python——适用于py3的pillow图像处理模块&plist组图分割

首先说明一下PIL库，PIL(Python Imaging Library)是Python一个强大方便的图像处理库，之前的博客也有用到过PIL模块的demo

不过可惜的是PIL库只支持到Python 2.7。python3以后，我们可以用pillow模块来代替PIL，pillow是PIL的一个派生分支，大概是PIL low的意思

现如今已经发展成为比PIL本身更具活力的图像处理库，支持PIP安装：pip install Pillow及easy_install安装：easy_install Pillow，本文用的是anaconda4.3.0（很全面的python版本，500多M，反正我听说过的库基本都包含了....）​​https://baike.baidu.com/item/anaconda/20407441#viewPageContent​​

anaconda4.3.0内置的pillow版本是4.0.0

首先简单说几个常用的方法

和PIL一样，主要是Image类

from PIL import Image

打开图片：

Image = Image.open("fliename")

保存图片：

Image.save(outfile, "JPEG")

获取图片属性

Image.format, Image.size, Image.mode……

裁剪、粘贴、合并

box = Image.copy() 
box = (100, 100, 400, 400)
region = Image.crop(box)
Image.paste(region, box)

几何变换

Image.resize((200,200))、Image.rotate(90)、Image.transpose(Image.ROTATE_90)以及参数一大堆的Image.transform()

更多高级姿势参考官方文档​​http://pillow.readthedocs.io/en/latest/index.html​​

然后再说一下plist，plist是一种文件格式，其实就是XML标签类型的配置文件。在Mac OS中较为常用，也经常用于前端开发将小图合并成组图的相关配置文件。

以IOS infoplist视图打开

以xml文本视图打开

python中plistlib模块提供一系列对.plist文件操作的函数，在这里我们也可以写一个简单的tree_to_dict方法以key-dict的形式提取数据

def tree_to_dict(tree):  
    d = {}  
    for index, item in enumerate(tree):  
        if item.tag == 'key':  
            if tree[index+1].tag == 'string':  
                d[item.text] = tree[index + 1].text  
            elif tree[index + 1].tag == 'true':  
                d[item.text] = True  
            elif tree[index + 1].tag == 'false':  
                d[item.text] = False  
            elif tree[index+1].tag == 'dict':  
                d[item.text] = tree_to_dict(tree[index+1])  
    return d

以麻将资源图为例：

我们通过对应的plist文件来提取单个麻将，其pilst文件配置如下：

<plist version="1.0">
    <dict>
        <key>frames</key>
        <dict>
            <key>M_autumn.png</key>
            <dict>
                <key>frame</key>
                <string>{{79,886},{75,113}}</string>
                <key>offset</key>
                <string>{0,0}</string>
                <key>rotated</key>
                <false/>
                <key>sourceColorRect</key>
                <string>{{0,0},{75,113}}</string>
                <key>sourceSize</key>
                <string>{75,113}</string>
            </dict>
            <key>M_bamboo.png</key>
            <dict>
                <key>frame</key>
                <string>{{79,771},{75,113}}</string>
                <key>offset</key>
                <string>{0,0}</string>
                <key>rotated</key>
                <false/>
                <key>sourceColorRect</key>
                <string>{{0,0},{75,113}}</string>
                <key>sourceSize</key>
                <string>{75,113}</string>
            </dict>
            <key>M_bamboo_1.png</key>
            <dict>
                <key>frame</key>
                <string>{{424,656},{75,113}}</string>
                <key>offset</key>
                <string>{0,0}</string>
                <key>rotated</key>
                <false/>
                <key>sourceColorRect</key>
                <string>{{0,0},{75,113}}</string>
                <key>sourceSize</key>
                <string>{75,113}</string>
            </dict>
            <key>M_bamboo_2.png</key>
            <dict>
                <key>frame</key>
                <string>{{309,618},{75,113}}</string>
                <key>offset</key>
                <string>{0,0}</string>
                <key>rotated</key>
                <true/>
                <key>sourceColorRect</key>
                <string>{{0,0},{75,113}}</string>
                <key>sourceSize</key>
                <string>{75,113}</string>
            </dict>             ……省略

 读取freames再循环读取每个frame参数，再利用PIL里的几个函数：

crop裁剪

rotate/transpose方法进行角度旋转（plist配置参数rotated）

paste粘贴

save保存

完整python代码：

#!python  
import os,sys,plistlib
from xml.etree import ElementTree  
from PIL import Image  
  
def tree_to_dict(tree):  
    d = {}  
    for index, item in enumerate(tree):  
        if item.tag == 'key':  
            if tree[index+1].tag == 'string':  
                d[item.text] = tree[index + 1].text  
            elif tree[index + 1].tag == 'true':  
                d[item.text] = True  
            elif tree[index + 1].tag == 'false':  
                d[item.text] = False  
            elif tree[index+1].tag == 'dict':  
                d[item.text] = tree_to_dict(tree[index+1])  
    return d  
  
def gen_png_from_plist(plist_filename, png_filename):  
    file_path = plist_filename.replace('.plist', '')  
    big_image = Image.open(png_filename)  
    root = ElementTree.fromstring(open(plist_filename, 'r',encoding='UTF-8').read())  
    plist_dict = tree_to_dict(root[0])  
    to_list = lambda x: x.replace('{','').replace('}','').split(',')  
    for k,v in plist_dict['frames'].items():  
        print('v:',v)
        rectlist = to_list(v['frame'])  
        width = int( rectlist[3] if v['rotated'] else rectlist[2] )  
        height = int( rectlist[2] if v['rotated'] else rectlist[3] )  
        box=(   
            int(rectlist[0]),  
            int(rectlist[1]),  
            int(rectlist[0]) + width,  
            int(rectlist[1]) + height,  
            )  
        sizelist = [ int(x) for x in to_list(v['sourceSize'])]  
        rect_on_big = big_image.crop(box)  
  
        if v['rotated']:  
            #rect_on_big = rect_on_big.rotate(90)  
            rect_on_big=rect_on_big.transpose(Image.ROTATE_90)
  
        result_image = Image.new('RGBA', sizelist, (0,0,0,0))  
        if v['rotated']:  
            result_box=(  
                ( sizelist[0] - height )//2,  
                ( sizelist[1] - width )//2,  
                ( sizelist[0] + height )//2,  
                ( sizelist[1] + width )//2  
                )  
        else:  
            result_box=(  
                ( sizelist[0] - width )//2,  
                ( sizelist[1] - height )//2,  
                ( sizelist[0] + width )//2,  
                ( sizelist[1] + height )//2  
                )  
        print('rect_on_big:',rect_on_big,'result_box:',result_box)
        result_image.paste(rect_on_big, result_box, mask=0)  
  
        if not os.path.isdir(file_path):  
            os.mkdir(file_path)  
        outfile = (file_path+'/' + k).replace('gift_', '')  
        print(outfile,"generated")
        result_image.save(outfile)  
  
if __name__ == '__main__':     
    #filename = sys.argv[1] 
    filename =  'mahjongs'

    plist_filename = filename + '.plist'  
    png_filename = filename + '.png'  
    if (os.path.exists(plist_filename) and os.path.exists(png_filename)):  
        gen_png_from_plist( plist_filename, png_filename )  
    else: 
        print("make sure you have boith plist and png files in the same directory")

1.读取文件时提示"UnicodeDecodeError: 'gbk' codec can't decode byte 0x80 in position 205: illegal multibyte sequence"

编码格式问题，在open参数后面加上encoding='UTF-8'

root = ElementTree.fromstring(open(plist_filename, 'r',encoding='UTF-8').read())

2.paste 执行报错，integer argument expected, got float

因为在Python 3中，要得到一个整数的结果，你需要使用//而不是/

所以计算box时将/改为//

if v['rotated']:  
            result_box=(  
                ( sizelist[0] - height )//2,  
                ( sizelist[1] - width )//2,  
                ( sizelist[0] + height )//2,  
                ( sizelist[1] + width )//2  
                )  
        else:  
            result_box=(  
                ( sizelist[0] - width )//2,  
                ( sizelist[1] - height )//2,  
                ( sizelist[0] + width )//2,  
                ( sizelist[1] + height )//2  
                )

3.paste执行报错，images do not match

由于之前旋转错误导致尺寸异常，将rect_on_big = rect_on_big.rotate(90)换成rect_on_big=rect_on_big.transpose(Image.ROTATE_90)就可以了

可能是这个版本的rotate方法有问题。

if v['rotated']:  
            #rect_on_big = rect_on_big.rotate(90)  
            rect_on_big=rect_on_big.transpose(Image.ROTATE_90)

运行截图：

输出文件：

环境：python3.6&Anaconda4.3.0&VS2017