reads cached data from Notion.app's localStorage, collects incomplete todo items, and writes them to todo.html

notion_todo.py

#!/usr/bin/env python3

# This is written for OSX, python3, and the current (2018-12-04) version of the Notion app
# It writes incomplete todo items, found in Notion's localStorage, to a "todo.html" file, 
# with links back to the Notion app

import os
import sqlite3
import json
from html import escape
from collections import defaultdict

# load cached blocks from Notion desktop's localStorage
blocks = {}
path = os.path.expanduser(
    "~/Library/Application Support/Notion/Local Storage/https_www.notion.so_0.localstorage")
c = sqlite3.Connection(path)
for k, v in c.execute("select key,value from itemtable where key like 'LRU:LocalRecordStore3:%'"):
    _, _, uuid, table = k.split(':')
    data = json.loads(v.decode('utf16')).get('value', {}).get('value')
    if data:
        blocks[uuid] = data


def page_name(id):
    "fetch name of page as text or 'Missinge Page'"
    page = blocks.get(id, {})
    if not page:
        return "Missing Page"
    return prop(page, 'title')


tags = {'b': 'b', 'i': 'em', 'c': 'code'}


def rich_prop(block, key):
    "renders rich property to html"
    parts = block.get('properties', {}).get(key, [])
    rval = []
    for part in parts:
        text = part[0]
        if len(part) > 1:
            for seg in part[1]:
                if seg[0] == 'p': text = page_name(seg[1])
                if seg[0] == 'a': text = ['a', {'href': seg[1]}, text]
                tag = tags.get(seg[0])
                if tag:
                    text = [tag, text]
        rval.append(text)
    return rval


def prop(block, key):
    "returns text from a property (sans page names)"
    if block is not None:
        parts = block.get('properties', {}).get(key, [])
        return ''.join(part[0] for part in parts)


def get_page(block):
    "return id of containing page or None"
    while block != None:
        if block.get('type') == 'page':
            return block.get('id')
        block = blocks.get(block.get('parent_id'))

def link(pid,bid=None):
    "link to block in page"
    rval = 'notion://-/'+pid.replace('-','')
    if bid:
        return rval + '#' + bid.replace('-','')
    return rval

def todo(b, pid):
    "render todo item"
    bid = b.get('id')
    title = rich_prop(b, 'title')
    return ['div', ['a', {'href': link(pid,bid)},"\u2610"], *title]


def render_html(node):
    "turn a pile of lists and dicts into html"
    tag, *rest = node
    attr = []
    body = []
    for x in rest:
        if isinstance(x, dict):
            for k, v in x.items():
                attr.append(f' {k}="{escape(v)}"')
        elif isinstance(x, list):
            body.append(render_html(x))
        else:
            body.append(escape(str(x)))
    return ''.join(['<', tag, *attr, '>', *body, '</', tag, '>'])


# collect todo items by page.
pages = defaultdict(list)
for id, b in sorted(blocks.items()):
    if b.get('type') == 'to_do':
        checked = prop(b, 'checked') == 'Yes'
        if not checked:
            pid = get_page(b)
            pages[pid].append(b)

# render todo items
div = ['div', {'class': 'content'}]
# for now, sort by id to make order stable
for pid, items in sorted(pages.items()):
    title = prop(blocks.get(pid), 'title')
    items = [todo(item, pid) for item in items]
    div.append(['h3', title])
    div.append(['div', *items])

css = '''
body { font-family: sans-serif; }
div.content { width: 800px; margin: auto; }
a { text-decoration: none; color: #555; }
'''

out = render_html(['html',
                   ['head', ['style', css]],
                   ['body', div]])

print(out, file=open('todo.html', 'w'))
print('wrote todo.html')