spinoza_project/source/frontend/utils.py

from queue import SimpleQueue
from dotenv import load_dotenv
import re
from langchain.callbacks.base import BaseCallbackHandler

job_done = object()  # signals the processing is done


class StreamingGradioCallbackHandler(BaseCallbackHandler):
    """Callback handler for streaming. Only works with LLMs that support streaming."""

    def __init__(self, q):
        self.q = q

    def on_llm_start(self, serialized, prompts, **kwargs) -> None:
        """Run when LLM starts running."""
        while not self.q.empty():
            try:
                self.q.get(block=False)
            except SimpleQueue.empty:
                continue

    def on_llm_new_token(self, token, **kwargs) -> None:
        """Run on new LLM token. Only available when streaming is enabled."""
        self.q.put(token)

    def on_llm_end(self, response, **kwargs) -> None:
        """Run when LLM ends running."""
        self.q.put(job_done)

    def on_llm_error(self, error, **kwargs) -> None:
        """Run when LLM errors."""
        self.q.put(job_done)


def add_gradio_streaming(llm):
    q = SimpleQueue()
    job_done = object()  # signals the processing is done
    llm.callbacks = [StreamingGradioCallbackHandler(q)]
    return llm, q


def gradio_stream(llm, prompt):
    thread = Thread(target=llm.predict, kwargs={"text": prompt})
    thread.start()
    text = ""
    while True:
        next_token = q.get(block=True)  # Blocks until an input is available
        if next_token is job_done:
            break
        text += next_token
        time.sleep(0.03)
        yield text
    thread.join()


def get_source_link(metadata):
    return metadata["file_url"] + f"#page={metadata['content_page_number'] + 1}"


def make_html_source(source, i, score, config):
    meta = source.metadata
    if meta["file_source_type"] == "AFP":
        return f"""
        <div class="card" id="doc{i}">
            <div class="card-content">
                <h2>Doc {i} - {meta['file_title']} - {meta['file_type']} AFP</h2>
                <p>{source.page_content}</p>
            </div>
            <div class="card-footer">
                <span>{meta['file_source_type']}</span>
                <span>Relevance Score : {round(100*score,1)}%</span>
            </div>
        </div>
        """

    if meta["file_source_type"] == "Presse":
        if meta["file_url"] != "none":
            return f"""
            <div class="card" id="doc{i}">
                <div class="card-content">
                    <h2>Doc {i} - {meta['file_title']} - {meta['file_publisher']}</h2>
                    <p>{source.page_content}</p>
                </div>
                <div class="card-footer">
                    <span>{meta['file_source_type']}</span>
                    <span>Relevance Score : {round(100*score,1)}%</span>
                    <a href={meta['file_url']} target="_blank">
                        <span role="img" aria-label="Open PDF">🔗</span>
                    </a>
                </div>
            </div>
            """
        else:
            return f"""
            <div class="card" id="doc{i}">
                <div class="card-content">
                    <h2>Doc {i} - {meta['file_title']} - {meta['file_publisher']}</h2>
                    <p>{source.page_content}</p>
                </div>
                <div class="card-footer">
                    <span>{meta['file_source_type']}</span>
                    <span>Relevance Score : {round(100*score,1)}%</span>
                </div>
            </div>
            """

    if meta["file_url"]:
        return f"""
        <div class="card" id="doc{i}">
            <div class="card-content">
                <h2>Doc {i} - {meta['file_title']} - Page {meta['content_page_number'] + 1}</h2>
                <p>{source.page_content.replace(config["passage_preprompt"], "")}</p>
            </div>
            <div class="card-footer">
                <span>{meta['file_source_type']}</span>
                <span>Relevance Score : {round(100*score,1)}%</span>
                <a href="{get_source_link(meta)}" target="_blank">
                    <span role="img" aria-label="Open PDF">🔗</span>
                </a>
            </div>
        </div>
        """
    else:
        return f"""
        <div class="card" id="doc{i}">
            <div class="card-content">
                <h2>Doc {i} - {meta['file_title']} - Page {meta['content_page_number'] + 1}</h2>
                <p>{source.page_content.replace(config["passage_preprompt"], "")}</p>
            </div>
            <div class="card-footer">
                <span>{meta['file_source_type']}</span>
                <span>Relevance Score : {round(100*score,1)}%</span>
            </div>
        </div>
        """


def parse_output_llm_with_sources(output):
    content_parts = re.split(
        r"[\[(]?(Doc\s?\d+(?:,\s?Doc\s?\d+)*|doc\s?\d+(?:,\s?doc\s?\d+)*|Doc\s\d+)[\])?]",
        output,
    )
    parts = []
    for part in content_parts:
        if part.lower().startswith("doc"):
            subparts = part.split(",")
            subparts = [
                subpart.lower().replace("doc", "").strip() for subpart in subparts
            ]
            subparts = [
                f"""<a href="#doc{subpart}" class="a-doc-ref" target="_self"><span class='doc-ref'><sup>{subpart}</sup></span></a>"""
                for subpart in subparts
            ]
            parts.append("".join(subparts))
        else:
            parts.append(part)
    content_parts = "".join(parts)

    return content_parts


def clear_text_box(textbox):
    return ""


def add_text(chatbot, text):
    chatbot = chatbot + [(text, None)]
    return chatbot, text


def init_env():
    try:
        load_dotenv()
    except:
        pass