import networkx as nx
# 創(chuàng)建有向圖
G = nx.DiGraph() 
# 有向圖之間邊的關(guān)系
edges = [("B1", "B"), ("B2", "B"), ("C1", "C"), ("C2", "C"), ("D1", "D"), ("D2", "D"), ("D", "A"), ("C", "A"), ("B", "A")]
for edge in edges:
    G.add_edge(edge[0], edge[1])
pagerank_list = nx.pagerank(G, alpha=1)
print("pagerank值是：", pagerank_list)

# -*- coding: utf-8 -*-
# 用 PageRank 挖掘希拉里郵件中的重要任務(wù)關(guān)系
import pandas as pd
import networkx as nx
import numpy as np
from collections import defaultdict
import matplotlib.pyplot as plt
# 數(shù)據(jù)加載
emails = pd.read_csv("./input/Emails.csv")
# 讀取別名文件
file = pd.read_csv("./input/Aliases.csv")
aliases = {}
for index, row in file.iterrows():
    aliases[row['Alias']] = row['PersonId']
# 讀取人名文件
file = pd.read_csv("./input/Persons.csv")
persons = {}
for index, row in file.iterrows():
    persons[row['Id']] = row['Name']
# 針對(duì)別名進(jìn)行轉(zhuǎn)換        
def unify_name(name):
    # 姓名統(tǒng)一小寫(xiě)
    name = str(name).lower()
    # 去掉, 和 @后面的內(nèi)容
    name = name.replace(",","").split("@")[0]
    # 別名轉(zhuǎn)換
    if name in aliases.keys():
        return persons[aliases[name]]
    return name
# 畫(huà)網(wǎng)絡(luò)圖
def show_graph(graph, layout='spring_layout'):
    # 使用 Spring Layout 布局，類(lèi)似中心放射狀
    if layout == 'circular_layout':
        positions=nx.circular_layout(graph)
    else:
        positions=nx.spring_layout(graph)
    # 設(shè)置網(wǎng)絡(luò)圖中的節(jié)點(diǎn)大小，大小與 pagerank 值相關(guān)，因?yàn)?pagerank 值很小所以需要 *20000
    nodesize = [x['pagerank']*20000 for v,x in graph.nodes(data=True)]
    # 設(shè)置網(wǎng)絡(luò)圖中的邊長(zhǎng)度
    edgesize = [np.sqrt(e[2]['weight']) for e in graph.edges(data=True)]
    # 繪制節(jié)點(diǎn)
    nx.draw_networkx_nodes(graph, positions, node_size=nodesize, alpha=0.4)
    # 繪制邊
    nx.draw_networkx_edges(graph, positions, edge_size=edgesize, alpha=0.2)
    # 繪制節(jié)點(diǎn)的 label
    nx.draw_networkx_labels(graph, positions, font_size=10)
    # 輸出希拉里郵件中的所有人物關(guān)系圖
    plt.show()
# 將寄件人和收件人的姓名進(jìn)行規(guī)范化
emails.MetadataFrom = emails.MetadataFrom.apply(unify_name)
emails.MetadataTo = emails.MetadataTo.apply(unify_name)
# 設(shè)置遍的權(quán)重等于發(fā)郵件的次數(shù)
edges_weights_temp = defaultdict(list)
for row in zip(emails.MetadataFrom, emails.MetadataTo, emails.RawText):
    temp = (row[0], row[1])
    if temp not in edges_weights_temp:
        edges_weights_temp[temp] = 1
    else:
        edges_weights_temp[temp] = edges_weights_temp[temp] + 1
# 轉(zhuǎn)化格式 (from, to), weight => from, to, weight
edges_weights = [(key[0], key[1], val) for key, val in edges_weights_temp.items()]
# 創(chuàng)建一個(gè)有向圖
graph = nx.DiGraph()
# 設(shè)置有向圖中的路徑及權(quán)重 (from, to, weight)
graph.add_weighted_edges_from(edges_weights)
# 計(jì)算每個(gè)節(jié)點(diǎn)（人）的 PR 值，并作為節(jié)點(diǎn)的 pagerank 屬性
pagerank = nx.pagerank(graph)
# 將 pagerank 數(shù)值作為節(jié)點(diǎn)的屬性
nx.set_node_attributes(graph, name = 'pagerank', values=pagerank)
# 畫(huà)網(wǎng)絡(luò)圖
show_graph(graph)

# 將完整的圖譜進(jìn)行精簡(jiǎn)
# 設(shè)置 PR 值的閾值，篩選大于閾值的重要核心節(jié)點(diǎn)
pagerank_threshold = 0.005
# 復(fù)制一份計(jì)算好的網(wǎng)絡(luò)圖
small_graph = graph.copy()
# 剪掉 PR 值小于 pagerank_threshold 的節(jié)點(diǎn)
for n, p_rank in graph.nodes(data=True):
    if p_rank['pagerank'] < pagerank_threshold: 
        small_graph.remove_node(n)
# 畫(huà)網(wǎng)絡(luò)圖,采用circular_layout布局讓篩選出來(lái)的點(diǎn)組成一個(gè)圓
show_graph(small_graph, 'circular_layout')