Confopy

Asserts the linguistic and structural quality of scientific texts.

Confopy is a command-line tool that accepts one or multiple PDF documents and prints textual reports. Currently it only works for German papers.

Name origin: Confopy := Conform + Python

Installation

Installation using pypi (preferred)

sudo pip install -U Confopy

Launch Confopy with

confopy --help
confopy -r document your_paper.pdf

Manual installation

Dependencies:

sudo apt-get install python-pdfminer

sudo pip install -U lxml
sudo pip install numpy==1.6.2
sudo pip install pyyaml nltk==3.0.0
sudo pip install pyenchant==1.6.5
sudo pip install pattern==2.6

Launch Confopy with

python confopy/ --help
python confopy/ -r document your_paper.pdf

Usage

$ confopy -h
usage: confopy [-h] [-l LANGUAGE] [-lx] [-ml] [-o OUTFILE] [-r REPORT] [-rl]
               [-ul] [-vl] [-x]
               [file [file ...]]

Language and structure checker for scientific documents.

positional arguments:
  file                  Document file to analyze (PDF).

optional arguments:
  -h, --help            show this help message and exit
  -l LANGUAGE, --language LANGUAGE
                        Language to use for PDF extraction and document
                        analysis. Default: de
  -lx, --latex          Tell the specified report to format output as LaTeX
                        (if supported by the report).
  -ml, --metriclist     Lists all available metrics by language and exits.
  -o OUTFILE, --outfile OUTFILE
                        File to write the output too. Default: terminal
                        (stdout).
  -r REPORT, --report REPORT
                        Analyses the given document according to the specified
                        report.
  -rl, --reportlist     Lists all available reports by language and exits.
  -ul, --rulelist       Lists all rules and exits.
  -vl, --validate       Validates a given XML against the XSD for the Confopy
                        data model.
  -x, --xml             Converts the PDF file(s) to Confopy XML (structure
                        orientated).

Getting a corpus

Confopy needs a corpus (collection of language data) to run.

For German (TIGER treebank):

Automated download:

1. Go to 
   <your python package directory>/confopy/localization/de/corpus\_de/
2. Execute the script
   tiger_dl_patch.py
   within that folder

Manual download:

1. Go to: 
   http://www.ims.uni-stuttgart.de/forschung/ressourcen/korpora/TIGERCorpus/license/htmlicense.html
2. Accept the license and download TIGER-XML Release 2.2: 
   http://www.ims.uni-stuttgart.de/forschung/ressourcen/korpora/TIGERCorpus/download/tigercorpus-2.2.xml.tar.gz
3. Unpack the archive into confopy/localization/de/corpus\_de/
4. Run the patch tiger\_release\_aug07.corrected.16012013\_patch.py in the same folder
5. Verify that the generated file is named exactly like in confopy/config.py

Python 3

• The package python-pdfminer only works with python 2.4 or newer, but not with python 3

Known Issues and Workarounds

enchant.errors.DictNotFoundError: Dictionary for language 'de_DE' could not be found

Install the German aspell package. E.g. on Ubuntu 16.04:

sudo apt install aspell-de

Unicode errors

• Configure terminal to use unicode!
• For Python devs: http://docs.python.org/2/howto/unicode.html#the-unicode-type
• Convert the TIGER Treebank file "tiger_release_aug07.corrected.16012013.xml" to utf-8 encoding before using Confopy!