visualise.py

# A python script to visualise the author relationships in a bibtex database
# Very basic bibtex parsing, but no dependencies (besides using graphvis for graphing the dot file)
# Author: Glen Robertson
# Date: 4 Oct 2011

BIBTEX_FILENAME = "Database.bib"
DOT_FILENAME = "graph.dot"

import re

bibtexFile = open(BIBTEX_FILENAME, 'r')
contents = bibtexFile.read()
bibtexFile.close()

edges = []

# Remove special character notation like in: "Andrea Fre{\ss}mann and Stefanie Br{\"u}ninghaus"
contents = re.sub(r"(\{\\.)(.)(\})", r"\2", contents)
# Split out each bibtex entry (at each @ symbol, will cause problems if @ in an entry)
# Probably would be better to match brackets but that still would fail for quoted brackets.
entries = re.findall("@\w*\{(\w*),([^@]*)", contents)
for entry in entries:
    try:
        # entry[0] is the bibtexkey
        # entry[1] is the content for that entry
        # Find authors in the form: author = {Andrea Fresmann and Stefanie Bruninghaus}
        authors = re.findall("author = \{([^\}]+)\}", entry[1])
        # Should return a list with one element always, so [0]
        if len(authors) > 0:
            authors = authors[0]
        else:
            # Try using editor as author
            authors = re.findall("editor = \{([^\}]+)\}", entry[1])
            if len(authors) > 0:
                authors = authors[0]
            else:
                raise Exception("Cannot find author or editor in bibtex file")
        # Split out individual authors to a list
        # Convert all whitespace to a single space
        authors = re.sub("\s+", " ", authors)
        authors = re.split(" and ", authors)
    except:
        # Debugging
        print("Failed on " + entry[0])
        print(" Entry[1] was: " + entry[1])
        continue
    # Convert from "Andrea Fresmann" to "Fresmann, A.". Avoid single-word authors.
    for i in range(len(authors)):
        if " " in authors[i] and not "," in authors[i]:
            pos = authors[i].rfind(" ")
            authors[i] = authors[i][pos+1:] + ", " + authors[i][:pos]
        # Get the first letter of the firstname
        pos = authors[i].find(",")
        authors[i] = authors[i][:pos+3] + "."
    # Record edges
    for i in range(len(authors)):
        for j in range(i+1, len(authors)):
            edges.append((authors[i], authors[j]))
# Write out the dot file for graphvis
dotFile = open(DOT_FILENAME, 'w')
dotFile.write("/* Warning: This file is generated by 'visualise.py' and will be overwritten! */\r\n")
dotFile.write("/* %d entries found in database */\r\n" % len(entries))
dotFile.write("graph G {\r\n")
for edge in edges:
    dotFile.write('"{0}" -- "{1}";\r\n'.format(edge[0], edge[1]))
dotFile.write("}\r\n")
dotFile.close()