# coding:utf-8
'''
step1：将描述中的机构名提取出来
step2：切分好后转化为向量
step3：计算相似度
'''

import codecs
import re
import jieba as jb
import numpy as np

#获取数据
def getInfo(name):
    f = codecs.open(name+'.txt','rb')
    return f.readlines()

#提取机构名
def extractOrg(data):
    orgList = []
    for line in data:
        extractRule = re.compile(u"[\u4e00-\u9fa5]+(?:研究员|副会长|副所长|副教授|会长|所长|教授|讲师|副院长|院长|大学|[部所院会室系科])")
        orgList.append(extractRule.findall(line.decode('utf-8')))
    return orgList

#切分机构名并向量化
def splitOrg(orgList):
    wordList = []
    vocabulary = []
    for line in orgList:
        words = []
        for item in line:
            words.extend(list(jb.cut(item)))
        wordList.append(words)

    vocabulary = reduce(lambda x,y:x+y,wordList)
    size = len(vocabulary)

    wordMatList = []
    #和单词表比照得到向量表示
    for w in wordList:
        wordArr= np.zeros(size)
        for index,word in enumerate(vocabulary):
            if  word in w:
                wordArr[index] = 1
        wordMatList.append(wordArr)

    return np.mat(wordMatList)


#计算相似度
def calcSim(dataMat):
    sim = []
    length = len(dataMat)
    for i in range(length):
        irow = np.zeros(length)
        for l in range(length):
            irow[l]=np.linalg.norm(dataMat[i]-dataMat[l])
        sim.append(irow)
    return sim

def getSimItem(sim):
    loc = []
    for i in range(len(sim)):
        min=sim[i].min()
        curloc =[]
        for t,v in enumerate(sim[i]):# 经试验发现，这里的相似基本上都是欧氏距离为0
            if v==min:#and t!=i这里去掉是为了下面分组的时候的相似具有传递性假设更合理
                curloc.append(t)
        loc.append(curloc)
    return loc

#给相似度高的聚合，假设相似具有传递性
def dividObj(loc):
    isize = len(loc)
    tempList = range(isize)
    dividList = []
    for i in range(isize):
        if len(tempList) > 0:
            divid = []
            if len(loc[i]) > 0:
                print loc[i]
                for n in loc[i]:
                    if
                    tempList.remove(n)
                    divid.append(n)
                dividList.append(divid)
            else:
                continue
        else:
            break  # 说明所有描述都分好类了，跳出循环

    print '总共有', len(dividList), '个同名学者，对于他们的描述分别是： '
    for i in dividList:
        print '第', i, '个是----->'
        for f in i:
            print content[f]



if __name__ =='__main__':
    content = getInfo(u'王德胜')
    data = extractOrg(content)
    simMat = calcSim(splitOrg(data))
    loc = getSimItem(simMat)




    # for a in range(len(loc)):
    #     print '第',a,'条描述是：',content[a],'与其最接近的有:'
    #     if len(loc[a])>0:
    #         for b in loc[a]:
    #             print '--->第',b,'条描述。分别为： '
    #             print content[b]
    #     else:
    #         print '没有与其相似的描述'