freedos-devel

[Freedos-devel] Converting WordStar to HTML

[Jim H. <jh...@fr...>]

If you've been exploring WordStar since author Robert J Sawyer's
re-packaging of WordStar for DOS, you may need to convert your new
WordStar files to another format afterwards.

WordStar has a very simple file format, so I published an article last
week on Both.org about how to convert WordStar files to HTML.
https://www.both.org/?p=6898

Someone asked me if I could share the code from the article in a
GitHub repo, so I have copied the code and adapted a README.md file
here:

https://github.com/freedosproject/wshtml

It's available under the MIT License.

Re: [Freedos-devel] Converting WordStar to HTML

[Rugxulo <ru...@gm...>]

You forgot to include "sample.ws" (or any other test suite).


On Mon, Aug 12, 2024 at 3:37 PM Jim Hall via Freedos-devel
<fre...@li...> wrote:
>
> I published an article last week on Both.org about how to convert WordStar files to HTML.
>
> Someone asked me if I could share the code from the article in a GitHub repo.

Re: [Freedos-devel] Converting WordStar to HTML

[Jim H. <jh...@fr...>]

Jim Hall wrote:
> >
> > I published an article last week on Both.org about how to convert WordStar files to HTML.
> >
> > Someone asked me if I could share the code from the article in a GitHub repo.
>

Rugxulo wrote:
>
> You forgot to include "sample.ws" (or any other test suite).
>


Oops, forgot about that (a lot going on lately).

I've added SAMPLE.WS and a few others in a 'wordstar' folder.

Re: [Freedos-devel] Converting WordStar to HTML

[Rugxulo <ru...@gm...>]

Hi,

On Fri, Aug 16, 2024 at 9:16 AM Jim Hall via Freedos-devel
<fre...@li...> wrote:
>
> Jim Hall wrote:
> > >
> > > I published an article last week on Both.org about how to convert WordStar files to HTML.
>
> I've added SAMPLE.WS and a few others in a 'wordstar' folder.

I'm mildly curious about your choice of C dialect.

Although C++ is beyond me (in most ways), so I know very little there.
(Like many languages, it has a long history and several dialects. I
barely wrote a very simple C++98 program with DJGPP a few months ago.)

You seem to use ANSI C, or C89/C90 (as it's sometimes called).

I like your BOOL typedef (a clever solution I've also used). I also am
still sympathetic to C89 (and DOS compilers).

Having said that, "bool" is also a feature of (semi-modern) C++ and
C99. (Well, OpenWatcom supports "bool" but TurboC++ does not, IIRC.)
So are // line comments. And "for (int i=0;" declarations. (But
"inline" support may be too limited to be useful.)

Your article says you used GCC, which by default (last I checked) is
"-std=gnu17", so it supports "bool" and // as well by default. (Okay,
C++ supports bool by default. C99 needs <stdbool.h>. Obviously. I'm
mainly testing OW's C++ here, basically implying that these features
also are available in C99.)

Are you trying to strictly stay compatible to C89? Or is it just
habit? Since it works in OpenWatcom (as C++ or C99 with <stdbool.h>)
and DJGPP, I would personally use such features. (Okay, I wouldn't,
but I *could*. It depends on how many compilers and OSes you want to
target.) In other words, in an abandoned DOS landscape, I'm halfway
seriously suggesting using old C++ as if C99.

Heck, you could make a simple #define, and the line comments could be
trivially converted to C89 style with Sed (or a very simple util).

#if !__cplusplus || defined __TURBOC__
typedef enum {false,true} BOOL;
#else
#define BOOL bool
#endif

sed -e "s,//\(.*\),/*\1*/," wshtml.c99 >wshtml.c89

* (UNCMNTC: uncomment C, it's a 3 kb .C source, public domain)
* http://cpmarchives.classiccmp.org//cpm/mirrors/cbfalconer.home.att.net/download/uncmntc.zip

See what I mean? You don't strictly have to do without everything. You
have options.

Okay, you barely mentioned GCC: "gcc -o wshtml wshtml.c". Thanks to
GNU Make's built-in rules, you can probably simplify that to just
"make wshtml".

You hardcoded C escape sequences (char literals) as numbers instead:

case 0x09: /* tab */
case 0x0a: /* new line */
case 0x0c: /* page feed */
case 0x0d: /* carr rtn */

But you could just use '\t' '\n' '\f' '\r':

* https://stanislavs.org/helppc/c_escape_sequences.html

I don't know why you use "return 1;" and "return 0;" instead of using
<stdlib.h> EXIT_SUCCESS and EXIT_FAILURE . (It's ANSI standard, and
IIRC the only DOS compiler who omits it is Pacific C.)

fmt.bold = false; fmt.dblstrike = false; fmt.underline = false;

You can initialize all of those at once:

fmt.bold = fmt.dblstrike = fmt.underline = false;

C99 and C++ also let you put "size_t i;" into the "for (i = 0 ..."
statement -> "for (size_t i = 0; ..."

Oh, even C89 lets you combine these two statements (at the start of a block):

"BYTE low; low = ch & 0x7f;"  ->  "BYTE low = ch & 0x7f;"

Is "if (len > 0)" even necessary? Will "for (i = 0; i < len ..." even
iterate if loop count is less than 1? It probably?? shouldn't.

You could simplify all of these:

'if (fmt.bold) { fputs("<b>",out); } else { fputs("</b>",out); }'

'fputs(fmt.bold ? "<b>" : "</b>",out);'

You could also combine these two statements:

'pfile = fopen(argv[1],"rb");' "if (pfile==NULL) {" ->  'if
((pfile=fopen(argv[1],"rb"))==NULL) {'

There are some other suggestions, but they are mostly about personal taste?

N.B. I tested most of this with OpenWatcom 1.9 as C++ (under DOS), and
the SAMPLE.WS output matched the C (C89) version's output. (The .EXE
was barely 102 bytes bigger.)