toolero

• Added write_clean_csv() for writing data frames to CSV with clean column names. Applies janitor::clean_names() if column names are not already clean and reports affected columns via cli feedback.
• Added check_project() for auditing a project directory against toolero conventions. Checks for expected folders, an .Rproj file, renv.lock, a git repository, a README, a .gitignore, and hidden files such as .RData or .Rhistory. Operates in two modes: a cli report (default) or a tibble return for programmatic use (error = FALSE).
• Added qmd_to_r() for extracting R code chunks from any .qmd file into a standalone .R script via knitr::purl(). The output path defaults to the same directory as the input with the extension replaced. The documentation argument controls how much context is preserved in the extracted script.
• Added generate_project_config() for writing a skeleton YAML project configuration file pre-filled with the standard toolero folder structure. Intended to be edited by the user and passed to init_project() via the new config argument. filename is required and explicit; path defaults to ".". An overwrite argument (default FALSE) guards against accidental replacement of an existing config. The file extension is normalized to .yml regardless of what is supplied.
• Added Palmer Penguins attribution (Horst, Hill & Gorman, 2020) to the template .qmd, the create_qmd() roxygen @details section, and a provenance note in inst/extdata/.

Breaking changes

• init_project(): the standard folder structure has been revised to better reflect research workflow conventions established by The Carpentries and UW-Madison Libraries. The new standard set is data-raw/, data/, scripts/, output/figures/, output/tables/, and reports/. The previous set (data/, data-raw/, images/, plots/, results/, scripts/, docs/, R/) is no longer created by default.
• init_project(): extra_folders has been renamed to custom_folders. The argument now supports a dplyr-select-like syntax: bare names add folders (e.g. "models"), names prefixed with "-" suppress creation of that folder from the resolved set (e.g. "-output/figures"). Suppression removes only the named leaf -- parent directories are preserved. Duplicate additions emit an informational message and are skipped; references to non-existent folders via "-" emit a warning.
• create_qmd(): no longer copies styles.css and header.html from the package into the project. Custom styling is now controlled exclusively by the new use_style argument. Projects that relied on create_qmd() copying UW-branded assets should use init_project(uw_branding = TRUE) to scaffold those files, then pass use_style = TRUE to create_qmd() to wire them into the YAML.
• create_qmd(): sample data is now copied into data-raw/ instead of data/, consistent with init_project()'s folder structure.

New features (continued from above)

• init_project(): added config argument. When supplied, the folder list in the YAML file replaces the built-in standard structure entirely. custom_folders is still applied on top of the config-derived set. Configs are produced by generate_project_config() and can be stored in the user home directory for reuse across project types.
• create_qmd(): added include_examples argument (default TRUE). When TRUE, copies a sample dataset (sample.csv) into data-raw/, a placeholder logo (logo.png) into assets/, and uses a worked example template with a params block referencing the sample data. When FALSE, creates a blank skeleton .qmd with only the YAML header and a setup chunk -- no sample data, no logo, no example analysis block.
• create_qmd(): added use_style argument (default FALSE). Accepts FALSE (no custom styling), TRUE (scans assets/ for .css and .html files), or a directory path (scans that directory instead). When exactly one .css file is found, it is added as css: in the YAML. When exactly one .html file is found, it is added as include-before-body:. If multiple files of either type are found, the function errors and asks the user to specify which one to use via yaml_data. This decouples styling from init_project() and supports non-UW branding workflows.
• Added inst/templates/skeleton.qmd -- a minimal Quarto template used when include_examples = FALSE. Contains the YAML header, a setup chunk with library(toolero), and a single placeholder heading.
• Added inst/templates/logo.png -- a placeholder logo image copied into assets/ when include_examples = TRUE. Reads "your logo goes here" so the user knows to replace it with their own branding.
• write_by_group(): group_col now accepts a character vector of column names, enabling grouping by more than one column at once. Sanitized filenames join multiple columns with -- (e.g. group_col = c("species", "sex") on an Adelie male produces adelie--male.csv); only combinations actually present in the data produce files, not the full cross-product of possible values. When manifest = TRUE, the manifest gains one column per grouping variable (holding the raw, unsanitized value) in addition to a composite group_value column joining the raw values with " | ". Single-column calls are unaffected -- filenames, manifest schema, and behavior are unchanged from previous versions.
• write_by_group(): added drop_na argument (default TRUE). Rows with a missing value in any grouping column are dropped before splitting, with a cli message reporting how many rows were dropped and from which column(s) -- this was previously silent, undocumented behavior inherited from split(). Set drop_na = FALSE to instead treat missing values as their own group rather than dropping them.

Bug fixes

• create_qmd(): use_style = TRUE now correctly copies rci-banner.png from inst/assets/ into the project assets/ directory. Previously the banner was only copied inside the include_examples block and was silently omitted when use_style = TRUE was combined with include_examples = FALSE.
• create_qmd(): filename argument now normalizes the file extension to .qmd via fs::path_ext_set(). Passing "my-document" and "my-document.qmd" both produce my-document.qmd; a double extension is never added.
• init_project(): path construction now uses fs::path() throughout rather than glue::glue("{path}/{folder}"), ensuring correct behavior on all platforms.
• init_project(): branding files are now copied from inst/assets/ rather than inst/extdata/, consistent with the rest of the package.