def get_corres_wikipedia_algo_id(page):
wikilinks = [linktitle
for (linksite, linktitle) in list(page.iwlinks())
if linksite == 'wp']
if len(wikilinks) == 0:
# no any wiki links
rd.sadd('rosetta-mapping-error-no-wiki-links', page.page_title)
return None
# first, try wikilinks that has titles similar to the task name,
# these links are sorted by confidence of fuzzy matching
for link in get_sorted_similar_links(page.page_title, wikilinks):
# check if indexed
id = get_id_of_corresponding_algorithm(link, page.page_title)
if id is None:
# try to index this algorithm
wikipage = get_wiki_page(link)
id = index_corresponding_algorithm(wikipage, link, page.page_title)
if id is None:
continue
rd.hset('rosetta-mapping-success', page.page_title,
json.dumps([id]))
rd.sadd('rosetta-mapping-similars-success', page.page_title)
return [id]
# then, if none of the links is similar to the task name,
# 1, store the task description
# 2, relate the implementation with ALL wiki algorithms pages
# mentioned in description
ids = list()
for link in wikilinks:
wikipage = get_wiki_page(link)
if wikipage is not None and is_algorithm_page(wikipage):
# check if indexed
id = get_id_of_corresponding_algorithm(link, page.page_title)
if id is None:
# try to index this algorithm
wikipage = get_wiki_page(link)
id = index_corresponding_algorithm(wikipage, link,
page.page_title)
if id is None:
continue
ids.append(id)
if len(ids) > 0:
rd.hset('rosetta-mapping-success', page.page_title,
json.dumps(ids))
return ids
rd.sadd('rosetta-mapping-error-undefinable-wikilinks', page.page_title)
return None
def index_wiki_page(title, depth, visited):
print 'looking at page %s, at depth %d:' % (title, depth)
algo_id = -1
cate_id = -1
if title in visited and not UPDATING_WIKI:
# don't need to revisit any page if we are not updating for new algos
print 'visited'
return get_ids_of_visited_wiki_page(title)
if pw.is_category_title(title): # is category page
if depth < pw.MAX_CATEGORY_DEPTH:
page = wiki.categorypage(title)
if page is None:
print '-> category not found'
mark_visited(title, visited)
return (algo_id, cate_id)
print '-> category'
child_algo_ids = list()
child_cate_ids = list()
for member in page.categorymembers:
(child_algo_id,
child_cate_id) = index_wiki_page(member, depth + 1, visited)
if child_algo_id != -1:
child_algo_ids.append(child_algo_id)
if child_cate_id != -1:
child_cate_ids.append(child_cate_id)
if len(child_algo_ids) == 0 and len(child_cate_ids) == 0:
# if not algorithm category, igore
mark_visited(title, visited)
return (-1, -1)
# add self to category table, and update cate_id
cate_id = index_wiki_category_entry(page, child_algo_ids,
child_cate_ids)
else: # is member page
page = pw.get_wiki_page(title)
if page is None:
print '-> member page not found'
mark_visited(title, visited)
return (algo_id, cate_id)
if pw.is_algorithm_page(page):
print '-> algorithm page'
# add this algorithm to algorithm table
algo_id = index_wiki_algorithm_entry(page, title, visited)
else:
print '-> member page of other stuff'
mark_visited(title, visited)
return (algo_id, cate_id)
def index_wiki_page(title, depth, visited):
print 'looking at page %s, at depth %d:' % (title, depth)
algo_id = -1
cate_id = -1
if title in visited and not UPDATING_WIKI:
# don't need to revisit any page if we are not updating for new algos
print 'visited'
return get_ids_of_visited_wiki_page(title)
if pw.is_category_title(title): # is category page
if depth < pw.MAX_CATEGORY_DEPTH:
page = wiki.categorypage(title)
if page is None:
print '-> category not found'
mark_visited(title, visited)
return (algo_id, cate_id)
print '-> category'
child_algo_ids = list()
child_cate_ids = list()
for member in page.categorymembers:
(child_algo_id, child_cate_id) = index_wiki_page(member,
depth + 1, visited)
if child_algo_id != -1:
child_algo_ids.append(child_algo_id)
if child_cate_id != -1:
child_cate_ids.append(child_cate_id)
if len(child_algo_ids) == 0 and len(child_cate_ids) == 0:
# if not algorithm category, igore
mark_visited(title, visited)
return (-1, -1)
# add self to category table, and update cate_id
cate_id = index_wiki_category_entry(page,
child_algo_ids, child_cate_ids)
else: # is member page
page = pw.get_wiki_page(title)
if page is None:
print '-> member page not found'
mark_visited(title, visited)
return (algo_id, cate_id)
if pw.is_algorithm_page(page):
print '-> algorithm page'
# add this algorithm to algorithm table
algo_id = index_wiki_algorithm_entry(page, title, visited)
else:
print '-> member page of other stuff'
mark_visited(title, visited)
return (algo_id, cate_id)
def index_wiki_category_entry(page, algo_ids, subcate_ids):
# page title -> category title
# e.g. 'Category:abcd' -> 'abcd'
title = page.title[9:]
result = es.get(index=INDEX_NAME,
doc_type='category',
id=normalize(title),
ignore=404)
if not result['found']:
body = {'name': title, 'algorithms': algo_ids, 'children': subcate_ids}
# try to get category description from corresponding page
corres_page = pw.get_wiki_page(title)
if corres_page is None:
# if there's no corresponding page,
# get category description from category page itself
corres_page = pw.get_wiki_page(page.title)
if corres_page.summary == '':
try:
corres_page = pw.get_wiki_page(corres_page.links[0])
except KeyError:
# if there is no links, KeyError will be raised
# set corres_page to None,
# because empty summary will cause error
corres_page = None
if corres_page is not None:
body['tag_line'] = get_tag_line(corres_page.summary)
body['description'] = corres_page.summary
retval = es.index(index=INDEX_NAME,
doc_type='category',
id=normalize(title),
body=body)
return retval['_id']
else:
print page.title, 'updated with new algos/cates!'
def index_wiki_category_entry(page, algo_ids, subcate_ids):
# page title -> category title
# e.g. 'Category:abcd' -> 'abcd'
title = page.title[9:]
result = es.get(index=INDEX_NAME, doc_type='category',
id=normalize(title), ignore=404)
if not result['found']:
body = {
'name': title,
'algorithms': algo_ids,
'children': subcate_ids
}
# try to get category description from corresponding page
corres_page = pw.get_wiki_page(title)
if corres_page is None:
# if there's no corresponding page,
# get category description from category page itself
corres_page = pw.get_wiki_page(page.title)
if corres_page.summary == '':
try:
corres_page = pw.get_wiki_page(corres_page.links[0])
except KeyError:
# if there is no links, KeyError will be raised
# set corres_page to None,
# because empty summary will cause error
corres_page = None
if corres_page is not None:
body['tag_line'] = get_tag_line(corres_page.summary)
body['description'] = corres_page.summary
retval = es.index(index=INDEX_NAME, doc_type='category',
id=normalize(title), body=body)
return retval['_id']
else:
print page.title, 'updated with new algos/cates!'
def nel_title_suggest(page_title, auto_suggest=True):
wikipage = get_wiki_page(page_title, auto_suggest)
if wikipage is not None:
# check if indexed
id = get_id_of_corresponding_algorithm(page_title, page_title)
if id is None:
# try to index this algorithm
id = index_corresponding_algorithm(wikipage, page_title,
page_title)
if id is not None:
rd.hset('rosetta-mapping-success', page_title, json.dumps([id]))
rd.sadd('rosetta-mapping-success-wikipedia-autosuggest',
page_title)
safe_print(id)
print '--second'
return [id]
def nel_title_suggest(page_title, auto_suggest=True):
wikipage = get_wiki_page(page_title, auto_suggest)
if wikipage is not None:
# check if indexed
id = get_id_of_corresponding_algorithm(page_title, page_title)
if id is None:
# try to index this algorithm
id = index_corresponding_algorithm(wikipage, page_title,
page_title)
if id is not None:
rd.hset('rosetta-mapping-success', page_title,
json.dumps([id]))
rd.sadd('rosetta-mapping-success-wikipedia-autosuggest',
page_title)
safe_print(id)
print '--second'
return [id]
def nel_wikilinks_match_all(wikilinks, page_title):
ids = list()
for link in wikilinks:
wikipage = get_wiki_page(link)
if wikipage is not None and is_algorithm_page(wikipage):
# check if indexed
id = get_id_of_corresponding_algorithm(link, page_title)
if id is None:
# try to index this algorithm
id = index_corresponding_algorithm(wikipage, link, page_title)
if id is None:
continue
ids.append(id)
if len(ids) > 0:
rd.hset('rosetta-mapping-success', page_title, json.dumps(ids))
rd.sadd('rosetta-mapping-success-all-algo-links', page_title)
safe_print(ids)
print '--all-link'
return ids
def nel_wikilinks_match_all(wikilinks, page_title):
ids = list()
for link in wikilinks:
wikipage = get_wiki_page(link)
if wikipage is not None and is_algorithm_page(wikipage):
# check if indexed
id = get_id_of_corresponding_algorithm(link, page_title)
if id is None:
# try to index this algorithm
id = index_corresponding_algorithm(wikipage, link,
page_title)
if id is None:
continue
ids.append(id)
if len(ids) > 0:
rd.hset('rosetta-mapping-success', page_title,
json.dumps(ids))
rd.sadd('rosetta-mapping-success-all-algo-links', page_title)
safe_print(ids)
print '--all-link'
return ids
def nel_title_crosswikis(page_title):
query = "SELECT cprob, entity FROM queries WHERE anchor = %s"
suggested_wikilinks = list(session.execute(query, [page_title]))
suggested_wikilinks = sorted(suggested_wikilinks, key=lambda tup: tup[0])
if len(suggested_wikilinks) > 0:
# get the most confident link
toplink = suggested_wikilinks[0][1]
wikipage = get_wiki_page(toplink.replace('_', ' '))
if wikipage is not None:
# check if indexed
id = get_id_of_corresponding_algorithm(toplink, page_title)
if id is None:
# try to index this algorithm
id = index_corresponding_algorithm(wikipage, toplink,
page_title)
if id is not None:
rd.hset('rosetta-mapping-success', page_title,
json.dumps([id]))
rd.sadd('rosetta-mapping-success-crosswikis', page_title)
safe_print(id)
print '--third'
return [id]
def nel_title_crosswikis(page_title):
query = "SELECT cprob, entity FROM queries WHERE anchor = %s"
suggested_wikilinks = list(session.execute(query, [page_title]))
suggested_wikilinks = sorted(suggested_wikilinks,
key=lambda tup: tup[0])
if len(suggested_wikilinks) > 0:
# get the most confident link
toplink = suggested_wikilinks[0][1]
wikipage = get_wiki_page(toplink.replace('_', ' '))
if wikipage is not None:
# check if indexed
id = get_id_of_corresponding_algorithm(toplink, page_title)
if id is None:
# try to index this algorithm
id = index_corresponding_algorithm(wikipage, toplink,
page_title)
if id is not None:
rd.hset('rosetta-mapping-success', page_title,
json.dumps([id]))
rd.sadd('rosetta-mapping-success-crosswikis',
page_title)
safe_print(id)
print '--third'
return [id]
def nel_wikilinks_fuzzy(wikilinks, page_title):
if len(wikilinks) == 0:
# no any wiki links
rd.sadd('rosetta-mapping-error-no-wiki-links', page_title)
else:
# first, try wikilinks that has titles similar to the task name,
# these links are sorted by confidence of fuzzy matching
for link in get_sorted_similar_links(page_title, wikilinks):
# check if indexed
id = get_id_of_corresponding_algorithm(link, page_title)
if id is None:
# try to index this algorithm
wikipage = get_wiki_page(link)
if wikipage is not None:
id = index_corresponding_algorithm(wikipage, link,
page_title)
safe_print(id)
if id is not None:
rd.hset('rosetta-mapping-success', page_title,
json.dumps([id]))
rd.sadd('rosetta-mapping-similars-success', page_title)
safe_print(id)
print '--first'
return [id]
def nel_wikilinks_fuzzy(wikilinks, page_title):
if len(wikilinks) == 0:
# no any wiki links
rd.sadd('rosetta-mapping-error-no-wiki-links', page_title)
else:
# first, try wikilinks that has titles similar to the task name,
# these links are sorted by confidence of fuzzy matching
for link in get_sorted_similar_links(page_title, wikilinks):
# check if indexed
id = get_id_of_corresponding_algorithm(link, page_title)
if id is None:
# try to index this algorithm
wikipage = get_wiki_page(link)
if wikipage is not None:
id = index_corresponding_algorithm(wikipage, link,
page_title)
safe_print(id)
if id is not None:
rd.hset('rosetta-mapping-success', page_title,
json.dumps([id]))
rd.sadd('rosetta-mapping-similars-success', page_title)
safe_print(id)
print '--first'
return [id]
