[Sommaire] [Imprimer]

   1  #!/usr/bin/python
   2  
   3  """Zope 2 spellchecker web service for Kupu"""
   4  
   5  COMMAND = 'aspell -a'
   6  
   7  import popen2, re
   8  
   9  try:
  10      from Globals import ClassSecurityInfo
  11  except ImportError:
  12      pass
  13  else:
  14      # hmmm... Zope 2...
  15      __allow_access_to_unprotected_subobjects__ = 1
  16  
  17  class SpellChecker:
  18      """Simple spell checker, uses ispell (or aspell) with pipes"""
  19  
  20      __allow_access_to_unprotected_subobjects__ = 1
  21  
  22      reg_unknown = re.compile('^& (.*?) \d* \d*: (.*)$', re.U)
  23      reg_unknown_no_replacement = re.compile('^\# (.*?) \d*.*$', re.U)
  24  
  25      def __init__(self):
  26          self.chout, self.chin = popen2.popen2(COMMAND)
  27          # throw away intro
  28          self.read_line()
  29  
  30      def __del__(self):
  31          self.chout.close()
  32          self.chin.close()
  33  
  34      def check(self, text):
  35          """checks a line of text
  36          
  37              returns None if spelling was okay, and an HTML string with words 
  38              that weren't recognized marked (with a span class="wrong_spelling")
  39          """
  40          result = {}
  41          for line in text.split('\n'):
  42              line = line.strip()
  43              if line:
  44                  self.write_line(line)
  45              while 1:
  46                  resline = self.read_line()
  47                  if not resline.strip():
  48                      break
  49                  if resline.strip() != '*':
  50                      match = self.reg_unknown.match(resline)
  51                      have_replacement = True
  52                      if not match:
  53                          match = self.reg_unknown_no_replacement.match(resline)
  54                          have_replacement = False
  55                      assert match, 'Unknown formatted line: %s' % resline
  56                      word = match.group(1)
  57                      if result.has_key(word):
  58                          continue
  59                      replacements = []
  60                      if have_replacement:
  61                          replacements = match.group(2).split(', ')
  62                      result[word] = replacements
  63          return result
  64  
  65      def read_line(self):
  66          buf = []
  67          while 1:
  68          char = self.read_char()
  69              if not char:
  70                  return ''        
  71              if char == '\n':
  72                  return ''.join(buf)
  73              buf.append(char)
  74  
  75      def write_line(self, line):
  76          try:
  77              self.chin.write('%s\n' % line)
  78              self.chin.flush()
  79              return
  80          except IOError:
  81              self.reconnect()
  82              self.chin.write('%s\n' % line)
  83              self.chin.flush()
  84              return
  85          raise
  86  
  87      def read_char(self):
  88          try:
  89              return self.chout.read(1)
  90          except IOError:
  91              self.reconnect()
  92              return self.chout.read(1)
  93          raise
  94  
  95      def reconnect(self):
  96          try:
  97              self.chout.close()
  98          except IOError:
  99              pass
 100          try:
 101              self.chin.close()
 102          except IOError:
 103              pass
 104          self.chout, self.chin = popen2.popen2(COMMAND)
 105  
 106  def format_result(result):
 107      """convert the result dict to XML"""
 108      buf = ['<?xml version="1.0" encoding="UTF-8" ?>\n<spellcheck_result>']
 109      for key, value in result.items():
 110          buf.append('<incorrect><word>')
 111          buf.append(key)
 112          buf.append('</word><replacements>')
 113          buf.append(' '.join(value))
 114          buf.append('</replacements></incorrect>')
 115      buf.append('</spellcheck_result>')
 116      return ''.join(buf)
 117  
 118  if __name__ == '__main__':
 119      c = SpellChecker()
 120      while 1:
 121          line = raw_input('Enter text to check: ')
 122          if line == 'q':
 123              break
 124          ret = c.check(line)
 125          if ret is None:
 126              print 'okay'
 127          else:
 128              print ret