Skip to content

Command Glossary v0

Note: This is the historical Phase 0 glossary. For the current command index, see command-reference.md.

This glossary defines the first small command surface for TabDat-Explore. It is intentionally high level. Detailed syntax, output, and error behavior should be written in a command contract before implementation.

Inspection And Loading

use

Load a dataset into the single active dataset slot.

Phase 1 focus: load Parquet or Stata .dta from a local path. HTTP/HTTPS .dta URLs are also supported; lazy mode stays Parquet-only.

describe

Show dataset structure, including columns, types, and basic shape.

Phase 1 focus: describe the active dataset.

summarize

Show descriptive statistics for one or more variables.

Phase 1 focus: summarize numeric columns for the active dataset.

codebook

Show compact column-level profiling, including type, missingness, and representative values.

Phase 3 focus.

count

Count rows in the active dataset, optionally after a filter in a later phase.

Phase 3 focus.

Preview the first rows of the active dataset.

Phase 3 focus.

tail

Preview the last rows of the active dataset where the backend can support it predictably.

Phase 3 focus.

Transformation

keep

Keep selected variables or, in a later phase, rows matching a condition.

Phase 3 focus.

drop

Drop selected variables or, in a later phase, rows matching a condition.

Phase 3 focus.

rename

Rename one variable while preserving data and type.

Phase 3 focus.

generate

Create a new variable from an expression.

Phase 3 focus after expression parsing exists.

Summaries

tabulate

Produce one-way, two-way, or multi-level frequency tables. Enhanced tabulations can fill cells with one aggregate value via values(<var>) stat(count|mean|sum|min|max).

Phase 3 focus.

SQL Escape Hatch

sql

Run a result-producing SQL query over the active dataset exposed as active.

Phase 4 focus.

Scripting

run

Run a TabDat script file in the current session.

Phase 8 focus: one command per non-empty line, whole-line comments, multiline SQL blocks, relative nested paths, recursion rejection, and deterministic script transcripts.

Deferred Commands

The following commands are important but should remain outside the v0 command set until the core workflow is stable:

  • replace
  • select
  • collapse
  • by:
  • histogram
  • scatter
  • bar
  • run
  • set