src/inout/io.py

   1 import os,sys
   2
   3
   4 def create_dir_if_necessary(path):
   5     if len(path) > 0 and not os.path.exists(path):
   6         os.makedirs(path)
   7
   8 def safe_chdir(path):
   9     #try:
  10     #    path = os_encode(path_unicode)
  11     #except UnicodeError:
  12     #    #logger.log('Error decoding path %s'%repr(path_unicode))
  13     #    print 'Error decoding path %s'%repr(path_unicode)
  14     #    path = path_unicode
  15     create_dir_if_necessary(path)
  16     os.chdir(path)
  17
  18 def create_file(file_path):
  19     dir = os.path.dirname(os_encode(file_path))
  20     create_dir_if_necessary(dir)
  21     encoded_file_path = os_encode(file_path)
  22     file_name = os.path.join(dir,os.path.basename(encoded_file_path))
  23
  24     f = file(file_name,'w')
  25     return f
  26
  27 def getfilesystemencoding():
  28     try:
  29         # sys.getfilesystemencoding() was introduced in Python v2.3
  30         return sys.getfilesystemencoding()
  31     except AttributeError:
  32         return 'ascii'
  33
  34 def os_encode(s):
  35     # Encoding can be None, meaning the default system encoding should
  36     # be used.
  37     enc = getfilesystemencoding()
  38     if enc:
  39         return s.encode(enc)
  40     else:
  41         return s
  42
  43 def os_decode(s):
  44     # Encoding can be None, meaning the default system encoding should
  45     # be used.
  46     enc = getfilesystemencoding()
  47     if type(s) == unicode or not enc:
  48         return s
  49     else:
  50         return unicode(s,enc)
  51
  52 def write(file_path,content):
  53     f = create_file(file_path)
  54     f.write(os_encode(content))
  55     f.close()
  56     from log.logging import logger
  57     logger.log(u'Wrote %s to %s'%(content,os.path.abspath(file_path)))
  58
  59 def list_dir(root,recursive=False,filter=None):
  60     encoded_root = os_encode(root)
  61     if not os.path.exists(encoded_root):
  62         return []
  63     all_files_and_dirs = []
  64     for name in os.listdir(encoded_root):
  65         file_name = os_decode(os.path.join(encoded_root,name))
  66         if recursive and os.path.isdir(os_encode(file_name)):
  67             all_files_and_dirs.extend(list_dir(file_name, True,filter))
  68         if (not filter) or filter(file_name):
  69             all_files_and_dirs.append(file_name)
  70     return all_files_and_dirs
  71
  72 def guess_encoding(data):
  73     #from logging import logger
  74     encodings = ['ascii','utf-8','utf-16']
  75     successful_encoding = None
  76     if data[:3]=='\xEF\xBB\xBF':
  77         data = data[3:]
  78
  79     for enc in encodings:
  80         if not enc:
  81             continue
  82         try:
  83             decoded = unicode(data, enc)
  84             successful_encoding = enc
  85             break
  86         except (UnicodeError, LookupError):
  87             pass
  88     if successful_encoding is None:
  89         raise UnicodeError('Unable to decode input data %s. Tried the'
  90             ' following encodings: %s.' %(repr(data), ', '.join([repr(enc)
  91                 for enc in encodings if enc])))
  92     else:
  93         #logger.log('Decoded %s to %s'%(repr(data),repr(decoded)),6)
  94         return (decoded, successful_encoding)
  95
  96
  97 def read_text_from_file(unicode_file_name):
  98     from log.logging import logger
  99     file_name = os_encode(unicode_file_name)
 100 #    logger.log(u'Reading from %s'%os.path.abspath(file_name).decode('utf-8'))
 101     f=file(file_name,'r')
 102     raw=f.read()
 103     f.close()
 104     (text,encoding)=guess_encoding(raw)
 105
 106     return text
 107 def parse_file_to_line_list(unicode_complete_path):
 108     text = read_text_from_file(unicode_complete_path)
 109     lines = text.splitlines()
 110     return lines
 111 def is_dir(unicode_path):
 112     return os.path.isdir(os_encode(unicode_path))