nlp = spacy.load("en_core_web_sm")
track = None
vol = None
with open(FILE_NAME + ".tsv", 'w') as f:
writer = csv.writer(f,
delimiter='\t',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
writer.writerow(
["Title", "Authors", "Pdf", "Presentation", "Volume_Name", "Notes"])
for p in page_lines:
#get the track
if is_centered(p):
print("New track name:", track)
track = is_centered(p)
continue
raw_text = unspace(p.get_text())
pdf = None
title = None
if has_pdf(p):
pdf = has_pdf(p)
pdf = urljoin(URL, pdf)
#print(pdf)
presentation = None
if "presentation" in p.text:
nlp = spacy.load("en_core_web_sm")
track = None
vol = None
with open(FILE_NAME + ".tsv", 'w') as f:
writer = csv.writer(f,
delimiter='\t',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
writer.writerow(
["Title", "Authors", "Pdf", "Presentation", "Volume_Name", "Notes"])
for p in page_lines:
#get the track
if is_centered(p):
print("New track name:", track)
# track = is_centered(p)
continue
raw_text = unspace(p.get_text())
pdf = None
title = None
if has_pdf(p):
pdf = has_pdf(p)
pdf = urljoin(URL, pdf)
#print(pdf)
presentation = None
if "presentation" in p.text: