Going further:

• Pipelines Can be used to check validity of returned data http://doc.scrapy.org/en/latest/topics/item-pipeline.html

CLOUD-BASED SCRAPING

• ScrapingHub, Scrapy Cloud
• Morph.io

Example deploy on ScrapingHub

• Login/Register on https://app.scrapinghub.com/ Use Google account

• If not done already, run $ pip install shub

• $ shub login Provide your API as found on https://app.scrapinghub.com/account/apikey This writes the API key into a local file: less ~/.scrapinghub.yml Not specific to that repository shub logout to remove it

• Create a new project, note its ID

• cd to the root of a scrapy project directory

• $ shub deploy Provide the ID of the recently created project

• From the web app, run spider Arguments can be provided

• Items are found under the Items tab Free project: one concurrent spider, max 2.5GB of data storage, data retention 7 days Items can be downloaded individually or in batch (big green Items button on top) Can also be accessed through API calls http://doc.scrapinghub.com/api/overview.html e.g. curl -u 1e2490bfc15d4e6089e4b842364b5cd1: “https://storage.scrapinghub.com/items/85589/1/1?format=xml”

• Jobs can be scheduled, etc.

They have a GUI (work in progress) to define elements and build spiders: Portia

CHECK IF ENOUGH TIME FOR LEGAL ASPECTS

ALTERNATIVE: MORPH.IO

• Port of ScraperWiki Classic
• Open source
• Project supported by the OpenAustralia Foundation
• Can be run locally, but also on cloud platform morph.io “Heroku for scrapers”- very convenient - works with GitHub requires a working install of git requires ruby or python + several libraries for local running
• For this reason, not chosen for this workshop
• Platform runs the scraper and also stores data Data can easily be shared, downloaded in structured format and accessed through dedicated API
• Demo if enough time

CHECK IF ENOUGH TIME FOR LEGAL ASPECTS

SCRAPE DATA FROM PDFs

Use Tabula - Free, open source software - Available for Mac, Windows, Linux - Runs in browser (much like OpenRefine) https://github.com/tabulapdf/tabula

This example uses a library that converts PDF to XML, then does the extraction.

http://www.bl.uk/reshelp/atyourdesk/docsupply/help/replycodes/dirlibcodes/ https://morph.io/ostephens/british_library_directory_of_library_codes British Library maintains list of library codes for its Document Supply service. List of codes in a PDF file -> UNSTRUCTURED