read_ods() and list_ods_sheets() can also be used to process flat ods files. read_fods() and list_fods_sheets() are still available, but not as the so-called “common interface.”read_ods() and read_fods() have two arguments - trim_ws and n_max which are the same as the arguments of readxl::read_excel().cpp11 versionappend and update of write_ods in C++Significant speed improvement; also xml2 is no longer a dependency.
col_names / row_namesPrior the previous stable release, <= 1.9.0, reading single-row / single-column with col_names = TRUE / row_names = TRUE produced errors.
In v1.9.0 (and the stable version v2.0.0 on CRAN), reading single-row / single-column (F)ODS with col_names = TRUE / row_names = TRUE will override the two parameters and return a non-empty data frame. This behaviour is consistent with other data reading R functions (see #146) such as readxl::read_xlsx(), readr::read_csv(), data.table::fread(), and openxlsx::read.xlsx(). For these functions, either a empty or zero-row data.frame is returned.
We changed this behaviour. The following will return a zero-row data.frame by default.
However, the previous behaviour is in the stable release and backward compatibility is needed. If you need that previous behaviour, please set the options("readODS.v200" = TRUE)
write_ods and write_fods allow list of data framesFix #56; and it is now the same as writexl::write_xlsx().
list_fods_sheets() and read_fods() cannot accept ~ as pathlist_fods_sheets() can’t guard non-fods XML disguised as fodsods_sheetsSee discussion #133
col_types can be character (“shorthand”) or listfix #135 and the review by Dr Ruedni
# Specifying col_types as shorthand, the third column as factor; other by guessing
read_ods("starwars.ods", col_types = "??f")
# Specifying col_types as list
read_ods("starwars.ods", col_types = list(species = "f"))write_fods is available, fix #103
write_odswrite_ods has been partially rewritten in C++ #140
To ensure UTF-8 everywhere, fix #107
Bump requirement to R>=3.6
dttm column was incorrectly written with one more column. It’s now fixed.
write_ods(na_as_string) behaviourThe default for na_as_string was getOption("write_ods_na", default = FALSE) in v1.8. The default now is FALSE, but it writes NA as blank cell (instead of the buggy behaviour of writing NA in the original type, which is rendered as 0 by LibreOffice for numeric types. see #79). This behaviour is compatible with the default of writexl::write_xlsx.
The behaviour of na_as_string = TRUE is the same as in v1.8: writes NA as string.
get_num_sheets_in_{f}ods()The descendant of getNrOfSheetsInODS() is not very useful. If you really need to have the similar function:
write_ods()overwrite: always TRUEverbose: always FALSEwrite_ods()as_tibble and .name_repair as arguments. If as_tibble is true, outputs as a tibble using tibble::as_tibble() passing on .name_repair (default being "unique"). By default as_tibble is set to TRUE.check_names argument. All name repairs are now dealt with using vctrs::vec_as_names(). This will significantly change the default names given to outputs. (Names in the style of check_names = TRUE can be obtained by setting .name_repair = minimal, although this is not advised)ods_sheets to v3There are many reverse dependencies using ods_sheets.
NEWS.md file to track changes to the package.These have been deprecated for several years.
include_external_data as an argument (FALSE by default). This hides stored data from external sources not normally accessible to the user.range argument, e.g. Range = "Sheet2!A2:B7". If this and the sheets argument are given, this is preferred.NA.as_tibble and .name_repair as arguments. If as_tibble is true, outputs as a tibble using tibble::as_tibble() passing on .name_repair (default being "check_unique").read_fods(), list_fods_sheets(), get_num_sheets_in_fods(). These work the same way as their analogue ODS functions. The extension does not need to be fods, however they do need to conform to the OO specification.