Script to convert markdowns

Gemfile

@@ -24,6 +24,7 @@ gem "thruster", require: false
 gem "bcrypt", "~> 3.1.7"
 gem "geared_pagination", "~> 1.2"
 gem "rqrcode"
+gem "redcarpet"
 gem "rouge"
 gem "jbuilder"
 gem "actiontext-lexical", github: "basecamp/actiontext-lexical", branch: "initial"
@@ -52,4 +53,3 @@ group :test do
   gem "webmock"
   gem "vcr"
 end
-

Gemfile.lock

@@ -26,15 +26,16 @@ GIT
 
 GIT
   remote: https://github.com/crmne/ruby_llm.git
-  revision: 8412d51e869f1ee8e5229831a9a0bb5e9b8f3847
+  revision: 87a86f70cb64e23d5b76994edf8080618f07e071
   specs:
-    ruby_llm (1.2.0)
+    ruby_llm (1.3.0rc1)
       base64
       event_stream_parser (~> 1)
-      faraday (~> 2)
-      faraday-multipart (~> 1)
-      faraday-net_http (~> 3)
-      faraday-retry (~> 2)
+      faraday (>= 1.10.0)
+      faraday-multipart (>= 1)
+      faraday-net_http (>= 1)
+      faraday-retry (>= 1)
+      marcel (~> 1.0)
       zeitwerk (~> 2)
 
 GIT
@@ -205,7 +206,7 @@ GEM
       addressable (>= 2.5.0)
     globalid (1.2.1)
       activesupport (>= 6.1)
-    hashdiff (1.1.2)
+    hashdiff (1.2.0)
     i18n (1.14.7)
       concurrent-ruby (~> 1.0)
     image_processing (1.14.0)
@@ -327,6 +328,7 @@ GEM
     rdoc (6.14.0)
       erb
       psych (>= 4.0.0)
+    redcarpet (3.6.1)
     regexp_parser (2.10.0)
     reline (0.6.1)
       io-console (~> 0.5)
@@ -479,6 +481,7 @@ DEPENDENCIES
   puma (>= 5.0)
   rails!
   rails_structured_logging!
+  redcarpet
   rouge
   rqrcode
   rubocop-rails-omakase

script/convert_markdowns_to_rich_text.rb

@@ -0,0 +1,116 @@
+#!/usr/bin/env ruby
+
+require_relative "../config/environment"
+require "redcarpet"
+require "nokogiri"
+
+class ActionText::Markdown < ApplicationRecord
+  belongs_to :record, polymorphic: true
+end
+
+class MarkdownToActionTextConverter
+  ATTACHMENT_URL_REGEX = %r{/u/(?<slug>[^\/\s\)]+)}
+
+  def initialize(html)
+    @doc = Nokogiri::HTML::DocumentFragment.parse(html)
+    @attachments = []
+  end
+
+  def convert
+    process_images
+    process_links
+    [ @doc.to_html, @attachments ]
+  end
+
+  private
+
+    def process_images
+      @doc.css("img").each do |img|
+        src = img["src"].presence
+        if src && match = src.match(ATTACHMENT_URL_REGEX)
+          if (attachment = find_attachment(match[:slug]))
+            puts "REEMPLAZO #{build_attachment_node(attachment).to_html}"
+            img.replace(build_attachment_node(attachment))
+            @attachments << attachment
+          end
+        end
+      end
+    end
+
+    def process_links
+      @doc.css("a").each do |link|
+        href = link["href"].presence
+
+        if href && match = href.match(ATTACHMENT_URL_REGEX)
+          if (attachment = find_attachment(match[:slug]))
+            link.replace(build_attachment_node(attachment))
+            @attachments << attachment
+          end
+        end
+      end
+    end
+
+    def build_attachment_node(attachment)
+      html = ActionText::Attachment.from_attachable(attachment).to_html
+      fragment = Nokogiri::HTML::DocumentFragment.parse(html)
+
+      node = fragment.at_css("action-text-attachment")
+      node["url"] = Rails.application.routes.url_helpers.rails_blob_path(attachment.blob, only_path: true)
+
+      fragment
+    end
+
+    def find_attachment(slug)
+      ActiveStorage::Attachment.find_by(slug: slug)
+    end
+end
+
+class RedcarpetRenderer
+  def self.render(markdown)
+    renderer = Redcarpet::Render::HTML.new
+    markdowner = Redcarpet::Markdown.new(renderer,
+      autolink: true,
+      tables: true,
+      fenced_code_blocks: true,
+      strikethrough: true,
+      superscript: true,
+    )
+    markdowner.render(markdown.to_s)
+  end
+end
+
+ApplicationRecord.with_each_tenant do |tenant|
+  puts "Processing tenant: #{tenant}"
+
+  ActionText::Markdown.find_each do |markdown|
+    next unless markdown.record
+
+    html = RedcarpetRenderer.render(markdown.content.to_s)
+    converter = MarkdownToActionTextConverter.new(html)
+    rich_text_html, attachments = converter.convert
+
+    rich_text = ActionText::RichText.create!(
+      name: markdown.name,
+      record: markdown.record,
+      body: rich_text_html
+    )
+
+    attachments.each do |attachment|
+      attachment.update!(record: rich_text)
+    end
+
+    puts "✓ Created rich text for #{markdown.record_type}##{markdown.record_id} (#{markdown.name})"
+  rescue => e
+    warn "✗ Failed to process markdown ##{markdown.id}: #{e.class} - #{e.message}"
+  end
+end
+
+# create_table "action_text_markdowns", force: :cascade do |t|
+#   t.text "content", default: "", null: false
+#   t.datetime "created_at", null: false
+#   t.string "name", null: false
+#   t.integer "record_id", null: false
+#   t.string "record_type", null: false
+#   t.datetime "updated_at", null: false
+#   t.index ["record_type", "record_id"], name: "index_action_text_markdowns_on_record"
+# end