Skip to contents

tests

tests.Rmd

Motivation

The Shiny modules in dfdiffs (upload, select, and compare) keep their reactive logic in server functions. These are tested with shiny::testServer(), which runs a module’s server function without starting an app or a browser: the test sets inputs with session$setInputs(), and then checks the reactive values (or rendered outputs) the module returns. The base-R comparison engine (compare_values(), diff_values_by_key(), and the join helpers) is tested directly with plain test_that() blocks, no Shiny session needed.

The tests live in tests/testthat/. Run them with:

devtools::test()

There are currently 8 test files containing 36 tests (test_that() blocks) and 82 expectations.

Test inventory

Each row is one test_that() block. The first column gives the file and the test name (the string passed to test_that()), the second column is the app feature or function the test protects, and the third describes what the test does and expects.

Test name / file Feature Test
test-mod_upload.R
mod_upload_server() reads an uploaded base file into base_data() 		Upload module: base file Uploads InitialData.csv as base_file (with base_new_name = "InitialData") and expects base_data() to be a data frame with 5 rows and a subject_id column.
test-mod_upload.R
mod_upload_server() reads an uploaded compare file into comp_data() 		Upload module: compare file Uploads ChangedData.csv as comp_file (with comp_new_name = "ChangedData") and expects comp_data() to be a data frame with 5 rows and a subject_id column.
test-mod_upload.R
mod_upload_server() reads an uploaded xlsx sheet into base_data() 		Upload module: xlsx file Writes InitialData to a temp .xlsx file with openxlsx::write.xlsx(), uploads it as base_file with a sheet name, and expects base_data() to match InitialData’s row count and columns.
test-mod_upload.R
mod_upload_server() uses base_new_name for base_name() when supplied 		Upload module: dataset naming Uploads InitialData.csv with base_new_name = "InitialData" and expects base_name() to return "InitialData".
test-mod_select.R
mod_select_server() builds joined base/compare data from selected columns 		Select module: join column Passes InitialData and ChangedData in as the uploaded data, selects every column, and sets by = c("subject_id", "record"). Expects base_join_col_data() and comp_join_col_data() to be data frames with the same number of rows as their source data, containing join_column, join_source, and data_source columns (with data_source equal to the dataset names).
test-mod_select.R
mod_select_server() falls back to a row-by-row comparison with no ‘by’ columns 		Select module: row-by-row comparison Selects every column but leaves by empty. Expects base_join_col_data() to have no join_column, but to keep the data_source column.
test-mod_compare.R
mod_compare_server() detects changed values via the ‘join_column’ path 		Compare module: changed data Builds joined base/compare data with create_join_column() (on subject_id and record) and passes it in as the selected data. Expects compare_cols() to include join_column, changed_data() to contain diffs and diffs_byvar (with at least one row in diffs_byvar), and new_data() and deleted_data() to have 0 rows.
test-mod_compare.R
mod_compare_server() download report works via the ‘join_column’ path 		Compare module: report download With joined base/compare data as the selected data, reads the downloaded .xlsx file and expects it to contain the “New Data”, “Deleted Data”, “Changed Data”, and “Review Changes” sheets, with the “Review Changes” sheet including a join_column.
test-mod_compare.R
mod_compare_server() download report works via the row-by-row path 		Compare module: report download (no join) With no join column in the selected data, expects compare_cols() to exclude join_column, and the downloaded report’s “Review Changes” sheet to include a rownumber column instead.
test-compare_data.R
compare_data() returns new, deleted, and changed comparison tables 		compare_data() Runs compare_data() on Roster2021/Roster2022 with by = "subject_id" and expects a named list of new_data, deleted_data, changed_num_diffs, changed_var_diffs, with new_data/deleted_data non-empty data frames.
test-create_comparison_report.R
create_comparison_report() writes all six sheets to the xlsx file 		create_comparison_report() Runs create_comparison_report() on Roster2021/Roster2022 and expects the output file to exist with all six expected sheets, and the “New Data” sheet to be non-empty.
test-launch_app.R
(four tests, one per launcher) 		App launchers Calls launch_app(), launch_app_dev(), launch_upload_demo(), and launch_select_demo() and expects each to return a shiny.appobj without error.
test-join_helpers.R
(seven tests) 		select_cols(), anti_join_base(), left_join_base() Covers column subsetting (including a missing-column error), multi-column-key matching, NA-in-key handling, and left-join column attachment/NA fill for unmatched rows.
test-compare_values.R
(fourteen tests) 		compare_values(), diff_values_by_key() Covers NA-vs-NA and NA-vs-value handling, exact string comparison, numeric tolerance (default and custom tolerance/scale), integer/double comparison, Date vs. same-instant POSIXct (equal) and different POSIXct (not equal), factor/character label comparison (default and strict_factor = TRUE), vectorization, and diff_values_by_key()’s row-matching, zero-diff, and class-mismatch behavior.

Test data

The tests use the package’s own datasets, so they don’t depend on files outside of dfdiffs:

  • InitialData and ChangedData (from data/) are the base and compare data for the select and compare tests.
  • Roster2021/Roster2022 are the base and compare data for the compare_data()/create_comparison_report() tests.
  • inst/extdata/csv/InitialData.csv and inst/extdata/csv/ChangedData.csv are the files uploaded in the CSV upload tests.

Call structure

Each Shiny-module test targets a module’s server function; the call trees below show the dfdiffs functions those tests exercise indirectly (for example, the compare module tests also run create_new_data(), create_deleted_data(), and create_modified_data()). They were generated from the package source with stackcallr (pak::pak("mjfrigaard/stackcallr")); only functions defined in dfdiffs are shown.

library(stackcallr)
call_tree_dir("R", root = "mod_upload_server")
call_tree_dir("R", root = "mod_select_server")
call_tree_dir("R", root = "mod_compare_server")
█─mod_upload_server
├─█─upload_data
│ └─load_flat_file
├─base_react_theme
└─comp_react_theme
█─mod_select_server
├─select_cols
├─base_react_theme
├─comp_react_theme
├─info_react_theme
└─█─create_join_column
  └─select_cols
█─mod_compare_server
├─%nin%
├─info_react_theme
├─█─create_new_data
│ ├─anti_join_base
│ ├─select_cols
│ ├─rename_join_col
│ └─create_new_column
├─new_react_theme
├─█─create_deleted_data
│ ├─anti_join_base
│ ├─select_cols
│ ├─rename_join_col
│ └─create_new_column
├─deleted_react_theme
├─█─create_modified_data
│ ├─select_cols
│ ├─█─diff_values_by_key
│ │ ├─compare_values
│ │ └─class_diffs
│ ├─rename_join_col
│ └─create_new_column
├─select_cols
├─changed_react_theme
└─left_join_base

test-compare_data.R, test-create_comparison_report.R, test-join_helpers.R, and test-compare_values.R call their target functions directly and don’t go through a Shiny module, so there’s no call tree to show for them beyond what’s already in the getting-started vignette.

What isn’t tested

The current tests cover the reactive logic of the three modules, the base-R comparison engine, compare_data(), create_comparison_report(), and a build smoke test for every launch_*() entry point. These areas don’t have tests yet: