makebib.py

def uniq(seq):
    set = {}
    map(set.__setitem__, seq, [])
    return set.keys()

#------------------------------------------------------------------------

from string import join

datfile = 'rawbib.dat'
output  = 'bibliography.tex'

lines = open(datfile).readlines()

nocomments = []
for line in lines:
    if line.find('%') == -1:
        nocomments.append(line)
lines = nocomments

entry_lines = []
count = 0
for line in lines:
    if line.startswith('\\bibitem'):
        entry_lines.append(count)
    count += 1

nentry = len(entry_lines)
print "Read %d entries in %s." % (nentry, datfile)

entries = {}
for i in range(nentry):
    i0 = entry_lines[i]
    if i < nentry-1:
        i1 = entry_lines[i+1]-1
    else:
        i1 = len(lines)
#    print "%d/%d, %d/%d\n" % (i, nentry, i1, len(lines))
    thisEntry = join(lines[i0:i1])

    # Find the identifier
    tempStr = thisEntry.split(']')[1]
    tempStr = tempStr.split('{')[1]
    ident = tempStr.split('}')[0]

    # If not in database, store it.
    if not entries.has_key(ident):
        entries[ident] = thisEntry

sorted_ids = sorted(entries, key=str.lower)

fptr = open(output, 'w')
fptr.write('\\bibliography{suthesis}\n')
fptr.write('\\addcontentsline{toc}{chapter}{Bibliography}\n')
fptr.write('\n')
fptr.write('\\begin{thebibliography}{}\n')
fptr.write('\n')
for ident in sorted_ids:
    fptr.write(entries[ident])
    fptr.write('\n')
fptr.write('\n')
fptr.write('\\end{thebibliography}{}\n')
fptr.close()

print "Wrote %d entries with unique identifiers in %s" % \
      (len(sorted_ids), output)