converts Dynalist's OPML so it can be imported into Workflowy

dynalist_to_workflowy.py

from tkinter import Tk, messagebox
from tkinter.filedialog import askopenfilename
from pathlib import Path
import xml.etree.ElementTree as ET
import re
from datetime import datetime
import html


color_list = ["red", "orange", "yellow", "green", "sky", "purple"]

# user settings:
convert_dynalist_urls = False  # whether to convert [markdown](links) that contain dynalist urls


def convert_opml(opml_string):
    """converts Dynalist's OPML so it can be imported into Workflowy"""
 
    root = ET.fromstring(opml_string)
    for outline in root.iter("outline"):

        # main text
        text = outline.get("text")
        if text is not None:
            outline.set("text", convert_text(text))

        # note text
        note = outline.get("_note")
        if note is not None:
            outline.set("_note", convert_text(note))

        # colour
        color_num = outline.get("colorLabel")
        if color_num is not None:
            color = color_list[int(color_num) - 1]
            text = outline.get("text")
            outline.set("text", f"<span class='colored bc-{color}'>{text}</span>")

        # complete
        complete = outline.get("complete")
        if complete is not None:
            # create workflowy's "_complete" attribute
            outline.attrib["_complete"] = outline.attrib["complete"]
            del outline.attrib["complete"]  # delete dynalist's "complete" attribute

    opml_output = ET.tostring(root, encoding="utf8").decode("utf8")
    return opml_output


def convert_text(string):
    # bold text
    if "**" in string:
        string = convert_formatting(string, "**", "b")

    # italic text
    if "__" in string:
        string = convert_formatting(string, "__", "i")

    # dates
    if "!(" in string and ")" in string:
        string = convert_dates(string)

    # markdown links
    if "[" in string and "](" in string and ")" in string:
        string = convert_markdown_links(string)

    # dynalist's highlights to workflowy's yellow
    if "==" in string:
        yellow_class = "<span class='colored bc-yellow'>text</span>"
        string = re.sub(r'==(.+?)==', lambda match: yellow_class.replace('text', match.group(1)), string)

    return string


def convert_formatting(string, markdown, html_tag):
    pattern = rf"{re.escape(markdown)}(.*?){re.escape(markdown)}"
    output = re.sub(pattern, rf"<{html_tag}>\1</{html_tag}>", string)
    return output


def convert_dates(string):
    """
    converts dates and date ranges such as !(YYYY-MM-DD) to:
    <time startYear='YYYY' startMonth='MM' startDay='DD'>Day, Month DD, YYYY</time>
    """
    def repl(match):
        # split the match into start and end dates
        dates = [datetime.strptime(date, "%Y-%m-%d") for date in match.group(1).split(" - ")]

        time_tag = "<time"
        for i, date in enumerate(dates):
            prefix = 'start' if i == 0 else 'end'  # add start or end attributes depending on the index
            time_tag += (f" {prefix}Year='{date.year}'"
                         f" {prefix}Month='{date.month}'"
                         f" {prefix}Day='{date.day}'")

        time_tag += ">"  # add date(s) in the format: Day, Month DD, YYYY
        time_tag += " - ".join([f"{date.strftime('%a, %b')} {date.day}, {date.year}" for date in dates])
        time_tag += "</time>"
        return time_tag

    output = re.sub(r"!\((.*?)\)", repl, string)
    return output


def convert_markdown_links(string):
    def repl(match):
        text = match.group(1)
        url = match.group(2)
        if url.startswith("https://dynalist.io/d/") and convert_dynalist_urls is False:
            return match.group(0)  # return original markdown link
        else:
            return f"<a href='{url}'>{text}</a>"

    output = re.sub(r"\[(.*?)\]\((.*?)\)", repl, string)
    return output


def pick_opml_file():
    # create a tkinter window
    root = Tk()
    root.withdraw()
    # set defaults
    default_folder = Path.home() / "downloads"
    file_types = [("OPML files", "*.opml")]
    # show file picker
    selected = askopenfilename(initialdir=default_folder, filetypes=file_types)
    root.destroy()  # close tkinter
    if selected:
        return Path(selected)


def document_name(opml_string):
    root = ET.fromstring(opml_string)
    # get the first 'outline' element
    first_outline = next(root.iter("outline"), None)
    if first_outline is not None:
        doc_name = first_outline.get('text')
        return doc_name


def main():
    file = pick_opml_file()
    if file:
        with open(file, "r", encoding="utf-8") as f:
            opml_string = f.read()
        # convert
        new_opml = convert_opml(opml_string)
        # get document name
        doc_name = document_name(opml_string)
        # remove characters not allowed in a filename
        doc_name = re.sub(r'[\\/:*?"<>|]', '', doc_name)
        # create new file path
        filename = file.stem.replace("dynalist", "workflowy")
        new_file = file.parent / f"{filename}_{doc_name}.opml"
        # save
        with open(new_file, "w", encoding="utf-8") as f:
            f.write(new_opml)


if __name__ == "__main__":
    main()

usage:

• export a dynalist document as opml to the downloads folder
• run this script
• pick the .opml file you want to convert
• copy the contents of the converted file to the clipboard and paste it into workflowy