OSDN Git Service

5cbeb19ee95d05bbf31f8c30cf223cd175e18dce
[stux/ultron.git] / src / main / Python / DataReader.py
1 import os
2 import Data
3 import yaml
4
5 class DataReader:
6     DATA_SOURCE =""
7
8     def __init__(self):
9         with open('config\\toolconf.yml','r') as config:
10             confdata = yaml.load(config)
11             self.DATA_SOURCE = confdata['dataPath']
12
13     def createLearningData(self):
14         data = []
15
16         for file in self.walkDataDirectory(self.DATA_SOURCE):
17             data.append(self.read(file))
18
19         return data
20
21     def walkDataDirectory(self,directory):
22         for root,dirs,files in os.walk(directory):
23             yield root
24             for file in files:
25                 yield os.path.join(root,file)
26
27     def read(self,file):
28         result = Data.Data()
29         if(os.path.isfile(file)):
30             datafile = open(file)
31             orgdata = datafile.readlines()
32             data = []
33
34             datafile.close()
35             result.setLabel(int(orgdata[0].strip()))
36
37             for i in range(1,len(orgdata)):
38                 data.append(orgdata[i].split())
39             result.setData(data)
40
41         return result