A Little Help

A Little Help

[D. Cooper Stevenson]

Dear All,

I've worked hard on importing ASCII data from GIS. Here's the good news:

  * Exporting GIS data into ASCII may be automated using just one command

  * The ASCII file format is simple

Here's the bad news: 

  * I haven't written (but do respect) the C language in a long time and it 
shows; I've had a tough go of it getting the application to work.

I think it's time to ask for help.

As I mentioned above, the GIS ASCII file format is simple. Here's a 
description:

 The first 7 or so lines represent the header of the file. The most relevant 
numbers for our purposes are the "rows" and "cols" numbers. The last line is 
the actual elevation (or landcover) data. Here's an example:

  north: 3980319.16466812
  south: 3978824.85093895
  east: 443960
  west: 442296
  rows: 747
  cols: 832
  10 10 10 9 9 ...........

Each number represents a terrain type. Well, that's not exactly true, but it's 
close. 1-3 say, is beach, 4-6 is plains and so on.
 
So if you want an area of 54 x 42 cells, you would want to sample a "box" of 
13 x 20 cells. 

In other words:

  1) Sample the first 13 numbers across by 20 lines down. A "mode" calculation 
(the number that shows up the most)  is best. An average will do I think. 
Export this number to a terrain file; this is the first cell type.

  2) Jump to the next "box," i.e. column 14-27 while still on lines 1-20 and 
output it to a file. This is the second cell type.

  3) Repeat until the file is finished.

Here is my code so far:

  http://www.xconq.org/ascii/parsegis.c

Here is the actual GIS file (Huge: 21M):

  http://www.xconq.org/ascii/seattle.arx

If you can offer a little help I'd be greatfull.


-Coop

Re: A Little Help

[Eric McDonald]

D. Cooper Stevenson wrote:

>   * I haven't written (but do respect) the C language in a long time and it 
> shows; I've had a tough go of it getting the application to work.
> 
> I think it's time to ask for help.
> 
> As I mentioned above, the GIS ASCII file format is simple. Here's a 
> description:
> 
>  The first 7 or so lines represent the header of the file. The most relevant 
> numbers for our purposes are the "rows" and "cols" numbers. The last line is 
> the actual elevation (or landcover) data. Here's an example:
> 
>   north: 3980319.16466812
>   south: 3978824.85093895
>   east: 443960
>   west: 442296

If you don't need these values, you might want to just toss them. That 
way you wouldn't have to tell your parser to ignore them.

>   rows: 747
>   cols: 832

You'll probably want to keep this information. The number of rows is 
probably a better termination condition than EOF. And the number of 
columns is useful for determining how many times to call strtok(3), for 
example.

Here is a vague handwaving outline of how one might go about some of 
this. This is not guaranteed to be the best way. Just some spewing off 
the top of my head...:

/* Some definitions earlier on. */
#define BUFSIZE 256
#define DATAWIDTH 5
char data [BUFSIZE]
int numrows = numcols = 0;

/* Some code.... */

/* Part of your parser. (One possible way; there are others. Assumes you 
are inside some sort of loop.) */
if (numrows && numcols)
   break;
fgets(data, BUFSIZE, ifp);
if (!strncmp("rows: ", data, 6)) {
   numrows = atoi(data+6);
   continue;
}
else if (!strncmp("cols: ", data, 6)) {
   numcols = atoi(data+6);
   continue;
}

/* Some code...? */

/* Another part of your parser. */
data2 = malloc(numcols*DATAWIDTH);
boxrowsz = numcols / BOXWIDTH;
elevbox = malloc(numcols * BOXHEIGHT * sizeof(int));
for (i = 0; i < numrows; ++i) {
   fgets(data2, numcols * DATAWIDTH, ifp);
   memset(elevrow, 0, numcols * sizeof(int));
   for (j = 0; j < numcols; ++j) {
     if (!j)
       nextdata = strtok(data2, " ");
     else
       nextdata = strtok(NULL, " ");
     /* Stuff the data into the appropriate elev row box and slot(s) 
within the box. */
     /* You can use / and % (div and mod) to help determine which box to 
stuff things in. */
   }
   if (i && (0 == (i % BOXHEIGHT)))
      /* Take median, mean, or whatever. */
}

/* More code.... */

/* You can use existing functions in libconq.a or maybe even just 
libconqlow.a to help in writing out Xconq-readable forms, including 
layer data. */

/* Hope this helps,
    Eric */

Re: A Little Help

[Eric McDonald]

Let me clear out some vestigial junk and clarify a few things:

> elevbox = malloc(numcols * BOXHEIGHT * sizeof(int));

elevboxrow = malloc(numcols * BOXHEIGHT * sizeof(int));

>   memset(elevrow, 0, numcols * sizeof(int));

Not necessary and certainly not in this form.

>     /* You can use / and % (div and mod) to help determine which box to 
> stuff things in. */

Probably not necessary. Just put directly into column and row. Reduction 
to medians or means is the only place where the row of boxes truly matters.

Eric

Re: A Little Help

[D. Cooper Stevenson]

Thanks for this help, Eric. I'll continue working on it keep you and this list 
abreast of progress. My goal is to have this done by the end of this weekend.


-Coop

On Monday 11 October 2004 19:14, Eric McDonald wrote:
> D. Cooper Stevenson wrote:
> >   * I haven't written (but do respect) the C language in a long time and
> > it shows; I've had a tough go of it getting the application to work.
> >
> > I think it's time to ask for help.
> >
> > As I mentioned above, the GIS ASCII file format is simple. Here's a
> > description:
> >
> >  The first 7 or so lines represent the header of the file. The most
> > relevant numbers for our purposes are the "rows" and "cols" numbers. The
> > last line is the actual elevation (or landcover) data. Here's an example:
> >
> >   north: 3980319.16466812
> >   south: 3978824.85093895
> >   east: 443960
> >   west: 442296
>
> If you don't need these values, you might want to just toss them. That
> way you wouldn't have to tell your parser to ignore them.
>
> >   rows: 747
> >   cols: 832
>
> You'll probably want to keep this information. The number of rows is
> probably a better termination condition than EOF. And the number of
> columns is useful for determining how many times to call strtok(3), for
> example.
>
> Here is a vague handwaving outline of how one might go about some of
> this. This is not guaranteed to be the best way. Just some spewing off
> the top of my head...:
>
> /* Some definitions earlier on. */
> #define BUFSIZE 256
> #define DATAWIDTH 5
> char data [BUFSIZE]
> int numrows = numcols = 0;
>
> /* Some code.... */
>
> /* Part of your parser. (One possible way; there are others. Assumes you
> are inside some sort of loop.) */
> if (numrows && numcols)
>    break;
> fgets(data, BUFSIZE, ifp);
> if (!strncmp("rows: ", data, 6)) {
>    numrows = atoi(data+6);
>    continue;
> }
> else if (!strncmp("cols: ", data, 6)) {
>    numcols = atoi(data+6);
>    continue;
> }
>
> /* Some code...? */
>
> /* Another part of your parser. */
> data2 = malloc(numcols*DATAWIDTH);
> boxrowsz = numcols / BOXWIDTH;
> elevbox = malloc(numcols * BOXHEIGHT * sizeof(int));
> for (i = 0; i < numrows; ++i) {
>    fgets(data2, numcols * DATAWIDTH, ifp);
>    memset(elevrow, 0, numcols * sizeof(int));
>    for (j = 0; j < numcols; ++j) {
>      if (!j)
>        nextdata = strtok(data2, " ");
>      else
>        nextdata = strtok(NULL, " ");
>      /* Stuff the data into the appropriate elev row box and slot(s)
> within the box. */
>      /* You can use / and % (div and mod) to help determine which box to
> stuff things in. */
>    }
>    if (i && (0 == (i % BOXHEIGHT)))
>       /* Take median, mean, or whatever. */
> }
>
> /* More code.... */
>
> /* You can use existing functions in libconq.a or maybe even just
> libconqlow.a to help in writing out Xconq-readable forms, including
> layer data. */
>
> /* Hope this helps,
>     Eric */

Website Update Request

[Elijah Meeks]

I love xconq.org, but...

The poll doesn't seem to work (At least for me).

Can we have pages for the individual xconq games? 
That way we can explain storyline, strategy and put up
screenshots.

Thanks,
Elijah



		
__________________________________
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.
http://promotions.yahoo.com/new_mail 

Re: Website Update Request

[Eric McDonald]

Elijah Meeks wrote:
> I love xconq.org, but...
> 
> The poll doesn't seem to work (At least for me).

I reactivated the surveys module, so this might be fixed now.

> Can we have pages for the individual xconq games?

I don't see why not. If you make the pages, you might to create a master 
index page so that we don't have dozens of links from the home page.

Eric

Re: Website Update Request

[D. Cooper Stevenson]

On Sat, 2004-10-16 at 11:22, Eric McDonald wrote:

[snip]
> 
> > Can we have pages for the individual xconq games?

I can give you "author" access, Elijah, so that you can create and post
the content yourself. When you're ready to post, we can link it to the
main page. I think Eric's plan to create a single "games link" is the
best way to go. Let me know if you're interested and would like a login.


-Coop 

Re: Website Update Request

[Elijah Meeks]

> I can give you "author" access, Elijah, so that you
> can create and post
> the content yourself. When you're ready to post, we
> can link it to the
> main page. I think Eric's plan to create a single
> "games link" is the
> best way to go. Let me know if you're interested and
> would like a login.

That sounds fine, as long as I'm working with
templates.  I have no artistic talent whatsoever. 
None.  Zero.  Nada.  Zilch.