env


stop_generation = False

def stream_from_vllm(messages):
    global stop_generation
    client = OpenAI(
        base_url="https://router.huggingface.co/v1",
        api_key=env.str("HF_API_KEY"),
    )


    response = client.chat.completions.create(
        model="THUDM/GLM-4.1V-9B-Thinking:novita",
        messages=messages,
        temperature=0.01,
        stream=True,
        max_tokens=8000
    )

    for chunk in response:
        if stop_generation:
            break

        if chunk.choices and chunk.choices[0].delta:
            delta = chunk.choices[0].delta
            yield delta


class GLM4VModel:
    def _strip_html(self, text: str) -> str:
        return re.sub(r"<[^>]+>", "", text).strip()

    def _wrap_text(self, text: str):
        return [{"type": "text", "text": text}]

    def _image_to_base64(self, image_path):
        with open(image_path, "rb") as image_file:
            encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
            ext = Path(image_path).suffix.lower()
            if ext in ['.jpg', '.jpeg']:
                mime_type = 'image/jpeg'
            elif ext == '.png':
                mime_type = 'image/png'
            elif ext == '.gif':
                mime_type = 'image/gif'
            elif ext == '.bmp':
                mime_type = 'image/bmp'
            elif ext in ['.tiff', '.tif']:
                mime_type = 'image/tiff'
            elif ext == '.webp':
                mime_type = 'image/webp'
            else:
                mime_type = 'image/jpeg'

            return f"data:{mime_type};base64,{encoded_string}"

    def _pdf_to_imgs(self, pdf_path):
        doc = fitz.open(pdf_path)
        imgs = []
        for i in range(doc.page_count):
            pix = doc.load_page(i).get_pixmap(dpi=180)
            img_p = os.path.join(tempfile.gettempdir(), f"{Path(pdf_path).stem}_{i}.png")
            pix.save(img_p)
            imgs.append(img_p)
        doc.close()
        return imgs

    def _ppt_to_imgs(self, ppt_path):
        tmp = tempfile.mkdtemp()
        subprocess.run(
            ["libreoffice", "--headless", "--convert-to", "pdf", "--outdir", tmp, ppt_path],
            check=True,
        )
        pdf_path = os.path.join(tmp, Path(ppt_path).stem + ".pdf")
        return self._pdf_to_imgs(pdf_path)

    def _files_to_content(self, media):
        out = []
        for f in media or []:
            ext = Path(f).suffix.lower()
            if ext in [".mp4", ".avi", ".mkv", ".mov", ".wmv", ".flv", ".webm", ".mpeg", ".m4v"]:
                out.append({"type": "video_url", "video_url": {"url": upload_to_s3(f)}})
            elif ext in [".jpg", ".jpeg", ".png", ".gif", ".bmp", ".tiff", ".webp"]:
                out.append({"type": "image_url", "image_url": {"url": upload_to_s3(f)}})
            elif ext in [".ppt", ".pptx"]:
                for p in self._ppt_to_imgs(f):
                    out.append({"type": "image_url", "image_url": {"url": upload_to_s3(p)}})
            elif ext == ".pdf":
                for p in self._pdf_to_imgs(f):
                    out.append({"type": "image_url", "image_url": {"url": upload_to_s3(p)}})
        return out

    def _stream_fragment(self, reasoning_content: str = "", content: str = "", skip_think: bool = True):
        think_html = ""
        answer_md = ""

        if reasoning_content and not skip_think:
            reasoning_content_clean = reasoning_content.strip()
            think_html = (
                "### 💭 Thinking\n"
                "<details open>\n"
                "<summary>Click to expand</summary>\n\n"
                f"{reasoning_content_clean}\n"
                "</details>\n"
            )

        if content:
            answer_md = content.strip()

        return think_html + "\n\n" + answer_md


    def _build_messages(self, raw_hist, sys_prompt):
        msgs = []
        if sys_prompt.strip():
            msgs.append({"role": "system", "content": [{"type": "text", "text": sys_prompt.strip()}]})
        for h in raw_hist:
            if h["role"] == "user":
                msgs.append({"role": "user", "content": h["content"]})
            else:
                raw = re.sub(r"<details.*?</details>", "", h["content"], flags=re.DOTALL)
                clean_content = self._strip_html(raw).strip()
                if clean_content:
                    msgs.append({"role": "assistant", "content": self._wrap_text(clean_content)})
        return msgs

    def stream_generate(self, raw_hist, sys_prompt: str, *, skip_special_tokens: bool = False):
        global stop_generation
        stop_generation = False
        msgs = self._build_messages(raw_hist, sys_prompt)
        reasoning_buffer = ""
        content_buffer = ""

        try:
            for delta in stream_from_vllm(msgs):
                if stop_generation:
                    break
                if hasattr(delta, 'reasoning_content') and delta.reasoning_content:
                    reasoning_buffer += delta.reasoning_content
                elif hasattr(delta, 'content') and delta.content:
                    content_buffer += delta.content
                else:
                    if isinstance(delta, dict):
                        if 'reasoning_content' in delta and delta['reasoning_content']:
                            reasoning_buffer += delta['reasoning_content']
                        if 'content' in delta and delta['content']:
                            content_buffer += delta['content']
                    elif hasattr(delta, 'content') and delta.content:
                        content_buffer += delta.content

                yield self._stream_fragment(reasoning_buffer, content_buffer)

        except Exception as e:
            error_msg = f"Error during streaming: {str(e)}"
            yield self._stream_fragment("", error_msg)


glm4v = GLM4VModel()

sys_prompt = """Instructions:
Extract only "BILL OF METERIAL" table containing columns same as it is!
colums: (POSITION, DESCRIPTION, N PIECES, MATERIAL (like SA 516 Gr.70N or SA 105 N), DIMENSIONS(like 1700 I.D. X 2045H 50 THK.), WT.Kgs

Ignore title blocks, revision notes, drawing numbers, and general annotations outside the "BILL OF METERIAL".
If a page contains multiple tables, extract only those explicitly related to BILL OF METERIAL.

Preserve the row and column's order and structure as it is!
Do not include any surrounding decorative lines or borders—only. 
give clean tabular data.
output format: markdown table format with following columns (POSITION, DESCRIPTION, N PIECES, MATERIAL, DIMENSIONS(like 1700 I.D. X 2045H 50 THK.) and WT.Kgs)"""

def extract_table_from_file(file):
    if file is None:
        return "Please upload a file."

    payload = glm4v._files_to_content([file.name])
    raw_hist = [{"role": "user", "content": payload}]
    
    full_response = ""
    yield "<h2>🌀 Processing...</h2>\n"
    try:
        for chunk in glm4v.stream_generate(raw_hist, sys_prompt):
            full_response = chunk
            yield full_response
    except Exception as e:
        yield f"<div style='color: red;'>Error: {html.escape(str(e))}</div>"

theme = gr.themes.Ocean(
    primary_hue="gray",
)

with gr.Blocks(title="demo", theme=theme) as demo:
    gr.Markdown(
        "<div style='text-align:center; margin-bottom:20px;'><h1> PDF Extraction Demo</h1></div"
    )
    with gr.Row():
        with gr.Column():
            up = gr.File(label="Upload File", type="filepath")
            format_selector = gr.Radio(choices=["CSV", "JSON"], label="Output Format", value="CSV")
            submit_btn = gr.Button("Submit", variant="primary")
        with gr.Column():
            output_markdown = gr.Markdown(label="Extracted Table")

    submit_btn.click(
        extract_table_from_file,
        inputs=[up],
        outputs=[output_markdown],
       
    )

if __name__ == "__main__":
    demo.launch()