datalog-theme 0.10.0 → 0.11.0

data/_plugins/page_dates.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require "time"
+
+module Datalog
+  # The date a page was published, when it has one. Jekyll gives a collection
+  # document without a `date:` the time of the build, so `page.date` cannot
+  # tell a dated page from an undated one, and an undated package, dataset or
+  # notebook was "published" again by every build (#411). A post's date, from
+  # its file name, and any `date:` in front matter are real.
+  module PageDates
+    module_function
+
+    # The page's date, or nil when it has none or its date is the build's.
+    def published(page, build_time)
+      date = page.respond_to?(:[]) && !page.is_a?(String) ? page["date"] : nil
+      return if date.nil? || date.to_s.strip.empty?
+
+      time = date.respond_to?(:to_time) ? date.to_time : Time.parse(date.to_s)
+      return if build_time && time.to_i == build_time.to_i
+
+      date
+    rescue ArgumentError
+      nil
+    end
+  end
+
+  module PageDateFilters
+    # `{{ page | published_date | localize_date }}`, nil for an undated page.
+    def published_date(page)
+      site = @context.registers[:site]
+      PageDates.published(page, site&.time)
+    end
+  end
+end
+
+Liquid::Template.register_filter(Datalog::PageDateFilters)

data/_plugins/references.rb

@@ -196,15 +196,29 @@ module Datalog
         raise Liquid::ArgumentError, "{% figure id=\"#{id}\" %} needs alt text for its image"
       end
 
-      src = "#{context.registers[:site].config['baseurl'].to_s.chomp('/')}#{src}" if src.start_with?("/")
+      baseurl = context.registers[:site].config["baseurl"].to_s.chomp("/")
+      src = "#{baseurl}#{src}" if src.start_with?("/")
       caption = References.inline(References.markdown(context, super))
       classes = ["datalog-figure", attributes["class"]].compact.join(" ")
 
       alt = CGI.escapeHTML(attributes["alt"])
+      dark = dark_source(attributes, baseurl)
       %(<figure class="#{CGI.escapeHTML(classes)}" id="#{id}" data-ref-target="#{id}" data-ref-kind="figure">) +
-        %(<img src="#{CGI.escapeHTML(src)}" alt="#{alt}" loading="lazy" decoding="async">) +
+        %(<img src="#{CGI.escapeHTML(src)}" alt="#{alt}" loading="lazy" decoding="async"#{dark}>) +
         %(<figcaption><span class="datalog-ref-label" data-ref-for="#{id}"></span> #{caption}</figcaption></figure>\n)
     end
+
+    # The same figure exported for a dark page, as `dark_src`. The image
+    # pipeline turns it into a <source media="(prefers-color-scheme: dark)">
+    # with the variants it built for it. The alt text stays on the one <img>:
+    # it is the same figure, and describing it twice would be wrong.
+    def dark_source(attributes, baseurl)
+      dark = attributes["dark_src"].to_s.strip
+      return "" if dark.empty?
+
+      dark = "#{baseurl}#{dark}" if dark.start_with?("/")
+      %( data-dark-src="#{CGI.escapeHTML(dark)}")
+    end
   end
 
   class TableTag < Liquid::Block

data/_plugins/related_posts.rb

@@ -0,0 +1,149 @@
+# frozen_string_literal: true
+
+module Datalog
+  module RelatedPosts
+    # Select recommendations once after Jekyll has read and normalized series.
+    # Each field has an inverted index, so a post visits matching candidates
+    # rather than comparing itself with every other post on the site.
+    class Index
+      FIELDS = %w[tags categories keywords].freeze
+      DEFAULT_WEIGHTS = { "tags" => 3.0, "categories" => 0.35, "keywords" => 1.5, "text" => 0.0 }.freeze
+
+      def initialize(site)
+        @posts = site.posts.docs
+        settings = site.config["related_posts"].is_a?(Hash) ? site.config["related_posts"] : {}
+        @limit = Integer(settings.fetch("limit", 3))
+        @min_score = Float(settings.fetch("min_score", 0.1))
+        @weights = DEFAULT_WEIGHTS.merge(string_keys(settings["weights"] || {}))
+        @weights.transform_values! { |value| Float(value) }
+        @terms = {}
+        @index = Hash.new { |hash, field| hash[field] = Hash.new { |terms, term| terms[term] = [] } }
+        @lookup = Hash.new { |hash, key| hash[key] = [] }
+        @excluded = @posts.to_h { |post| [post, post.data["related"] == false] }
+        index_posts
+      end
+
+      def assign!
+        results = @posts.to_h do |post|
+          [post, post.data["related_posts"] == false ? [] : recommendations(post)]
+        end
+        results.each { |post, recommendations| post.data["related"] = recommendations.map { |item| summary(item) } }
+      end
+
+      def recommendations(post)
+        selected = []
+        used_series = {}
+        manual = Array(post.data["related_posts"]).map { |pick| resolve(post, pick) }
+        manual.each do |candidate|
+          validate_pick!(post, candidate)
+          add(selected, used_series, candidate)
+        end
+        return selected.first(@limit) if selected.size >= @limit
+
+        ranked = scores(post).sort_by { |candidate, score| [-score, -candidate.date.to_i, candidate.url] }
+        ranked.each do |candidate, score|
+          next if score < @min_score || !eligible?(post, candidate)
+
+          add(selected, used_series, candidate)
+          break if selected.size >= @limit
+        end
+        selected
+      end
+
+      private
+
+      def index_posts
+        @posts.each do |post|
+          @terms[post] = {}
+          fields = FIELDS + (@weights["text"].positive? ? ["text"] : [])
+          fields.each do |field|
+            terms = field == "text" ? text_terms(post) : terms_for(post.data[field])
+            @terms[post][field] = terms
+            terms.each { |term| @index[field][term] << post }
+          end
+          [post.url, post.relative_path, post.data["title"]].compact.each do |key|
+            @lookup[key.to_s.strip.downcase] << post
+          end
+        end
+      end
+
+      def scores(post)
+        scores = Hash.new(0.0)
+        @terms.fetch(post).each do |field, terms|
+          weight = @weights.fetch(field)
+          next if weight.zero?
+
+          terms.each do |term|
+            matches = @index[field][term]
+            rarity = field == "tags" ? Math.log((@posts.size + 1.0) / (matches.size + 1)) + 1.0 : 1.0
+            matches.each { |candidate| scores[candidate] += weight * rarity unless candidate.equal?(post) }
+          end
+        end
+        scores
+      end
+
+      def resolve(post, pick)
+        key = pick.to_s.strip.downcase
+        matches = @lookup[key].uniq
+        return matches.first if matches.one?
+
+        problem = matches.empty? ? "matches no post" : "matches several posts; use a URL or path"
+        raise Jekyll::Errors::FatalException, "#{post.relative_path} related_posts pick #{pick.inspect} #{problem}"
+      end
+
+      def validate_pick!(post, candidate)
+        return if eligible?(post, candidate)
+
+        raise Jekyll::Errors::FatalException,
+              "#{post.relative_path} related_posts pick #{candidate.url.inspect} is the post itself, " \
+              "excluded, or already appears in its series navigation"
+      end
+
+      def eligible?(post, candidate)
+        own_series = series_id(post)
+        !candidate.equal?(post) && !@excluded[candidate] && (own_series.nil? || series_id(candidate) != own_series)
+      end
+
+      def add(selected, used_series, candidate)
+        id = series_id(candidate)
+        return if selected.include?(candidate) || (id && used_series[id])
+
+        selected << candidate
+        used_series[id] = true if id
+      end
+
+      def series_id(post)
+        value = post.data["series"]
+        value["id"] if value.is_a?(Hash)
+      end
+
+      def terms_for(value)
+        Array(value).map { |term| term.to_s.strip.downcase }.reject(&:empty?).uniq
+      end
+
+      def text_terms(post)
+        text = "#{post.data['title']} #{post.data['excerpt']}"
+        text.downcase.scan(/[[:alnum:]]{3,}/).uniq
+      end
+
+      def string_keys(value)
+        value.is_a?(Hash) ? value.transform_keys(&:to_s) : {}
+      end
+
+      def summary(post)
+        excerpt = post.data["excerpt"] || (post.excerpt if post.respond_to?(:excerpt))
+        { "title" => post.data["title"], "url" => post.url, "teaser" => post.data["teaser"],
+          "date" => post.date, "excerpt" => excerpt.to_s, "tags" => post.data["tags"] }
+      end
+    end
+  end
+
+  class RelatedPostsGenerator < Jekyll::Generator
+    safe true
+    priority :low
+
+    def generate(site)
+      RelatedPosts::Index.new(site).assign!
+    end
+  end
+end

data/_plugins/search_sections.rb

@@ -0,0 +1,110 @@
+# frozen_string_literal: true
+
+require "nokogiri"
+
+module Datalog
+  # Splits a rendered document into the passages a reader would recognise: the
+  # text under each h2 or h3, keyed by the id the heading already carries.
+  #
+  # A search result used to point at a whole page. For the long methods posts
+  # this theme is built for, that hands the reader a second search problem:
+  # find the paragraph. The headings and their ids were already in the page —
+  # kramdown writes them, the contents list links to them — and the index was
+  # the only thing that never recorded which passage sat under which (#336).
+  module SearchSections
+    HEADINGS = %w[h2 h3].freeze
+    # Their text is markup, not prose: a MathJax configuration block or a
+    # JSON-LD blob is not something anyone searches for.
+    IGNORED = %w[script style template].freeze
+    # What cannot sit in a URL fragment. A page whose Liquid has not run when
+    # the index renders can carry a heading id of "{{ group_name | slugify }}",
+    # and a link to that fragment would go nowhere; the section still indexes,
+    # pointing at the page.
+    UNUSABLE_ANCHOR = %r{[\s"'<>{}#%\\/]}
+
+    module_function
+
+    # Takes the document's own rendered HTML, not the finished page, so the
+    # headings a layout puts around the article — "About this post", "Reading
+    # notes" — cannot turn up as a section of every result.
+    def split(html)
+      sections = [section_for(nil)]
+      collect(Nokogiri::HTML5.fragment(html.to_s), sections)
+      sections.filter_map { |section| finish(section) }
+    end
+
+    # Walks in document order, opening a section at every heading. An element
+    # holding a heading somewhere inside it is walked into rather than taken
+    # whole, so a wrapper around part of the article does not swallow the
+    # boundaries within it.
+    def collect(node, sections)
+      node.children.each do |child|
+        next unless child.element? || child.text?
+        next if child.element? && IGNORED.include?(child.name)
+
+        if child.element? && HEADINGS.include?(child.name)
+          sections << section_for(child)
+        elsif child.element? && child.css(HEADINGS.join(", ")).any?
+          collect(child, sections)
+        else
+          sections.last[:parts] << prose(child)
+        end
+      end
+    end
+
+    # An element taken whole may still hold a script or a style somewhere
+    # inside it — a visualization block carries the code that draws it — and
+    # that code is not prose. Liquid's `strip_html` drops those blocks with
+    # their contents, so the flattened text of the page does not have them
+    # either.
+    def prose(node)
+      return node.text unless node.element? && node.css(IGNORED.join(", ")).any?
+
+      copy = node.dup
+      copy.css(IGNORED.join(", ")).each(&:remove)
+      copy.text
+    end
+
+    def section_for(heading)
+      {
+        title: heading ? squeeze(heading.text) : "",
+        # A heading with no id — a site that turns kramdown's auto_ids off —
+        # still indexes; its section simply links to the page.
+        anchor: heading && present(heading["id"]),
+        level: heading ? heading.name[1].to_i : 0,
+        parts: []
+      }
+    end
+
+    # The lead paragraphs before the first heading are a section of their own,
+    # untitled; so is the whole of a page that has no headings at all.
+    def finish(section)
+      content = squeeze(section[:parts].join(" "))
+      return if section[:title].empty? && content.empty?
+
+      { "title" => section[:title], "anchor" => section[:anchor],
+        "level" => section[:level], "content" => content }
+    end
+
+    def squeeze(text)
+      text.to_s.gsub(/\s+/, " ").strip
+    end
+
+    def present(value)
+      value = value.to_s.strip
+      return if value.empty? || value.match?(UNUSABLE_ANCHOR)
+
+      value
+    end
+  end
+
+  # `{{ content | markdownify | search_sections }}` in the search index, over
+  # the same HTML the flattened text is taken from.
+  module SearchSectionsFilter
+    def search_sections(html)
+      SearchSections.split(html)
+    end
+  end
+end
+
+Liquid::Template.register_filter(Datalog::SearchSectionsFilter)

data/_plugins/site_identity.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+module Datalog
+  # The site as an entity of its own, for the `WebSite` node that
+  # meta/schema.html writes on the homepage. Search engines read that node as
+  # the strongest statement of a site's name, and its `alternateName` as the
+  # other names people know it by, such as a handle (#322).
+  #
+  # Off unless `site_identity` in _config.yml asks for it, so a site that wrote
+  # its own WebSite JSON-LD does not end up with two that disagree:
+  #
+  #   site_identity: true               # the site's title is its name
+  #   site_identity:
+  #     name: Jane Doe                  # default: title
+  #     alternate_names: [janedoe]      # or alternate_name: janedoe
+  module SiteIdentity
+    module_function
+
+    # Nil when the site has not asked for it, else its name and other names.
+    def resolve(site)
+      config = Authors.value(site, "site_identity")
+      config = {} if config == true
+      return unless config.is_a?(Hash) && config["enabled"] != false
+
+      name = config["name"].to_s.strip
+      name = Authors.value(site, "title").to_s.strip if name.empty?
+      return if name.empty?
+
+      others = Authors.list(config, "alternate_names", "alternate_name")
+      { "name" => name, "alternate_names" => (others - [name]).uniq }
+    end
+  end
+
+  module SiteIdentityFilters
+    # `{% assign identity = site | site_identity %}`, nil when it is off.
+    def site_identity(_input = nil)
+      SiteIdentity.resolve(@context["site"])
+    end
+  end
+end
+
+Liquid::Template.register_filter(Datalog::SiteIdentityFilters)

data/_plugins/social_cards.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+require_relative "authors"
+require_relative "i18n"
+require_relative "image_optimizer"
+require_relative "../lib/datalog/social_cards"
+
+# Draws the share cards after every other generator, so the pages that
+# generators add are there, and the series and authors are resolved.
+class Datalog::SocialCardsGenerator < Jekyll::Generator
+  safe true
+  priority :lowest
+
+  def generate(site)
+    Datalog::SocialCards.generate(site)
+  end
+end

data/_sass/_academic-dashboard.scss

@@ -1,5 +1,6 @@
-// The academic dashboard (_includes/components/academic-dashboard.html): citation
-// metrics, bibliography export, and the submission, calendar and funding lists.
+// The academic dashboard (_includes/components/academic-dashboard.html), which
+// the demo renders at /academic/: profiles, citation metrics, the bibliography
+// export, and the submission, calendar and funding lists.
 // _components.scss loads this only when features.academic_dashboard is on.
 
 @use 'sass:map';
@@ -13,21 +14,6 @@
   gap: 2.5rem;
 }
 
-.academic-dashboard__header {
-  text-align: center;
-  max-width: 60ch;
-  margin: 0 auto;
-}
-
-.academic-dashboard__header h1 {
-  font-size: clamp(1.8rem, 3vw, 2.4rem);
-  margin-bottom: 0.5rem;
-}
-
-.academic-dashboard__header p {
-  color: var(--muted-foreground, #4a5568);
-}
-
 .academic-dashboard__profiles {
   background: var(--surface-card, #fff);
   border-radius: 1.25rem;
@@ -47,10 +33,10 @@
 }
 
 .academic-profile {
-  border: 1px solid rgba(12, 35, 64, 0.12);
+  border: 1px solid var(--color-border);
   border-radius: 1rem;
   padding: 1.5rem;
-  background: rgba(12, 35, 64, 0.03);
+  background: var(--surface-elevated);
   display: grid;
   gap: 1rem;
 }
@@ -69,7 +55,7 @@
 .academic-profile a {
   font-size: 0.875rem;
   font-weight: 500;
-  color: var(--accent, #3277f6);
+  color: var(--accent);
 }
 
 .academic-profile dl {
@@ -97,12 +83,13 @@
   gap: 1.5rem;
   border-radius: 1.25rem;
   padding: 2rem;
-  background: rgba(50, 119, 246, 0.08);
+  background: rgba($color-accent-primary, 0.08);
 }
 
 .citation-metrics__totals {
   display: grid;
   gap: 1rem;
+  margin: 0;
 }
 
 @media (min-width: 40rem) {
@@ -112,7 +99,7 @@
 }
 
 .citation-metrics .metric {
-  background: rgba(255, 255, 255, 0.8);
+  background: var(--surface-card);
   border-radius: 1rem;
   padding: 1rem;
   text-align: center;
@@ -133,7 +120,7 @@
 }
 
 .citation-metrics__details {
-  background: rgba(255, 255, 255, 0.92);
+  background: var(--surface-card);
   border-radius: 1rem;
   padding: 1rem 1.5rem;
 }
@@ -164,7 +151,7 @@
   gap: 1rem;
   padding: 0.5rem 0.75rem;
   border-radius: 0.75rem;
-  background: linear-gradient(90deg, rgba(50, 119, 246, 0.15) var(--timeline-value, 35%), transparent 0);
+  background: linear-gradient(90deg, rgba($color-accent-primary, 0.15) var(--timeline-value, 0%), transparent 0);
 }
 
 .bibliography-export {
@@ -177,9 +164,9 @@
   min-height: 16rem;
   font-family: var(--font-mono, "Fira Code", monospace);
   border-radius: 0.75rem;
-  border: 1px solid rgba(12, 35, 64, 0.2);
+  border: 1px solid var(--color-border);
   padding: 1.25rem;
-  background: rgba(12, 35, 64, 0.04);
+  background: var(--surface-elevated);
 }
 
 .bibliography-export__actions {
@@ -221,10 +208,10 @@
 .opportunity-list li,
 .mentorship-list li,
 .workflow-list li {
-  border: 1px solid rgba(12, 35, 64, 0.12);
+  border: 1px solid var(--color-border);
   border-radius: 1rem;
   padding: 1.25rem;
-  background: rgba(255, 255, 255, 0.92);
+  background: var(--surface-card);
 }
 
 .status-badge {
@@ -233,8 +220,8 @@
   gap: 0.4rem;
   padding: 0.2rem 0.65rem;
   border-radius: 999px;
-  background: rgba(50, 119, 246, 0.15);
-  color: var(--accent-foreground, #0b1d3d);
+  background: rgba($color-accent-primary, 0.15);
+  color: var(--color-accent-text);
   font-size: 0.85rem;
   font-weight: 600;
 }
@@ -255,6 +242,7 @@
   gap: 1.25rem;
 }
 
+
 @media (min-width: 48rem) {
   .networking-grid {
     grid-template-columns: repeat(2, minmax(0, 1fr));

data/_sass/_base.scss

@@ -23,6 +23,7 @@ html {
 }
 
 body {
+  color-scheme: light;
   margin: 0;
   font-family: $font-family-sans;
   color: $color-ink-900;
@@ -33,6 +34,7 @@ body {
 }
 
 body.dark-mode {
+  color-scheme: dark;
   color: $dark-ink-900;
   background-color: $dark-surface-100;
 }

data/_sass/_components.scss

@@ -143,11 +143,11 @@ body.dark-mode .btn--ghost:hover {
   transition: all 0.2s ease;
 
   &--primary {
-    background: var(--color-accent, #3b82f6);
-    color: white;
+    background: var(--color-accent);
+    color: var(--accent-foreground, white);
 
     &:hover {
-      background: var(--color-accent-dark, #2563eb);
+      background: var(--color-accent-dark);
     }
   }
 
@@ -968,11 +968,15 @@ body.dark-mode {
     box-shadow: 0 20px 48px rgba($dark-ink-900, 0.32);
   }
 
+  .notebook-article__label {
+    color: $dark-accent-primary;
+  }
+
   .notebook-article__summary,
   .notebook-article__meta,
   .notebook-sidebar__metadata dd,
   .notebook-sidebar__stats li {
-    color: rgba($dark-ink-200, 0.9);
+    color: $dark-ink-500;
   }
 
   .notebook-article__meta span,
@@ -986,34 +990,28 @@ body.dark-mode {
   }
 
   .notebook-cell {
-    background: rgba($dark-accent-primary, 0.14);
+    background: rgba($dark-accent-primary, 0.22);
     box-shadow: inset 0 0 0 1px rgba($dark-ink-900, 0.35);
   }
 
+  // Code and outputs sit a shade darker than their cell, as in the light
+  // theme. $dark-ink-900 is the lightest ink, so it lightened them instead.
   .notebook-cell__code {
-    background: rgba($dark-ink-900, 0.45);
+    background: rgba($dark-surface-100, 0.45);
   }
 
   .notebook-output {
-    background: rgba($dark-ink-900, 0.35);
+    background: rgba($dark-surface-100, 0.35);
     color: $dark-ink-050;
   }
 
   .notebook-output__warning {
-    color: rgba($dark-ink-100, 0.8);
-  }
-
-  .notebook-output--widget {
-    color: $dark-accent-secondary;
+    color: $dark-ink-300;
   }
 
   .notebook-output-plotly,
   .notebook-output-vega {
-    background: rgba($dark-ink-900, 0.32);
-  }
-
-  .notebook-cell {
-    background: rgba($dark-accent-primary, 0.22);
+    background: rgba($dark-surface-100, 0.32);
   }
 
   .reference-list li::before,
@@ -1112,7 +1110,7 @@ body.dark-mode {
 
   div {
     padding: 0.75rem;
-    background: rgba(50, 119, 246, 0.08);
+    background: rgba($color-accent-primary, 0.08);
     border-radius: 0.75rem;
     text-align: center;
   }
@@ -1134,7 +1132,7 @@ body.dark-mode {
 
   body.dark-mode & {
     div {
-      background: rgba(50, 119, 246, 0.2);
+      background: rgba($dark-accent-primary, 0.2);
     }
 
     dt {
@@ -1168,7 +1166,7 @@ body.dark-mode {
     align-items: center;
     padding: 0.25rem 0.75rem;
     border-radius: 999px;
-    background: rgba(50, 119, 246, 0.1);
+    background: rgba($color-accent-primary, 0.1);
     color: inherit;
     text-decoration: none;
     font-size: 0.8125rem;
@@ -1176,16 +1174,16 @@ body.dark-mode {
 
   a:hover,
   a:focus-visible {
-    background: rgba(50, 119, 246, 0.2);
+    background: rgba($color-accent-primary, 0.2);
   }
 
   body.dark-mode & a {
-    background: rgba(50, 119, 246, 0.25);
+    background: rgba($dark-accent-primary, 0.25);
   }
 
   body.dark-mode & a:hover,
   body.dark-mode & a:focus-visible {
-    background: rgba(50, 119, 246, 0.35);
+    background: rgba($dark-accent-primary, 0.35);
   }
 }