Re: hdf_dump_cb data loss

Thanks.  Probably just didn't think about nodes without values or
sub-nodes as that important, but seems no reason not to include them.

I'll put this in the next release.

Brandon

On 08/22/07 Nikolai Kondrashov uttered the following other thing:
> Hello, all!
> 
> I'm not sure whether it is a bug or just design inconsistency, but
> hdf_dump_cb (used, for example, by hdf_write_file) ignores nodes without
> both value and children.
> 
> So, if you read a file like that:
> 
> fruit {
>   apple {
>     color = green
>   }
>   orange {
>   }
> }
> 
> then "fruit.orange" node will be created, but will not have neither value
> (i.e. it will be NULL) nor children.
> 
> But if you will write this dataset into a file using hdf_write_file, you
> will end up with this:
> 
> fruit {
>   apple {
>     color = green
>   }
> }
> 
> Note: "fruit.orange" is lost.
> 
> Below is a patch, which seems to fix it. It also should fix a case of lost
> attributes on a node without value, and also extra space character after
> attributes.
> ---:<---
> --- util/neo_hdf.c      (revision 41766)
> +++ util/neo_hdf.c      (working copy)
> @@ -1146,7 +1146,7 @@
>                             void *rock, DUMPF_CB dump_cbf)
>  {
>    NEOERR *err;
> -  char *p, op;
> +  char *p;
>    char ml[10] = "\nEOM\n";
>    int ml_len = strlen(ml);
>    char whsp[256] = "";
> @@ -1163,10 +1163,8 @@
> 
>    while (hdf != NULL)
>    {
> -    op = '=';
> -    if (hdf->value)
> +    if (hdf->value || hdf->attr || !hdf->child)
>      {
> -      if (hdf->link) op = ':';
>        if (prefix && (dtype == DUMP_TYPE_DOTTED))
>        {
>         err = dump_cbf(rock, "%s.%s", prefix, hdf->name);
> @@ -1204,11 +1202,22 @@
>           }
>           attr = attr->next;
>         }
> -       err = dump_cbf(rock, "] ");
> +       err = dump_cbf(rock, "]");
>         if (err) return nerr_pass(err);
>        }
> -      if (strchr (hdf->value, '\n'))
> +      if (!hdf->value)
>        {
> +       if (dtype == DUMP_TYPE_DOTTED)
> +         err = dump_cbf(rock, " {}\n");
> +       else
> +         err = dump_cbf(rock, " {\n%s}\n", whsp);
> +      }
> +      else if (hdf->link)
> +      {
> +       err = dump_cbf(rock, " : %s\n", hdf->value);
> +      }
> +      else if (strchr (hdf->value, '\n'))
> +      {
>         int vlen = strlen(hdf->value);
> 
>         while (strstr(hdf->value, ml) || ((vlen > ml_len) && !strncmp(hdf->value + vlen - ml_len + 1, ml, strlen(ml) - 1)))
> @@ -1223,7 +1232,7 @@
>        }
>        else
>        {
> -       err = dump_cbf(rock, " %c %s\n", op, hdf->value);
> +       err = dump_cbf(rock, " = %s\n", hdf->value);
>        }
>        if (err) return nerr_pass (err);
>      }
> --->:---
> 
> Thanks for your attention :)
> 
> Sincerely,
> Nick
> 
> 
> 
>  
> Yahoo! Groups Links
> 
> 
> 

-- 
 "We [the Net] get portrayed in a crappy light...  This time it's a cult.  
  Usually, it's that we're all child pornographers."  
                    -- Internet Defender, _Washington Post_, 3-29-97
                                           http://www.fiction.net/blong/