From 9e8acf257d2335c05fcd3a7aa56816f5739233a2 Mon Sep 17 00:00:00 2001
From: Hendrik Kleinwaechter <hendrik.kleinwaechter@gmail.com>
Date: Mon, 11 Dec 2023 23:36:22 +0100
Subject: [PATCH] Validate website format (#304)

* Validate website format

This validates the format of the website and throws an error if
something is fishy.

* Remove -j flag

* clean build

* Clean cache in between
---
 .github/workflows/release-book-website.yml | 32 +++++++++++-----------
 .github/workflows/test-book-website.yml    | 16 +++++------
 website/modify_build.rb                    | 17 ++++++++++++
 3 files changed, 41 insertions(+), 24 deletions(-)

diff --git a/.github/workflows/release-book-website.yml b/.github/workflows/release-book-website.yml
index 85a4600..03c3f32 100644
--- a/.github/workflows/release-book-website.yml
+++ b/.github/workflows/release-book-website.yml
@@ -49,22 +49,6 @@ jobs:
           aws_bucket: ${{ secrets.AWS_BUCKET_BOOK }}
           source_dir: book/release
           destination_dir: release
-      - name: Bake the website
-        uses: addnab/docker-run-action@v3
-        with:
-          image: ${{ env.IMAGE }}
-          options: -v ${{ github.workspace }}:/app
-          run: |
-            cd /app/book
-            make website
-      - name: Release baked website to S3
-        uses: shallwefootball/s3-upload-action@master
-        with:
-          aws_key_id: ${{ secrets.AWS_KEY_ID }}
-          aws_secret_access_key: ${{ secrets.AWS_SECRET_ACCESS_KEY}}
-          aws_bucket: ${{ secrets.AWS_BUCKET_WEBSITE }}
-          source_dir: website/static_website_html
-          destination_dir: static_html_root
       - name: Upload book Artifacts
         uses: actions/upload-artifact@v3
         with:
@@ -73,6 +57,22 @@ jobs:
             book/book_serif/book.log
             book/book_serif/book.pdf
             book/book-epub/book.epub
+      - name: Bake the website
+        uses: addnab/docker-run-action@v3
+        with:
+          image: ${{ env.IMAGE }}
+          options: -v ${{ github.workspace }}:/app
+          run: |
+            cd /app/book
+            make mrproper && make website
+      - name: Release baked website to S3
+        uses: shallwefootball/s3-upload-action@master
+        with:
+          aws_key_id: ${{ secrets.AWS_KEY_ID }}
+          aws_secret_access_key: ${{ secrets.AWS_SECRET_ACCESS_KEY}}
+          aws_bucket: ${{ secrets.AWS_BUCKET_WEBSITE }}
+          source_dir: website/static_website_html
+          destination_dir: static_html_root
       - name: Upload website Artifacts
         uses: actions/upload-artifact@v3
         with:
diff --git a/.github/workflows/test-book-website.yml b/.github/workflows/test-book-website.yml
index 8afd33f..24c9e2e 100644
--- a/.github/workflows/test-book-website.yml
+++ b/.github/workflows/test-book-website.yml
@@ -40,14 +40,6 @@ jobs:
           run: |
             cd /app/book
             make -j build_serif_pdf build_ebook
-      - name: Test building website
-        uses: addnab/docker-run-action@v3
-        with:
-          image: ${{ env.IMAGE }}
-          options: -v ${{ github.workspace }}:/app
-          run: |
-            cd /app/book
-            make -j website
       - name: Upload book Artifacts
         uses: actions/upload-artifact@v3
         with:
@@ -56,6 +48,14 @@ jobs:
             book/book_serif/book.log
             book/book_serif/book.pdf
             book/book-epub/book.epub
+      - name: Test building website
+        uses: addnab/docker-run-action@v3
+        with:
+          image: ${{ env.IMAGE }}
+          options: -v ${{ github.workspace }}:/app
+          run: |
+            cd /app/book
+            make mrproper && make website
       - name: Upload website Artifacts
         uses: actions/upload-artifact@v3
         with:
diff --git a/website/modify_build.rb b/website/modify_build.rb
index e7604d4..8f527fe 100644
--- a/website/modify_build.rb
+++ b/website/modify_build.rb
@@ -5,6 +5,8 @@ require 'nokogiri'
 # several optimisations on the HTML. Nokogiri is used to facilitate the
 # modifications.
 
+class InvalidWebsiteFormat < StandardError; end
+
 class ModifyBuild
   HOST = "https://www.the-sourdough-framework.com".freeze
 
@@ -15,6 +17,8 @@ class ModifyBuild
   def build
     build_latex_html
     create_sitemap
+  rescue InvalidWebsiteFormat => e
+    raise e
   end
 
   private
@@ -42,6 +46,7 @@ class ModifyBuild
 
   def modify_file(filename)
     orig_text = File.read(filename, encoding: "UTF-8")
+    validate_file(orig_text)
     text = fix_double_slashes(orig_text)
     text = fix_navigation_bar(text)
     text = fix_titles(text)
@@ -101,6 +106,18 @@ class ModifyBuild
     text.gsub(/\/\//, "/")
   end
 
+  # Sometimes for whatever reason the make4ht input produces files that are
+  # improperly formatted. This validator will go through the files and do a
+  # couple of basic checks to see if the files are in the format we expect. If
+  # not an exception is caused.
+  def validate_file(text)
+    doc = build_doc(text)
+    stylesheets = doc.css("link[rel='stylesheet']").map{|attr| attr["href"] }
+    has_all_styles = %w(book.css style.css).all? { |required_stylesheet| stylesheets.include?(required_stylesheet) }
+    raise InvalidWebsiteFormat.new("No style tag style.css found in the website") unless has_all_styles
+    true
+  end
+
   def fix_navigation_bar(text)
     doc = build_doc(text)
     elements = [doc.search('.chapterToc'), doc.search('.sectionToc'), doc.search('.subsectionToc')].flatten