fix(export): inline public/ images in static HTML export

[mscolnick]

Markdown like mo.md("![alt](public/image.png)") previously produced a
broken <img src="public/image.png"> reference in the exported standalone
HTML — opening the file outside the notebook directory would 404 on every
image. The static HTML export only inlined ./@file/ virtual files, not
relative public/ paths.

This change adds replace_public_files_with_data_uris and runs it during
Exporter.export_as_html (alongside the existing virtual-file pass) so
<img>, <audio>, <video>, and <source> src attributes pointing at
the notebook's public/ folder are embedded as base64 data URIs. The same
10 MB inline limit used for virtual files applies; oversized files are
left as-is. Resolved paths are validated to stay inside the public
directory, rejecting ../ traversal and symlink escapes — matching the
runtime serve_public_file containment check.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
marimo-docs	Ready	Preview, Comment	May 20, 2026 3:53pm

[Copilot]

Pull request overview

This PR fixes static HTML export so that media references like public/image.png (commonly produced by markdown such as ![alt](public/image.png)) are inlined into the exported standalone HTML as base64 data URIs, making exported files self-contained outside the notebook directory.

Changes:

• Added replace_public_files_with_data_uris to inline public/ (and ./public/) src references for common media tags while enforcing containment checks against traversal/symlink escapes.
• Updated Exporter.export_as_html to run the new public-file inlining pass in addition to the existing virtual-file inlining.
• Added unit and integration-style regression tests covering inlining behavior and security constraints (traversal/symlink escape).

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
tests/_server/export/test_exporter.py	Adds export-level regression tests for inlining public/ assets and blocking traversal.
tests/_convert/common/test_dom_traversal.py	Adds unit tests for public/ HTML attribute inlining, including security-focused cases.
marimo/_server/export/exporter.py	Runs public-folder inlining during HTML export and refactors traversal of string outputs.
marimo/_convert/common/dom_traversal.py	Implements public/ path resolution + inlining into data URIs with containment checks.

[cubic-dev-ai]

3 issues found across 4 files

Architecture diagram

sequenceDiagram
    participant User as User (Browser)
    participant Exporter as Exporter (Server)
    participant DomTraversal as dom_traversal
    participant FileSystem as Notebook Public Dir
    participant Output as HTML Snapshot

    Note over User,Output: Static HTML Export Flow (public/ image inlining)

    User->>Exporter: Request: export_as_html(filename, app, session_view)
    Exporter->>Exporter: Resolve public_dir = Path(filename).parent / "public"
    Exporter->>Exporter: _inline_virtual_files(session_snapshot)
    Exporter->>Exporter: _iter_html_data_strings() yields (data_dict, mime, html_str)
    Exporter->>DomTraversal: replace_virtual_files_with_data_uris(html_str)
    DomTraversal-->>Exporter: processed_html, replaced_files
    Exporter->>Exporter: Update data_dict[mime] = processed_html

    Exporter->>Exporter: _inline_public_files(session_snapshot, public_dir)
    alt public_dir exists
        Exporter->>Exporter: Iterate HTML strings via _iter_html_data_strings()
        Exporter->>Exporter: Check if "public/" in data
        Exporter->>DomTraversal: replace_public_files_with_data_uris(html_str, public_dir)
        DomTraversal->>DomTraversal: Parse HTML attributes (img, audio, video, source)
        DomTraversal->>DomTraversal: Match src values against ^(?:\./)?public/(.+)$
        alt Match found
            DomTraversal->>FileSystem: _resolve_public_file(public_dir, relpath)
            alt File exists & is inside public_dir
                FileSystem-->>DomTraversal: Resolved Path
                DomTraversal->>FileSystem: Read file bytes
                alt File size <= max_inline_bytes (10MB)
                    FileSystem-->>DomTraversal: bytes
                    DomTraversal->>DomTraversal: base64 encode + build_data_url
                    DomTraversal-->>Exporter: Replaced src with data URI
                    Exporter->>Exporter: Add to replaced set
                else File exceeds size limit
                    DomTraversal-->>Exporter: Original src preserved (skip)
                end
            else File missing or outside public_dir
                FileSystem-->>DomTraversal: None (reject)
                DomTraversal-->>Exporter: Original src preserved
            end
        else No match (external URL, other paths)
            DomTraversal-->>Exporter: No change
        end
        Exporter->>Exporter: Update data_dict[mime] = processed_html
    else public_dir does not exist
        Exporter->>Exporter: Skip (no-op)
    end

    Exporter->>Exporter: Generate full HTML document with inlined assets
    Exporter-->>User: Standalone HTML string (self-contained)

Loading

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

1 issue found across 4 files (changes from recent commits).

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[kirangadhave]

🚀 great fix!

[github-actions]

🚀 Development release published. You may be able to view the changes at https://marimo.app?v=0.23.7-dev67