I tried out this1 cool tool for extracting amazon kindle notes. It is way better than emailing myself a pdf of notes like I had done in the past. Then I had some script I lost for converting from pdf to markdown. This project just pulls directly from amazon2 which I didn’t even know was a place your notes go! I used this bookmarklet tool from Prime Radiant Inc and pulled my notes, but then I tried looking at my kindle Clippings file directly also, which looks like is even easier.
However, when I compared my nearly 2 megabyte dump from read.amazon , with my My Clippings.txt at just under 1 megabyte, I realized, right, I have switched kindles in the past and so the 2 megabyte one has more notes. I found 18 books in my clippings and 57 in my historical file. So that sort of adds up.
clippings directly
Counted unique books from notes
import re
from pathlib import Path
clippings = Path("/Volumes/Kindle/documents/My Clippings.txt").read_text()
notes = clippings.split("==========")
def extract_book(s):
m = re.match(r"^(.*)[(]", s.strip().strip("\ufeff"))
if m:
return m.groups()[0].strip()
books = [extract_book(x) for x in notes if extract_book(x) is not None]
len(notes), len(books), len(set(books))
notes structured
def extract_entry(s):
entry = s.strip().strip("\ufeff")
m = re.match(r"^(?P<title>.*)[(](?P<author>.*)[)]\n- Your (?P<type>Highlight|Note|Bookmark) on (page (?P<page>[\dlvixcd]+) [|] )?Location (?P<location>\d+(-\d+)?) [|] Added on (?P<timestamp>[^\n]+)(\n\n)?(?P<content>.*)", entry, re.DOTALL)
return m.groupdict() if m else {"raw": entry}
Kind of a funny side note haha my regex above was failing for some of my notes, in particular like,
Meditations (Modern Library) (Aurelius, Marcus)
- Your Highlight on page xlviii | Location 632-635 | Added on Thursday, November 27, 2025 8:48:32 PM
elaborate scene setting that we expect in a true dialogue, but we do find in a number of entries a kind of internal debate in which the questions or objections of an imaginary interlocutor are answered by a second, calmer voice which corrects or rebukes its errors. The first voice seems to represent Marcus’s weaker, human side; the second is the voice of philosophy.
so I thought ahh ok the title can also have a parenthesis, “Meditations (Modern Library)”, but no haha that was not the problem in my regex. The problem was my page regex, (?P<page>\d+) did not capture roman numerals. Of course the only uncaptured notes in my kindle notes with roman numeral pages were for a Roman author 😂.