ReviewBeautifier.py

import sys
import re
import os
import numpy

rParts = [
    ("Your Assessment of this Paper's Contribution to HCI", "contribution"),
    ("Overall Rating", "rating"),
    ("Expertise", "expertise"),
    ("The Review", "review"),
    ("The Meta-Review", "metareview")]

def parseReviews(filename):
    text = open(filename, 'r', encoding='utf8').read()

    rData = dict(reviews=[])
    rData['conf'] = text.splitlines()[0]

    titleQuery = re.search(r"Reviews of submission #([0-9]+): \"((.|\r\n|\n)*?\")", text)
    if titleQuery:
        rData['number'] = titleQuery.group(1)
        rData['title'] = titleQuery.group(2)

    splitExpr = r"(%s)" % "|".join([p[0] for p in rParts])

    reviews = re.split(r"[-]+ Submission [0-9]+, Review ([0-9]+) [-]+", text)
    for rNo, rText in zip(reviews[1::2], reviews[2::2]):
        review = dict(number=rNo)
        
        rTypeQuery = re.search(r"Reviewer:\s+([a-z]+)", rText)
        if rTypeQuery:
            review['type'] = rTypeQuery.group(1)

        parts = re.split(splitExpr, rText)
        for partHeader, rPart in zip(parts[1::2], parts[2::2]):
            key = [p[1] for p in rParts if p[0] == partHeader][0]
            review[key] = rPart.strip()

        review['ratingScore'] = float(re.search(r"([1-5]\.[05])", review['rating']).group(1))
        review['expertiseScore'] = float(re.search(r"([1-4])", review['expertise']).group(1))

        rData['reviews'].append(review)

    ratings = [r['ratingScore'] for r in rData['reviews']]
    expertise = [r['expertiseScore'] for r in rData['reviews']]

    rData['scores'] = ratings
    rData['avgScore'] = numpy.mean(ratings)
    rData['stdDevScore'] = numpy.std(ratings)
   
    return rData

def outputLaTeX(reviews):
    scriptDir = os.path.dirname(os.path.realpath(__file__))
    filename = "Submission %s - All Reviews.tex" % reviews['number']
    
    with open(os.path.join(scriptDir, filename), 'w', encoding='utf8') as out:
        intro = r"""
\documentclass[a4paper, twocolumn]{report}

\usepackage[vmargin=2.5cm,lmargin=2.5cm,rmargin=2.5cm,columnsep=20pt]{geometry}
\setlength{\parindent}{0pt}
\usepackage{times}
\usepackage[utf8]{inputenc}

\newenvironment{review}[3]{
\textbf{Review: #1 \hfill Rating: #2 \hfill Expertise: #3 \hfill\hfill}
\par\vspace{-17pt}
\rule[-10pt]{\linewidth}{1pt}
\rule[0pt]{\linewidth}{1pt}
}{
\par\vspace{-11pt}
\rule[-5pt]{\linewidth}{1pt}
\rule[5pt]{\linewidth}{1pt}
\par\vspace{0.2cm plus 0.2cm}
}

\newenvironment{summary}{
  \itshape
}{
  \par\rule[3pt]{\linewidth}{1pt}
}
\newenvironment{improve}{
  \par\rule[3pt]{\linewidth}{1pt}
  \textbf{Areas for improvement:}
}{}
\begin{document}
"""
        outro = r"""
\end{document}"""

        title = r"""
\twocolumn[
    \begin{center}
        {\LARGE %s \par}
        {\large %s \par}
        {\rule[4pt]{\linewidth}{1pt}\par}
    \end{center}
]
"""

        out.write(intro)

        def tex_escape(s):
            return s.replace('%', '\\%').replace('#', '\\#').replace('&', '\\&')

        titleString = "(%s) %s" % (reviews['number'], reviews['title'])
        subTitleString = "Scores: %s, Mean Score: %1.1f, Std-Dev: %1.2f" % (str(reviews['scores']), reviews['avgScore'], reviews['stdDevScore'])
        out.write(title % (titleString, subTitleString))

        for review in reviews['reviews']:
            out.write("\\begin{review}{%s (%s)}{%s}{%s}\n" % (review['number'], review['type'], review['rating'], review['expertise']))

            if 'contribution' in review:
                out.write("\\begin{summary}\n")
                out.write(tex_escape(review['contribution']))
                out.write("\\end{summary}\n")

            if 'review' in review:
                out.write(tex_escape(review['review']))
            if 'metareview' in review:
                out.write(tex_escape(review['metareview']))

            out.write("\\end{review}\n")
        
        out.write(outro)

if __name__ == '__main__':    
    if len(sys.argv) != 2:
        print("call with review text file as parameter")
        raw_input()
    else:
        reviews = parseReviews(sys.argv[1])
        outputLaTeX(reviews)