stepmod_file_annotator.rb in stepmod-utils-0.3.31

- old
+ new

@@ -19,23 +19,24 @@
       def initialize(express_file:, stepmod_dir: nil)
         @express_file = express_file
         @resource_docs_cache = resource_docs_schemas(stepmod_dir)
         @stepmod_dir = stepmod_dir || Dir.pwd
         @added_bibdata = {}
+        @images_references = {}
 
         @schema_name = Expressir::Express::Parser.from_file(express_file)
                                                  .schemas
                                                  .first
                                                  .id
       end
 
       def resource_docs_schemas(stepmod_dir)
-        filepath = File.join(stepmod_dir, "data/resource_docs/*/resource.xml")
+        filepath = File.join(stepmod_dir, "data", "resource_docs", "*", "resource.xml")
 
         schemas = {}
         Dir.glob(filepath).each do |resource_docs_file|
-          match = resource_docs_file.match("data/resource_docs/([^/]+)/resource.xml")
+          match = resource_docs_file.match("data[/\]resource_docs[/\]([^/\]+)[/\]resource.xml")
           resource_docs_dir = match.captures[0]
 
           resource_docs = Nokogiri::XML(File.read(resource_docs_file)).root
           resource_docs.xpath("schema").each do |schema|
             schemas[schema["name"]] = resource_docs_dir
@@ -53,10 +54,11 @@
                                       descriptions_base)
 
         output_express = File.read(express_file)
         converted_description = ""
         base_linked = ""
+        processed_images_cache = {}
 
         if File.exist?(descriptions_file)
           descriptions = Nokogiri::XML(File.read(descriptions_file)).root
           added_resource_descriptions = {}
 
@@ -80,10 +82,14 @@
             if description.text.strip.length.positive? && resource_docs_dir.nil?
               converted_description << convert_from_description_text(
                 descriptions_file, description
               )
             end
+
+            schema_base_dir = resource_docs_cache[base_linked]
+            add_images_references(converted_description, schema_base_dir, processed_images_cache)
+
             # Add converted description from exact linked path
             if resource_docs_dir && added_resource_descriptions[description["linkend"]].nil?
               output_express << convert_from_resource_file(resource_docs_dir,
                                                            stepmod_dir, description["linkend"], descriptions_file)
               added_resource_descriptions[description["linkend"]] = true
@@ -110,19 +116,41 @@
                             )
                           else
                             converted_description
                           end
 
-        sanitize(output_express)
+        {
+          annotated_text: sanitize(output_express),
+          images_references: @images_references,
+        }
       rescue StandardError => e
         puts "[ERROR]!!! #{e.message}"
         puts e.backtrace
       end
 
       private
 
       def sanitize(file_content)
         file_content.gsub("(*)", "(`*`)")
+      end
+
+      def add_images_references(description, schema_base_dir, processed_images_cache)
+        referenced_images = description.scan(/image::(.*?)\[\]/).flatten
+
+        referenced_images.each do |referenced_image|
+          next unless schema_base_dir
+
+          image_file_path = File.join("resource_docs", schema_base_dir, referenced_image)
+          new_image_file_path = referenced_image
+
+          if processed_images_cache[new_image_file_path]
+            processed_images_cache[new_image_file_path] = true
+            next
+          end
+
+          processed_images_cache[new_image_file_path] = true
+          @images_references[image_file_path] = new_image_file_path
+        end
       end
 
       def convert_from_description_text(descriptions_file, description)
         Dir.chdir(File.dirname(descriptions_file)) do
           wrapper = "<ext_descriptions>#{description}</ext_descriptions>"