I’ll start off by admitting I hadn’t used the I18n functionality but had read through it a few times in the past, but I did just whip up a test to make sure it worked as I thought.
Short Answer: I don’t think you can do exactly what you are hoping.
The reason is that Hugo won’t generate a file without an existing file in the content/ directory. It’s one-to-one in my understanding (came up recently). So without a page2.ru.md file, no corresponding page2/index.html file will be created (assuming you’re using prettyURLs).
Workarounds: A few ideas come to mind.
Basic Option
What lang.Merge can do for you is automatically create links to the EN pages for any missing RU pages. But they would have english menus etc. So with your given files
{{ range .Site.RegularPages }}
<li><a href="{{.Permalink}}">{{.Title}}</a></li>
{{ end }}
Will output 2 links on an en list page, but only 1 on the same ru list page. If you changed it to be
{{ $pages := .Site.RegularPages }}
{{ range .Site.Home.Translations }}
{{ $pages = $pages | lang.Merge .Site.RegularPages }}
{{ range $pages }}
<li><a href="{{.Permalink}}">{{.Title}}</a></li>
{{ end }}
{{ end }}
The ru list page would show two links, one to the ru page1, and the second to the en page2. It does not automatically create a ru page2 with the content from page1.md. That’s probably the confusion from the documentation, and not what you were trying to achieve.
Better Option (probably)
What you could do is have a script outside hugo which creates a link to the missing file, which hugo would then see and process correctly.
On the command line, I created a link (an alias) from page2.md to page2.ru.md and hugo successfully created the page2.ru/index.html file correctly. I tested this on macOS using both hard (ln page2.md page2.ru.md) and symbolic (ln -s page2.md page2.ru.md) commands and they worked.
What didn’t work is an alias created via the Finder either with File → Make Alias or option-command-dragging. This is probably because those are not unix based links, they are actual files containing some info the Finder interprets as an alias, but hugo doesn’t.
The easiest way I see to do this is use a script outside hugo. The basic steps would be:
get a list of all items recursively of the content directory
for each item:
if it's a .md file AND a matching .ru.md file DOES NOT exist:
create a link
Honestly fairly trivial unix type script, thought my shell scripting skills are not strong, there’s lots of examples of this type of script. Run it before building hugo, and you’re all set. You could probably easily add a step to log any links it finds to a file so you have a list of content that isn’t translated if there was a process to work through them.
[Edit] Just had another thought. The script could copy the files instead of creating the link. Advantages:
- you have real files ready to be translated if you wanted
Cons:
- if the content in the en version changes, they will become out of sync
- loose the ability to easily list the files that are missing translations unless the script also echoed some parameter into the front matter marking them as a copy. Then grep from the command line or a “hidden” list page in hugo could find all content with that parameter.
I’d use a link, but your workflow/use case might benefit from the other.
Alternative
One last option comes to mind, but this seems extremely impractical. Instead of file based translations, hugo supports individual string translations. Although I don’t have a i18n translation file to test with, the docs are pretty clear:
If a string does not have a translation for the current language, Hugo will use the value from the default language.
But to make use of this, the entire content of the missing pages would have to be managed in the i18n strings file, which seems extremely cumbersome, and maybe not even possible since I can’t test. You’d have to get the content from i18n and then process it for markdown, etc etc.