openstax_content 1.2.0 → 1.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 606de08c8e8a83ca196e43eef4d8159f4bddd9d94bf5a357af6e5e8109ed57ad
-  data.tar.gz: 4ef85c1b6bba83ef5a2605d76a8fc76fd3ce63dc80d84a3030a2f1943a56251d
+  metadata.gz: 640f5b30f23af9597543e0e6239b3def597b456ae05b85c9657e6380d5ad6623
+  data.tar.gz: ba19461ea0e3700f2f8c752c4dcf2fe5808bf8a798b22707fec47d9857f081c8
 SHA512:
-  metadata.gz: 6f994e45ad7f63e9d2ac8e2e143ef6b23a2688fee58c7de54a230cda84723bfa43eb1febeade0687e007d3453d1c970f7c4c6dad04cd75cee5ff0b9bc01003b2
-  data.tar.gz: 345ec56a1b9aecf201612f67ee3b283f22928afba02ec1ddeb2ad9f6e4243ab95f33049bb1f7892b570515538588c3a97859e4aa009f5f24038bf57fcf6aff0b
+  metadata.gz: a5589b01cfa98a740916cd47274171ada0265429960ca68657124f1d0c6c98e551542b3027db5278845e3ef10b9191d54b74d84cf4fd8db7355c28f2e3722d5e
+  data.tar.gz: 0e02160d69f9789030a33bd0f84da38eb0922bf3bb06c54fcc6740bdb7649ae1bbd72c62f130ce3e49947f73a711d477380c3e6eb36fc2b40c6a10ba0b344d58

data/README.md

@@ -26,45 +26,11 @@ s3_access_key_id and s3_secret_access_key are optional (you can use AWS instance
 
 ## Usage
 
-### Approved Book List (to get approved books and approved versions)
+### Approved Book List (to get books and page slugs)
 ```rb
 abl = OpenStax::Content::Abl.new
-approved_books = abl.approved_books
-approved_versions = abl.approved_versions
-```
-
-### S3 Bucket Listing (to get latest archive and book versions)
-```rb
-s3 = OpenStax::Content::S3.new
-if s3.bucket_configured?
-  latest_archive_version = s3.ls.last
-  latest_book_ids = s3.ls latest_archive_version
-  chosen_book = latest_book_ids.sample
-  book_uuid, book_version = chosen_book.split('@', 2)
-  book = OpenStax::Content::Book.new(
-    archive_version: latest_archive_version, uuid: book_uuid, version: book_version
-  )
-end
-```
-
-### Archive (to create archive links, load content and get book and page slugs)
-```rb
-archive = OpenStax::Content::Archive.new latest_archive_version
-
-book_id = "#{book_uuid}@#{book_version}"
-page_id = "#{book_id}:#{page_uuid}"
-
-book_url = archive.url_for book_id
-page_url = archive.url_for page_id
-
-book_json = archive.fetch book_id
-page_json = archive.fetch page_id
-
-book_hash = archive.json book_id
-page_hash = archive.json page_id
-
-book_slug = archive.slug book_id # or book_uuid
-page_slug = archive.slug page_id # or "#{book_uuid}:#{page_uuid}"
+books = abl.books
+slugs = abl.slugs_by_page_uuid
 ```
 
 ### Fragment Splitter (to split pages and create interactive readings)

data/lib/openstax/content/abl.rb

@@ -2,6 +2,10 @@ require_relative 'archive'
 require_relative 'book'
 
 class OpenStax::Content::Abl
+  # Check back this many archive versions
+  # If there are more than this number of archive versions still building, errors will happen
+  DEFAULT_MAX_ARCHIVE_ATTEMPTS = 5
+
   def initialize(url: nil)
     @url = url
   end
@@ -14,68 +18,85 @@ class OpenStax::Content::Abl
     @body_string ||= Faraday.get(url).body
   end
 
-  def body_hash
-    @body_hash ||= JSON.parse(body_string, symbolize_names: true)
+  def body_array
+    @body_array ||= JSON.parse(body_string, symbolize_names: true)
   end
 
   def digest
     Digest::SHA256.hexdigest body_string
   end
 
-  def latest_approved_version_by_collection_id(archive: OpenStax::Content::Archive.new)
-    {}.tap do |hash|
-      body_hash[:approved_versions].each do |version|
-        next if version[:min_code_version] > archive.version
+  def books(archive: OpenStax::Content::Archive.new)
+    body_array.filter { |book| book[:code_version] <= archive.version }.map do |book|
+      OpenStax::Content::Book.new(
+        archive: archive,
+        uuid: book[:uuid],
+        version: book[:commit_sha][0..6],
+        min_code_version: book[:code_version],
+        slug: book[:slug],
+        committed_at: book[:committed_at]
+      )
+    end
+  end
+
+  def each_book_with_previous_archive_version_fallback(max_attempts: DEFAULT_MAX_ARCHIVE_ATTEMPTS, &block)
+    raise ArgumentError, 'no block given' if block.nil?
+    raise ArgumentError, 'given block must accept the book as its first argument' if block.arity == 0
+
+    books = OpenStax::Content::Abl.new.books
+    attempt = 1
+
+    until books.empty?
+      previous_version = nil
+      previous_archive = nil
+      retry_books = []
+
+      books.each do |book|
+        begin
+          block.call book
+        rescue StandardError => exception
+          raise exception if attempt >= max_attempts
+
+          # Sometimes books in the latest archive fails to load (when the new version is still building)
+          # Retry with an earlier version of archive, if possible
+          previous_version ||= book.archive.previous_version
 
-        existing_version = hash[version[:collection_id]]
+          if previous_version.nil?
+            # There are no more earlier archive versions
+            raise exception
+          else
+            previous_archive ||= OpenStax::Content::Archive.new version: previous_version
 
-        next if !existing_version.nil? &&
-                (existing_version[:content_version].split('.').map(&:to_i) <=>
-                 version[:content_version].split('.').map(&:to_i)) >= 0
+            retry_book = OpenStax::Content::Book.new(
+              archive: previous_archive,
+              uuid: book.uuid,
+              version: book.version,
+              slug: book.slug,
+              min_code_version: book.min_code_version,
+              committed_at: book.committed_at
+            )
 
-        hash[version[:collection_id]] = version
+            # If the book requires an archive version that hasn't finished building yet, don't include it
+            retry_books << retry_book if retry_book.valid?
+          end
+        end
       end
+
+      books = retry_books
+      attempt += 1
     end
   end
 
-  def approved_books(archive: OpenStax::Content::Archive.new)
-    # Can be removed once we have no more CNX books
-    version_by_collection_id = latest_approved_version_by_collection_id(archive: archive)
-
-    body_hash[:approved_books].flat_map do |approved_book|
-      if approved_book[:versions].nil?
-        # CNX-hosted book
-        version = version_by_collection_id[approved_book[:collection_id]]
-
-        next [] if version.nil?
-
-        approved_book[:books].map do |book|
-          OpenStax::Content::Book.new(
-            archive: archive,
-            uuid: book[:uuid],
-            version: version[:content_version].sub('1.', ''),
-            slug: book[:slug],
-            style: approved_book[:style]
-          )
-        end
-      else
-        # Git-hosted book
-        approved_book[:versions].flat_map do |version|
-          next [] if version[:min_code_version] > archive.version
-
-          commit_metadata = version[:commit_metadata]
-
-          commit_metadata[:books].map do |book|
-            OpenStax::Content::Book.new(
-              archive: archive,
-              uuid: book[:uuid],
-              version: version[:commit_sha][0..6],
-              slug: book[:slug],
-              style: book[:style]
-            )
-          end
+  def slugs_by_page_uuid(max_attempts: DEFAULT_MAX_ARCHIVE_ATTEMPTS)
+    @slugs_by_page_uuid ||= {}.tap do |hash|
+      each_book_with_previous_archive_version_fallback(max_attempts: max_attempts) do |book|
+        book.all_pages.each do |page|
+          hash[page.uuid] ||= []
+          hash[page.uuid] << { book: book.slug, page: page.slug }
         end
       end
+
+      hash.each { |uuid, slugs| slugs.uniq! }
     end
   end
 end

data/lib/openstax/content/book.rb

@@ -1,18 +1,24 @@
+require 'forwardable'
 require_relative 'book_part'
 
 class OpenStax::Content::Book
   extend Forwardable
 
-  attr_reader :archive, :uuid, :version, :slug, :style
+  attr_reader :archive, :uuid, :version, :slug, :min_code_version, :committed_at
 
-  def initialize(archive:, uuid:, version:, url: nil, hash: nil, slug: nil, style: nil)
+  def initialize(archive:, uuid:, version:, url: nil, hash: nil, min_code_version: nil, slug: nil, committed_at: nil)
     @archive = archive
     @uuid = uuid
     @version = version
     @url = url
     @hash = hash
+    @min_code_version = min_code_version
     @slug = slug
-    @style = style
+    @committed_at = committed_at
+  end
+
+  def valid?
+    min_code_version.nil? || min_code_version <= archive.version
   end
 
   def url
@@ -51,5 +57,5 @@ class OpenStax::Content::Book
     @root_book_part ||= OpenStax::Content::BookPart.new(hash: tree, is_root: true, book: self)
   end
 
-  def_delegator :root_book_part, :all_pages
+  def_delegators :root_book_part, :all_book_parts, :all_pages
 end

data/lib/openstax/content/book_part.rb

@@ -45,6 +45,12 @@ class OpenStax::Content::BookPart
     end
   end
 
+  def all_book_parts
+    @all_book_parts ||= parts.flat_map do |part|
+      part.is_a?(OpenStax::Content::BookPart) ? [ part ] + part.all_book_parts : []
+    end
+  end
+
   def all_pages
     @all_pages ||= parts.flat_map do |part|
       part.is_a?(OpenStax::Content::Page) ? [ part ] : part.all_pages

data/lib/openstax/content/fragment_splitter.rb

@@ -1,4 +1,5 @@
 require 'nokogiri'
+require 'ostruct'
 require_relative 'fragment/reading'
 require_relative 'custom_css'
 
@@ -76,7 +77,7 @@ class OpenStax::Content::FragmentSplitter
     args = { node: node, labels: labels }
     args[:reference_view_url] = reference_view_url \
       if fragment_class.is_a? OpenStax::Content::Fragment::Reading
-    fragment = fragment_class.new args
+    fragment = fragment_class.new **args
     fragment unless fragment.blank?
   end
 

data/lib/openstax/content/version.rb

@@ -1,5 +1,5 @@
 module OpenStax
   module Content
-    VERSION = '1.2.0'
+    VERSION = '1.5.0'
   end
 end

data/lib/openstax_content.rb

@@ -1,8 +1,8 @@
 module OpenStax
   module Content
     class << self
-      attr_accessor :abl_url, :archive_path, :bucket_name, :domain, :exercises_search_api_url,
-                    :logger, :s3_region, :s3_access_key_id, :s3_secret_access_key
+      attr_accessor :abl_url, :archive_path, :bucket_name, :domain, :exercises_search_api_url, :logger, :s3_region,
+                    :s3_access_key_id, :s3_secret_access_key
 
       def configure
         yield self

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: openstax_content
 version: !ruby/object:Gem::Version
-  version: 1.2.0
+  version: 1.5.0
 platform: ruby
 authors:
 - Dante Soares
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-04-26 00:00:00.000000000 Z
+date: 2025-05-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: aws-sdk-s3
@@ -183,7 +183,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.4
+rubygems_version: 3.5.21
 signing_key:
 specification_version: 4
 summary: Ruby bindings to read and parse the OpenStax ABL and the content archive