def show_cluster_node(gid):
"""
"""
cluster_db = ClusterDBModule()
cluster_edge = ClusterEdgeModule()
cluster_similarity = NovelSimilarityModule()
cluster_node = cluster_edge.cluster_node_collection(gid)
virtual_novel_node = cluster_similarity.virtual_novel_node_generate(cluster_node)
book_name = cluster_node.book_name.encode('GBK', 'ignore')
pen_name = cluster_node.pen_name.encode('GBK', 'ignore')
print('gid: {0}, book_name: {1}, pen_name: {2}'.format(gid, book_name, pen_name))
print(', '.join('%s: %d' % (chapter.chapter_title.encode('GBK', 'ignore'), chapter.rank) for chapter in virtual_novel_node.chapter_list))
return virtual_novel_node
def run(self, update_edge = False):
"""
"""
self.logger.info('novel cluster node module start')
for site_id in xrange(self.start_site_id, self.end_site_id + 1):
if not self.site_dict.has_key(site_id):
continue
update_time = self.proc_time_dict['dir_fmt_info{0}'.format(site_id)]
self.novel_node_generate(site_id, update_time)
self.novel_gid_list = {}.fromkeys(self.novel_gid_list).keys()
if update_edge is True and len(self.novel_gid_list) > 0:
cluster_edge_module = ClusterEdgeModule()
cluster_edge_module.run(self.novel_gid_list)
self.logger.info('novel cluster node module end')
return True
def check_chapter_title_clean(gid):
"""
检查一本书的标题过滤 和 虚拟目录效果
"""
cluster_edge = ClusterEdgeModule()
cluster_similarity = NovelSimilarityModule()
novel_clean = NovelCleanModule()
cluster_node = cluster_edge.cluster_node_collection(gid)
if cluster_node is False:
return False
print('book_name: {0}, pen_name: {1}'.format(cluster_node.book_name, cluster_node.pen_name))
for novel_node in cluster_node.novel_node_list:
for chapter in novel_node.chapter_list:
chapter.chapter_title = chapter.raw_chapter_title
novel_clean.novel_chapter_clean(novel_node)
print(novel_node.dir_url)
print(', '.join('%s' % chapter.chapter_title.encode('GBK', 'ignore') for chapter in novel_node.chapter_list))
print(', '.join('%s' % chapter.raw_chapter_title.encode('GBK', 'ignore') for chapter in novel_node.chapter_list))
def cluster_edge_module():
"""
"""
novel_module = ClusterEdgeModule()
novel_module.run()
