On 11 Nov., 19:28, "bruno.postle" <[email protected]> wrote:

> The #-hugin lines are used by the hugin GUI to store non-critical
> metadata as key=value pairs.  Your parser should at least  preserve
> them, though we don't have a spec.

No spec... sigh :-(
A common problem.
So what non-critical metadata are there, apart from the crop factor?
Can I rely on these data preceding the line they belong to?
If they are key-value pairs, that's a consolation, because at least
it's easy to parse
without touching semantics. I'll include some code in my parser.

>
> The #-imagefile lines are not used by hugin, they might be something
> to do with ptgui.

so yet another cook spoiling the broth. This comes from bad design.
Someone starts out with an ad-hoc approach, hey, let's just use one-
letter-tags and put the value right after them, great idea, scans
fast, all I need. Comes time and suddenly there aren't enough letters
in the alphabet that even faintly resemble the tag's purpose. Ah,
heck, let's just use letter combinations. Someone else comes along,
thinks, hey this is ugly, I'll use comments and key-value-pairs. So by
now there are three different ways of doing it. If whoever has dreamt
up the pto file format in the first place had thought ahead, they'd
sure come up with something more timeproof.
Maybe it's time for a new format? Like, XML-based? After all we're
only talking metadata, so processing time shouldn't be that critical,
and there are very good tools to parse XML efficiently (at least in
Python...). The pto format by now is a bit unruly, but not beyond
parsability, so it'd be a snap to write a format converter to a future-
proof format that could be thoroughly thought-out and phased in as an
optional alternative, and the parser should be much easier to write as
well if it could rely on an underlying format like XML, where every
value is typed and the whole parse finishes with a nice clean tree
structure.
>
> Have you looked at Panotools::Script which has a perl parser for .pto
> files? Perl and python are  not so different...
>
My parser currently retains every bit of information in the source
file, but it only really processes non-comment lines. I reckon your
package Panotools::Script::Line does similar things, seems like you're
not looking at these non-critical metadata either? I haven't really
looked much at libpano's own parser, though enough to have seen that
it scans with full semantic awareness (like, this member is such-and-
such, so scanf it as a so-and-so)...
I thought about just using that scanner, but having to emulate all
it's infrastructure in Python would have been much more effort than
writing something from scratch, which is the cleaner solution anyway.
After all it wasn't that much work, and since parser-building is one
of my hobbies, it was even fun.
I found perl quite different from Python. This may be because I don't
know perl, but I feel the syntax has too many non-alphanumeric
characters (like, dollar signs are banned from Python syntax...). It
did look terse, though. But I think I also made more of an effort to
extract every bit of meaning that could be gleaned without actually
touching semantics.

with regards
Kay

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Reply via email to