def get_nips_url(volume_or_year):
nips_proceedings_repository = 'https://papers.nips.cc'
webpage_text = load_webpage(nips_proceedings_repository)
volumes_list = extract_html_tag(webpage_text, 'a')
nips_pattern = re.compile(
r'(?:Advances in )?Neural Information Processing Systems (?:(?P<volume>\d{1,2}) )?\(NIPS (?P<year>\d{4})\)',
re.IGNORECASE)
nips_by_year = {}
nips_by_volume = {}
for v in volumes_list:
extract = nips_pattern.search(v.contents[0])
if extract:
year = extract.group('year')
year = year.strip() if year is not None else year
volume = extract.group('volume')
volume = volume.strip() if volume is not None else volume
url = nips_proceedings_repository + v.get('href').strip()
if year is not None:
nips_by_year[year] = (url, volume, year)
if volume is not None:
nips_by_volume[volume] = (url, volume, year)
book_url = nips_by_year.get(volume_or_year)
if book_url is None:
book_url = nips_by_volume.get(volume_or_year)
if book_url is None:
raise Exception(
'Unknown NIPS volume or year {}'.format(volume_or_year))
else:
return book_url
else:
return book_url
def get_nips_proceedings(volume_or_year):
nips_book_url, volume, year = get_nips_url(volume_or_year)
meta_filename = '_metadata.pkl'
proceedings_source = 'nips'
proceedings_name = year.replace(' ', '_').strip()
proceedings_dir = os.path.join(ROOT_TEMPDIR, proceedings_source,
proceedings_name)
meta_file = os.path.join(proceedings_dir, meta_filename)
if os.path.exists(meta_file):
print('Pickle found: {}\nReading pickle'.format(meta_file))
with open(meta_file, 'rb') as pf:
nips_data = pickle.load(pf)
else:
create_global_tempdir()
create_dir(proceedings_dir)
webpage_text = load_webpage(nips_book_url)
nips_data = parse_nips_proceedings(webpage_text, year)
with open(meta_file, 'wb') as pf:
pickle.dump(nips_data, pf)
return nips_data
def get_aaai_proceedings(year):
# http://www.aaai.org/Library/AAAI/aaai-library.php
meta_filename = '_metadata.pkl'
proceedings_source = 'aaai'
proceedings_name = year.replace(' ', '_').strip()
proceedings_dir = os.path.join(ROOT_TEMPDIR, proceedings_source,
proceedings_name)
meta_file = os.path.join(proceedings_dir, meta_filename)
if os.path.exists(meta_file):
print('Pickle found: {}\nReading pickle'.format(meta_file))
with open(meta_file, 'rb') as pf:
aaai_data = pickle.load(pf)
else:
create_global_tempdir()
create_dir(proceedings_dir)
aaai_proceedings_url = 'http://www.aaai.org/Library/AAAI/aaai{}contents.php'.format(
proceedings_name[2:4])
webpage_text = load_webpage(aaai_proceedings_url)
aaai_data = parse_aaai_proceedings(webpage_text, year)
with open(meta_file, 'wb') as pf:
pickle.dump(aaai_data, pf)
return aaai_data
def parse_old_aaai(paper_url):
paper_webpage = load_webpage(paper_url)
paper_soup = soup_up(paper_webpage)
try:
paper_soup = paper_soup.find_all('div', {'id': 'abstract'})[0]
except IndexError:
print('\nSkipping (unreadable web page): {}'.format(paper_url))
return None, None, None, None, None
title_and_url = paper_soup.find_all('h1')[0]
title_and_url_ = title_and_url.find_all('a')
if title_and_url_:
title_and_url = title_and_url_[0]
# Title
title = title_and_url.text.strip()
# PDF url and file name
pdf_url = title_and_url.get('href')
pdf_filename = None
if pdf_url is not None:
base_dir = os.path.dirname(paper_url)
https_in = False
if base_dir.startswith('https'):
https_in = True
base_dir = base_dir[7:]
else:
https_in = False
base_dir = base_dir[6:]
pdf_url = os.path.normpath(os.path.join(base_dir, pdf_url))
if https_in:
pdf_url = 'https:/' + pdf_url
else:
pdf_url = 'http:/' + pdf_url
pdf_filename = os.path.basename(pdf_url)
else:
# Title
title = title_and_url.text.strip()
pdf_url = None
pdf_filename = None
paper_p = paper_soup.find_all('p')
# Track/ info
info = paper_p[2].contents[-1].encode('utf-8').decode('utf-8')
# Authors
authors = paper_p[0].text.strip()
authors = [a.strip() for a in authors.split(',')]
# Abstract
abstract = paper_p[1].text.strip()
return info, title, authors, pdf_url, pdf_filename
def get_nips_paper(url):
nips_proceedings_repository = 'https://papers.nips.cc'
pdf_url, pdf_filename, zip_sup_url, zip_sup_filename = None, None, None, None
paper_page = load_webpage(url)
paper_page_a = extract_html_tag(paper_page, 'a')
for a in paper_page_a:
a_contents = a.contents[0].strip().lower()
if a_contents == '[pdf]':
pdf_url = nips_proceedings_repository + a.get('href').strip()
pdf_filename = os.path.basename(pdf_url)
elif a_contents == '[supplemental]':
zip_sup_url = nips_proceedings_repository + a.get('href').strip()
zip_sup_filename = os.path.basename(zip_sup_url)
return pdf_url, pdf_filename, zip_sup_url, zip_sup_filename
def parse_new_aaai(paper_url):
paper_webpage = load_webpage(paper_url)
paper_soup = soup_up(paper_webpage)
try:
# Track/ info
track = paper_soup.find_all('div',
{'id': 'breadcrumb'})[0].find_all('a')
info = track[-2].contents[0].strip()
# Title
title = paper_soup.find_all('div', {'id': 'title'})
title = title[0].contents[0].strip()
# Authors
authors = paper_soup.find_all('div', {'id': 'author'})
authors = [a.strip() for a in authors[0].text.split(',')]
# Abstract
abstract = paper_soup.find_all('div', {'id': 'abstract'})
abstract = abstract[0].find_all('div')[0].text.strip()
# PDF url and file name
pdf = paper_soup.find_all('div', {'id': 'paper'})
pdf_url = None
pdf_filename = None
for p in pdf[0].find_all('a'):
if 'pdf' in p.text.lower():
pdf_url = p.get('href')
break
if pdf_url is not None:
pdf_url = pdf_url.replace('/view/', '/download/')
pdf_filename = '{}.pdf'.format('-'.join(pdf_url.split('/')[-2:]))
return info, title, authors, pdf_url, pdf_filename
except IndexError:
print('\nSkipping (unreadable web page): {}'.format(paper_url))
return None, None, None, None, None
def get_mlr_proceedings(volume):
meta_filename = '_metadata.pkl'
proceedings_source = 'mlr'
proceedings_name = volume.replace(' ', '_').strip()
proceedings_dir = os.path.join(ROOT_TEMPDIR, proceedings_source,
proceedings_name)
meta_file = os.path.join(proceedings_dir, meta_filename)
if os.path.exists(meta_file):
print('Pickle found: {}\nReading pickle'.format(meta_file))
with open(meta_file, 'rb') as pf:
mlrp_data = pickle.load(pf)
else:
create_global_tempdir()
create_dir(proceedings_dir)
mlr_proceedings_url = 'http://proceedings.mlr.press/{}/'.format(volume)
webpage_text = load_webpage(mlr_proceedings_url)
mlrp_data = parse_mlr_proceedings(webpage_text, volume)
with open(meta_file, 'wb') as pf:
pickle.dump(mlrp_data, pf)
return mlrp_data
