summaryrefslogtreecommitdiff
path: root/doc/forum/transition_from_handwritten_html_to_ikiwiki.mdwn
blob: a8d04a0ad5a82d30f684c585aaabfabab42fc537 (plain)

I'm trying to convert hand written html site to ikiwiki and maintain url compatibility. html plugin with indexpages=1 converts all dir_name/index.html correctly to dir_name urls with wiki/css based content, but somedir/somefile.html files are only accessible as somedir/somefile/. Non .html files seem to accessible with their full paths, for example somedir/pic.jpg from hand written html can be accessed by same path under ikiwiki.

How to make somedir/somefile.html accessible as somedir/somefile.html under ikiwiki?

Thanks,

-Mikko

Hello! The options you need to investigate are --usedirs and --no-usedirs. The default --usedirs takes any source page foo (regardless of its format, be it markdown or html) and converts it into a destination page foo/index.html (URL foo/). By comparison, --no-usedirs maps the source file onto a destination file directly: src/foo.html becomes dest/foo.html, src/bar.mdwn becomes dest/bar.html, etc.

It sounds like you want --no-usedirs, or the corresponding usedirs => 0, option in your setup file. See [[usage]] for more information. -- [[Jon]]

Thanks, usedirs seems to be just the thing I need.

-Mikko

Actually usedirs didn't do exactly what I want. The old site contains both somedir/index.html and somedir/somename.html files. With html plugin and indexpages=1 the somedir/index.html pages are accessed correctly but somedir/somefile.html files not.

With usedirs => 0, somedir/somename.html pages are accessed correctly but somedir/index.html pages are not. Actually the handwritten somedir/index.html files were removed on a rebuild:

$ ikiwiki -setup blog.setup -rebuild -v
...
removing test2/index.html, no longer built by test2

Is there a way for both index.html and somename.html raw html files to show up through ikiwki?

-Mikko

I think you want usedirs => 0 and indexpages => 0?

What IkiWiki does is to map the source filename to an abstract page name (indexpages alters how this is done), then map the abstract page name to an output filename (usedirs alters how this is done).

The three columns here are input, abstract page name, output:

usedirs => 0, indexpages => 0:
    a/index.html -> a/index -> a/index.html
    a/b.html -> a/b -> a/b.html
usedirs => 1, indexpages => 0:
    a/index.html -> a/index -> a/index/index.html
    a/b.html -> a/b -> a/b/index.html
usedirs => 0, indexpages => 1:
    a/index.html -> a -> a.html
    a/b.html -> a/b -> a/b.html
usedirs => 1, indexpages => 1:
    a/index.html -> a -> a/index.html
    a/b.html -> a/b -> a/b/index.html

The abstract page name is what you use in wikilinks and pagespecs.

What I would suggest you do instead, though, is break your URLs once (but put in Apache redirections), to get everything to be consistent; I strongly recommend usedirs => 1 and indexpages => 0, then always advertising URLs that look like http://www.example.com/a/b/. This is what ikiwiki.info itself does, for instance. --[[smcv]]

Thanks for the explanation. usedirs => 0 and indexpages => 0 does the trick, but I'll try to setup mod_rewrite from foo/bar.html to foo/bar in the final conversion.

-Mikko

That's roughly what I do, but you can do it with Redirect and RedirectMatch from mod_alias, rather than fire up rewrite. Mind you I don't write a generic rule, I have a finite set of pages to redirect which I know. -- [[Jon]]

I'm getting closer. Now with usedirs => 1 and raw html pages, ikiwiki transforms foo/index.html to foo/index/index.html. Can ikiwiki be instructed map foo/index.html to page foo instead that foo/index?

-Mikko

If you don't already have a foo.html in your source, why not just rename foo/index.html to foo.html? With usedirs, it will then map to foo/index.html. Before, you had 'foo/' and 'foo/index.html' as working URLS, and they will work after too.

If you did have a foo.html and a foo/index.html, hmm, that's a tricky one. -- [[Jon]]

We may be going round in circles - that's what indexpages => 1 does :-) See the table I constructed above, which explains the mapping from input files to abstract page names, and then the mapping from abstract page names to output files. (I personally think that moving your source pages around like Jon suggested is a better solution, though. --[[smcv]]