if 0 {[Richard Suchenwirth] 2003-06-03 - The following proc was written to extract the text content from a (Microsoft) Word document, via [dde] and the clipboard. It's still not pretty or very stable, but after hours of wrestling with fuzzy documentation, I now wikify it as a first shot.

The procedure is called with the filename of a Word document, and returns the reaped text, after some while (I could not think of other synchronization rather than to wait much...) A known issue is that some characters (apostrophe, "...") get specially encoded, and the output contains a question mark in these positions. Tk is needed so we can use the clipboard.

Hints on how to do this better are greatly appreciated!
}
 proc doc2txt fn {
    package require dde
    package require Tk; # because we need [selection]
    eval exec [auto_execok start] [list $fn] &

    #Loop to wait until Word is really there, and ready to talk
    set word ""
    while {$word==""} {
	set word [dde services Winword System]
	after 200
    }
    after 1000 ;# wait for the window to load...
    dde execute Winword System {[EditSelectAll]}
    dde execute Winword System {[EditCopy]}
    set res [selection get -selection CLIPBOARD]
    dde execute Winword System {[FileExit 2]}
    set res
 }
----
[LES]: I can't do it with DDE. DDE is so crude. But [COM] and [optcl] can produce a very good result. First, create and save this Word macro in your Normal.dot:

 Sub wordreaper()
 
 Dim myRange As Range
 
 With Word.Application
     If .Windows.Count > 0 Then
         Set myRange = ActiveDocument.Content
     End If
 End With
 
 Open "c:\windows\desktop\extract.txt" For Append As #1
 Print #1, myRange
 Close #1
 
 End Sub

Then, run this Tcl code:

 set  _docpath  {c:\windows\desktop\some.doc}
 
 package require optcl
 set  ::hWORD  [ optcl::new  word.application ]
 set  ::hDOC  [ $::hWORD  -with documents  open  $_docpath ]
 $::hWORD  run  wordreaper
 $::hDOC Close
 $::hWORD Quit
 
The macro in Normal.dot is often necessary because "translating" the entire macro into some way that [optcl] can send entirely on its own is very difficult, if possible at all. I also have no idea of how one could pass the path to "extract.txt" to the macro instead of hard-coding it.
----
[Arts and crafts of Tcl-Tk programming]