I have a nested yaml data file for my jekyll blog like so:
Nickname1:
name: John
address: Johnstreet 1
Nickname2:
name: Rudolf
address: Rudolfstreet 1
I use this to print additional information after each post, which I specify in the front matter. This works fine.
I now wish to create a site that lists all entries of this data file. This should in theory be easy:
{% for nickname in site.data.NAMEOFFILE %}
<li> {{ nickname.address }} </li>
{% endfor %}
However, this does not work because the YAML data is nested. The data has to remain nested, and each entry needs to have a different nickname. The problem with that is of course that I am unable to loop over all entries, because they are all differently named.
Can you folks help me out here? Is there any way in which I can achieve this without changing the nested structure of my data file? Changing the nested structure would break large parts of my site.
Jekyll data files can be set up in two formats: list or dictionary (not the official terminology, but that's what I call them and what helps me understand them).
Lists are used for iteration. Dictionaries are used for looking up an individual item and shouldn't be used in iteration.
// list.yml
- nickname: Nickname1
name: John
address: Johnstreet 1
- nickname: Nickname2
name: Rudolf
address: Rudolfstreet 1
...
// usage
{% for person in site.data.list %}
<li> {{ person.address }} </li>
{% endfor %}
// dictionary.yml
Nickname1:
name: John
address: Johnstreet 1
Nickname2:
name: Rudolf
address: Rudolfstreet 1
...
// usage
{% assign person = site.data.dictionary['Nickname1'] %}
<ul>
<li> {{ person.address }} </li>
</ul>
I have the same data in two different files: one in a list format and one in a dictionary format. That lets me iterate or do specific lookups whenever I need. Drawbacks are that you have duplicated data across two different files and need to maintain consistency whenever you make changes.
To solve your specific problem, I would make another data file with the data formatted into a list so that you can iterate through the data. This means you don't need to change the file with the nested structure and can avoid breaking the site.
Jekyll docs example on 'dictionary' usage
Jekyll docs example on 'list' usage
When you iterate over an object, the iteration variable is a two-element array with the key at index 0 and the value at index 1:
<ul>
{% for nickname in site.data.people -%}
<li>
{{ nickname[0] }}: name {{ nickname[1].name }},
address {{ nickname[1].address }}
</li>
{% endfor %}
</ul>
renders (after whitespace cleanup) as
<ul>
<li>Nickname1: name John, address Johnstreet 1</li>
<li>Nickname2: name Rudolf, address Rudolfstreet 1</li>
</ul>
Weirdly, I can't find this described anywhere in the Liquid docs, neither on the Jekyll site nor in the official documentation. The only reference I've seen is on the platformOS documentation page.
Related
I want to loop over all of the collections in my Jekyll site, and within that I want to sort and list all of the pages in that collection.
Based on this stackoverflow answer, I can loop through all the collections and items:
{% for collection in site.collections %}
<h2>Items from {{ collection.label }}</h2>
<ul>
{% assign pages = site[collection.label] %}
{% for item in pages %}
<li>{{ item.title }}</li>
{% endfor %}
</ul>
{% endfor %}
For a specific collection, I can also sort the update field from the frontmatter:
{% assign sorted = site.programming | sort: 'update' %}
But if I try to apply this to the 1st example, it fails:
{% assign pages = site[collection.label] | sort: 'update' %}
This gives a rather generically useless error:
Liquid Exception: Liquid error (line 30): comparison of Array with Array failed in index.md
Error: Liquid error (line 30): comparison of Array with Array failed
Error: Run jekyll build --trace for more information.
I'm guessing that somehow site[collection.label] returns something different than site.programming, but I'm not sure what, or how to solve this.
EDIT: I tried using collection.docs instead of site[collection.label] and got the same error.
The cause of this error turned out not to be the difference between collection.docs and site[collection.label]. Rather, in one of the collections, there was an item where the frontmatter did not contain the update field. As a result, it couldn't sort by this field because not all items had it. Similarly, you get the same error if the types are not comparable in all cases. In this case, they were all supposed to be dates; if one was not a valid date format, it failed.
I still think this is a terribly ambiguous error message.
As outlined in Jekyll's Collections docs, I've added an additional field to my collections. Here's part of my _config.yml:
collections:
algebra1-2:
title: Algebra 1-2
output: true
How can I access the title field in pages and layouts?
I've tried {{ page.collection.title }} (and a dozen other permutations) with no luck. However, I have found that {{ page.collection }} will render algebra1-2.
Edit: I'm not checking whether or not the page belongs to the collection, and also am trying to avoid conditionals and looping over all pages, as this answer recommends.
I found that this is possible using the following syntax: {{ site.collections[1].<field> }}.
So, given the example above, {{ site.collections[1].title }} will evaluate to Algebra 1-2.
I want something like:
---
my_var: "foo.bar.baz"
---
{{- site.data.header[page.my_var] -}}
Unfortunately, it doesn't work…
I know that I can use site.data.header.foo.bar.bazor site.data.header["foo"]["bar"]["baz"], but it's not that.
Also I know that I can split page.my_var to substrings and use them then as site.data.header["foo"]["bar"]["baz"]. But this increases Jekyll build time.
I guess I need some Ruby plugin, but don't know one and I don't know Ruby to write one.
If you know such plugin or can help me write one or know some native workaround, this would be sooo great!
Help :)
Also I know that I can split page.my_var to substrings and use them then as site.data["foo"]["bar"]["baz"]. But this increases Jekyll build time.
I'm not sure you can do this. And even if you can, I'm not sure that there is a real performance impact.
But you can do this :
_data/foo.yml
bar:
baz: 1
biz: 2
buz: 3
beer:
baz: 1
biz: 2
buz: 3
Then :
---
my_var: "foo.bar.baz"
---
{% assign keys = page.my_var | split: "." %}
{% assign object = site.data %}
{% for key in keys %}
{% assign object = object[key] %}
{% endfor %}
{{ object }}
What are hoping to do has been blocked by design.
site.data["foo.bar.baz"] implies that you need to have a data file named foo.bar.baz.yml or foo.bar.baz.json, ... etc.
But Jekyll sanitizes data files' names and therefore the resulting object would only have a key named "foobarbaz"
A plugin that purposely removes this sanitization can written but it wouldn't appear to be a benign or safe to end-users. So the chances of finding such a plugin in the wild is slim..
config.yml defines collectio eggs
collections:
eggs:
output: true
A folder _eggs has a document with front matter
I can access the collection label so:
{{ site.collections[0].label }}
which returns "eggs" but not so
{{ site.eggs.label }}
which returns nothing as does this:
{{ site.eggs }}
The documentation about collections at https://jekyllrb.com/docs/collections/#liquid-attributes does not make much sense to me: "The collections are also available under site.collections, with the metadata you specified". In an issue at github the authors say, that the collections field was (silently?) dropped (https://github.com/jekyll/jekyll/issues/4392).
I am currently evaluating Jekyll and this causes doubts where it stable, has up to date docs and has other pitfalls ahead.
Do I misunderstand the docs? Why does the above access to collection metadata not work.
The metadata of each collection is available with site.collections, that means, it will return an array of the collections with its metadata.
If one access a collection directly, like site.eggs, there won't be metadata available, but an array of all the collection files, i.e. all the files in the _eggs folder.
example
To display the contents of site.eggs you can iterate over each file, consider having the following file in /_eggs/item.yml
---
title: "Jekyll is awesome"
---
Then you can display it in /index.yml like:
{% for egg in site.eggs %}
{{egg.title}}
{% endfor %}
Output:
Jekyll is awesome
I have started a Jekyll based blog with a theme jekyll-now. I am new to Jekyll and how it operates (especially Liquid). I understand that new posts need to be as follows: 2014-10-04-Hello-World.md. But I really don't understand how I could order these.
My first thought is that they order by date so two posts on the same date would order randomly. But is this not the case? Is there a way of ordering posts chronologically. OR at least having a post archive page?
There is an example in the official Jekyll documentation how to create a basic post archive page:
Displaying an index of posts
Bonus: For a prettier archive page (grouped by year or year/month), see this answer.
You're right, I can't find anything in the docs where it says how the posts are ordered, but in fact Jekyll does order them chronologically, with the most recent post first (you can see this if you try the examples I linked above).
To sort them the other way (the oldest post first), you can use the reversed keyword, according to the Liquid documentation:
{% for post in site.posts reversed %}
However, I don't know how two posts on the same date are ordered, because I don't write that much posts, so I never had that problem :-)
You have to try that yourself.
Just faced the same problem and solved with this solution:
https://groups.google.com/forum/#!topic/jekyll-rb/8QCIzevauSU
Add a date field to the YAML Front Matter of a post, like so:
date: 2010-09-15 14:40:45
e.g. if you have 2 posts on 2014/12/31, you can add date: 2014-12-31 00:30:00 to latest_post.md, and date: 2014-12-31 00:10:00 to older_post.md.
You can add time zone (e.g. date: 2014-12-31 00:10:00 +08:00) if needed
I want to document my struggle into this post so it may help other users. You need to do two changes:
Open your posts and add weight. e.g., weight:100
Open your html file for the menu where you want the sorted posts. For Java/J2EE menu I have java.html file at the root path of my project.
Then, add the {% assign pages_list = pages_list | sort:"weight" %} line as shown in the below code. This will sort by weight.
{% for category in site.categories %}
{% if category[0] contains 'java' %}
<h3 id="{{ category[0] }}-ref">{{ category[0] | join: "/" }}</h3>
<ul>
{% assign pages_list = category[1] %}
{% assign pages_list = pages_list | sort:"weight" %}
{% include JB/pages_list %}
</ul>
{% endif %}
{% endfor %}
Jekyll just string-compares post paths when sorting, which is why the date format is year-month-day. Posts are internally also collections and you can see the sorting being invoked in reader.rb:
# Sorts posts, pages, and static files.
def sort_files!
site.collections.each_value { |c| c.docs.sort! }
site.pages.sort_by!(&:name)
site.static_files.sort_by!(&:relative_path)
end
So it's using generic ruby methods to sort and implements the comparator in document.rb:
# Compare this document against another document.
# Comparison is a comparison between the 2 paths of the documents.
#
# Returns -1, 0, +1 or nil depending on whether this doc's path is less than,
# equal or greater than the other doc's path. See String#<=> for more details.
def <=>(other)
return nil unless other.respond_to?(:data)
cmp = data["date"] <=> other.data["date"]
cmp = path <=> other.path if cmp.nil? || cmp.zero?
cmp
end
Which means it first compares the dates and only checks the text if needed.
The date is special only if it somehow wasn't found (no metadata). For drafts it falls back to the file modification time, for the rest to the site time.
So if you want to force a different ordering of posts from the same day, craft the start of the title in the filename to alphabetically sort first. Eg. 2020-01-01-a.md will come after 2020-01-01-b.md if you're listing posts in descending order.