Merge pull request #35 from /issues/5

Implement regularize

DESCRIPTION

@@ -16,7 +16,7 @@ LazyData: true
 Suggests:
     knitr,
     rmarkdown,
-    testthat
+    testthat (>= 2.1.0)
 Imports:
     cytominer,
     docopt(>= 0.4.5),

NAMESPACE

@@ -4,6 +4,7 @@ export(aggregate)
 export(annotate)
 export(normalize)
 export(preselect)
+export(regularize)
 export(sample)
 importFrom(magrittr,"%<>%")
 importFrom(magrittr,"%>%")

R/regularize.R

@@ -0,0 +1,57 @@
+#' Regularize backend file
+#'
+#' \code{regularize} regularizes backend file.
+#'
+#' @param input_file     Input file. CSV or SQLite only.
+#' @param output_file    Output file. Same format as input_file.
+#' @importFrom magrittr %>%
+#' @importFrom magrittr %<>%
+#' @export
+regularize <- function(input_file, output_file) {
+
+  input_extension <- tools::file_ext(input_file)
+
+  output_extension <- tools::file_ext(output_file)
+
+  if(input_extension != output_extension) {
+    stop("Input and output filetypes are different.")
+  }
+
+  if(!(input_extension %in% c("csv", "sqlite"))) {
+    stop("Unsupported filetype. Only csv or sqlite is supported.")
+  }
+
+  if(input_extension == "sqlite") {
+    file.copy(input_file, output_file)
+
+    db <- DBI::dbConnect(RSQLite::SQLite(), output_file)
+
+    image <- DBI::dbGetQuery(db, "SELECT * from Image;")
+
+    futile.logger::flog.info(
+      "Stripping Image_ from column names of Image table...")
+
+    image %<>%
+      setNames(names(image) %>%
+                 stringr::str_remove_all("^Image_"))
+
+    DBI::dbRemoveTable(db, "Image")
+
+    DBI::dbWriteTable(db, "Image", image)
+
+    image <- DBI::dbGetQuery(db, "SELECT * from Image;")
+
+    DBI::dbDisconnect(db)
+  } else {
+
+    input_data <- suppressMessages(readr::read_csv(input_file))
+
+    input_data %<>%
+      setNames(names(input_data) %>%
+                 stringr::str_replace_all("^Image_Metadata", "Metadata"))
+
+    input_data %>% readr::write_csv(output_file)
+
+  }
+
+}

inst/scripts/cytotools_annotate

@@ -6,7 +6,7 @@ extends <- methods::extends
 'annotate
 
 Usage:
-  annotate.R -b <id> -p <id> [-c <str> -d -j <file> -m <str> -o <file> -w <file>]
+  cytotools_annotate -b <id> -p <id> [-c <str> -d -j <file> -m <str> -o <file> -w <file>]
 
 Options:
   -h --help                               Show this screen.

inst/scripts/cytotools_normalize

@@ -6,7 +6,7 @@ extends <- methods::extends
 'normalize
 
 Usage:
-cytotools_normalize [-i <file> -o <file> -s <query> -g -q <file> -c <str> -m <str> -r <str> -b <id> -p <id> -w <dir> -j <str> -k <str> -l <str>]
+  cytotools_normalize [-i <file> -o <file> -s <query> -g -q <file> -c <str> -m <str> -r <str> -b <id> -p <id> -w <dir> -j <str> -k <str> -l <str>]
 
 Options:
   -h --help                                Show this screen.

inst/scripts/cytotools_preselect

@@ -6,7 +6,7 @@ extends <- methods::extends
 'preselect
 
 Usage:
-  preselect.R -i <file> -r <list> [-b <id>] [-n <n>] [-s <query>] [-c <n>] [-w <dir>]
+  cytotools_preselect -i <file> -r <list> [-b <id>] [-n <n>] [-s <query>] [-c <n>] [-w <dir>]
 
 options:
   -h --help                        Show this screen.

inst/scripts/cytotools_regularize.R

@@ -0,0 +1,19 @@
+#!/usr/bin/env Rscript
+
+# https://github.com/tidyverse/dplyr/issues/1760
+extends <- methods::extends
+
+'regularize
+
+Usage:
+  cytotools_regularize -i <file> -o <file>
+
+Options:
+  -h --help                   Show this screen.
+  -i <file> --input=<file>    Input file. CSV or SQLite only.
+  -o <file> --output=<file>   Output file. Same format as input_file' -> doc
+
+opts <- docopt::docopt(doc)
+
+cytotools::regularize(input_file = opts[["input"]],
+                      output_file = opts[["output"]])

inst/scripts/cytotools_sample

@@ -6,7 +6,7 @@ extends <- methods::extends
 'sample
 
 Usage:
-  sample -b <id> -f <pattern> [-n <n>] -o <file> [-w dir]
+  cytotools_sample -b <id> -f <pattern> [-n <n>] -o <file> [-w dir]
 
 Options:
   -h --help                             Show this screen.

tests/testthat/test-regularize.R

@@ -0,0 +1,65 @@
+context("regularize")
+
+test_that("`regularize` regularizes SQLite file", {
+  futile.logger::flog.threshold(futile.logger::WARN)
+
+  output_file <- tempfile("SQ00015116_regularized", fileext = ".sqlite")
+
+  input_file <-
+    system.file(
+      "extdata", "backend", "batch0", "SQ00015116",
+      "SQ00015116.sqlite",
+      package = "cytotools"
+    )
+
+  regularize(input_file, output_file)
+
+  db <- DBI::dbConnect(RSQLite::SQLite(), input_file)
+
+  input_image <- DBI::dbGetQuery(db, "SELECT * from Image;")
+
+  DBI::dbDisconnect(db)
+
+  db <- DBI::dbConnect(RSQLite::SQLite(), output_file)
+
+  output_image <- DBI::dbGetQuery(db, "SELECT * from Image;")
+
+  DBI::dbDisconnect(db)
+
+  input_image %<>%
+    setNames(names(input_image) %>%
+               stringr::str_remove_all("^Image_"))
+
+  expect_equal(output_image, output_image)
+
+  file.remove(output_file)
+
+})
+
+test_that("`regularize` regularizes CSV file", {
+  futile.logger::flog.threshold(futile.logger::WARN)
+
+  output_file <- tempfile("SQ00015116_regularized", fileext = ".csv")
+
+  input_file <-
+    system.file(
+      "extdata", "backend", "batch0", "SQ00015116",
+      "SQ00015116.csv",
+      package = "cytotools"
+    )
+
+  regularize(input_file, output_file)
+
+  input_data <- readr::read_csv(input_file)
+
+  output_data <- readr::read_csv(output_file)
+
+  input_data %<>%
+    setNames(names(input_data) %>%
+               stringr::str_remove_all("^Image_"))
+
+  expect_equal(input_data, output_data)
+
+  file.remove(output_file)
+
+})