"[" Brackets can't appear in field name

[FlorianFritz]

Description

Hi! I tried to generate a x-y-Line-Graph with altair from a pandas DataFrame with two columns named "Force in [N]" and "Travel in [inches]".

When I use the default altair renderer in a JupyterLab Notebook it fails with a Javascript error very similar to this bug report vega/vega-lite#8153

When I use the png renderer (which uses vl-convert) the graph does not render, just the canvas without a line.

This is the SPEC: with_brackets.json

Excerpt of SPEC:

{
  // OMITTED //
  "encoding": {
    "x": {
      "field": "Travel in [inches]",
      "type": "quantitative"
    },
    "y": {
      "field": "Force in [N]",
      "type": "quantitative"
    }
  },
  // OMITTED //
}

When I remove all brackets the graph renders as expected. SPEC: without_brackets.json

Excerpt of SPEC:

{
  // EVERYTHING LIKE ABOVE //
  "encoding": {
    "x": {
      "field": "Travel in inches",
      "type": "quantitative"
    },
    "y": {
      "field": "Force in N",
      "type": "quantitative"
    }
  },
  // OMITTED //
}

When I escape the brackets [ ] with \\[ and \\] in the encoding section (like mentioned in the vega-lite issue) the graph renders as expected but the axis labels contain one escaping backslash: Force in \[N\]

SPEC: with_brackets_escaped.json

Excerpt of SPEC:

{
  // EVERYTHING LIKE ABOVE //
  "encoding": {
    "x": {
      "field": "Travel in \\[inches\\]",
      "type": "quantitative"
    },
    "y": {
      "field": "Force in \\[N\\]",
      "type": "quantitative"
    }
  },
  // OMITTED //
}

Used Versions

Sourcecode: VegaError.ipynb.txt

pandas 2.3.0
altair 5.5.0
vl_convert 1.7.0
RendererRegistry.enable('png')