Restructure CI build to only fail for changed pages

Author: WasabiFan

_config.yml

@@ -10,6 +10,7 @@ exclude:
   - cibuild.sh
   - .travis.yml
   - maintainer-notes.md
+  - check-build-html.rb
 
 gems:
   - jekyll-mentions

check-build-html.rb

@@ -0,0 +1,81 @@
+require "jekyll"
+require "html-proofer"
+
+$site_source_dir = ARGV[0]
+$site_dest_dir = ARGV[1]
+$target_range = ARGV[2]
+
+$only_proof_subset = !!$target_range && !$target_range.empty?
+
+$proofer_runner = HTMLProofer::Runner.new([$site_dest_dir], {
+    :type => :directory,
+    :check_html => true,
+    :allow_hash_href => true,
+    :url_ignore => [
+        /example\.com/,
+        /https:\/\/github\.com\/myuser\/myrepo/,
+        /robosnap\.net/,
+        /https:\/\/github.com\/ev3dev\/ev3dev\.github\.io\/edit\/.*/,
+        /warmcat\.com/,
+        /robosnap\.net/,
+        /01\.org/
+    ]
+})
+
+begin
+    $proofer_runner.run
+rescue => e
+    puts e.message
+end
+
+failures = $proofer_runner.instance_variable_get :@failures
+
+changed_dest_files = []
+if $only_proof_subset
+    $site = Jekyll::Site.new(Jekyll.configuration({
+        "source" => $site_source_dir + "/",
+        "config" => $site_source_dir + "/_config.yml",
+        "destination" => $site_dest_dir,
+        "safe" => true
+    }))
+    $site.read
+
+    $jekyll_file_map = {}
+    def save_entity_mapping(jekyll_entity)
+        current_dest_file = File.expand_path(jekyll_entity.destination $site.dest)
+        $jekyll_file_map[File.expand_path(jekyll_entity.path)] = current_dest_file
+    end
+
+    $site.pages.each do |page|
+        save_entity_mapping page
+    end
+
+    $site.posts.docs.each do |post|
+        save_entity_mapping post
+    end
+
+    $site.static_files.each do |file|
+        save_entity_mapping file
+    end
+
+    changed_source_files = `git -C "#{$site_source_dir}" diff --name-only #{$target_range} --`.split("\n")
+    if $?.success?
+        changed_dest_files = changed_source_files.map { |file| $jekyll_file_map[File.expand_path(file)] }.compact
+    else
+        $only_proof_subset = false
+    end
+    
+end
+
+fatal_failures = failures.select { |failure| !$only_proof_subset || (changed_dest_files.include? File.expand_path(failure.path)) }
+
+if fatal_failures.empty?
+    puts "No fatal failures found. There may be non-fatal failures printed above."
+    exit 0
+else
+    puts "The following fatal failures were found:"
+    fatal_failures.each do |failure|
+        puts "\t" + failure.to_s
+    end
+    exit 1
+end

cibuild.sh

@@ -26,28 +26,21 @@ fi
 echo "Building site ------------------------------------"
 bundle exec jekyll build --trace
 
-# credit: code snippet borrowed from jekyllrb.com website source
-IGNORE_HREFS=$(ruby -e 'puts %w{
-    example.com
-    https:\/\/github\.com\/myuser\/myrepo
-    .*revolds-whitepaper\.pdf
-    https:\/\/github.com\/ev3dev\/ev3dev\.github\.io\/edit\/.*
-    warmcat.com
-    robosnap.net
-    01.org
-    alldatasheet.com
-}.map{|h| "/#{h}/"}.join(",")')
-
-# Explanation of ignored sites:
-# - example.com and github.com/myuser/myrepo are fake/example links
-# - The edit on github pages don't exist when you create a page, so ignoring them.
-#   They are automatically generated anyway.
-#  -revolds-whitepaper.pdf causes problems on travis but is really a good link (for now)
-# - warmcat.com works locally but fails on travis for some reason related to ssl
-
 echo "Validating HTML ----------------------------------"
+
+SOURCE_DIR="$(readlink -f .)"
+DEST_DIR="."
+
+if [[ -z $TRAVIS || ($TRAVIS = true && $TRAVIS_EVENT_TYPE = "cron") ]]; then
+    COMMIT_RESTRICTION_RANGE=""
+else
+    COMMIT_RESTRICTION_RANGE="$TRAVIS_COMMIT_RANGE"
+fi
+
+echo "Commit range target: $COMMIT_RESTRICTION_RANGE"
+
 # We want to use the publish script so that we can implement other transformations in the future
-ruby publish.rb --no-fix-links --test "htmlproofer ./ --url-ignore $IGNORE_HREFS --check-html --allow-hash-href"
+ruby publish.rb --no-fix-links --test "ruby $(readlink -f check-build-html.rb) '$SOURCE_DIR' '$DEST_DIR' $COMMIT_RESTRICTION_RANGE"
 
 # If the site build succeeded but we found BOMs, we want to fail the build
 if [ $FOUND_BOM == true ]