Copying HTML files to output without processing them

[liyanage]

The documentation for FILES_TO_COPY says directories are allowed:

A list of files (or directories) to copy

But when I try it, it errors out at that point:

CRITICAL: (21, 'Is a directory')
CRITICAL: [Errno 21] Is a directory: u'/Users/liyanage/Documents/websites/foo.com-pelican/content/software/bar/'

[justinmayer]

Hi Marc! Very sorry for the delay in responding. Your question has been somewhat obsoleted by the release of Pelican 3.3, which removed FILES_TO_COPY and instead relies on STATIC_PATHS and EXTRA_PATH_METADATA. See the Path Metadata section of the docs for more details.

As am aside, I've been using your Mac OS X tools for many, many years, so it's a pleasure to see your interest in Pelican. If there's anything further we do to assist, please don't hesitate to ask! 😺

[liyanage]

Heh, small world, glad to hear my OS X packages were useful to you.

Good to hear the new version has a new feature to handle this, I will try that. Many thanks!

[liyanage]

I just tried the STATIC_PATHS option, and while this does indeed no longer error out with directories, it seems to attempt to process the .html files in the directory, which I don't want:

[localhost] local: pelican content -s pelicanconf.py
WARNING: Could not process software/xxx/genindex.html
'NoneType' object has no attribute 'lower'
WARNING: Could not process software/xxx/index.html
'NoneType' object has no attribute 'lower'

[justinmayer]

Pelican supports HTML files as source input, along with Markdown, reST, and Asciidoc formats. I believe that particular error should be converted to a warning as of 35375b1, which was just merged and is not yet in a shipped release.

But the larger question remains... Why do you have .html files in your source content?

[liyanage]

In my content folder I have a directory with already fully pre-generated HTML, generated outside Pelican by Sphinx. I would like that entire directory to be copied over to the output directory without any changes, so I listed it in the STATIC_PATHS array.

Am I using it wrong?

[justinmayer]

As @wilbur-ma mentioned in #1157, try adding the following to your settings file:

READERS = {"html": None}

[liyanage]

That works, thanks!

Perhaps as a feature request, it seems that this setting is global. It might be nice to be able to scope it to parts of the input tree, in case I ever want to have Pelican preprocess .html files in other parts of the source tree.

Feel free to close this if you think such a feature would complicate matters too much.

[justinmayer]

I can see how some folks might have a need for that. How do you think that might be implemented? And would that be something you would want to work on?

[liyanage]

Yes if I find some time and/or need this ability in the project I'm about to set up with Pelican, I'll try to add it.

Perhaps by extending the system so it accepts an array or dictionary of source path match patterns and dictionaries as value for the READER item. This would be in addition to what it accepts today, of course.

[foresto]

Would setting PAGE_EXCLUDES=['software/xxx'] and/or ARTICLE_EXCLUDES=['software/xxx'] solve this problem? (Along with STATIC_PATHS, of course.)

[MartinThoma]

I think I might have the same problem: http://stackoverflow.com/q/37104625/562769 - can anybody solve it?

[Scheirle]

@MartinThoma move your html5 folder in your content folder (symlink works too) and only use STATIC_PATHS = [..., 'html5']

[MartinThoma]

@Scheirle I tried it and I got

ERROR: Skipping html5/regression/regression.htm: could not find information about 'NameError: date'
ERROR: Skipping html5/regression/README.md: could not find information about 'NameError: title'

[Scheirle]

@MartinThoma use ARTICLE_EXCLUDES and maybe PAGE_EXCLUDES to exclude the html5 folder.

Alternatively move all your articles in a sub folder and use ARTICLE_PATHS, so pelican only looks in the given folder for articles.

[leotrs]

This is solved by setting both STATIC_PATHS and PAGE/ARTICLE_EXCLUDES at the same time.

I admit it's annoying, but at least this is no longer an issue.

[justinmayer]

I'm sure this use case could be handled better, but there are at least avenues to achieve the desired result. Follow-up enhancements are of course welcome.

[mattgilbertnet]

Just ran into this issue. The fix in this thread worked, but I would think any static paths would be excluded (from pages and articles processing) by default. Being static and all.

[avaris]

@mattgilbertnet not necessarily