def get_or_create_index(path, schema, src):
"""Get or create an Index."""
index = open_dir(path) if exists_in(path) else create_in(path, schema)
indexed_titles = set(field['title'] for field in gen_indexed_fields(index))
corpus = Corpus(src)
documents = set(corpus.gen_documents())
update_index(index.writer(), indexed_titles, documents)
return index
def get_or_create_index(path, schema, src):
"""Get or create an Index."""
index = open_dir(path) if exists_in(path) else create_in(path, schema)
indexed_titles = set(field['title'] for field in gen_indexed_fields(index))
corpus = Corpus(src)
documents = set(corpus.gen_documents())
update_index(index.writer(), indexed_titles, documents)
return index
def get_out_links():
corpus = Corpus('src')
links = {}
for doc in corpus.gen_documents():
text = doc.content
try:
refs = get_references(text)
except:
continue
else:
# print(doc.filename, refs)
out_links = []
for ref in set(refs):
try:
canonical_name = get_canonical_name(ref)
while not canonical_name.endswith('html'):
canonical_name = get_canonical_name(canonical_name)
except KeyError:
# print('KeyError:', doc.title, ref)
pass
else:
out_links.append(canonical_name)
links[doc.filename] = set(out_links)
return links