file_writer.py

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