From a formatted sheet to a tidy table you can trust

1 · Learn the move · Recipe pattern

Everything in this area composes into one fixed recipe, and the recipe's rule is that each step's output is checked before the next step runs. Survey: the structure report - headers, units, non-data rows, formatting-as-meaning. Parse: with explicit, evidence-based settings - encoding, delimiter, decimal - never defaults. Separate: data rows from titles, subtotals, and notes, with every excluded row logged by number and reason, because an unlogged exclusion is a silent edit. Scan: the coercion families - dates-from-genes, stripped zeros, notation collapse. Join: sample metadata with full count reconciliation. Verify: control values read true, row counts add up, every column carries a unit. The product is two files, not one: the tidy table, and the log that says exactly what was done to produce it - which is the difference between cleaned data and data someone once fiddled with.

Run the full sheet-to-tidy recipe on this file, one step at a time, stopping after each for my confirmation: (1) SURVEY - structure report; (2) PARSE - explicit settings with evidence; (3) SEPARATE - data vs non-data rows, every exclusion logged by row and reason; (4) SCAN - auto-coercion families; (5) JOIN - to [sample sheet] with count reconciliation; (6) VERIFY - controls: [known values], counts, units complete. Final output: the tidy table spec AND the processing log. No step runs while the previous has an unresolved flag. File: [paste]