GHSA-3c48-9r95-hqhr json
Untrusted labels can inject SVG attributes, CSS values, and terminal escape sequences
published 3mo ago modified 30d ago also RUSTSEC-2026-0345
- cvss
- —
- not scored
- cisa kev
- Not listed
- not in CISA's catalog
- epss
- —
- no score, EPSS covers CVEs only
- public exploits
- 0
- none known
Summary
Affected versions of `kuva` wrote data-derived strings straight into rendered output without escaping them for the target format. Applications that plot untrusted input (column headers, category and group names, series labels) were affected in three places.
**1. SVG attribute injection / stored XSS (CWE-79).** When SVG interactivity was enabled via `Layout::with_interactive`, group and category labels were formatted directly into `data-group` and `data-x` attributes through the raw `Primitive::GroupStart.extra_attrs` string, which the SVG backend emitted verbatim. The backend's escaping helper covered text nodes and `<title>` but not attributes. A label containing a double quote could close the attribute and add another, for example an event handler. Opening the resulting SVG in a browser, or inlining it into an HTML page, executed the injected script. Reachable from scatter, line, bar, and strip plots, and from legends.
**2. Unvalidated CSS and style attribute values (CWE-74).** `Color::Css` wrote arbitrary strings into `fill` and `stroke`, `stroke-dasharray` was emitted raw from the public dash-pattern builders, and the SVG root `font-family` and `fill` plus the background rectangle `fill` were written raw from layout and theme configuration. Applications that forwarded user-supplied style values into the plotting API could have those values break out of the attribute they were written into.
**3. Terminal escape sequence injection (CWE-150).** `TerminalBackend` stored label characters unfiltered and emitted each one unconditionally when producing its ANSI string. Control characters in a label, including ESC, OSC sequences, and carriage returns, were written straight to the operator's terminal. Depending on the emulator this allows forged or misleading output, window title and clipboard manipulation, and reaching terminal emulator vulnerabilities that are triggered by escape sequences.
The fix routes every data-derived string through format-appropriate escaping. A single `escape_attr` helper XML-escapes `&`, `<`, `>`, `"`, and `'`, and is applied at each raw-attribute site: the `extra_attrs` interpolations, the `Color::Css` catch-all, and the SVG backend's `stroke-dasharray`, `font-family`, and background `fill` writes. `TerminalBackend::set_char` replaces control characters with U+FFFD before they enter the character grid, so one cell stays one column rather than the grid shifting, and the ANSI writer emits the grid unchanged.
Patched in 0.5.0, with backports to 0.4.1 for the 0.4.x line and 0.3.1 for users still on 0.1.5 through 0.3.0. No versions were yanked.
Users who cannot upgrade should sanitize labels before passing them to `kuva`, and avoid interactive SVG output for untrusted data.
Note that `Scene::defs`, `Scene::scripts`, and `Primitive::GroupStart.extra_attrs` remain raw escape hatches by design. They are documented as trusted-only and are not covered by this advisory; populating them from user data is equivalent to executing that data.
Patch / fix
- crates.io kuva 0.3.1, 0.4.1
Is GHSA-3c48-9r95-hqhr in your dependencies? Fencer shows which of your repositories are affected as soon as an advisory lands, and opens the fix as a pull request.
Check with FencerAffected packages
| ecosystem | package | affected | fixed in |
|---|---|---|---|
| crates.io | kuva | >=0.1.5, <0.3.1 >=0.4.0, <0.4.1 |
0.3.1 0.4.1 |
References
-
github.com
/Psy-Fer/kuva/commit/2b53ceb796bbb0e9bb537836ad19faac53a51852
web
-
github.com
/Psy-Fer/kuva/commit/2219f5931913f79f3b9be47690b8999a3d8b8bfb
web
-
github.com
/Psy-Fer/kuva/commit/76049ff5133dfb796ba759c2c5e486779d3d30c5
web
-
crates.io
/crates/kuva
package
-
rustsec.org
/advisories/RUSTSEC-2026-0345.html
advisory
-
github.com
/Psy-Fer/kuva/releases/tag/v0.4.1
web
-
github.com
/Psy-Fer/kuva/releases/tag/v0.3.1
web
Sources
Fencer Intel keeps one entry per vulnerability: the CVE id wins, then GHSA, and records that alias each other are merged, except that two CVEs always keep separate entries. The same data is available as JSON.