SF Google Sheet Range

Where its sibling SF Google Sheet Cell pulls one cell, SF Google Sheet Range grabs a whole rectangle of them and joins the values into one string. Think: a row of character traits, a column of style tags, or a small lookup table that gets flattened into your prompt. It's the same "live spreadsheet as a content source" idea, but for the case where you need several cells at once and would rather not wire up fifty single-cell nodes.

How it works

Like the Cell node, this uses the gviz/tq CSV export endpoint - no Google API key, no OAuth, just a public sheet URL. It fetches the tab's full contents as CSV, then walks your range left to right, top to bottom, joining each cell with a delimiter. The output is a single STRING named text. If you set a range with empty cells, they still appear as separators - A, B, , D - which is usually what you want for prompt building, but worth knowing before you build a workflow around it.

The inputs that matter

• url and sheet_gid - the full sheet link, and the #gid= value for the tab (default 0 is the first tab). Same rules as the Cell node.
• start_row / start_col / end_row / end_col - the corners of your range, all inclusive. Rows are 1-indexed, columns are letters (A, B, ..., AA). A1:B3 becomes start_row=1, start_col=A, end_row=3, end_col=B.
• delimiter - what goes between cells, default ", ". You can type it or wire it from another string node, which is handy when you want a newline between items instead of a comma.

The node validates your range before fetching: start_row greater than end_row (or start column after end column) throws a clear error instead of returning nonsense.

Installing it

It's part of the SF ComfyUI Nodes pack from Stillfront. ComfyUI Manager → search "SF ComfyUI Nodes" → Install → restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/Stillfront/comfyui-sf-nodes.git
cd comfyui-sf-nodes
pip install -r requirements.txt

No model downloads - the only dependency that matters here is requests.

Common issues

The usual suspect is sharing. The sheet must be Anyone with the link → Viewer; anything else returns an HTTP error, not a helpful one. Also mind that the range is inclusive on both ends - a common first-run surprise is getting one more row or column than you intended because you read end_row as exclusive. And since it fetches live on every execution, it's a network call per run; put it somewhere in the graph that executes once, not inside a per-frame loop, unless you enjoy rate limits.