v0.9.0

New features

• The summary_rows() and grand_summary_rows() functions have been rewritten extensively to allow more flexibility in providing aggregation information to fns. The documentation for both functions will walk you through all the new ways you can express your aggregations. The formatting of cells in new summary rows is now easier thanks to the new fmt argument. We can provide a single formatting expression or a number of them in a list. The dedicated documentation section for formatting can walk you through all of this. Furthermore, we now have the option to place summary rows at top or bottom of a group (in summary_rows()) or table (in grand_summary_rows()) with the new side argument (#177, #270, #292, #545, #581, #784, #785, #840). (#1159, #1171)

• The data_color() function has been completely overhauled to make it easier to use and also more powerful. It gains a rows argument to better constrain where the colorization should take place. Instead of having to use functions from scales to define the coloring scheme, you can now use the method argument to handle most cases (otherwise, the new fn argument allows for a color-mapping function to be used). The new palette argument allows for flexible input of color palettes. One or more adjacent columns can receive the coloring treatment by way of the new target_columns argument. Coloring can now also be applied in a row-wise fashion by using direction = "row". Despite all of these changes, older code should still work with the revamped function (#116, #633, #637, #1103, #1119). (#1147)

• We now have a much improved cols_merge() function! It has gained a rows argument for limiting which body cells get merged together. The pattern argument now has a more advanced syntax. It now has the following delimiters and rules: (1) { } for arranging single column values in a row-wise fashion, and (2) << >> to surround spans of text that will be removed if any of the contained { } yields a missing value. The new << >> delimiters will help in cases where NA values are present and should be omitted during the merging process (#1167). (#1144, #1168)

• The cols_label() function has been improved by allowing formula expressions to be given to .... With two-sided formulas (e.g., <LHS> ~ <RHS>), the left-hand side corresponds to selections of columns and the right-hand side evaluates to single-length values for the label to apply. Named arguments in the format of <column name> = <label> are still also valid as input for simple mappings of column name to label text. Also, there's a .fn argument, which takes a function to be applied to all provided label values (#1146). (#1255, #1256, #1265)

• Interactive HTML tables can now be generated in gt. Options for enabling interactive table rendering are available in tab_options() and also in opt_interactive(). The collection of use_* arguments in opt_interactive() allow for control of pagination, global search, filtering, sorting, and more. The documentation within opt_interactive() demonstrates how to generate HTML tables with interactive elements (#207). (#902, #1252, #1261, #1278, #1280)

• A new function, cols_label_with(), has been added for relabeling columns via functions. It allows for such functions to be applied to arbitrary sets of columns, taking advantage of tidyselect. It's like dplyr::rename_with() for your gt table, but it doesn't need unique names for labeling (#701). (#1158, thank you @christopherkenny!)

• We added fmt_spelled_num() and vec_fmt_spelled_num() so that numeric values could be transformed to spelled out numbers. Any values from 0 to 100 can be spelled out according to a given locale value. For example, the value 23 will be rendered as "twenty-three" if the locale is an English-language one (or, not provided at all); should a Swedish locale be provided (e.g., "sv"), the output will instead be "tjugotre". (#1198)

• The fmt_index() and vec_fmt_index() functions were added in so that numeric values could undergo transformation to index values, which are usually based on letters (depending on the locale chosen). For example, the value 5 will be rendered as "E" if the locale is an English-language one. The characters chosen for indexing here are based on character sets intended for ordering (often leaving out characters with diacritical marks). (#1189)

• Should body cells contain URLs, the new fmt_url() function can be used to make them navigable links. There are several options provided that control how the links should be styled (conventional underline style or button-like text) (#609). (#1205, #1220, #1221)

• To more easily insert graphics into body cells, we have added the fmt_image() function. This allows for one or more images to be placed in the targeted cells. The cells need to contain some reference to an image file, either: (1) complete http/https or local paths to the files; (2) the file names, where a common path can be provided via path; or (3) a fragment of the file name, where the file_pattern argument helps to compose the entire file name and the path argument provides the path information. (#1240)

• The new fmt_flag() function helps with inserting a flag icon (or multiple) in body cells. The input cells need to contain 2-letter ISO 3166-1 country codes (e.g., Mauritius with the "MU" country code). This function will parse the targeted body cells insert the appropriate flag icon. Multiple flags can be included per body cell by separating country codes with commas. (#1241, #1244, #1251)

• With fmt_bins(), you can format bin or interval syntax (returned from cut()) to a form that presents better in a display table. It's possible to format the values of the intervals with the fmt argument, and, the separator can be modified with the sep argument. (#1250)

• The new fmt_auto() function has been added to automatically apply formatting of various types in a way that best suits the data table provided. For example, the function will attempt to format numbers such that they are condensed to an optimal width, and, currency values can be detected (by currency codes embedded in the column name) and formatted in the correct way (#248). (#1176)

• Three text_*() functions were added to better enable transformations of formatted text in various table locations: text_replace(), text_case_when(), and text_case_match(). (#1238)

• We can use any of 15 system font stacks available in gt! These are families of themed system fonts that are widely available in different computer systems. Examples include 'System UI' (the default font stack), 'Humanist', and 'Neo-Grotesque'. The new helper function system_fonts() can be used to access the various font stack and is useful when specifying font values in the cell_text() function. The opt_table_font() function has been upgraded to include a stack argument (which accepts keywords for the different font stacks). (#1275)

• We now have a new family of functions that let you deal with a multitude of gt tables, all at once. You can create a gt_group object with multiple tables via the gt_group() function, or, split a single table into multiple with gt_split(). There are some advantages to having a group of tables bundled together like this. You could apply options that pertain to all tables yet still access the individual tables to give them their own specialized modifications. They all print together at once too! For HTML, each table will be separated by a line break whereas in paginated formats (e.g., RTF, Word, etc.) the tables are separated by page breaks. (#1216, #1219, #1226, #1248, #1249, #1262, #1264)

• There is now a choice of which Markdown engine to use for Markdown rendering (in fmt_markdown() and vec_fmt_markdown()) and a default default choice. The markdown package can be chosen instead of commonmark to perform conversions of Markdown to HTML and LaTeX (and it's now the default). One of many benefits to this change is the parsing of subscripts (using ~x~) and superscripts (using ^x^) (#725). (#1254)

• Added the footnote formatting options footnotes.spec_ref and footnotes.spec_ftr in tab_options() and exposed these options further in a new function called opt_footnote_spec(). Specifications for footnote formatting can be produced for the footnote marks that alight to the targeted text within cells from various locations in the table, or, for the footnote marks that appear in the table footer. (#1268, #1271)

• The towny dataset was added. It provides a fun, population-based dataset with place names, URLs, coordinate information, and population figures with repetition (values for different census years). (#1184, #1199)

• The metro dataset is new. It has 308 rows, where each corresponds to a different Paris Metro station (#1231).

• The rx_adsl and rx_addv datasets have been added. These clinical trial toy datasets are both featured in a new vignette that explains how gt can be used for table generation in the Pharma space (#1145, #1200; thank you @alex-lauer!).

Documentation enhancements

• Addition of a new vignette (Case Study: Clinical Tables) that uses the rx_adsl and rx_addv datasets. (#1145, #1217, #1224; all contributed by @alex-lauer)

• All examples in the documentation were modernized by using the |> instead of the %>%. A lot of explanatory text was improved through rewriting. The ordering of documentation sections was improved. And functions that had very little documentation (like text_transform()) were greatly improved (#1087). (#1166, #1185, #1192)

• Several small documentation and housekeeping updates. (#1133, #1134, #1135, #1136, #1190, #1206, #1218, #1242, #1246, #1258, #1260, #1266, #1270, #1276, #1281)

• Added a citation file for the package. (#1128, #1277)

Minor improvements and bug fixes

• The fmt_scientific() and fmt_engineering() (and their vec_fmt_*() analogues) now have a exp_style argument for defining the exponent notation. By default this is "x10n" but other options include using a single letter (e.g., "e", "E", etc.), a letter followed by a "1" to signal a minimum digit width of one, or "low-ten" for using a stylized "10" marker. These functions now also have the force_sign_m and force_sign_n arguments for forcing + signs in the mantissa or the exponent. (#1178)

• We now have right-to-left (RTL) text handling and gt will automatically set the appropriate alignment for RTL text in HTML output. This is done through detection of RTL script characters. (#697). (#1202)

• For any columns that are labeled (i.e., have a label attribute), gt will automatically use those labels as column labels (#656). (#1230)

• Tables in the Word output format now have much better Markdown support (#1057, #1067, #1204). (#1142, #1263, #1274)

• Added the "double" option for a border style in cell_borders() (#1132). (#1137)

• The countrypops dataset was updated with recent (as of January 2023) World Bank data that revises population estimates and brings the final year up to 2021. All examples, tests, and articles using the dataset were also updated. (#1191)

• Fixed a conflict between tab_spanner_delim() and cols_label() (#1130). (#1173)

• Added a fix for a summary_rows() problem where summary rows that were generated with only a single group would not display the summary rows at all. (#1188)

• HTML output tables no longer have two <thead> elements when they display a header and column labels (#1235). (#1237)

• Incorporated a CSS fix for superscripted footnote marks. Prior to the change, footnote marks were never styled within gt tables rendered in Quarto documents. (#1210)

• We now remove knitr flags when saving an RTF file through the gtsave() function (#674). (#1153)

• Fixed LaTeX rendering of tables using the xelatex engine by removing the unneeded "amsmath" LaTeX package from the default set of LaTeX dependencies. This set is now modifiable through the gt.latex_packages option (#751). (#1279)

• Fixed some border-related issues for gt tables published in R Markdown and Quarto documents (#1120). (#1234)

• Added the quarto.use_bootstrap and quarto.disable_processing options in tab_options() to control whether Quarto will add Bootstrap classes to a gt table or ignore a gt table. (#1222, #1227)

• Fixed the unintended appearance of a bottom border in the column spanner row for some CSS environments. (#1209)

• Fixed the SCSS class .gt_footnote_marks to ensure that the footnote mark is at a consistent height from the baseline in different CSS environments. (#1187)

• Fixed an error in the SCSS class .gt_footnote where the wrong types of padding were specified. (#1236)

• Added frequency levels to cli_abort() warnings throughout package (#1160). (#1169)