import sys
import urllib
import json
from wekeypedia.wikipedia_page import WikipediaPage as Page, url2title, url2lang
from wekeypedia.wikipedia_network import WikipediaNetwork
from wekeypedia.exporter.nx_json import NetworkxJson
source = "List_of_geometry_topics"
hl = WikipediaNetwork("")
hl.keywords = hl.get_page_links(source)
hl.build()
exporter = NetworkxJson(hl.graph)
exporter.nx_export("adjacency", "_data/network.json")
# print hl.keywords
print ""
def fetch_page(source):
print "📄 fetching: %s" % source.encode('utf-8-sig')
p = Page()
r = p.fetch_from_api_title(source.strip(), { "redirects":"true", "rvparse" : "true", "prop": "info|revisions", "inprop": "url", "rvprop": "content" })
with open("path_points/%s.json" % (source), "w") as f:
json.dump(r, f)
with codecs.open("_path_points/%s.md" % (source), "w", "utf-8-sig") as file:
da_list = []
da_list = map(parse_source_line, open(da_source, "r"))
return da_list
def store_mapping(mapping):
m = Mapping(mapping)
m.csv("%s.mapping.csv" % (sys.argv[1]))
def get_wikipedia_network_structure(concepts):
wkn = WikipediaNetwork(concepts)
wkn.build()
store_mapping(wkn.mapping)
return wkn.graph
if len(sys.argv) > 1:
source = sys.argv[1]
concepts = open_source_list(source)
graph = get_wikipedia_network_structure(concepts)
for export_type in ["node_link", "adjacency"]:
target_name = "%s.%s.json" % (source, export_type)
exporter = NetworkxJson(graph)
exporter.nx_export(export_type, target_name)