jekyll: enable sitemap and local search for dev environment

Signed-off-by: CrazyMax <crazy-max@users.noreply.github.com>

.github/workflows/publish.yml

@@ -18,17 +18,20 @@ jobs:
           JEKYLL_ENV=development
           DOCS_AWS_REGION=us-east-1
           if [ "${{ github.ref }}" = "refs/heads/master" ]; then
+            DOCS_URL="https://docs-stage.docker.com"
             DOCS_S3_BUCKET="docs.docker.com-stage-us-east-1"
             DOCS_S3_CONFIG="_website-config-docs-stage.json"
             DOCS_LAMBDA_FUNCTION_CACHE="arn:aws:lambda:us-east-1:710015040892:function:docs-stage-cache-invalidator"
             DOCS_SLACK_MSG="Successfully promoted docs-stage from master. https://docs-stage.docker.com/"
           elif [ "${{ github.ref }}" = "refs/heads/published" ]; then
             JEKYLL_ENV=production
+            DOCS_URL="https://docs.docker.com"
             DOCS_S3_BUCKET="docs.docker.com-us-east-1"
             DOCS_S3_CONFIG="_website-config-docs.json"
             DOCS_LAMBDA_FUNCTION_CACHE="arn:aws:lambda:us-east-1:710015040892:function:docs-cache-invalidator"
             DOCS_SLACK_MSG="Successfully published docs. https://docs.docker.com/"
           elif [ "${{ github.ref }}" = "refs/heads/lab" ]; then
+            DOCS_URL="https://docs-lab.docker.com"
             DOCS_S3_BUCKET=""
             DOCS_LAMBDA_FUNCTION_CACHE=""
             DOCS_S3_CONFIG="_website-config-docs-lab.json"
@@ -37,6 +40,7 @@ jobs:
             exit 1
           fi
           echo "JEKYLL_ENV=$JEKYLL_ENV" >> $GITHUB_ENV
+          echo "DOCS_URL=$DOCS_URL" >> $GITHUB_ENV
           echo "DOCS_AWS_REGION=$DOCS_AWS_REGION" >> $GITHUB_ENV
           echo "DOCS_S3_BUCKET=$DOCS_S3_BUCKET" >> $GITHUB_ENV
           echo "DOCS_S3_CONFIG=$DOCS_S3_CONFIG" >> $GITHUB_ENV

Dockerfile

@@ -10,7 +10,7 @@ ARG RUBY_VERSION=2.7.6
 ARG BUNDLER_VERSION=2.3.13
 
 ARG JEKYLL_ENV=development
-ARG DOMAIN=docs.docker.com
+ARG DOCS_URL=http://localhost:4000
 
 # Base stage for building
 FROM ruby:${RUBY_VERSION}-alpine AS base
@@ -44,27 +44,18 @@ COPY --from=vendored /out /
 # After building with jekyll, fix up some links
 FROM gem AS generate
 ARG JEKYLL_ENV
-ARG DOMAIN
+ARG DOCS_URL
 ENV TARGET=/out
 RUN --mount=type=bind,target=.,rw \
   --mount=type=cache,target=/src/.jekyll-cache <<EOT
 set -eu
-if [ "${JEKYLL_ENV}" = "production" ]; then
-  (
-    set -x
-    bundle exec jekyll build --profile -d ${TARGET} --config _config.yml,_config_production.yml
-    sed -i "s#<loc>/#<loc>https://${DOMAIN}/#" "${TARGET}/sitemap.xml"
-  )
-else
-  (
-    set -x
-    bundle exec jekyll build --trace --profile -d ${TARGET}
-    mkdir -p ${TARGET}/js
-    echo '[]' > ${TARGET}/js/metadata.json
-  )
-fi
+CONFIG_FILES=_config.yml$([ "$JEKYLL_ENV" = "production" ] && echo ",_config_production.yml" || true)
+(
+  set -x
+  bundle exec jekyll build --profile -d ${TARGET} --config ${CONFIG_FILES}
+)
 find ${TARGET} -type f -name '*.html' | while read i; do
-  sed -i "s#\(<a[^>]* href=\"\)https://${DOMAIN}/#\1/#g" "$i"
+  sed -i "s#\(<a[^>]* href=\"\)${DOCS_URL}/#\1/#g" "$i"
 done
 EOT
 

_config.yml

@@ -30,7 +30,6 @@ exclude:
   - datacenter
   - docker-bake.hcl
   - ee
-  - js/metadata.json
   - LICENSE
   - Makefile
   - README.md
@@ -67,12 +66,15 @@ compose_switch_version: "1.0.4"
 # which we show a badge (currently: API v1.40, or "Docker 19.03").
 min_api_threshold: 1.40
 
+# Enable search autocompletion (requires metadata.json to be generated)
+local_search: true
+
 # List of plugins to enable for local development builds. Mostly the same as
-# for production, but without the "jekyll-sitemap" plugin, which is not needed
-# for previewing, and excluding saves some time to build
+# for production.
 plugins:
   - jekyll-redirect-from
   - jekyll-relative-links
+  - jekyll-sitemap
 
 # Assets
 #
@@ -87,10 +89,7 @@ sass:
 # Set default options / metadata for some paths.
 #
 # Setting options here prevents having to repeat the same option in front-matter
-# on every page. Avoid using wildcards, such as "path: engine/api/v1.*", as
-# limitations in Jekyll cause those to introduce a _severe_ impact on build-time,
-# affecting generation of (e.g.) sitemap.xml and metadata.json, resulting in the
-# total build to take 60 seconds longer to build (!).
+# on every page.
 defaults:
   # Default one for development builds (local and PR previews)
   # sitemap is disabled here but not for production in _config_production.yml
@@ -99,7 +98,6 @@ defaults:
       type: "pages"
     values:
       layout: docs
-      sitemap: false
       toc_min: 2
       toc_max: 4
 

_config_production.yml

@@ -17,20 +17,12 @@ exclude:
 google_analytics: GTM-WL2QLG5
 polldaddy_id:     8453675
 
-# Enable search autocompletion (requires metadata.json to be generated)
-local_search: true
-
-plugins:
-  - jekyll-redirect-from
-  - jekyll-relative-links
-  - jekyll-sitemap
-
 # Assets
 #
 # For production/deploy, we build css with the "compressed" format, to produce
 # smaller files.
 sass:
-  style:    compressed
+  style: compressed
 
 collections:
   samples:
@@ -39,10 +31,7 @@ collections:
 # Set default options / metadata for some paths.
 #
 # Setting options here prevents having to repeat the same option in front-matter
-# on every page. Avoid using wildcards, such as "path: engine/api/v1.*", as
-# limitations in Jekyll cause those to introduce a _severe_ impact on build-time,
-# affecting generation of (e.g.) sitemap.xml and metadata.json, resulting in the
-# total build to take 60 seconds longer to build (!).
+# on every page.
 #
 # The list below is used for *production* deploys, and overrides the one defined
 # in "_config.yml", which is used for local builds and pull-request previews.

_plugins/update_sitemap.rb

@@ -0,0 +1,19 @@
+Jekyll::Hooks.register :site, :post_write do |site|
+  beginning_time = Time.now
+  Jekyll.logger.info "Starting plugin update_sitemap.rb..."
+
+  sitemap_path = File.join(site.dest, 'sitemap.xml')
+  # DEPLOY_URL is from Netlify for preview of sitemap on PR
+  # https://docs.netlify.com/configure-builds/environment-variables/#deploy-urls-and-metadata
+  docs_url = ENV['DEPLOY_URL'] || ENV['DOCS_URL'] || 'http://localhost:4000'
+
+  if File.exist?(sitemap_path)
+    sitemap_file = File.read(sitemap_path)
+    replace = sitemap_file.gsub!("<loc>/", "<loc>#{docs_url}/")
+    Jekyll.logger.info "  Replacing '<loc>/' with '<loc>#{docs_url}/' in #{sitemap_path}"
+    File.open(sitemap_path, "w") { |file| file.puts replace }
+  end
+
+  end_time = Time.now
+  Jekyll.logger.info "done in #{(end_time - beginning_time)} seconds"
+end

docker-bake.hcl

@@ -1,6 +1,9 @@
 variable "JEKYLL_ENV" {
   default = "development"
 }
+variable "DOCS_URL" {
+  default = "http://localhost:4000"
+}
 variable "DOCS_SITE_DIR" {
   default = "_site"
 }
@@ -8,6 +11,7 @@ variable "DOCS_SITE_DIR" {
 target "_common" {
   args = {
     JEKYLL_ENV = JEKYLL_ENV
+    DOCS_URL = DOCS_URL
   }
   no-cache-filter = ["generate"]
 }

docker-compose.yml

@@ -10,6 +10,7 @@ services:
     build:
       args:
         - JEKYLL_ENV
+        - DOCS_URL
       context: .
     image: docs/docstage
     ports: