file_writer.py

   1 import constants
   2 import os
   3
   4 def remove_tricky_unicode(x):
   5     try:
   6         x = x.decode('utf-8')
   7         x = x.replace("\u2018", "'").replace("\u2019", "'")
   8         x = x.replace("\u201c", '"').replace("\u201d", '"')
   9         x = x.replace("\u2e3a", "-").replace("\u2014", "-")
  10     except:
  11         pass
  12     return x
  13
  14 class file_writer:
  15     def __init__(self, filename):
  16         self.full_filename = os.path.join(constants.pages_dir,
  17                                           filename)
  18         self.f = open(self.full_filename, 'wb')
  19         self.xforms = [ remove_tricky_unicode ]
  20
  21     def add_xform(self, xform):
  22         self.xforms.append(xform)
  23
  24     def write(self, data):
  25         for xform in self.xforms:
  26             data = xform(data)
  27         self.f.write(data.encode('utf-8'))
  28
  29     def done(self):
  30         self.f.close()
  31
  32     def close(self):
  33         self.done()
  34
  35 # Test
  36 #def toupper(x):
  37 #    return x.upper()
  38 #
  39 #fw = file_writer("test")
  40 #fw.add_xform(toupper)
  41 #fw.write(u"This is a \u201ctest\u201d. \n")
  42 #fw.done()