Automate the creation of IWPT 1989 proceedings and metadata

ingest_IWPT_1989.py

#!/bin/python

import re
import subprocess

MATCHLINE = re.compile(r'^(</li>)?<li>(.*)[,.].*\.\s+(.*)\; p\. ([0-9]+)-([0-9]+)\.$')

counter = 0

ID='W89-01%02i'

last_page = 0

OFFSET = 10
LAST_CORRECT_PAGE = 374

def correct_page(x):
    if x > LAST_CORRECT_PAGE:
        if x % 2 == 0:
            return x + 1
        else:
            return x - 1
    else:
        return x

with open('iwpt89_toc.html') as f:
    for line in f:
        m = MATCHLINE.match(line)
        if m:
            counter += 1
            authors = list(map(lambda x: x[:-1] if x[-1] in '*' else x, re.split(' and | &amp; |, ', m.group(2))))
            title = m.group(3)
            pages = int(m.group(4)), int(m.group(5))
            if counter == 29:
                pages = pages[0], 285
            # print(pages, last_page)
            assert pages[0] == last_page + 1
            last_page = pages[1]
            s  = '@inproceedings{' + ID % counter + ",\n"
            s += '\t' + 'author = {' + ' and '.join(', '.join(reversed(author.rsplit(maxsplit=1))) for author in authors) + '},\n'
            s += '\t' + 'title = {' + title + '},\n' 
            s += '\tpages = {' + '%i--%i' % pages + '}\n'
            s += '\tbooktitle = {Proceedings of the First International Workshop on Parsing Technologies},\n'
            s += '\taddress = {Pittsburgh, Pennsylvania, USA},\n'
            s += '\tmonth = aug,\n'
            s += '\tyear = 1989,\n'
            s += '}\n'
            with open('IWPT1989-anthology/' + ID % counter + '.bib', 'w') as bib:
                print(s, file=bib)

            pdf_pages = map(str,[OFFSET + correct_page(x) for x in range(pages[0], pages[1] + 1)])
            subprocess.call(['pdftk', 'IWPT_1989.pdf', 'cat'] + list(pdf_pages) + ['output', 'IWPT1989-anthology/' + ID % counter + '.pdf'])
            

    assert last_page == 467