PureBoard

Code: Alles auswählen

;/ Silbentrennung
;/ Algorithmus von Frankling Mark Liang (1983)
;/ basierend "hyphenator für HTML" (Mathias Nater, Zürich)
;/ Mai 2009 Thorsten Hoeppner (Thorsten1867)


Structure HyphenStructure
  chars.s
  pattern.s
EndStructure

Global NewList HyphPattern.HyphenStructure()

Procedure LoadPattern(PatternFile.s)
  ClearList(HyphPattern())
  If ReadFile(0, PatternFile.s)
    While Eof(0) = 0 
      pattern$ = Trim(ReadString(0,#PB_UTF8))
      AddElement(HyphPattern())
      HyphPattern()\chars = StringField(pattern$,1,":")
      HyphPattern()\pattern = StringField(pattern$,2,":")
    Wend 
    CloseFile(0)
  Else
    MessageRequester("Lade Trennungsmuster", "Datei konnte nicht eingelesen werden.", 0)
  EndIf
EndProcedure

Procedure.s InsertString(string.s, idx.b, chars.s)
  If idx > 1
    result.s = Left(string, idx-1)+chars+Mid(string, idx)
  ElseIf idx >= Len(string)
    result.s = string + chars
  Else
    result.s = chars + string
  EndIf
  ProcedureReturn result
EndProcedure

Procedure.s HyphenateWord(wort.s)
  Define.b WLen, WIdx
  Define.s Char, LWort
  
  wort = "_"+wort+"_"
  LWort = LCase(wort)
  WLen = Len(wort)
  Dim Hypos.s(WLen)
  
  ForEach HyphPattern() ; Pattern auswerten
    WortIdx = FindString(LWort, HyphPattern()\chars, 1)
    If WortIdx
      digits = 1
      For i=1 To Len(HyphPattern()\pattern)
        Char.s = Mid(HyphPattern()\pattern, i, 1)
        If Asc(Char) >= 48 And Asc(Char) <= 57 ; Ist Zahl (0-9)
          If i = 1
            z = WortIdx
          Else
            z = WortIdx + i - digits
          EndIf
          If Not Hypos(z) Or Hypos(z) < Char ; kleinere Zahl überschreiben
            Hypos(z) = Char
          EndIf
          digits + 1
        EndIf
      Next
      Debug "Pattern: "+HyphPattern()\pattern+" ("+Str(WortIdx)+")"
    EndIf
  Next

  inserted.b = 0 ; Trennmuster erzeugen
  For i = 3 To WLen-2
    If Val(Hypos(i))
      wort = InsertString(wort, i + inserted, Hypos(i))
      inserted + 1
    EndIf
  Next  ;}
  
  wort.s = RemoveString(wort, "_") ; Wortanfang/-ende entfernen
  Debug "Trennmuster: "+wort
  For i = 1 To 9 ; Trennstellen ermitteln (ungerade Zahlen)
    If (i % 2) ; Trennstelle
      wort = ReplaceString(wort, Str(i), "|")
    Else ; keine Trennstelle
      wort = RemoveString(wort, Str(i))
    EndIf
  Next
  ProcedureReturn wort
EndProcedure

LoadPattern("HyphenPattern.de")

wort$ =  InputRequester("Silbentrennung", "Wort eingeben:", "")
result$ = HyphenateWord(wort$)
MessageRequester("Silbentrennung", ReplaceString(result$, "|", " - "), #MB_OK)