Get or set global multi-reviewer review-tracking configuration

Decouples project-specific reviewer identities from the review-tracking mechanics in docx_review_extract, mirroring how pipeline_config decouples study name and stage labels from the core compute-graph classes: the number and names of reviewers are data supplied by the calling project, not hardcoded here.

Usage

review_config(
  fork_reviewers = NULL,
  signoff_reviewers = NULL,
  documents_sheet_mode = NULL,
  csv_reviewer_columns = NULL,
  pipeline_catalog = NULL,
  stem_aliases = NULL,
  analysis_dir = NULL,
  input_dirs = NULL,
  fork_mode = NULL,
  reset = FALSE
)

Arguments

fork_reviewers: Character vector of fork-reviewer identifiers. An id is used as a column name, as a fork-workbook file suffix and (with documents_sheet_mode = "per_reviewer") in a sheet name, so it is validated: ids must be unique (ignoring case) and consist of 1-20 letters, digits or underscores; they must not be the name of a built-in tracker column (for example file, author or resolved) and must not end in _comment. A violation is an error, and nothing is changed.
signoff_reviewers: Character vector of document-level sign-off reviewer identifiers (same rules as fork_reviewers; an id cannot be both a fork and a sign-off reviewer). Defaults to none.
documents_sheet_mode: One of "columns" (default; a single "Documents" sheet with one column, or column pair, per reviewer) or "per_reviewer" (a shared "Documents" sheet with counts and resolved only, plus one additional "Documents_<id>" sheet per reviewer carrying just that reviewer's status/comment columns). This applies to the master workbook; a fork workbook always has one plain "Documents" sheet.
csv_reviewer_columns: Which reviewer columns the CSV exports (comments.csv, tracked_changes.csv, suggested_changes.csv) contain. "fork" (default) keeps the fork reviewers' sign-off and comment columns but leaves out the sign-off reviewers' columns and the Word author column; "none" leaves out every reviewer column and the author; "all" writes every column, including the sign-off reviewers' and the author, which is what the CSVs held before this option existed.
pipeline_catalog: The analysis stages the reviewed documents belong to, used to label (pipeline_stage) and order documents. A list of stages, each a list with a stem (the document's file name without extension, date and reviewer initials, compared in lower case) and optionally heading, name and stage (the label shown; default "Heading <heading>: <name>", else the name, else the stem); a data frame with those columns also works. The order of the list is the order of the documents. The package ships no stages: unless a catalog (or a reports_to_render.xlsx manifest in analysis_dir) is supplied, every document gets the stage "Unknown / Extra". A file whose stem is not in the catalog is matched to the most similar stem when they are at least 60 percent alike. Use list() to clear.
stem_aliases: Named character vector of file-name stems (as typed, for example a recurring typo) and the catalog stems they stand for. Defaults to none.
analysis_dir: Folder, relative to the working directory (or to the TEMPLECBE_ANALYSIS_ROOT environment variable) or absolute, that holds reports_to_render.xlsx and the .qmd sources shown on the docXwalk sheet. Defaults to none: no manifest or source index is looked up. Use character(0) to clear.
input_dirs: Character vector of folders (relative to the working directory or TEMPLECBE_ANALYSIS_ROOT, or absolute) in which docx_review_extract looks, in order, for .docx files when input_dir is not given. Defaults to none, so the working directory is used. Use character(0) to clear.
fork_mode: Who sees whom in the fork workbooks. "redacted" (default) gives each reviewer a fork without the other fork reviewers' sign-off and note columns (single-blind review). "collaborative" keeps every fork reviewer's columns in every fork, so reviewers can read each other's notes and confirm consensus. Either way the sign-off reviewers' columns and the Word author never go into a fork, and a reviewer's edits to another reviewer's columns are not merged back: only the fork's own columns are.
reset: If TRUE, forget everything set earlier (every argument above) and return to the defaults before applying any other argument given in the same call; the arguments are validated first, so a rejected call changes nothing. review_config(reset = TRUE) alone just restores the defaults. Defaults to FALSE.

Value

A list containing the current configuration options.

Details

Two reviewer categories are supported:

fork_reviewers: reviewers who sign off on individual comments and suggested changes. Each gets a personal fork workbook (review_tracker_<id>.xlsx) that holds only that reviewer's own columns: no other reviewer's columns, no sign-off reviewer's columns and no Word author names appear on any sheet, and the "Documents" sheet carries just the counts and that reviewer's own rolled-up column (there is no "Documents_<id>" sheet in a fork). In a fork the resolved columns reflect that reviewer's own sign-offs only; in the master workbook a document's resolved status is TRUE only when every fork reviewer has signed off on every comment and suggested change in that document.
signoff_reviewers: reviewers who sign off once per document (e.g. a final PI read), tracked on the master workbook's "Documents" sheet (a status column plus a free-text comment column). They are not counted toward resolved, and they never appear in a fork workbook or, by default, in the CSV exports (see csv_reviewer_columns). The master's Comments and SuggestedChanges sheets also carry their column pair, so that anything typed there is kept.

Existing review_tracker_*.xlsx files continue to work: column data is preserved by header union regardless of configuration, but for the app's reviewer-aware behavior (rollups, resolved status, fork filtering) to line up with a file's existing columns, keep configuring the same reviewer identifiers used when that file was created.