Build the term-grammar / model-matrix bridge for one submodel

extract_terms() is the prerequisite for every renderer. It bridges R formula terms <-> model-matrix columns <-> biological symbols. Without this layer, equations silently break on factor contrasts, interactions, offsets, and transformations.

For each model-matrix column (plus offsets), it returns one row giving the term label, source variable, role classification, factor contrast level, transformation applied, biological symbol, coefficient symbol, and a ready-to-splice LaTeX fragment.

Supported roles in v0.1: intercept, predictor, factor_contrast, interaction, transformation, offset. Random-effect terms are detected upstream and removed before extract_terms() runs in v0.1.

Usage

extract_terms(
  formula,
  data,
  submodel,
  symbols = NULL,
  coefficient_family = "beta"
)

Arguments

formula: One-sided or two-sided formula for the submodel.
data: Data frame used for fitting.
submodel: Character. Name of the submodel, e.g. "mu", "sigma".
symbols: Named character vector mapping variable names to LaTeX symbols. Missing variables receive auto-generated symbols.
coefficient_family: Character. Coefficient symbol family used in coefficient_symbol, e.g. "beta" for mu, "gamma" for log sigma.

Value

A tibble with columns: submodel, term_label, variable, role, contrast_level, transform, symbol, coefficient_symbol, latex_term.