iPhone Notes extractor

iphone_notes_converter.py

"""
This takes the sqlite 'notes' database from an iphone and dumps it to
flat text files in the current directory, preserving modification
times.

Uses 'lynx' (separate program) to dump html to plain text.
"""

import sqlite3
import datetime
import re
import subprocess
import os
import time

encoding = 'utf8'


def text_from_html(html):
    lynx = subprocess.Popen('lynx --dump -stdin', shell=True, close_fds=True,
                            stdin=subprocess.PIPE, stdout=subprocess.PIPE)
    text, err = lynx.communicate(input=html)
    if err or (lynx.returncode != 0):
        raise RuntimeError("lynx failed with code %s and output: %r" % (lynx.returncode, err))
    # Lynx leaves some extra leading space.
    text = re.sub(r'^   ', '', text, flags=re.MULTILINE)
    return text

def handle_row(row):
    title = re.sub('\W+', ' ', row['title'])
    fname = re.sub('\W+', '_', title) + '.txt'
    output = open(fname, 'w')
    output.write(('title: %s\n' % title).encode(encoding))

    created = datetime.datetime(2001, 1, 1) + datetime.timedelta(seconds=row['creation_date'])
    output.write('created: %s\n' % created.ctime())
    modtime = datetime.datetime(2001, 1, 1) + datetime.timedelta(seconds=row['modification_date'])
    output.write('modified: %s\n' % modtime.ctime())
    output.write('\n\n')
    html = row['data'].encode('utf8')
    output.write(text_from_html(html))
    output.close()
    # Fix file timestamps.
    atime = time.time()
    mtime = time.mktime(modtime.timetuple())
    os.utime(fname, (atime, mtime))

def main(db):
    conn = sqlite3.connect(db)
    conn.row_factory = sqlite3.Row
    cursor = conn.cursor()
    cursor.execute('SELECT * FROM Note, note_bodies ON Note.ROWID = note_bodies.note_id')
    for row in cursor.fetchall():
        handle_row(row)

if __name__ == '__main__':
    import sys
    db = sys.argv[1]
    main(db)