Parse chapters for KingReader.apk

chapters.py

#!/usr/bin/python
# coding: utf-8

import sqlite3 as sqlite
import os.path
import struct
import re
from binascii import hexlify

PIN = re.compile(r'[ 　]*[^ 　]+(经卷|品)第[一二三四五六七八九十百千万]+($|\(|之)', re.U)
HUI = re.compile(r'^第[一二三四五六七八九十百千万]+回', re.U)

def toNum(n, length):
    return '{0:0{width}x}'.format(n, width=length)

def toByte(f):
    return hexlify(struct.pack('>f', f)).decode()

def toUnicode(s):
    return ''.join('{0:04x}'.format(ord(c)) for c in s)

def split(s, length):
    return [s[i:i+length] for i in range(0, len(s), length)]

def getTOC(bid, path):
    if path.startswith('/mnt/sdcard/external_sd/大正藏'.encode('utf-8')):
        ptn = PIN
    else:
        ptn = HUI
    owner = toNum(bid, 16)
    f = open(path, 'r')
    fsize = os.path.getsize(path)
    offset = 0
    arr = [
        [],
    ]
    while True:
        lline = f.readline()
        if len(lline)==0: break
        line = lline.strip()
        if ptn.match(line):
            row = [
                toNum(offset, 16),
                owner,
                toByte(offset * 1.0 / fsize),
                toNum(len(line), 8),
                toUnicode(line)
            ]
            arr.append(row)
            print(line)
        offset += len(lline.encode('utf8'))
    if arr == [[]]: return ''
    arr[0].append(toNum(len(arr) - 1, 8))
    s = ''.join(''.join(x) for x in arr)
    ret = 'aced0005'
    for item in split(s, 2048): # 1k of length
        length = '7a' + toNum(len(item) // 2, 8)
        ret += length + item
    return ret

def main():
    con = sqlite.connect('/data/data/com.kingreader.framework/databases/kingreaderv3.db')
    cur = con.cursor()
    cur.execute('SELECT _id, fullpath FROM urls')
    pass
    for row in cur.fetchall():
        print(row)
        path = row[1].encode('utf-8')
        content = getTOC(row[0], path)
        cur.execute('''UPDATE urls SET chapters=X'%s' WHERE _id=%s''' % (content, row[0]))
    con.commit()
    cur.close()
    con.close()

def test():
    con = sqlite.connect('kingreaderv3.db')
    cur = con.cursor()
    content = getTOC(1, '../dzz/01/0157悲华经.txt')
    cur.execute('UPDATE urls SET chapters=? WHERE _id=?', (content, 78))
    con.commit()
    cur.close()
    con.close()

main()