FIX: Integrate BBCode through the native Markdown pipeline

INTEGRATION_NOTES.md

@@ -0,0 +1,101 @@
+# Discourse integration notes
+
+## Rendering model
+
+BBCode is registered through Discourse's Markdown plugin pipeline, using the
+same bundled BBob parser for server cooking and browser previews. A markdown-it
+core rule runs before normalization and detects whether the source needs custom
+BBCode processing. The `bbcode_enabled` setting and Markdown feature overrides
+control whether that rule is registered, including restricted cooks such as chat.
+
+Posts without custom BBCode retain native Markdown paragraphs, headings, lists,
+tables, whitespace, and code blocks. Discourse's `[b]`, `[i]`, `[u]`, and `[s]`
+tags also stay native on that path. BBCode shown only inside native Markdown
+code examples does not enable layout processing.
+
+For a post containing custom BBCode, BBob processes the source before Markdown.
+That whole post retains the plugin's legacy whitespace conventions: paragraph
+wrappers are suppressed and indented layout markup is not treated as a native
+indented code block. These differences apply to surrounding Markdown too; this
+implementation does not isolate processing to individual BBCode spans. Core
+inline tags nested in those posts use the BBob preset handlers.
+
+A renderer extension restores content hoisted by the parser, including code
+examples and custom styles. Restored code text is escaped, so HTML inside
+`[code]` or `[plain]` appears as source. Allowed HTML outside code examples still
+uses Discourse's normal rendering and sanitization. The integration does not
+replace Discourse's engine initialization or bypass its sanitizer.
+
+## Imported source
+
+- Use native Markdown lists and tables. This change adds no XenForo `[list]`,
+  `[*]`, `[table]`, `[tr]`, `[td]`, `[th]`, or `[tf]` handlers. Convert unsupported
+  imported syntax while preserving its contents; inspect complex tables individually.
+- Markdown headings remain active inside layout tags such as `[div]` and
+  `[nobr]`. Escape a line's leading hash as `\# label` when it is literal text.
+  Rebaking cannot determine the author's original intent.
+- This change adds no automatic repair of missing or misnested closing tags.
+  Malformed source can lose content during parsing; correct it before relying
+  on a rebake to reproduce the original layout.
+
+Use the composer's Markdown mode for BBCode. This change does not add a rich-text
+BBCode editor. External fonts, images, and optional Font Awesome kits still need
+available resources and their normal configuration.
+
+## Existing safeguards and behavior
+
+The current CSS containment rules, cross-process parser reset, spoiler fix, and
+rehosting of external image URLs used in BBCode CSS are retained. Initializers
+for fonts, icons, and BBCode highlighting respect the plugin's enabled setting.
+
+Containment and Discourse's sanitizer serve different purposes; this integration
+does not establish a sandbox for arbitrary user HTML, CSS, or scripts. BBScript
+is optional and remains disabled by default. Expanding that interpreter or
+introducing unrestricted HTML requires a separate security review.
+
+## Deployment and rebaking
+
+1. After changing `bbcode-src`, rebuild and include the generated bundle and
+   source map:
+
+   ```sh
+   pnpm install --frozen-lockfile
+   pnpm build
+   ```
+
+2. Deploy the full plugin. Restart Discourse web processes and Sidekiq so both
+   use the new renderer. Restart the frontend development build when adding or
+   removing plugin modules.
+3. Correct unsupported or malformed imported source, then compare representative
+   posts in the composer preview and after saving.
+4. Rebuild saved posts on the destination server from its Discourse environment:
+
+   ```sh
+   RAILS_ENV=production bundle exec rake posts:rebake
+   ```
+
+   For imported posts, follow the
+   [importer's verified rebake instructions](https://github.com/RpNation/discourse_xf_importer#readme),
+   using a fresh run ID after renderer changes. A previously completed checkpoint
+   does not establish that posts have been rebuilt with the new renderer.
+5. Check failed post IDs and Sidekiq retry/dead jobs, allow background processing
+   to finish, and inspect representative layouts. Command completion does not
+   prove that every imported design renders faithfully.
+
+Run full imports and site-wide rebakes on the migration servers, not on the
+local development computer.
+
+## Focused validation
+
+Inside a Discourse checkout with this plugin installed and assets built:
+
+```sh
+LOAD_PLUGINS=1 bin/rspec plugins/bbcode/spec/lib/pretty_text_spec.rb
+bin/qunit --target bbcode --filter BBCode
+```
+
+Check native Markdown and code examples, mixed BBCode layouts, enabled/disabled
+settings, sanitization, and preview/save parity. Run lint for changed files and
+review desktop and mobile layouts. Synthetic tests and browser emulation do not
+establish that all historical designs or Safari/iOS render correctly; retain a
+visual review of representative imported posts after source conversion.

README.md

@@ -4,9 +4,18 @@
 
 RpNation's Official BBCode Implementation for Discourse
 
-See: [https://www.rpnation.com]
+See [RpNation](https://www.rpnation.com).
 
-The goal of this repo and plugin is to provide users with the BBCode suite that they have grown accustomed to when it comes to using our site before our migration to Discourse and make sure that old posts rebake correctly. Above in the chart is marked our status on each BBCode which will hopefully co-exist in tandom even with the markdown/htlm versions provided in the box experience by the Discourse Software.
+This plugin adds RpNation's custom BBCode to Discourse while using native
+Discourse formatting where it already exists. Use Markdown for lists and tables,
+and Discourse's own inline bold, italic, underline, and strikethrough tags.
+Custom layout tags remain available for designs that need them.
+
+Imported XenForo posts may need source conversion before rebaking: this plugin
+does not add XenForo list/table syntax or automatically repair malformed tags.
+Markdown headings also work inside layout tags; write `\# label` when the hash
+should be literal. See the [integration notes](INTEGRATION_NOTES.md) for migration
+and rendering boundaries.
 
 ## Features/Planned
 
@@ -52,7 +61,8 @@ The goal of this repo and plugin is to provide users with the BBCode suite that
 - [x] Sides
 - [x] Tabs
 - [x] Accordions
-- [x] ~~Tables~~ now using markdown tables
+- [x] Native Discourse Markdown tables
+- [x] Native Discourse numbered and bulleted lists
 - [x] Center Block
 - [x] Background
 - [x] Border
@@ -105,6 +115,24 @@ For more, see the [Discourse Docker Guide](https://meta.discourse.org/docs?topic
 
 ## Architecture
 
-The architecture of this project is for all BBCode Parser related code to be contained in `/bbcode-src`, which would then be minified into a module and added to the appropriate location in `/assets/javascripts` to be used by the discourse plugin proper. This is to work around the weird way discourse requires libraries to be loaded in. There will be a Rollup config and github action setup to automate minifying and moving the module.
-
-Honestly, if anyone has a better solution, please send help.
+Parser and tag implementations live in `bbcode-src`. Run
+`pnpm install --frozen-lockfile` and `pnpm build` to regenerate the checked-in
+parser bundle and source map after source changes. Install the full plugin in
+Discourse's `plugins/bbcode` directory.
+
+The same parser bundle serves server cooking and browser previews. A registered
+Markdown plugin uses a markdown-it core rule to detect custom BBCode and process
+it through BBob. Ordinary Markdown, core inline formatting, and Markdown code
+examples containing BBCode keep native behavior when no custom BBCode occurs
+outside those examples.
+
+Processing remains per post: a post containing custom BBCode uses the plugin's
+legacy whitespace and paragraph behavior throughout the post, including its
+surrounding Markdown. This is not isolation of each BBCode span. Rendered output
+still passes through Discourse's sanitizer.
+
+Deploying a parser change does not refresh saved cooked HTML. Restart Discourse
+web processes and Sidekiq, then follow the
+[deployment and rebake instructions](INTEGRATION_NOTES.md#deployment-and-rebaking)
+on the destination server. Existing CSS containment, parser resets, and CSS
+image rehosting remain in place. BBScript remains optional and disabled by default.

assets/javascripts/discourse/api-initializers/fontawesome.js

@@ -5,7 +5,7 @@ const postsMissingFontAwesome = [];
 
 export default apiInitializer((api) => {
   const siteSettings = api.container.lookup("service:site-settings");
-  if (!siteSettings.fontawesome_kit_url) {
+  if (!siteSettings.bbcode_enabled || !siteSettings.fontawesome_kit_url) {
     return;
   }
   window.FontAwesomeConfig = {

assets/javascripts/discourse/api-initializers/google-font.js

@@ -10,7 +10,7 @@ import { apiInitializer } from "discourse/lib/api";
 function addGoogleFont(post) {
   // Cleans up post if we're editing
   const priorLinks = post.querySelectorAll("link[data-rendered-gfont]");
-  priorLinks.forEach((oldLink) => post.removeChild(oldLink));
+  priorLinks.forEach((oldLink) => oldLink.remove());
 
   const elements = post.querySelectorAll("[data-font]");
   if (!elements.length) {
@@ -22,7 +22,7 @@ function addGoogleFont(post) {
     const data = e.getAttribute("data-font");
     if (
       !gFonts.includes(data) &&
-      data.startsWith("https://fonts.googleapis.com")
+      data.startsWith("https://fonts.googleapis.com/css2?")
     ) {
       frag.appendChild(linkBuilder(data));
       gFonts.push(data);
@@ -46,6 +46,10 @@ function linkBuilder(data) {
 }
 
 export default apiInitializer((api) => {
+  if (!api.container.lookup("service:site-settings").bbcode_enabled) {
+    return;
+  }
+
   api.decorateCookedElement((elem) => addGoogleFont(elem), {
     id: "add google font",
   });

assets/javascripts/discourse/api-initializers/highlight.js

@@ -340,6 +340,10 @@ function markdownHighlight(hljs) {
 }
 
 export default apiInitializer((api) => {
+  if (!api.container.lookup("service:site-settings").bbcode_enabled) {
+    return;
+  }
+
   api.registerHighlightJSLanguage("bbcode", bbcodeHighlight);
   api.registerHighlightJSLanguage("markdown-bbcode", markdownHighlight);
 });