Best way to use a server side language for development, but deploy to static HTML

Question

We have a client for whom we build a lot of template based sites. Ideally we would use something like kohana (http://www.kohanaphp.com/) to handle the templating and make doing site wide changes a breeze.

Unfortunately our client cannot (and will not) host any server side code (and before you ask, this will not change and hosting the files ourselves is not an option), so any files deployed to them must be HTML, Javascript, CSS, images and Flash only.

Is there a good way to develop in a framework environment like kohana to make the site manageable, but be able to deploy or export a HTML only version of the site (there is no dynamic aspect to the site such as searching that requires server side languages, and no database use)?

I guess it would be similar to spidering the site, but I'd like something a bit more reliable because some page assets are loaded dynamically with javascript.

Thanks

Answer 1

Consider alternative solutions, even if you don't think you can use them. It is a mistake to become blind to novel solutions because your client "can not handle server side languages" (paraphrased). Sometimes a requirement is a result of incorrect knowledge. For example, is the client really unable to host any server side languages because they can't host IIS or Apache - or is it because they don't have them and can't install any other application either? (If they can install an application, perhaps a solution could be to just provide a "small" webserver - perhaps XSP).

I also suggest you take a look at, eg, google "gears". I've heard pretty good things about that, though it might not be a good fit for your needs.

Failing that, you are probably best off ignoring classic server-side languages and doing it all on the client side, or using server side languages as an exporter. (Which will be a custom job). I think "exporting" a site using custom scripts is probably the best approach.

Finally, this question really does not have a single answer - as a programmer, it is your job to weigh different strengths and different solutions and choose the best one.

Answer 2

Hmm... If the client's server does not allow PHP or ASP I don't think they will go for CGI which is more insecure.

Well, you could make your pages in a dev app, load up those pages in a browser and then save the resultant html. Do that for all the pages and you're set.

Just make sure the links are pointing right.

Answer 3

I don't know of any specific solution to this but if you are willing to create a simple templating-system of your own in the language of your choice it's really only a matter of making a script to generate the pages. I had a site like that back in my teens. Anytime I uploaded something new I regenerated the site and uploaded the new pages. Back then I was using PERL.

The complexity of creating such a script depends a lot on what you're using to build the site. I was working with an entirely custom build that was built to work as it did. In this case I imagine you might want to write a simple template-system yourself to get the advantage of templates as well as being able to generate static HTML from it.

Answer 4

I don't know if this will be an option, but I would suggest CushyCms. This is an online CMS where you define content areas using CSS classes, and can edit the content online. You do need to provide FTP details, so this might be a deal breaker for you, but I thought it was worth mentioning.

Answer 5

Give WaveMaker Studio a try. It will solve your problem partially.

WaveMaker Studio has a kind of templating feature and it comes in community open source version.

HTH

Answer 6

I use Template Toolkit (Perl) and have a simple script that generates static files from the templates. This is great for the situation you are in (common navigation etc etc).

It comes with a ttree command that'll process a directory tree and put the results in another.

Here's the tt.rc file I use:

# ignore these files (regular expressions)
ignore = \.svn
ignore = ^#
ignore = ~$
ignore = .DS_Store
ignore = \.tt$

# if these template files change, reprocess everything
depend *=tpl/wrapper,tpl/defaults,style/default.html

# just copy these files, don't process as templates
copy = \.(gif|png|pdf|jpg)$

# verbose output
verbose

# recurse into subdirectories
recurse

# setup some defaults from tpl/defaults
pre_process = tpl/defaults

# process this file instead of the real file (see below how this is used)
process     = tpl/wrapper

# process files from src/, output to html/
# extra templates in lib/ (tpl/wrapper for example).
src  = src
dest = html
lib  = lib

A couple of special files, tpl/defaults is

[%-  page = {
            title = template.title,
            style = template.style or 'default.html'
    };

    base = INCLUDE tpl/base_uri;

    # don't include any whitespace from here...
    RETURN;
-%]

And tpl/wrapper is

[%- content = PROCESS $template;    
   IF !template.name.search('.html') OR page.style == 'none';
      content;
   ELSE;
      default_style_template = "style/" _ page.style;
      PROCESS $default_style_template;
   END;
%]

This will process the real template; put the results in the content variable and then process the style template (set with page.style in tpl/defaults; defaults to defaults.html).

the lib/style/default.html style file just needs to have

[% content %]

somewhere to include the real template; before and after that you can have the standard footer and headers.

You can read more about Template Toolkit at tt2.org.

Another option would be to use wget (or similar) in recursive mode to "mirror" pages generated by PHP on the development server; but I wouldn't recommend that.