OldCodes/ts/dicoparser.py

   1
   2 import shlex
   3
   4 DICO_EXT = '.dico'
   5
   6 class DicoParser:
   7         def __init__( self, theFileName, theStartAttr, theIndexAttr ):
   8                 if theFileName=='':
   9                         return
  10                 if not theFileName.endswith( DICO_EXT ):
  11                         theFileName += DICO_EXT
  12                 aFile = open( theFileName )
  13                 self.data = self.parse( aFile.read(), theStartAttr, theIndexAttr )
  14                 aFile.close()
  15
  16         def parse( self, theText, theStartAttr, theIndexAttr ):
  17                 aTokenList = self.parse_tokens( theText )
  18                 self.expand_values( aTokenList )
  19                 aList = self.convert_to_tuples( aTokenList )
  20                 return self.convert_to_blocks( aList, theStartAttr, theIndexAttr )
  21
  22         def parse_tokens( self, theText ):
  23                 theText = theText.replace( "''", "`" )
  24                 aLexer = shlex.shlex( theText )
  25                 aLexer.commenters = '/'
  26                 aLexer.quotes = '\'"'
  27                 aLexer.wordchars = aLexer.wordchars + '-.'
  28                 aTokenList = []
  29                 for aToken in aLexer:
  30                         #print aToken
  31                         if aToken[0]=="'" and aToken[-1]=="'":
  32                                 aToken = aToken[1:-1]
  33                         elif aToken[0]=='"' and aToken[-1]=='"':
  34                                 aToken = aToken[1:-1]
  35                         if aToken=='`':
  36                                 aToken = ''
  37                         aToken = aToken.replace( "`", "'" )
  38                         aToken = aToken.replace( "\n", "" )
  39                         aToken = self.simplify_spaces( aToken )
  40                         #print aToken
  41                         aTokenList.append( aToken )
  42                 return aTokenList
  43
  44         def simplify_spaces( self, theToken ):
  45                 return ' '.join( theToken.split() )
  46
  47         def expand_values( self, theList ):
  48                 for i in xrange( 0, len( theList ) ):
  49                         if '=' in theList[i] and theList[i]!='=':
  50                                 aTokenList = self.parse_tokens( theList[i] )
  51                                 aSubItemsList = self.convert_to_tuples( aTokenList )
  52                                 if( len( aSubItemsList ) == 0 ):
  53                                         print theList[i]
  54                                 theList[i] = aSubItemsList[0] # we assume that only one '=' in the subitem
  55
  56         def convert_to_tuples( self, theList ):
  57                 aPairsList = []
  58                 aKey = ''
  59                 i = 0
  60                 n = len( theList )
  61                 while i<n:
  62                         if theList[i]=='=' and i<n-1:
  63                                 aPairsList.append( (aKey, theList[i+1]) )
  64                                 i = i + 1
  65                         elif theList[i]==';' and len( aPairsList ) > 0 and i<n-1:
  66                                 aKey, aValue = aPairsList[-1]
  67                                 if isinstance( aValue, list ):
  68                                         aValue.append( theList[i+1] )
  69                                 else:
  70                                         aValue = [aValue, theList[i+1]]
  71                                 aPairsList[-1] = ( aKey, aValue )
  72                                 i = i + 1
  73                         else:
  74                                 aKey = theList[i]
  75                         i = i + 1
  76                 return aPairsList
  77
  78         def convert_to_blocks( self, theTuples, theStartAttr, theIndexAttr ):
  79                 aData = {}
  80                 aBlockKey = ''
  81                 aBlock = {}
  82                 for aKey, aValue in theTuples:
  83                         #print aKey
  84                         if aKey==theStartAttr:
  85                                 #print "__start__"
  86                                 if len(aBlock)>0:
  87                                         aData[aBlockKey] = aBlock
  88                                 aBlockKey = ''
  89                                 aBlock = {}
  90                         elif aKey==theIndexAttr:
  91                                 #print "__index__"
  92                                 aBlockKey = aValue
  93                                 #print aBlockKey
  94                         aBlock[aKey] = aValue
  95                         #print aData
  96                 if len(aBlock)>0:
  97                         aData[aBlockKey] = aBlock
  98                 return aData
  99
 100         def are_equal( self, theStrDico, theStrCata ):
 101                 return theStrDico == self.to_dico_str( theStrCata )
 102
 103         def to_dico_str( self, theStrCata ):
 104                 aCata = theStrCata.replace( '_', ' ' )
 105                 aCata = aCata.upper()
 106                 return aCata
 107
 108         def to_cata_str( self, theStrDico ):
 109                 aWordsList = theStrDico.split()
 110                 aCata = []
 111                 for aWord in aWordsList:
 112                         aWord = aWord.lower()
 113                         aWord = aWord[0].upper() + aWord[1:]
 114                         aCata.append( aWord )
 115                 return ' '.join( aCata )
 116
 117         def search_in_block( self, theBlock, theAttrTr ):
 118                 #print 'search_in_block:', theAttrTr
 119                 return theBlock[theAttrTr]
 120
 121         def search( self, theIndexText, theAttrTr, theChoiceText, theAttrCh, theAttrChTr ):
 122                 anIndexText = self.to_dico_str( theIndexText )
 123                 if not anIndexText in self.data:
 124                         return ''
 125
 126                 if theAttrCh=='':
 127                         return self.search_in_block( self.data[anIndexText], theAttrTr )
 128
 129                 aDataCh = self.search_in_block( self.data[anIndexText], theAttrCh )
 130                 aDataChTr = self.search_in_block( self.data[anIndexText], theAttrChTr )
 131                 if isinstance( theChoiceText, basestring ):
 132                         aChoiceText = self.to_dico_str( theChoiceText )
 133                 else:
 134                         aChoiceText = theChoiceText
 135                 #print 'Choice text:', aChoiceText
 136                 #print 'Choice data:', aDataCh
 137                 #print 'Choice tr data:', aDataChTr
 138                 if isinstance( aDataCh, basestring ) and aDataCh==aChoiceText:
 139                         return aDataChTr
 140
 141                 for i in xrange( 0, len(aDataCh) ):
 142                         if isinstance( aDataCh[i], tuple ):
 143                                 aKey, aValue = aDataCh[i]
 144                         elif isinstance( aDataCh[i], basestring ):
 145                                 aKey = ''
 146                                 aValue = aDataCh[i]
 147                         #print aKey, aValue
 148                         if aValue==aChoiceText:
 149                                 if isinstance( aDataChTr[i], tuple ):
 150                                         aKeyTr, aValueTr = aDataChTr[i]
 151                                 elif isinstance( aDataChTr[i], basestring ):
 152                                         aKeyTr = ''
 153                                         aValueTr = aDataChTr[i]
 154                                 return aValueTr
 155                 return ''
 156
 157         def translate( self, theIndexText, theAttrTr, theChoiceText='', theAttrCh='', theAttrChTr='' ):
 158                 aTrText = self.search( theIndexText, theAttrTr, theChoiceText, theAttrCh, theAttrChTr )
 159                 #print aTrText
 160                 return self.to_cata_str( aTrText )