Convert OpenAI Deep Research output to look nice for pasting in Google Docs

mdcite.py

# /// script
# dependencies = [
#   "pyperclip",
# ]
# ///

########
# README
# OpenAI Deep Research is great. When you copy from the Chat into, say, a Google Doc,
# the citations are all messed up (either in place or no citations if you select + copy)
# This script fixes that
# 
# INSTALLATION AND USAGE INSTRUCTIONS:
# $ pip install pyperclip
# $ brew install pandoc
# 
# Hit the copy button at the end of the deep research output
# Run this script: `$ python mdcite.py`
#
# This will make a temp HTML file and open it in your current browser
# You can then copy / paste the contents into a google doc

import pyperclip
import re
import tempfile
import subprocess
import webbrowser
import os


def process_markdown():
    # Get markdown from clipboard
    markdown = pyperclip.paste()

    # Add two spaces at the end of each line that's part of a list
    # This ensures proper line breaks in the HTML output
    lines = markdown.split("\n")
    for i in range(len(lines)):
        if re.match(r"^\s*[-*+]\s", lines[i]) or re.match(r"^\s*\d+\.\s", lines[i]):
            lines[i] = lines[i].rstrip() + "  "
    markdown = "\n".join(lines)

    # Dictionary to store unique citations and their numbers
    citations = {}
    citation_count = 1

    def replace_citation(match):
        nonlocal citation_count
        citation_text = match.group(1).strip()
        citation_link = match.group(2).strip()

        citation_key = (citation_text, citation_link)

        # If this citation hasn't been seen before, add it to our dictionary
        if citation_key not in citations:
            citations[citation_key] = citation_count
            citation_count += 1

        # Return a direct HTML link instead of a footnote reference
        return f'<a href="{citation_link}" title="{citation_text}">[{citations[citation_key]}]</a>'

    # Replace all citations in the text
    pattern = r"\(\[(.*?)\]\(([^)]+)\)\)"
    new_markdown = re.sub(pattern, replace_citation, markdown)

    # Add the references section at the end
    new_markdown += "\n\n---\n\n"
    for citation_key, number in citations.items():
        citation_text, citation_link = citation_key
        # Add numbered references with both text and link
        new_markdown += f"{number}. [{citation_text}]({citation_link})  \n"

    # Copy the new markdown back to clipboard
    pyperclip.copy(new_markdown)

    print(f"Processed {len(citations)} citations.")
    print("New markdown has been copied to clipboard.")

    # Create a temporary markdown file
    with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as md_file:
        md_file.write(new_markdown)
        md_path = md_file.name

    # Create a temporary HTML file path
    html_path = md_path.replace(".md", ".html")

    try:
        # Convert markdown to HTML using pandoc with additional options
        subprocess.run(
            [
                "pandoc",
                "-f",
                "markdown+lists_without_preceding_blankline",
                "-t",
                "html",
                "--wrap=none",
                "-o",
                html_path,
                md_path,
            ],
            check=True,
        )

        # Open the HTML file in the default browser
        webbrowser.open("file://" + html_path)

        print(f"Preview opened in your browser.")
    except subprocess.CalledProcessError:
        print(
            "Error: Failed to convert markdown to HTML. Please make sure pandoc is installed."
        )
    except Exception as e:
        print(f"Error opening preview: {str(e)}")
    finally:
        # Clean up temporary files after a delay (giving browser time to load)
        try:
            os.unlink(md_path)
        except:
            pass


if __name__ == "__main__":
    try:
        process_markdown()
    except Exception as e:
        print(f"An error occurred: {str(e)}")