feat: articles format option with markdown new default

docs/api/articles.md

@@ -0,0 +1,97 @@
+# Reading articles
+
+Every route that returns an article accepts the optional `format` query parameter and returns the article body in that format. Articles are stored as HTML (Tiptap output). Storage and the request bodies of `POST /v1/article/` and `PATCH /v1/article/{id}` stay HTML.
+
+## Routes that honour `format`
+
+- `GET /v1/article/`
+- `GET /v1/article/{id}`
+- `POST /v1/article/`
+- `PATCH /v1/article/{id}`
+- `PATCH /v1/article/{id}/publish`
+- `PATCH /v1/article/{id}/unpublish`
+
+Visibility is unchanged: anonymous and unverified callers only see online articles, and an offline article answers `404 Article not found`.
+
+## The `format` parameter
+
+| Value | `body` | `body_format` |
+| --- | --- | --- |
+| empty or absent | markdown string | `markdown` |
+| `markdown` | markdown string | `markdown` |
+| `html` | the stored HTML string | `html` |
+| `json` | JSON array of blocks | `json` |
+
+Values are case-insensitive. When `format` is repeated, the first value is used. Any other value answers `400` with the plain-text message `The format must be one of: markdown, html, json.` before any article is read or written. An invalid id is reported first, ahead of the format.
+
+Breaking change: the default used to be HTML. Clients that need HTML must send `?format=html`. The web interface does so.
+
+## Article object
+
+```json
+{
+  "id": 1,
+  "title": "Backyard Renovation Update",
+  "body": "## Plan\n\nReplace the **old** deck.",
+  "body_format": "markdown",
+  "is_online": true,
+  "created_at": "2026-03-12T09:14:00Z",
+  "updated_at": "2026-09-10T07:00:00Z"
+}
+```
+
+## Markdown output
+
+Headings, bold, italic, links, bullet and ordered lists, blockquotes, fenced code blocks, rules, images and inline code are converted. Anything else is reduced to its text. `script` and `style` content is removed, and the source HTML is never passed through; the only HTML in the output is the alignment wrapper described below. List indentation attributes are dropped. Links with a scheme other than http, https, mailto or tel are reduced to their text. Characters that markdown would read as formatting (`*`, `_`, `` ` ``, `[`, `]`, `<`, `\`) and line starts that would read as a heading, list, quote or rule (`#`, `-`, `+`, `>`, `1.`) are backslash-escaped, so plain text stays plain text. A line break becomes a backslash followed by a newline.
+
+### Text alignment
+
+Markdown has no syntax for alignment, so centered, right-aligned and justified paragraphs and headings are wrapped in a `div` with an inline style. The blank lines around the content make CommonMark parse the inside as markdown:
+
+```markdown
+<div style="text-align: center">
+
+A **centered** paragraph
+
+</div>
+```
+
+Left alignment is the default and produces no wrapper. Your markdown renderer must allow raw HTML for the alignment to show (`html: true` in markdown-it); otherwise the wrapper tags appear as text, and a client that does not want them can request `?format=json` instead, where the alignment is a field.
+
+## JSON block schema
+
+`body` is a native JSON array. Every block has a `type`.
+
+| Block | Fields |
+| --- | --- |
+| `heading` | `level` (1 to 6), `align` (only when `center`, `right` or `justify`), `children` (text nodes) |
+| `paragraph` | `align` (only when `center`, `right` or `justify`), `children` (text nodes, empty array for an empty paragraph) |
+| `list` | `ordered` (boolean), `items` (array of items, each an array of blocks) |
+| `blockquote` | `children` (blocks) |
+| `code` | `language` (empty string when unknown), `text` |
+| `hr` | none |
+| `image` | `src`, `alt` (empty string when missing) |
+
+Text node: `{"type":"text","text":"...","marks":[...]}`. `marks` is always an array, listed outermost first. A mark is `{"type":"bold"}`, `{"type":"italic"}` or `{"type":"link","href":"..."}`. A line break becomes a `"\n"` inside the text. Adjacent text with identical marks is merged. Inline code, underline, strike-through and other formatting degrade to plain text. An image inside a paragraph becomes its own block and splits the paragraph. An empty body is `[]`.
+
+Example for `<h2>Plan</h2><p>Replace the <strong>old</strong> deck.</p>`:
+
+```json
+[
+  {"type":"heading","level":2,"children":[{"type":"text","text":"Plan","marks":[]}]},
+  {"type":"paragraph","children":[
+    {"type":"text","text":"Replace the ","marks":[]},
+    {"type":"text","text":"old","marks":[{"type":"bold"}]},
+    {"type":"text","text":" deck.","marks":[]}
+  ]}
+]
+```
+
+## Errors
+
+| Status | Body |
+| --- | --- |
+| `400` | `The server expects the ID to be in the format of an unsigned 32-bit integer (uint32).` for an invalid id |
+| `400` | `The format must be one of: markdown, html, json.` for an unsupported format |
+| `404` | `Article not found` |
+| `500` | `Internal server error` when the body cannot be converted |

docs/release-notes.md

@@ -29,3 +29,4 @@ Each body lives in `docs/releases/<tag>.md`. Publish with `tea releases create -
 | `v2.5.0-beta` | v2.5.0-beta - Media library & image picker |
 | `v2.6.0-beta` | v2.6.0-beta - Signed updates and browser restart |
 | `v2.6.2-beta` | v2.6.2-beta - English and French interface |
+| `v2.7.0-beta` | v2.7.0-beta - Markdown and block article bodies |

docs/releases/v2.7.0-beta.md

@@ -0,0 +1,35 @@
+This release lets your website read article bodies as markdown, HTML or structured blocks, and fixes text alignment in the editor.
+
+## New
+
+**Article body formats**
+
+- Read an article body as markdown, HTML or a typed block array with the `format` query parameter, so a headless frontend can render it with markdown-it, comark or its own components.
+- Keep text alignment in every format: markdown wraps centered, right-aligned and justified blocks in a `div` with an inline style, and the block array carries an `align` field.
+- Keep plain text plain: text that merely looks like markdown, such as `# not a heading` or `- not a list`, is escaped in the markdown output.
+- See the Integration page snippets and [the article API reference](../api/articles.md) for the full contract.
+
+## Improvements and fixes
+
+- Fix the text alignment buttons in the article editor, which were disabled and showed a blocked cursor. Paragraphs and headings can be aligned left, center, right or justified again, and unaligned text stays free of inline styles.
+- The web interface requests HTML explicitly when it loads, saves, publishes, unpublishes or creates an article, so the editor and the image counts behave exactly as before.
+
+## API
+
+- Changed: `GET /v1/article/`, `GET /v1/article/{id}`, `POST /v1/article/`, `PATCH /v1/article/{id}`, `PATCH /v1/article/{id}/publish` and `PATCH /v1/article/{id}/unpublish` now return `body` as markdown by default. Pass `?format=html` to keep the previous HTML body.
+- Every article object now includes `body_format` (`markdown`, `html` or `json`), which names the format of `body`.
+- `?format=json` on the same routes returns `body` as a JSON array of typed blocks instead of a string.
+- An unsupported `format` answers `400` before anything is read or written.
+- Storage and the request bodies of `POST` and `PATCH` stay HTML.
+
+## Under the hood
+
+- Add the article body renderer port in the domain, with a markdown and JSON block adapter, and tests for every supported element.
+- Update `golang.org/x/crypto` to v0.57.0 and add `golang.org/x/net` as a direct dependency for HTML parsing.
+- Add `@tiptap/extension-text-align` to the web interface.
+
+## Upgrading
+
+- Replace the binary. No database migration or new environment variable is required.
+- Clients that read an article `body` as HTML must add `?format=html` to their article requests, because the default is now markdown.
+- A markdown renderer needs raw HTML enabled for text alignment to show, for example `html: true` in markdown-it.