Skip to content

okfde/bundesrat-scraper

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Bundesrat Voting Scraper

This repo collects the votes in the Bundesrat. For a website that presents the data, check out the Bundesrat Scraper Website as well as a live demo on Render.

The scraper and website, including the data, scraper and website code, are unofficial. The Bundesrat has nothing to do with it. There is no warranty that the scraped data is correct or complete or the website displays the correct information.

The plan:

  • bundesrat contains a scraper that gets the sessions and their agenda items (TOPs) and puts them in a file called sessions.json.
  • The actual voting behaviour of the states is located on the respective states website. The scrapers put the texts into a $STATE/sessions_tops.json file, together with the links to the original documents in $STATE/sessions_urls.json.
  • Each state has its own Scraper in the according folder in form of a Jupyter file.

Setup

Everything is tested unter Python 3.6.8, so please install anaconda to use this version of python. It is optimized for Arch Linux, but should work with other Linux Distros as well. The pdftohtml dependency is required for the pdfcutter tool to work.

  pip install --user jupyter
  yay -S anaconda --noconfirm #Or your package manager
  yay -S poppler --noconfirm #Or your package manager, need pdftohtml program
  source /opt/anaconda/bin/activate root
  conda create -n py368 python=3.6 ipykernel
  sudo conda install -y nb_conda_kernels
  conda activate py368
  sudo conda install -y lxml
  pip install --user pdfcutter requests lxml
  pip install --user wand #Only needed for the graphical debugger of pdfcutter
  sudo ipython3 kernel install #Otherwise Jupyter can't find kernel

Usage

To scrape the sessions and their agenda items, connect to the internet and open the bundesrat-scraper with:

  source /opt/anaconda/bin/activate py368
  jupyter notebook bundesrat-scraper/bundesrat/bundesrat_scraper.ipynb

, and start the code. If you have any problems with import pdfcutter inside Jupyter, then delete the kernel folder of Jupyter (Kernel path taken from jupyter notebook scraper.ipynb) and re-open Jupyter.

You want to do this if there was a new bundesrat session you want to scrape. Before this, you might have to delete the session.json file and the _cache folder.

To scrape the Abstimmungsverhalten of a state, do:

  source /opt/anaconda/bin/activate py368
  jupyter notebook bundesrat-scraper/$STATE/scraper.ipynb

, and start the code. If the bundesrat/session.json file was extended by a session, the Scraper will look for the Abstimmungsverhalten of the new sessions. You might want to disable any VPN because some states won't let you download their documents otherwise.

Environment

Files

  • Each state has its own Scraper in the according folder in form of a Jupyter file. The Jupyter file is a wrapper around the actual scraper_$STATE.py file. This file extends the PDFTextExtractor.py file, which is the code base for the Scrapers.
  • The Glossary.md file explains the used terminology used in the code and comments.
  • The MainBoilerPlate.py contains the code base for collecting the links to the documents of the states.
  • The helper.py file includes some common methods and adaptations of the pdfcutter library.
  • The selectionVisualizer.py file contains a method for doing graphical debugging of the pdfcutter without the need of using Jupyter, but your browser of choice (e.g. Firefox). See Tips section for more information on that.
  • If you want to know more about the files, look into the comments inside the according file.

Scraping

The scraping of the states behaviours consists of four parts:

  1. Collect the URLs to the documents (done by an extension of MainBoilerPlate.py/MainExtractorMethod)
  2. Determine the position of the TOPs in the documents (done by an extension of PDFTextExtractor.py/DefaultTOPPositionFinder). Some common implementations are available in the same file.
  3. Scrape the text between TOPs by the according rules (done by an extension of PDFTextExtractor.py/AbstractSenatsAndBRTextExtractor). Some common implementations are available in the same file.
  4. Put all three classes together (done by an extension of PDFTextExtractor.py/TextExtractorHolder), where you can also define different scraping rules for step 2 and 3 according to the current session and TOP.
  • If you want to know more about the files or see some example classes, look into the comments inside the according file.

Tips

Execute Scraper without Jupyter

If you want to execute the scraper without Jupyter, you can do

  jupyter nbconvert --to script scraper.ipynb

, remove the following lines from the beginning of the new python file:

  get_ipython().run_line_magic('load_ext', 'autoreload')
  get_ipython().run_line_magic('autoreload', '2')

, and execute the python file (after enabling anaconda).

Graphical Debug PDFCutter with Jupyter

Assuming you have a selection from your pdfcutter instance, you can execute the following code to see the selected text parts in a picture:

#import pdfcutter
#cutter=pdfcutter.PDFCutter(filename='./_cache/_download_136845')
debugger = cutter.get_debugger()
#selection = cutter.all().filter(...)
debugger.debug(page).draw(17) # Shows picture of the selected text of `selection` on page 17

Graphical Debug PDFCutter without Jupyter

Assuming you have a selection from your pdfcutter instance, you can execute the following code to see the selected text parts in a picture inside a browser (e.g. Firefox):

#import pdfcutter
#cutter=pdfcutter.PDFCutter(filename='./_cache/_download_136845')
import selectionVisualizer as dVis
#selection = cutter.all().filter(...)
dVis.showCutter(selection, pageNumber=17) # Shows picture of the selected text of `selection` on page 17

Get XML output from PDF

If you want to see the direct out that pdfcutter is working on (e.g. for seeing which words form a chuck or see coordinates for lines ), use

pdftohtml $DOCUMENT.pdf -xml

Debug Output Collection Document Links

If you want to see the PDF links the MainBoilerPlate file sends to the TextExtractorHolder (i.e. what's part of the session_urls.json file), use:

  print(list(MainExtractorMethod(None)._get_pdf_urls()))

Debug only one session

If you want to debug a scraping error inside one session (say 973), you can add the following to lines to the scraper Jupyter file (cell 8):

#  if str(num) in session_tops:
#    continue
  if str(num) != "973":
    continue

Alternatively, you can remove session 973 from the appropriate session_tops.json file and rerun the Jupyter file.

About

Voting behaviour in Bundesrat

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published