Pergunta antiga, mas estou fazendo a mesma coisa no momento usando o Pybtex library, que possui um analisador embutido:
from pybtex.database.input import bibtex
#open a bibtex file
parser = bibtex.Parser()
bibdata = parser.parse_file("myrefs.bib")
#loop through the individual references
for bib_id in bibdata.entries:
b = bibdata.entries[bib_id].fields
try:
# change these lines to create a SQL insert
print b["title"]
print b["journal"]
print b["year"]
#deal with multiple authors
for author in bibdata.entries[bib_id].persons["author"]:
print author.first(), author.last()
# field may not exist for a reference
except(KeyError):
continue
