Get book information (isbn or search) from real bookstores.
Web scraping to get book information
This library is to get book information. We can search with keywords, with the isbn, with an advanced search, and do pagination.
We get the data from existing websites. We scrape:
- for French books, http://www.librairie-de-paris.fr (also Decitre, but it’s less complete)
- for Spain: http://www.casadellibro.com
- for Germany: http://www.buchlentner.de
we get: the title and authors, the price, the publisher(s), the cover, etc
Import data from an ods or csv file
Another goal of this lib is to easily fetch a large amount of books you have listed in an ods or csv file.
If your file has an ‘isbn’ and a ‘quantity’ column, it’s easy, we will find all the book information.
If it only has the title and the publisher, it’s doable but error prone. We can still do it, but you shall do an inventory of your stock afterwards.
See the odsimport module. It gives back a json. It’s your responsibility to add what you want in your database (this is done in Abelujo https://gitlab.com/vindarel/abelujo).
Usable, but work in progress.
Accepted format and columns
We can read ods and csv files.
- a file with an “isbn” and “quantity” column,
- a file with columns “title”, “publisher”, “isbn” (optionnal in this case), “shelf”, “distributor”, “quantity”. There is no “price” column. “authors” is optionnal (it can help to fetch the correct book).
If the file has no headers, use the “odsettings.py” configuration file (in particular, to set the csv delimiter, either “,” or “;”).
Why not Amazon ?
Amazon kills the book industry and its employees. But moreover, we can add value to our results. We can link to a good and independent bookshop from within our application, we could command books from it, we could say if it has exemplaries in stock or not, etc. And… we learn a lot in doing this !
Technically speaking, the Amazon API web service can be too limitating and not appropriate. One must register to Amazon Product Advertising and to AWS, and the requests rate is limited to 1 request per second.
Why not Google books ?
It has very few data.
Why not the BNF (Bibliothèque Nationale de France) ?
Because, for bookshops, we need recent books (they enter the BNF database after a few months), up to date information. There isn’t a lot of tools either.
Install from pypi:
pip install bookshops
You can try this lib on the command line with the following commands: - livres: french books - libros: spanish books - bucher: german books - come and ask for more :)
and you get the above screenshot.
Options: (this may vary according to the scrapers, check them with -h) - -i or --isbn to ensure to get all the isbn. The command line tool won’t get them by default if they need to be fetched with another http request for each book. That depends on the websites.
As a library
But most of all, from within your program:
from bookshops.frFR.librairiedeparis.librairiedeparisScraper import Scraper as frenchScraper scraper = frenchScraper("search keywords") cards = scraper.search() # we get a list of dictionnaries with the title, the authors, etc.
Work in progress.
You can search ed:agone to search for a specific publisher.
We do pagination:
scraper = frenchScraper("search keywords", PAGE=2)
Develop and test
pip install -e .
Now you can edit the project and run the development version like the lib is meant to be run, i.e. with the entry_points: livres, libros, etc.
Bugs and shortcomings
This is webscraping, so it doesn’t go without pitfalls:
- the site can go down. It happened already.
- the site can change, it which case we would have to change our sraper too. This can be catched early with automated and frequent tests (work ongoing).
- german scraper: search by isbn
- German scraper
- multiprocessing for the german scraper (from 15 to 9s) (see issue #1)
- --isbn option for it
- french, spanish scrapers
- command line tool