@nodetool-ai/text-nodes

Text-processing nodes for NodeTool.

A pack of text nodes for NodeTool workflows: concatenate and template strings, filter streams, embed and transcribe, and assemble SVG documents. HTML and Markdown parsing moved to the @nodetool-ai/sandbox-html and @nodetool-ai/sandbox-markdown sandbox packs, importable from a Code node.

The string-manipulation, comparison, regex, parsing, and token-counting node classes were removed: nodetool.code.Code covers that ground in three lines of JavaScript. Saved workflows keep working — NODE_TYPE_MIGRATIONS in packages/protocol/src/graph.ts rewrites each removed type to a Code node with the equivalent body prefilled.

Install

npm install @nodetool-ai/text-nodes

Nodes

Composition

• nodetool.text.Concat, nodetool.text.Template, nodetool.text.Prompt, nodetool.text.Collect

Stream filters

• nodetool.text.FilterString, nodetool.text.FilterRegexString

Embeddings and speech

• nodetool.text.Embedding, nodetool.text.AutomaticSpeechRecognition

I/O

• nodetool.text.LoadTextAssets, nodetool.text.LoadTextFolder, nodetool.text.SaveText, nodetool.text.SaveTextFile

SVG

lib.svg.Document assembles an element list into a document; lib.svg.SVGToImage rasterizes one through the native sharp addon. The 12 element builders (Rect, Circle, Text, Gradient, …) were removed — an svg_element is a plain object, so a Code node builds the whole list in one place, where the array order is the paint order. The editor ships them as the svg-* snippets.

The lib.nlp.* nodes (tokenize, stem, classify, entities, sentiment, TF-IDF, phonetic match) were removed with their 2,766 lines of vendored algorithms. No replacement pack ships yet.

Links

• NodeTool
• GitHub