fix(api): apply_style fails fast on a non-Axes first positional

[dchaudhari7177]

Symptom

fr.apply_style("SCITEX_STYLE")   # string preset name in the ax slot
# → AttributeError: 'str' object has no attribute 'set_facecolor'

apply_style(ax, style=None) accepted any first positional, so a preset name passed where the Axes belongs sailed past the API boundary and crashed deep inside with an opaque AttributeError. Downstream try/except Exception wrappers silently swallowed it — the exact silent-fallback the issue calls out.

Fix

A TypeError at the boundary, per the issue's proposed shape:

if not isinstance(ax, matplotlib.axes.Axes):
    raise TypeError(
        f"apply_style(ax, style=...): ax must be a matplotlib Axes, "
        f"got {type(ax).__name__}. If you meant a style preset, pass it "
        f"as the style argument: apply_style(ax, style=...)."
    )

I pointed the hint at the existing style= argument rather than load_style('SCITEX'), since that auto-propagation is #159's scope and isn't in this repo yet — keeping this PR to just the apply_style half the issue explicitly separates out. Easy to reword to reference load_style when #159 lands.

Tests

test__style_manager.py gains two regression tests: the string first-positional raises TypeError matching must be a matplotlib Axes, and a real Axes still returns a positive float trace width. pytest tests/figrecipe/_api/ → 32 passed.

Fixes #160

[github-actions]

Please sign the SciTeX CLA before your contribution can be merged.
Comment I have read and agree to the SciTeX CLA. to sign.

---

I have read and agree to the SciTeX CLA.

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[ywatanabe1989]

Great catch, and a faithful implementation of exactly what #160 proposed — the guard, the message, and both tests are all right. We ran your tests against current main: 3/3 pass.

One case slipped through, and it is really on us: the issue's own snippet had the same blind spot, so you implemented precisely what we asked. figrecipe's own fr.subplots() returns a RecordingAxes composition wrapper (src/figrecipe/_wrappers/_axes.py — mixins delegating through __getattr__, not an Axes subclass). On current main, fr.apply_style(ax) with that wrapper works via delegation; with the strict isinstance guard it now raises TypeError. Our internal style path already unwraps first (src/figrecipe/_api/_subplots.py:219), and the same one-liner fixes it here — see the suggested change below.

Could you also add a third test pinning the wrapper case, so nobody re-breaks it?

def test_apply_style_accepts_figrecipe_wrapped_axes():
    import figrecipe as fr
    fig, ax = fr.subplots()
    try:
        result = fr.apply_style(ax)
        assert isinstance(result, float)
    finally:
        import matplotlib.pyplot as plt
        plt.close("all")

Everything else is merge-ready. (Separately, whenever convenient: the CLA bot's one-comment signature is the only other thing needed before merge — opening and review needed none.)

[ywatanabe1989]

Suggested change

	# Fail loud at the boundary: a non-Axes first positional (commonly a preset
	# name passed as `apply_style("SCITEX_STYLE")`) otherwise crashes deep
	# inside with an opaque `'str' object has no attribute ...` (#160).
	if not isinstance(ax, matplotlib.axes.Axes):
	# figrecipe's own subplots() returns a RecordingAxes wrapper (composition,
	# not an Axes subclass) — unwrap it first so `fig, ax = fr.subplots();
	# fr.apply_style(ax)` keeps working, matching the unwrap in _subplots.py.
	ax = getattr(ax, "_ax", ax)
	# Fail loud at the boundary: a non-Axes first positional (commonly a preset
	# name passed as `apply_style("SCITEX_STYLE")`) otherwise crashes deep
	# inside with an opaque `'str' object has no attribute ...` (#160).
	if not isinstance(ax, matplotlib.axes.Axes):

[dchaudhari7177]

Good catch — fixed. apply_style now unwraps to ax._ax before the isinstance guard (the same one-liner as _subplots.py:219), so a RecordingAxes wrapper from fr.subplots() passes through while a string/other non-Axes still raises TypeError. Added your suggested test_apply_style_accepts_figrecipe_wrapped_axes — the suite is now 4/4.

On the CLA: that's a legal sign-off my collaborator needs to make on their own account, so I've flagged it to them to complete.

[ywatanabe1989]

Perfect — the unwrap is exactly the idiom from _subplots.py, placed before the guard so wrappers pass and a stray string still fails loud, and the new wrapper test pins it. Approving.

For the CLA: our bot checks the commit author account, and your commits are authored as dipakchaudhari12717 — so that's the account that needs to post the one-line signing comment (here or on any of your three PRs; one signature covers all SciTeX repos). Once it's in, we'll merge all three. Thanks for the quick turnaround — three solid PRs on day one is a great start.

[ywatanabe1989]

@dchaudhari7177 — full note on #318 (covers all three). In short: the CI red here was on our side — I updated your branch to current main to fix it — and the only remaining step is the CLA sign-off comment from dipakchaudhari12717 (the commit author) on any one of the three, which covers this one. Sorry for the runaround, and thank you for the careful work. — Yusuke

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!