Python词云生成工具3：定制更多参数

文章目录

Python打造一个词云软件
显示分词结果

添加整型参数

我们所有的设置都放在了wcDct中，所以若想用更多的参数来定制词云，那么只需在wcDct中添加内容，例如下面这些整型参数

其次，WordCloud中有很多参数的数据类型都是整型，这些适用于Spinbox

参数	说明	合适的范围	步长
width	词云宽度	100-2000	10
height	词云高度	100-2000	10
min_font_size	最小文字尺寸	1-50	1
max_font_size	最大文字尺寸	10-1000	10
font_step	字体步长	1-20	1
max_words	最大单词数	10-500	10
min_word_length	最短单词长度	0-10	1
scale	图像缩放	默认是1

下面就是要向wcDct中添加的内容。

python 复制代码

wcDct = {
    "最小文字" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":1, "to":50, "increment":1},
        "default":4,
        "call" : "min_font_size"},
    "最大文字" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":10, "to":1e3, "increment":10},
        "default":400,
        "call" : "max_font_size"},
    "字体步长" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":1, "to":20, "increment":1},
        "default":10,
        "call" : "font_step"},
    "最短词长" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":0, "to":10, "increment":1},
        "default":1,
        "call" : "min_word_length"},
    "最多词数" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":10, "to":500, "increment":10},
        "default":200,
        "call" : "max_words"},
    "图像缩放" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":0.5, "to":5, "increment":0.1},
        "default":1,
        "call" : "scale"},
}

布尔型参数

然后是一些布尔类型的参数，适合用Checkbutton

参数	说明	类型	适用组件
repeat	是否重复单词	默认False	Checkbutton
include_numbers	是否包含数字	默认False
normalize_plurals	是否去掉词尾的s	默认True

python 复制代码

wcDct = {
    "单词重复" : {
        "ctrl": ttk.Checkbutton, 
        "paras" : {"width":10},
        "default": False,
        "call" : "repeat"},
    "包含数字" : {
        "ctrl": ttk.Checkbutton, 
        "paras" : {"width":10},
        "default": False,
        "call" : "include_numbers"},
    "去词尾s" : {
        "ctrl": ttk.Checkbutton, 
        "paras" : {"width":10},
        "default": False,
        "call" : "normalize_plurals"},
}

背景颜色

最后，还有一个背景颜色对话框

参数	说明	对话框类型	说明
background_color	背景色	颜色对话框	默认"black"

python 复制代码

wcDct = {
    "背景颜色" : {
        "ctrl": DialogButton,
        "paras" : {"height":5, "widthL":22, "widthR":8, "logtype":"颜色"},
        "call" : "background_color",
        "default" : "balck"},
}

更改之后的界面如下

词云生成逻辑

有了这些之后，还要修改词云生成逻辑，即调用这些参数所获得的值，最后根据上图中的参数，得到点云如下

源代码

所有源代码如下

python 复制代码

import tkinter as tk 
import tkinter.ttk as ttk
from tkinter.filedialog import (askopenfilename,
    askopenfilenames, askdirectory, asksaveasfilename)
from tkinter.colorchooser import askcolor

from threading import Thread

import numpy as np
import re
import csv
import jieba
from wordcloud import WordCloud

import os

class DialogButton(ttk.Frame):
    def __init__(self, master, 
        height, widthL, widthR, logtype, label=None, text=None, 
        frmDct={}, btnDct={}, enyDct={}, logDct={}):
        w = widthL + widthR
        super().__init__(master, 
            height=height, width = w, **frmDct)
        self.pack(fill=tk.X)

        self.text = tk.StringVar() if not text else text
        ttk.Entry(self, width=widthL, textvariable=self.text, 
            **enyDct).pack(side=tk.LEFT, fill = tk.X, padx=5)
        
        ttk.Button(self, width=widthR, 
            text=self.setLabel(logtype, label),
            command = self.Click, **btnDct).pack(side=tk.RIGHT)
        self.logtype = logtype
        self.logDct = logDct

    def setLabel(self, key, label=None):
        if label:
            return label
        labelDct = {
            "文件"   : "选择文件",
            "文件夹" : "选择路径",
            "多文件" : "选择多个文件",
            "保存" : "存储路径",
            "颜色"   : "选择颜色",
        }
        return labelDct[key]

    def Click(self):
        typeDct = {
            "文件"  : askopenfilename,
            "文件夹": askdirectory,
            "多文件": askopenfilenames,
            "保存"  : asksaveasfilename,
            "颜色"  : askcolor,
        }
        text = typeDct[self.logtype](**self.logDct)
        if self.logtype == "颜色":
            text = text[1]
        self.text.set(text)

    def get(self):
        return self.text.get()

    def set(self, txt):
        self.text.set(txt)

wcDct = {
    "词云宽度" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":100, "to":2000, "increment":10},
        "default":800,
        "call" : "width"},
    "词云高度" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":100, "to":2000, "increment":10},
        "default":450,
        "call" : "height"},
    "图像缩放" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":0.5, "to":10, "increment":0.1},
        "default":1,
        "call" : "scale"},
    "最小文字" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":1, "to":50, "increment":1},
        "default":4,
        "call" : "min_font_size"},
    "最大文字" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":10, "to":1e3, "increment":10},
        "default":400,
        "call" : "max_font_size"},
    "字体步长" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":1, "to":20, "increment":1},
        "default":10,
        "call" : "font_step"},
    "最短词长" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":0, "to":10, "increment":1},
        "default":1,
        "call" : "min_word_length"},
    "最多词数" : {
        "ctrl": ttk.Spinbox, 
        "paras" : {"width":10, "from_":10, "to":500, "increment":10},
        "default":200,
        "call" : "max_words"},
    "字体路径" : {"ctrl": DialogButton,
                 "paras" : {"height":5, "widthL":22, "widthR":8, "logtype":"文件"},
                 "call" : "font_path",
                 "default" : r"C:\Windows\Fonts\simhei.ttf"},
    "输入路径" : {"ctrl": DialogButton, "paras" : {"width":25},
                "paras" : {"height":5, "widthL":22, "widthR":8, "logtype":"文件"},},
    "输出路径" : {"ctrl": DialogButton, "paras" : {"width":25},
                "paras" : {"height":5, "widthL":22, "widthR":8, "logtype":"保存"},},
    "背景颜色" : {
        "ctrl": DialogButton,
        "paras" : {"height":5, "widthL":22, "widthR":8, "logtype":"颜色"},
        "call" : "background_color",
        "default" : "balck"},
    "单词重复" : {
        "ctrl": ttk.Checkbutton, 
        "paras" : {"width":10},
        "default": False,
        "call" : "repeat"},
    "包含数字" : {
        "ctrl": ttk.Checkbutton, 
        "paras" : {"width":10},
        "default": False,
        "call" : "include_numbers"},
    "去词尾s" : {
        "ctrl": ttk.Checkbutton, 
        "paras" : {"width":10},
        "default": False,
        "call" : "normalize_plurals"},
}


class DrawWords(ttk.Frame):
    def __init__(self, master, **options):
        super().__init__(master, **options)
        self.pack()
        self.words = None

        self.initWidgets()

    
    def initWidgets(self):
        frm = ttk.Frame(self)
        frm.pack(side=tk.LEFT, fill=tk.Y)
        self.initPara(frm)

        frm = ttk.LabelFrame(self, text="分词结果")
        frm.pack(fill=tk.BOTH, expand=True)
        self.txtSplit = tk.Text(frm)
        self.txtSplit.pack(side=tk.LEFT, fill=tk.BOTH, padx=5, pady=5, expand=True)
        self.addScroll(frm, self.txtSplit)
    
    def addScroll(self, frm, txt):
        scroll = ttk.Scrollbar(frm)
        scroll.pack(side=tk.RIGHT,fill=tk.Y)
        txt.config(yscrollcommand=scroll.set)
        scroll.config(command=txt.yview)

    def setOneCheck(self, frm, key):
        v = wcDct[key]      # 组件参数
        n = v["call"]       # 调用名
        self.vars[n] = tk.BooleanVar()
        self.vars[n].set(v["default"])
        self.checks[n] = v["ctrl"](frm, text=key,
            variable=self.vars[n], **v["paras"])
        self.checks[n].pack(side=tk.LEFT)
        

    def setOneSpinBox(self, frm, key):
        ttk.Label(frm, width=8, text=key).pack(side=tk.LEFT)
        v = wcDct[key]      # 组件参数
        n = v["call"]       # 调用名
        self.spins[n] = v["ctrl"](frm, **v["paras"])
        self.spins[n].set(v["default"])
        self.spins[n].pack(side=tk.LEFT)
    
    def setOneDiaButton(self, frmPara, key):
        frm = ttk.Frame(frmPara)
        frm.pack(side=tk.TOP, fill=tk.X)
        ttk.Label(frm, width=8, text=key).pack(side=tk.LEFT)
        v = wcDct[key]
        n = v["call"] if 'call' in v else key
        self.paths[n] = v["ctrl"](frm, **v['paras'])
        self.paths[n].pack(side=tk.LEFT)
        if 'default' in v:
            self.paths[n].set(v['default'])

    def setOneColButton(self, frm, key):
        frm = ttk.Frame(frmPara)
        frm.pack(side=tk.TOP, fill=tk.X)
        ttk.Label(frm, width=8, text=key).pack(side=tk.LEFT)
        v = wcDct[key]
        n = v["call"] if 'call' in v else key
        self.paths[n] = v["ctrl"](frm, **v['paras'])
        self.paths[n].pack(side=tk.LEFT)
        if 'default' in v:
            self.paths[n].set(v['default'])


    def initPara(self, frmPara):
        self.spins = {}
        self.checks = {}
        self.vars = {}
        keys = ["词云宽度", "词云高度", "最小文字", "最大文字", 
            "字体步长", "图像缩放", "最短词长", "最多词数"]
        for i,key in enumerate(keys):
            if i%2==0:
                frm = ttk.Frame(frmPara)
                frm.pack(side=tk.TOP, fill=tk.X, pady=2)
            self.setOneSpinBox(frm, key)
        
        keys = ["单词重复", "包含数字", "去词尾s"]
        for i,key in enumerate(keys):
            if i%4==0:
                frm = ttk.Frame(frmPara)
                frm.pack(side=tk.TOP, fill=tk.X, pady=2)
            self.setOneCheck(frm, key)

        self.paths = {}
        for key in ["背景颜色", "输入路径", "输出路径", "字体路径"]:
            self.setOneDiaButton(frmPara, key)
                
        frm = ttk.Frame(frmPara)
        frm.pack(side=tk.TOP, fill=tk.X)
        ttk.Button(frm, text="分词预览", 
            command=self.splitWords).pack(side=tk.LEFT)
        ttk.Button(frm, text="分词保存", 
            command=self.saveWords).pack(side=tk.LEFT)
        ttk.Button(frm, text="输出词云", 
            command=self.genWordCloud).pack(side=tk.LEFT)

    def splitWords(self):
        p = self.paths["输入路径"].get()
        with open(p, encoding='utf8') as f:
            text = f.read()
        words = jieba.lcut(text)
        self.words = [w for w in words if len(w)>1] # 取出长度大于1的词
        self.setSplit("\n".join(self.words))

    def saveWords(self):
        path = asksaveasfilename()
        with open(path) as f:
            f.write(self.txtSplit.get(1.0, 'end'))

    def genWordCloud(self):
        dct = {}
        keys = ['width', 'height', 'font_path', 'scale',
            'min_font_size', 'max_font_size', 'font_step', 
            'max_words', 'min_word_length', 'background_color',
            'repeat', 'include_numbers', 'normalize_plurals']
        for key in keys:
            if key in self.spins:
                dct[key] = int(self.spins[key].get())
            if key in self.paths:
                dct[key] = self.paths[key].get()
            if key in self.checks:
                dct[key] = self.vars[key].get()=='1'  
        print(dct)
        try:
            cloud = WordCloud(**dct)
        except Exception as e:
            print(e)
        txt = self.txtSplit.get(1.0, "end")
        txt = " ".join(txt.split("\n"))
        cloud.generate(txt)
        p = self.paths["输出路径"].get()
        if not (p.endswith('.png') or p.endswith('.svg')):
            p = p+".png"
        cloud.to_file(p)
    
    def setSplit(self, txt):
        self.txtSplit.delete(1.0, "end")
        self.txtSplit.insert("end", txt)
        self.txtSplit.see("end")

if __name__ == "__main__":
    root = tk.Tk()
    DrawWords(root).pack(side=tk.TOP, fill=tk.BOTH)
    root.mainloop()