python程式設計：計算詞頻的函數繪制圖形

源碼：

import turtle
 
##全局變量##
#詞頻排列顯示個數
count = 10
#單詞頻率數組-作為y軸資料
data = []
#單詞數組-作為x軸資料
words = []
#y軸顯示放大倍數-可以根據詞頻數量進行調節
yScale = 6
#x軸顯示放大倍數-可以根據count數量進行調節
xScale = 30
 
################# Turtle Start  ####################  
#從點(x1,y1)到(x2,y2)繪制線段
def drawLine(t, x1, y1, x2, y2):
    t.penup()
    t.goto (x1, y1)
    t.pendown()
    t.goto (x2, y2)
 
# 在坐标(x,y)處寫文字
def drawText(t, x, y, text):
    t.penup()
    t.goto (x, y)
    t.pendown()
    t.write(text)
 
def drawGraph(t):
    #繪制x/y軸線
    drawLine (t, 0, 0, 360, 0)
    drawLine (t, 0, 300, 0, 0)
 
    #x軸: 坐标及描述
    for x in range(count):
        x=x+1 #向右移一位,為了不畫在原點上
        drawText(t, x*xScale-4, -20, (words[x-1]))
        drawText(t, x*xScale-4, data[x-1]*yScale+10, data[x-1])
    drawBar(t)
 
#繪制一個柱體
def drawRectangle(t, x, y):
    x = x*xScale
    y = y*yScale#放大倍數顯示
    drawLine(t, x-5, 0, x-5, y)
    drawLine(t, x-5, y, x+5, y)
    drawLine(t, x+5, y, x+5, 0)
    drawLine(t, x+5, 0, x-5, 0)
     
#繪制多個柱體
def drawBar(t):
    for i in range(count):
        drawRectangle(t, i+1, data[i])    
################# Turtle End  ####################
 
         
#對文本的每一行計算詞頻的函數
def processLine(line, wordCounts):
    #用空格替換标點符号
    line = replacePunctuations(line)
    #從每一行擷取每個詞
    words = line.split() 
    for word in words:
        if word in wordCounts:
            wordCounts[word] += 1
        else:
            wordCounts[word] = 1
 
#空格替換标點的函數
def replacePunctuations(line):
    for ch in line:
        if ch in "~@#$%^&*()_-+=<>?/,.:;{}[]|\'""":
            line = line.replace(ch, " ")
    return line
 
def main():
    #使用者輸入一個檔案名
    filename = input("enter a filename:").strip()
    infile = open(filename, "r")
     
    #建立用于計算詞頻的空字典
    wordCounts = {}
    for line in infile:
        processLine(line.lower(), wordCounts)
         
    #從字典中擷取資料對
    pairs = list(wordCounts.items())
 
    #清單中的資料對交換位置,資料對排序
    items = [[x,y]for (y,x)in pairs] 
    items.sort() 
 
    #輸出count個數詞頻結果
    for i in range(len(items)-1, len(items)-count-1, -1):
        print(items[i][1]+"\t"+str(items[i][0]))
        data.append(items[i][0])
        words.append(items[i][1])
         
    infile.close()
     
    #根據詞頻結果繪制柱狀圖
    turtle.title('詞頻結果柱狀圖')
    turtle.setup(900, 750, 0, 0)
    t = turtle.Turtle()
    t.hideturtle()
    t.width(3)
    drawGraph(t)
         
#調用main()函數
if __name__ == '__main__':
    main()      

原文位址： 

http://www.icourse163.org/learn/BIT-268001?tid=1002001005#/learn/content?type=detail&id=1002613039&cid=1002856109