Update check-new-jdk workflow for JSON-first architecture

- Create individual category/slug.json files instead of editing data/snippets.json
- No longer create HTML files manually — generate.py handles that
- Add prev/next/related fields to snippet schema
- Run generate.py to verify the build before creating PR

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: brunoborges

.github/workflows/check-new-jdk.md

@@ -37,7 +37,8 @@ code comparisons showing old Java patterns next to their modern replacements.
 1. **Check for new OpenJDK releases.**
    - Fetch the OpenJDK project page at `https://openjdk.org/projects/jdk/` and identify the
      latest GA (General Availability) JDK release.
-   - Compare it against the site's current coverage. Read `data/snippets.json` to see which
+   - Compare it against the site's current coverage. Read the existing JSON snippet files in
+     the category subfolders (e.g., `language/*.json`, `concurrency/*.json`) to see which
      JDK versions are already covered.
    - If the latest GA release is already fully covered, stop and report "No new JDK release to cover."
 
@@ -49,33 +50,41 @@ code comparisons showing old Java patterns next to their modern replacements.
      unless they have a clear developer-facing usage pattern.
    - Also note any features that graduated from preview to final in this release.
 
-3. **Propose new snippets.**
-   - For each relevant new feature, draft a snippet entry in the same JSON format used in
-     `data/snippets.json`. Each snippet needs:
-     - `slug`: kebab-case URL slug
+3. **Create new snippet JSON files.**
+   - Each snippet is an individual JSON file at `{category}/{slug}.json`.
+   - Use an existing JSON file (e.g., `language/type-inference-with-var.json`) as a reference
+     for the schema. Each snippet needs:
+     - `id`: next sequential integer
+     - `slug`: kebab-case URL slug (must match the filename without `.json`)
      - `title`: human-readable title
-     - `category`: one of language, collections, strings, streams, concurrency, io, errors, datetime, security, tooling
+     - `category`: one of language, collections, strings, streams, concurrency, io, errors, datetime, security, tooling (must match the parent folder)
      - `difficulty`: beginner, intermediate, or advanced
      - `jdkVersion`: the JDK version where this became final (non-preview)
      - `oldLabel` / `modernLabel`: e.g., "Java 8" / "Java 26+"
      - `oldApproach` / `modernApproach`: short description of each approach
      - `oldCode` / `modernCode`: complete, compilable code snippets (concise, max ~12 lines each)
      - `summary`: one-sentence summary
      - `explanation`: 2-3 sentence explanation of why the modern approach is better
-     - `whyModernWins`: array of 3 objects with icon, title, desc
+     - `whyModernWins`: array of exactly 3 objects with `icon`, `title`, `desc`
      - `support`: version info string, e.g., "Finalized in JDK 26 (JEP NNN, Month Year)."
+     - `prev`: `category/slug` of the previous pattern, or `null` if first
+     - `next`: `category/slug` of the next pattern, or `null` if last
+     - `related`: array of exactly 3 `category/slug` strings for related patterns
 
-4. **Make the changes.**
-   - Add the new snippets to `data/snippets.json` (append to the array, assign sequential IDs).
-   - For each new snippet, create a `{slug}.html` article page following the exact same HTML
-     template structure as existing article pages (e.g., `records-for-data-classes.html`).
-   - Add a card for each new snippet to `index.html` inside the `#tipsGrid` div.
+4. **Update existing files.**
+   - Update the `next` field of the last existing snippet's JSON file to point to the first
+     new snippet, and set the new snippet's `prev` accordingly. Chain all new snippets together.
+   - Add a preview card for each new snippet to `index.html` inside the `#tipsGrid` div.
    - Update the snippet count wherever it appears in `index.html` (hero-badge, section-badge,
      structured data `numberOfItems`).
-   - Update prev/next navigation links on the last existing article page to point to the first
-     new page, and chain the new pages together.
+   - Do **NOT** edit `data/snippets.json` or any HTML files in category subfolders — these are
+     generated by `generate.py` and must not be modified directly.
 
-5. **Create a pull request.**
+5. **Verify the build.**
+   - Run `python3 generate.py` to regenerate all HTML pages and `data/snippets.json`.
+   - Confirm the new pages were generated successfully.
+
+6. **Create a pull request.**
    - The PR title should be: `[new-jdk] Add snippets for JDK {version} features`
    - The PR body should list each new snippet with its title and a one-line summary.
    - Mention which JEPs are covered and link to the OpenJDK release page.
@@ -84,7 +93,8 @@ code comparisons showing old Java patterns next to their modern replacements.
 
 - Only include features that are **final** (non-preview) in the new JDK release.
 - Label preview features as preview if you choose to include them, with "(Preview)" in the modernLabel.
-- Ensure all code in HTML files is properly HTML-escaped.
-- Do not modify existing snippets unless a feature graduated from preview to final.
+- All content goes in `{category}/{slug}.json` files — never edit generated HTML or `data/snippets.json`.
+- Run `python3 generate.py` after making changes to verify the build.
+- Do not modify existing snippet JSON files unless a feature graduated from preview to final.
 - If a previously preview feature is now final, update its `modernLabel` and `support` text
-  to remove the "(Preview)" label.
+  to remove the "(Preview)" label in its JSON file.