feat: Implement basic text overlay node (CORE-137)

[drozbay]

Implements a basic native Text Overlay node:

Currently uses standard built-in PIL default font and draws an overlay with an outline on top of the image or video at the desired position. Compatible with newline escape characters (\n).

Example workflow that creates a grid of test images with different settings:

droz_textoverlay_test_C137_1.json

Image example:

Video example:

[coderabbitai]

📝 Walkthrough

Walkthrough

A new file comfy_extras/nodes_text_overlay.py is added, implementing a TextOverlay ComfyUI node that renders text onto image tensors (including batches). It exposes inputs for text content, vertical position, horizontal alignment, font size as a percentage of image height, line spacing, text color, and outline color. Rendering uses PIL on a transparent RGBA layer with adaptive font sizing and optional background band, then alpha-composites the result onto the input tensor. A TextOverlayExtension registers the node, and nodes.py adds the file to the builtin extras load list.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 14.29% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely describes the main change: adding a basic text overlay node.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description check	✅ Passed	The description matches the new native Text Overlay node and its image/video overlay behavior.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@comfy_extras/nodes_text_overlay.py`:
- Around line 125-126: The text wrapping logic in the overlay renderer is
collapsing user-entered whitespace because `split()` is used in the loop inside
`nodes_text_overlay.py`, which destroys repeated spaces and tabs before drawing.
Update the wrapping path in the text processing function that iterates over
`raw_line` so it preserves literal spacing and tab characters from the original
caption instead of tokenizing on all whitespace; keep the existing `\t` handling
from the earlier text normalization intact while changing the wrapping
implementation to work with the raw line content.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: bb2ee910-d7f8-4730-87c9-8287e7f216af

📥 Commits

Reviewing files that changed from the base of the PR and between f197357 and d8711f0.

📒 Files selected for processing (2)

• comfy_extras/nodes_text_overlay.py
• nodes.py

[coderabbitai]

🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Preserve literal spacing when wrapping text.

Line 126 uses split(), which collapses repeated spaces and removes tabs. That rewrites the user’s caption before rendering, so indentation, double-spaces, and the escaped \t handling from Line 37 never survive to the image.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@comfy_extras/nodes_text_overlay.py` around lines 125 - 126, The text wrapping
logic in the overlay renderer is collapsing user-entered whitespace because
`split()` is used in the loop inside `nodes_text_overlay.py`, which destroys
repeated spaces and tabs before drawing. Update the wrapping path in the text
processing function that iterates over `raw_line` so it preserves literal
spacing and tab characters from the original caption instead of tokenizing on
all whitespace; keep the existing `\t` handling from the earlier text
normalization intact while changing the wrapping implementation to work with the
raw line content.