## A Demonstration of how one can use Python to select chapters of a Ebook based on their POV in title.¶

• The is is a simple demonstration, it only catches the basic case
• You'll likely need to customise it on a per-book basis
• For demonstration, I am using a particular ebook containing the first for books of George R.R. Matrin's Song of Ice and Fire
• It is your responsibility to ensure the legality of this in your local

#### The Libraries¶

We are using python3 today, but this code should work almost without change in python2. To libraries are required.

• ebooklib is for reading and writing the epubs as a whole -- they are basically Zip Archieves
• BeautifulSoup is for reading the HTML files within them

Both can be installed with pip.

We are also going to use the standard library component:

• re for regular expressions
In [22]:
import ebooklib
from ebooklib import epub
from bs4 import BeautifulSoup
import re


## What to Keep:¶

ebooklib.epub breaks the epub up into items. These are files with in the zip archieve. Generally most booked have one item (ie file), per chapter. That is the case for our book.

Of these items, there are three catagories of item we want to keep:

• items that are not chapters at all -- these could be pictures, or metadata or something else. we don't know.
• chapters that are universal, eg the prologue, the dedication or the appendix.
• chapter's that are about the character we are interested in
In [23]:
def is_not_chapter(item):
return item.get_type() != ebooklib.ITEM_DOCUMENT

##### Recognising univeral chapters¶

All the normal chapted in out case are named along the lines of: b01-c01 for book 1 (as it is a complation) chapter 1. Special chapters like the appendix don't follow this pattern. We can check for it with a regex

In [24]:
def is_univeral_chapter(chapter):
return not re.match("(b\d\d.c\d\d)",chapter.get_id())


## Is it about our character?¶

In this particular book all the character names are in the chapter headings. However it does represent them in two different ways. In some sections it is with a <h1> element, in others in is in a <p class="ct"> element. We'll check for both.

Notice this function is a higher order function that returns a function. That makes it work nice with filter -- useful for testing, if you've already stripped down to just the normal chapters.

filter(is_character("JON"), chapters)

In [25]:
def is_character(name):
def inner(chapter):
soup = BeautifulSoup(chapter.get_content())
lambda: soup.find_all(class_='ct')
]
else:
return False

return chapter_character_name == name

return inner


### Bring our conditions together¶

Another higher order function, again to make it work with filter. In this case it is a closure.

In [26]:
def keep_item(character_name):
is_our_charatacter = is_character(character_name)
def inner(item):
return (is_not_chapter(item)
or is_univeral_chapter(item)
or is_our_charatacter(item))
return inner



### Combine it all, with a read and a write¶

Also we'll modify the title, don't want to get them confused. There is also a helper function below to workout the new filename

In [27]:
def rewrite_book_by_character(filename, character):
book.items = list(filter(keep_item(character), book.items))
book.title+=": " + character + "POVs_ONLY"

new_filename = get_new_filename(filename,character)
epub.write_epub(new_filename, book, {})
return new_filename

def get_new_filename(filename,character):
import os.path
filename_base, ext = os.path.splitext(filename)
new_filename = filename_base +"_" + character+"_ONLY"+ext
return new_filename


## Git it a go¶

In [28]:
from IPython.display import FileLink
filename = rewrite_book_by_character('asoiaf01-04.epub', "JON")

Out[28]:
In [ ]: