XTL Cookbook

Focused, copy-paste recipes for common reporting workflows. Each recipe is a short markdown page with a scenario, the template cells, and the expected output.

The cookbook complements two existing surfaces:

• examples/ holds three runnable templates (renewal-report, sheet-per-region, multi-source-join) that exercise composed shapes end-to-end. Copy one as a starting point.
• spec/language.md is the normative reference for each function and directive. Use it when a recipe doesn't cover your exact case.

Recipes here favor "the smallest template that demonstrates X" over production realism — the goal is fast lookup when you remember the shape but forget the syntax.

Recipes

#	Recipe	What you'll learn
01	Getting started in 5 minutes	Template + data → output. Substitution and __config__.
02	Conditional cells	IF, IFEMPTY, comparison operators, truthiness.
03	Aggregates over rows	SUM, COUNT, AVERAGE, MIN, MAX — block vs whole-source.
04	One file per group	File grouping via output_file_pattern.
05	One sheet per group	Sheet grouping + list-based filtering.
06	Runtime inputs	__inputs__ for per-run values (month, region, etc.).
07	Multi-source + @join	__sources__, @source, @join.
08	XLOOKUP	Cross-source lookups.
09	Sort and Top-N	@sort (stable), @top, multi-key sort.
10	Styling and branding	tabColor, merged cells, numFmt, TEXT().
11	TEXT() formatting	Currency, dates, percentages; when to use numFmt vs TEXT().
12	Empty values in depth	IFEMPTY, empty-vs-zero traps, (blank), aggregates over sparse data.
13	Error handling for hosts	Catching XtlError, the code catalog, preview() for fail-fast.
14	__config__ as a value dictionary	Author-defined keys, type-awareness, __config__ vs __inputs__.
15	Composing directives	Execution order, multi-@filter AND, forbidden compositions.

How to read a recipe

Each recipe has the same shape:

1. Scenario — the operator's goal in one sentence.
2. __config__ — required keys.
3. Template cells — the smallest set of cells that produce the goal.
4. Data — a tiny input table.
5. Output — what convert() returns.
6. Notes — gotchas and pointers to the spec when you want more.

Conventions

• Cells use the A1 notation Excel uses, not [row, col].
• __config__ values are written as key = value for compactness; in a real template.xlsx they live in two columns (A: key, B: value).
• Source data is shown as a markdown table to keep the recipe terse. The real data.xlsx would have those rows in a worksheet matching source_sheet.

Running a recipe

The cookbook recipes are documentation-first — not every recipe ships with a runnable .xlsx pair. To try one:

1. Open Excel and create a new workbook.
2. Add the __config__ sheet with the keys listed.
3. Add the data sheet matching source_sheet.
4. Add the template sheet with the cells from the recipe.
5. Save as template.xlsx. Save the data as data.xlsx.
6. Run convert(templateBuffer, dataBuffer) (see README).

Or, faster: copy one of the runnable examples and adapt it.