Skip to contents

Changelog

Source: NEWS.md

autoslider.core 0.3.6.7

  • Placeholder slides (section dividers, cover and other non-TLG pages) can now be declared in a speclayout and placed by their order among the TLG entries with place_speclayout_placeholders(); no page numbers need to be typed. Built-in types are section (new append_section_slides()), title and custom (new append_custom_slides(), any layout, placeholders by label). On basic.pptx, section uses the “Section Header” layout, which has a title and a text box but no subtitle: a subtitle goes into the text box, before text. Text may contain {token}s filled from metadata. Layout roles (placeholder_layouts(), validate_placeholder_layouts()) and slide types (placeholder_types()) can be extended per call or through the autoslider.core.placeholder_layouts / autoslider.core.placeholder_types options, and append_placeholder_slides() / placeholder_slide() are the building blocks for new types. import_layouts() copies layouts between templates. Adds xml2 and zip to Imports. Moved from autoslideR.
  • generate_slides() no longer fails with “formal argument "lpp" matched by multiple actual arguments” when lpp/cpp reach it both through a spec entry and as an argument (as study scripts that pass every spec key to generate_slides() do). The spec value still wins; an lpp/cpp passed only as an argument now acts as a per-call default instead of raising an error.
  • A program that fails with an rlang/cli-style error (e.g. a dplyr/tidyselect “Can’t select columns that don’t exist”) no longer aborts generate_outputs(). These errors leave e$message empty, which made autoslider_error() fail inside the error handler; the handler now uses conditionMessage(), so the output becomes an autoslider_error with the full message and the remaining outputs still run. generate_slides() now also accepts a single failed output on its own (reporting and skipping it, as it does inside a list).
  • The “;” separator that flextable adds after an inline footnote now uses the same font size as the footnote text. Previously only the text was resized, so with a small font_size$footer the semicolon showed at the larger table default size.

autoslider.core 0.3.5

  • Minor bug fixing in tlg_template.R.

autoslider.core 0.3.4

CRAN release: 2026-09-29

  • Per-slide pagination: a spec entry may now carry an optional lpp: (lines per page) and cpp: (columns per page), overriding the deck-wide t_lpp/t_cpp (tables) and l_lpp/l_cpp (listings) arguments of generate_slides() for that slide only, so a short table and a long one can use different densities in the same deck. Entries without the fields are unchanged. Note that the gtsummary path still recomputes lpp from the slide height (#121), so a spec lpp does not yet affect it (#126).
  • Confidential footnotes now render at 8 pt by default, or follow the resolved body size when one is supplied. The size is applied on both the decorated and decor = FALSE slide paths, so no PPTX/XML post-processing is needed.
  • with_font_sizes() now forwards additional formatter arguments supplied by its caller.
  • Added t_mh_slide() integration and a reproducibly generated synthetic eg_admh example dataset for medical-history summaries.
  • Added add_ai_story() (and an add_ai_story MCP tool): a post-processing step that reads a generated .pptx, asks an LLM to tell the story of the tables, and inserts real content slides – a summary section at the front and a conclusions section at the end – instead of hidden speaker notes. get_ai_story() and add_story_slides() expose the LLM call and the (network-free) slide insertion separately. get_ellmer_chat() now also supports the "anthropic" platform. get_ai_story() falls back to plain-text JSON mode for providers that do not support native structured output (e.g. DeepSeek).
  • Per-slide font sizes: generate_slides() now reads an optional font_size: block (body/header/footer) and table_format from each output’s spec entry, and a deck-wide font_size argument. Sizes are applied via the new exported with_font_sizes() helper, which injects sizes into any table formatter (only forwarding sizes the formatter accepts).
  • autoslider_format() and black_format_tb() gain a footer_font_size argument.
  • to_flextable.data.frame(): font_size now defaults to NULL and the uniform font override is only applied when it is explicitly set, so font sizes coming from table_format are no longer overwritten. Note: plain data-frame slides that previously rendered at the hard-coded 9pt now follow the sizes from table_format unless font_size is supplied.
  • Added apply_tokens() to substitute {token} placeholders (e.g. {study}) in titles, footnotes and placeholder slides with values from a user-controlled metadata list, so study-level text can be driven from metadata instead of edited in the deck.
  • Added read_metadata() and an example metadata.yml (system.file("metadata.yml", package = "autoslider.core")) so {token} values can be kept in a yaml file. read_spec() now also accepts a metadata file path for its metadata argument and reads it for you, e.g. read_spec("spec.yml", metadata = "metadata.yml").

autoslider.core 0.3.3

CRAN release: 2026-08-29

  • Adding mcp server.
  • Fix gtsummary code, now supports NEST 2.0.

autoslider.core 0.3.2

CRAN release: 2026-01-12

  • Update dependency version.

autoslider.core 0.3.1

CRAN release: 2025-09-25

  • Fix vignette builder.

autoslider.core 0.3.0

  • Fix layout, adding assertion.

autoslider.core 0.2.9

  • Dependency version bump for officer version 0.7.0.

autoslider.core 0.2.8

CRAN release: 2025-08-25

  • Renamed AI functionality from footnotes to speaker notes in documentation and function references.
  • Added CRAN test skips to reduce test execution time.
  • Removed styler from suggests.

autoslider.core 0.2.7

CRAN release: 2025-07-25

  • Revert officer dependency on function layout_default.
  • Realign table to the slide centre.
  • Remove thinking messages.

autoslider.core 0.2.6

CRAN release: 2025-07-22

  • Added support for custom templates via symbolic links in inst/templates/.
  • Added vignette tlg_templates , adding_templates, use_LLM, and using_format.
  • Added support for use_templates for specified packages.
  • Added support for adding placeholder slides (e.g. title, section headers) using append_all_slides()
  • Enabled function generation from templates.
  • Enabled automatic plot title generation.
  • Split out specific plot functions (e.g., g_vs_slide) from g_mean_general().
  • Integrated AI insights to data analytics, written detailed instructions in vignette use_LLM.
  • Provided the option of making plots editable or fixed on generated slides.
  • Fixed ggplot graphs location and size on generated slides.

autoslider.core 0.2.5

CRAN release: 2025-06-13

  • Dropping dependencies level.
  • Adding ellmer dependency, enable AI prompt.
  • Bug fix in side by side plot rendering.
  • Code enhancement when using split_rows_by and analysis on the same variable multiple times, enhancing the table paths. Upstream enhancement from rtables.

autoslider.core 0.2.4.433

Miscellaneous

  • Align upstream packages (formatters/rtables/rlistings/tern) to the latest releases.
  • Allow structured header for side-by-side tables.
  • Fixed bugs in decoration.
  • Support gtsummary.
  • Support slides creation from rds files.

autoslider.core 0.2.3

CRAN release: 2025-01-17

Miscellaneous

  • Align upstream packages (formatters/rtables/rlistings/tern) to the latest releases.

autoslider.core 0.2.2

CRAN release: 2025-01-08

Miscellaneous

  • Plots in slides are now editable.
  • Keep indentation in tables.
  • Added vignette for open-sourcing.

autoslider.core 0.2.1

CRAN release: 2024-12-23

Miscellaneous

  • Trim l_ae_slide example.

autoslider.core 0.2.0

New features

  • Added new templates g_lb_slide, g_eg_slide, g_vs_slide, l_ae_slide and t_ds_slide.

Miscellaneous

  • Added vignette
  • Improved test coverage

autoslider.core 0.1.0

CRAN release: 2024-11-27

Miscellaneous

  • First release.