How to move typewriter floppy disk to a PC

July 13, 2007 5:58 AM

An 80s technology throwback question for you. A paralegal in our office has an IBM Wheelwriter 50 Series II (6788) typewriter that she uses for all her wordprocessing. It's from 1988. It has a little computer monitor attached to it that accepts 3.5" floppies. I'm trying to help her move to a PC, but she doesn't want to lose all the documents she has stored on these floppies. I've tried reading the floppies with MS Word and Wordperfect on a PC, but it comes up as nonsense. I've also tried using a utility called CanOpener and OmniFlop with no success. Any other ideas on a utility/process to move the documents over to a PC-readable format?
posted by banger100 to technology (22 comments total)

If you open the docs as plain text (use Notepad, Notepad2 etc) are the contents exclusively nonsense or is the text there too?

This is key, because if so you can use a script to batch process the documents, remove the formatting/document information and save the remaining text as RTF or similar.

Then it's just a case of tidying up to house style as and when.
posted by dance at 6:02 AM on July 13, 2007

No, it's all nonsense, garbly-gook, and wing-ding type characters.
posted by banger100 at 6:07 AM on July 13, 2007

BBEdit for Mac might be useful. But as dance said, you will probably have to do some hand-editing, or search-and-replace functions to get rid of the header and formatting code, if indeed the actual text is visable in the files.
posted by The Deej at 6:07 AM on July 13, 2007

Print the documents she has stored, scan/OCR them, then save as whatever format you need.
posted by LoraxGuy at 6:08 AM on July 13, 2007 [1 favorite]

Yeah, we were thinking we might have to scan the docs, but I was hoping to find a slicker, quicker way to do it.
posted by banger100 at 6:15 AM on July 13, 2007

I believe that IBM used EBCDIC coding, which explains the nonsense characters. It may be possible to save documents as ASCII, but I guess you'd have to open each one first.

Or there's this...
posted by No Mutant Enemy at 6:17 AM on July 13, 2007

Ooh, EBCDIC, good point. Here's an online EBCDIC-ASCII converter that you could run a couple files through, to see if Mutant Enemy's suggested app would be useful.
posted by rkent at 6:39 AM on July 13, 2007

On a unix-like system (including Mac):

dd if=file.txt of=newfile.txt conv=ascii

Converts the file from EBCDIC and reveals readable text... there's still a bunch of crap in there, but it's half the battle.
posted by buxtonbluecat at 6:44 AM on July 13, 2007

Oh, and I confirmed that the EBCDIC - ASCII conversion worked, with a sample file provided by the OP in other online forums where (s)he has posed the same question.
posted by buxtonbluecat at 6:47 AM on July 13, 2007

Wow. I'll give this conversion a try. Thanks for the suggestion. I'll let you know how it goes.
posted by banger100 at 7:04 AM on July 13, 2007

You folks are golden. The EBCDIC-ASCII conversion tool worked! The resulting file has all the text, but also a lot of nonsense that I'll need to strip out. Any easy way to strip out all the remaining nonsense?
posted by banger100 at 7:17 AM on July 13, 2007

Note that there are several different versions of EBCDIC, most of the online converters are going to be geared towards the System/360 variant I'd imagine. So you might want to try the files on a few different translators if you don't have immediate success.
posted by Mitheral at 7:26 AM on July 13, 2007

How many docs are you talking about? Are there that many that would require her to retype for months? It seems to me it might be quicker to retype what she needs on the PC than try to go through all the docs and edit the bad characters out and reformat them. Sometimes the technological solution isn't the easiest.
posted by JJ86 at 7:46 AM on July 13, 2007

It would be courteous to let us know that you had asked the question previously.

The file might be compatible with IBM's DisplayWrite word processing software, so you might search for [displaywrite conversion].

If all else fails you can use the UNIX strings command to strip out the junk. You'll get weird whitespace instead but the text will be intact. It'd be much easier than retyping.
posted by grouse at 7:51 AM on July 13, 2007

Based on the sample file, here's something that might work better than "strings" for stripping out the formatting characters and such. This should all be one line:

dd if=Docum001.txt conv=ascii | perl -pe 's/M\213/!XYZZY!/g; s/[\000\264\024\215\213\227\302][cwxABMNW]&*//g; s/[\x00-\x1f\x7f-\xff]//g; s/!XYZZY!/\n/g'
posted by hades at 10:12 AM on July 13, 2007

Oh, shoot, that's not quite right. I'll take another shot at it in a minute.
posted by hades at 10:22 AM on July 13, 2007

*I'm sorry, I should have mentioned that I had posted on another site.
*I'll do a search for DisplayWrite.
*It's been years since I've done anything on a Unix box. I only have a PC laptop and a PC desktop. I'm not even sure if there's a way to run Unix commands on a PC.

Thanks again everyone. I'm still searching for ways to strip out the junk characters.
posted by banger100 at 11:19 AM on July 13, 2007

Ok, this is better:

dd if=file.txt conv=ascii | perl -pe 's/^.*\x78\x1e(.*?)\x1e\x78/Document Title: $1\n\n/; s/\x8b[wBN].*?[wBN]\x8b//g; s/\x8bM.*?M\x8b/!XYZZY!/g; s/[\x80-\x9C]//g; s/!XYZZY!/\n/g' > newfile.txt

With only one sample file to work with, I may have made some bad guesses, so no guarantees. But if you can't find a converter that'll do it properly, that command might work well enough.

On a PC, you can run unix commands by installing Cygwin.
posted by hades at 11:42 AM on July 13, 2007

Or, if you don't mind converting one file at a time, I've thrown that command into a quick web interface: http://www.elsewhere.org/tmp/meficonvert/

(Upload the original EBCDIC files there, not the converted ASCII versions.)
posted by hades at 1:20 PM on July 13, 2007 [3 favorites]

Now that's a best motherfucking answer.
posted by grouse at 1:38 PM on July 13, 2007

Holy shite, hades. You're a friggin genius. That's a huge score.

Thanks to everyone, indeed, but hades wins the best mf answer award.
posted by banger100 at 2:09 PM on July 13, 2007

On the off chance that someone else has the same problem, here's what I eventually settled on as a "good enough" solution, or at least as good I was going to be able to do:

#!/usr/bin/perl
while (<>) {
$conv .= $_;
}

$conv =~ s/^\x8bO.\d=.*?\x00O//g;
$conv =~ s/.\x8bM.{0,7}M\x8b/\n/g;
$conv =~ s/\x8bB.{0,7}B\x8b/ /g;
$conv =~ s/\x8bw.{0,7}w\x8b//g;
$conv =~ s/\x8bN.{0,7}N\x8b//g;
$conv =~ s/\x8bR.{0,7}R\x8b//g;
$conv =~ s/\x8by.{0,7}y\x8b//g;
$conv =~ s/x\x0d.*?\x0dx/!DBLSPC!/g;
$conv =~ s/x\x13.*?\x13x/!SNGLSPC!/g;
$conv =~ s/\x8b\x78.(.*?).\x78\x8b/<$1 >\n/sg;
$conv =~ s/[\x80-\x9C]//g;

@tmp = split(/\n/, $conv);
for ($i = 0; $i <= $#tmp; $i++) {
  if ($tmp[$i] =~ s/!DBLSPC!//g) {
    $nl = "\n";
  }
  if ($tmp[$i] =~ s/!SNGLSPC!//g) {
    $nl = "";
  }
  print $tmp[$i] . $nl . "\n";
}


Pipe the output of dd conv=ascii to that, and it'll strip out most of the formatting code and interpret some of the rest, leaving much less to clean up by hand than just using strings would. I tried a few commercial converter demos, and was surprised that nothing seemed to be able to recognize these files, so maybe it'll be useful for someone else.
posted by hades at 2:11 PM on July 17, 2007