Source Code for Module jazzparser.utils.csv

 1  from __future__ import absolute_import 
 2  """Additional CSV reading/writing utilities. 
 3   
 4  Comma-separated value file reader and writer that wraps the standard  
 5  package to handle unicode. 
 6   
 7  This stuff comes straight from the example in the CSV package's  
 8  documentation. 
 9   
10  Original author: Skip Montanaro <skip@pobox.com> 
11   
12  @see: U{http://docs.python.org/library/csv.html} 
13   
14  """ 
15   
16 -class UTF8Recoder(object): 
17      """ 
18      Iterator that reads an encoded stream and reencodes the input to  
19      UTF-8. 
20       
21      """ 
22 -    def __init__(self, f, encoding): 
23          import codecs 
24          self.reader = codecs.getreader(encoding)(f) 
25   
26 -    def __iter__(self): 
27          return self 
28   
29 -    def next(self): 
30          return self.reader.next().encode("utf-8") 
31   
32 -class UnicodeCsvReader(object): 
33      """ 
34      A CSV reader which will iterate over lines in the CSV file "f", 
35      which is encoded in the given encoding. 
36       
37      """ 
38 -    def __init__(self, f, dialect='excel', encoding="utf-8", **kwds): 
39          import csv 
40          f = UTF8Recoder(f, encoding) 
41          self.reader = csv.reader(f, dialect=dialect, **kwds) 
42   
43 -    def next(self): 
44          row = self.reader.next() 
45          return [unicode(s, "utf-8") for s in row] 
46   
47 -    def __iter__(self): 
48          return self 
49   
50 -class UnicodeCsvWriter(object): 
51      """ 
52      A CSV writer which will write rows to CSV file "f", 
53      which is encoded in the given encoding. 
54       
55      """ 
56 -    def __init__(self, f, dialect='excel', encoding="utf-8", **kwds): 
57          import csv, cStringIO, codecs 
58          # Redirect output to a queue 
59          self.queue = cStringIO.StringIO() 
60          self.writer = csv.writer(self.queue, dialect=dialect, **kwds) 
61          self.stream = f 
62          self.encoder = codecs.getincrementalencoder(encoding)() 
63   
64 -    def writerow(self, row): 
65          self.writer.writerow([unicode(s).encode("utf-8") for s in row]) 
66          # Fetch UTF-8 output from the queue ... 
67          data = self.queue.getvalue() 
68          data = data.decode("utf-8") 
69          # ... and reencode it into the target encoding 
70          data = self.encoder.encode(data) 
71          # write to the target stream 
72          self.stream.write(data) 
73          # empty queue 
74          self.queue.truncate(0) 
75   
76 -    def writerows(self, rows): 
77          for row in rows: 
78              self.writerow(row) 
79