mongodb python image 图像存储读取

最近做一些数据库调研的工作，目标是实现影像更快的入库、出库、查询，并实现并行访问等操作。

将结果总结成一个mongoImg类，也算是小结吧。

'''
Created on 2013-8-6
class mongoInsert
@author: tree
'''
__metaclass__ = type

import os
from pymongo.database import Database
import time
import gridfs

class mongoImg(object):
    """mongoInsert is a class for inserting document
    
    
    """
    def __init__(self, database, dir):
        """Create a new instance of :class:mongoInsert
        :Parameters:
          - `database`: database to use
          - `dir` : directory of document 
          """
        if not isinstance(database, Database):
            raise TypeError("database must be an instance of Database")
        if len(dir) < 1:
            raise TypeError("dir must be an string of directory")
        
#         self.__con = Connection()
        self.__imgdb = database
        self.__imgfs = gridfs.GridFS (self.__imgdb)
        self.__dir = dir
        self.__filelist=[]

    #save filepath in list.txt
    def __dirwalk(self,topdown=True):
        """traverse the documents of self.__dir and save in self.__filelist
        """
        sum=0
        self.__filelist.clear()
        
        for root,dirs,files in os.walk(self.__dir,topdown):
            for name in files:
                sum+=1
                temp=os.path.join(root,name)
                self.__filelist.append(temp)
        print(sum)

    #insert image 
    def insert(self):
        """insert images in mongodb
        """
        self.__dirwalk()

        tStart = time.time()        
        for fi in self.__filelist:       
            with open (fi,'rb') as myimage:
                data=myimage.read()                  
                self.__imgfs.put(data, content_type = "jpg", filename =fi)
    
        tEnd =time.time ()
        print ("It cost %f sec" % (tEnd - tStart))
        
    #get image by filename
    def getbyname(self,filename,savepath):
        """get img from mongdb by filename
        """
        if len(savepath) < 1:
            raise TypeError("dir must be an string of directory")
        dataout=self.__imgfs.get_version(filename)
        try:
            imgout=open(savepath,'wb')
            data=dataout.read()
            imgout.write(data)
        finally:
            imgout.close()
        

 使用示例：也可以将数据库连接写在类内部

from pymongo import Connection
import mongoImg

filedir=r'D:image'
con = Connection()
db = con.imgdb
imgmongo=mongoImg.mongoImg(db,filedir)
imgmongo.insert()

感觉mongodb存储影像切片还是蛮快的，1w多个图片，大约100-200秒左右。

tip：

gridfs.GridFS.put 函数

put(data, **kwargs)
Put data in GridFS as a new file.

Equivalent to doing:

try:
    f = new_file(**kwargs)
    f.write(data)
    finally
f.close()

在存储读取图像时，犯了低级错误，将open得到的file实例当做数据存储，读取的时候怎么也读不出数据。。。囧

另外以字节流形式读取图像数据比较适合。

pipe = open('/dev/input/js0','rb')

如果以str形式存储的话，可能会出现UnicodeDecodeError错误，貌似是因为图像数据有些超出了python默认编码的存储区间。

ps：初学python 数据库操作也忘得差不多 欢迎大家批评和指正~