AutoHotkey Community

/*
Function : GetWordsFrequencies()
Argument : File > GetWordsFrequencies(File)  or   Clipboard > GetWordsFrequencies()
Author: TomXIII
*/

If 0 = 1 ;If a file is directly passed
{
   File = %1%
   GetWordsFrequencies(File)
}

GetWordsFrequencies(InFile="") ;Main function
{
   Text := GetText(InFile)
   Text := ListWords(Text)
   WordsList := SetWordsList(Text)
   UseWordsList(WordsList)
}

;=============================================

GetText(InFile)
{
   If FileExist(InFile)
   {
      SplitPath, InFile,,, FileExt
      If (FileExt<>"txt") ;Poor convert method, doesn't work often!
      {
         TempFile = %A_ScriptDir%\temp.txt
         FileRead, TempText, %InFile%
         FileDelete, %TempFile%
         FileAppend, %TempText%, %TempFile%
         FileRead, TempText, %TempFile%
      }
      Else
         FileRead, TempText, %InFile%
   }
   Else
      TempText := Clipboard
   Return, %TempText%
}

ListWords(TempText) ;Extract all words and separate them with "-"
{
   TempText := RegexReplace(TempText, "(`r|`n|`r`n)", " ")
   TempText := RegexReplace(TempText, "i)[^[:alpha:]éÉèÈàÀùÙâÂêÊîÎôÔûÛïÏëËüÜçÇ'æÆœŒ ]?", "") ; Special chars for French words
   TempText := RegexReplace(TempText, "[ ]+", "-")
   Return, %TempText%
}

SetWordsList(TempText)
{
   StringSplit, TempWord, TempText, -
   TempText = -%TempText%-
   Loop, %TempWord0%
   {
      CurrentTempWord := TempWord%A_Index%
      RegEx = i)-%CurrentTempWord%-
      TempText := RegexReplace(TempText, RegEx, "-", RplCount)
      If (RplCount and CurrentTempWord)
      {
         StrLen := StrLen(RplCount)
         Loop,% 3-StrLen
            RplCount := "0" . RplCount
         CurrentLine := RplCount . "`t   |`t" . CurrentTempWord
         TempList := TempList ? TempList . "`n" . CurrentLine : "Frequency`tWord`n____________________`n" . CurrentLine
      }
   }
   Sort, TempList, D`n R
   Return, %TempList%
}

UseWordsList(TempWordsList)
{
   OutFile = %A_ScriptDir%\Words_Frequencies.txt
   FileDelete, %OutFile%
   FileAppend, %TempWordsList%, %OutFile%
   Run, %OutFile%,,UseErrorLevel
   If ErrorLevel
      MsgBox, Error occured!
}

#Include %A_ScriptDir%\WordFrequencyCounter.ahk
File = %A_ScriptFullPath%
GetWordsFrequencies()

File = %A_ScriptFullPath%
FileRead, TempText, %File%
;~ TempText := RegexReplace( TempText, "(`r|`n|`r`n)", "-")
TempText := RegexReplace( TempText, "i)[^[:alpha:]éÉèÈàÀùÙâÂêÊîÎôÔûÛïÏëËüÜçÇ'æÆœŒ]+", ">`n<") ; éÉèÈàÀùÙâÂêÊîÎôÔûÛïÏëËüÜçÇ'æÆœŒ  Special chars for French words
TempText := RegexReplace( TempText, "[ ]+", "`n")
TempText := RegExReplace( TempText, "(^>`n*)|(`n<*$)", "")
LastText := TempText
Sort, TempText, D`n U
StringSplit, Word, TempText, `n
Loop, %Word0%
{
   CurrentWord := Word%A_Index%
   RegEx = i)%CurrentWord%
   RplCount := 0
   LastText := RegExReplace( LastText, RegEx, "", RplCount)
   StrLen := StrLen(RplCount)
   Loop,% 3-StrLen
      RplCount := "0" . RplCount
   CurrentWord := RegExReplace( CurrentWord, "(^<*)|(>*$)", "")
   CurrentLine := RplCount . "`t   |`t" . CurrentWord
   TempList := TempList ? TempList . "`n" . CurrentLine : CurrentLine
}
Sort, TempList, D`n R
TempList := "Frequency`tWord`n____________________`n" . TempList
OutFile = %A_ScriptDir%\Words_Frequencies.txt
FileDelete, %OutFile%
FileAppend, %TempList%, %OutFile%
Run, %OutFile%,,UseErrorLevel
If ErrorLevel
   MsgBox, Error occured!!!!!!!!

File = %A_ScriptFullPath%

GetWordsFrequencies(File)

GetWordsFrequencies(File="", MinLenght=0, MinFrequency=0)
{
   If File
      FileRead, TempText, %File%
   Else
      TempText := Clipboard
   TempText := RegexReplace( TempText, "i)[^[:alpha:]éÉèÈàÀùÙâÂêÊîÎôÔûÛïÏëËüÜçÇ'æÆœŒ]+", ">`n<")
   TempText := RegexReplace( TempText, "[ ]+", "`n")
   TempText := RegExReplace( TempText, "(^>`n*)|(`n<*$)", "")
   TempList := CreateWordsList(TempText, MinLenght, MinFrequency)
   UseWordsList(TempList)
}

CreateWordsList(TempText, MinLenght, MinFrequency)
{
   LastText := TempText
   Sort, TempText, D`n U
   StringSplit, Word, TempText, `n
   Loop, %Word0%
   {
      CurrentWord := Word%A_Index%
      RegEx = i)%CurrentWord%
      RplCount := 0
      LastText := RegExReplace( LastText, RegEx, "", RplCount)
      StrLen := StrLen(RplCount)
      Loop,% 3-StrLen
         RplCount := "0" . RplCount
      CurrentWord := RegExReplace( CurrentWord, "(^<*)|(>*$)", "")
      CurrentLine := RplCount . "`t   |`t" . CurrentWord
      Lenght := StrLen(CurrentWord)
      If (Lenght>=MinLenght and RplCount>=MinFrequency)
         TempList := TempList ? TempList . "`n" . CurrentLine : CurrentLine
   }
   Sort, TempList, D`n R
   TempList := "Frequency`tWord`n____________________`n" . TempList
   Return, %TempList%
}

UseWordsList(TempList)
{
   OutFile = %A_ScriptDir%\Words_Frequencies.txt
   FileDelete, %OutFile%
   FileAppend, %TempList%, %OutFile%
   Run, %OutFile%,,UseErrorLevel
   If ErrorLevel
      MsgBox, Error occured!!!!!!!!
}