😼> Text Replace

It's a find-and-replace node: pass it a text string plus a find value and a replace value, and it swaps every occurrence of find with replace before the text moves on. That's the entire feature set, and that's the point - it's the one you reach for instead of building out a Python-eval or regex node just to templatize a prompt.

The obvious use is a template prompt with a placeholder token - write your prompt once as "a photo of {subject} standing in a forest" and use Text Replace to swap {subject} for whatever value is coming out of another node upstream (a Text Pick Random Line, a batch loop, an API input). It also comes in handy for cleanup: stripping out a placeholder LoRA trigger word you don't want in the final saved caption, or normalizing inconsistent tag formatting (replacing underscores with spaces from a booru-tag source, for instance).

How it works

Straight string substitution - every instance of find inside text gets swapped for replace. It's not regex, so there's no pattern matching, capture groups, or wildcards; what you type in find is matched literally.

The inputs and outputs that matter

• text - the source string to operate on.
• find - the literal substring to search for.
• replace - what to substitute in its place. Leave it empty to just delete every occurrence of find instead of substituting something else.

One output: text, the result - chain it into a CLIP Text Encode, another Text Replace for a second substitution, or a Save Image filename.

How to install it

Via ComfyUI Manager: search "yanc" or "YANC - Yet Another Node Collection," install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/ALatentPlace/ComfyUI_yanc

then restart. Pure string manipulation, no extra dependencies or downloads.

Common issues & troubleshooting

Nothing gets replaced. Since matching is literal, capitalization and exact spacing both matter - find set to "Subject" won't match "subject" in the text, and "a cat" (double space) won't match "a cat" (single space). Check your find value character-for-character against what's actually in text.

You need multiple replacements. One Text Replace handles one find/replace pair. For a template with several placeholders ({subject}, {location}, {style}), chain multiple Text Replace nodes in sequence, each handling one placeholder - the output text of one feeds straight into the text input of the next.

You wanted pattern matching, not exact-text matching. If you need something like "replace any number" or "replace anything inside brackets," that's regex territory and this node doesn't do it - it's intentionally the simple literal-substitution tool, not a scripting node. For anything pattern-based you'd need a more general text/expression node elsewhere in the ecosystem.