allow for multiple trailing and/or leading whitespace in header (#63)

RMI-PACTA · Jan 15, 2024 · 8e2543d · 8e2543d
1 parent b0997c0
commit 8e2543d
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 4 deletions.
diff --git a/R/determine_headers.R b/R/determine_headers.R
@@ -59,9 +59,9 @@ determine_headers <- function(filepath) {
     num_of_cols <- length(headers)
 
     if (num_of_cols >= 3) {
-      isin_col <- grep(pattern = "^[[:space:]]?isin[[:space:]]?$", x = headers, ignore.case = TRUE, value = TRUE)
-      market_value_col <- grep(pattern = "^[[:space:]]?market[._ ]{0,1}value[[:space:]]?$", x = headers, ignore.case = TRUE, value = TRUE)
-      currency_col <- grep(pattern = "^[[:space:]]?currency[[:space:]]?$", x = headers, ignore.case = TRUE, value = TRUE)
+      isin_col <- grep(pattern = "^[[:space:]]*isin[[:space:]]*$", x = headers, ignore.case = TRUE, value = TRUE)
+      market_value_col <- grep(pattern = "^[[:space:]]*market[._ ]{0,1}value[[:space:]]*$", x = headers, ignore.case = TRUE, value = TRUE)
+      currency_col <- grep(pattern = "^[[:space:]]*currency[[:space:]]*$", x = headers, ignore.case = TRUE, value = TRUE)
 
       if (num_of_cols > 3) {
         investor_name_col <- grep(pattern = "investor", x = headers, ignore.case = TRUE, value = TRUE)

diff --git a/tests/testthat/test-determine_headers.R b/tests/testthat/test-determine_headers.R
@@ -22,7 +22,35 @@ test_that("column names with no underscore are properly determined", {
 test_that("column names with leading and lagging whitespace are properly determined", {
   proper_names <- c("investor_name", "portfolio_name", "isin", "market_value", "currency")
   lead_and_lag_whitespace <- withr::local_tempfile()
-  writeLines(" investor_name,portfolio_name , isin ,market_value,currency\nx,y,z,1,a", lead_and_lag_whitespace)
+  writeLines(" investor_name,portfolio_name , isin , market_value,currency \nx,y,z,1,a", lead_and_lag_whitespace)
+  expect_setequal(names(determine_headers(lead_and_lag_whitespace)), proper_names)
+})
+
+test_that("column names with leading and lagging whitespace are properly determined (double-padded)", {
+  proper_names <- c("investor_name", "portfolio_name", "isin", "market_value", "currency")
+  lead_and_lag_whitespace <- withr::local_tempfile()
+  writeLines("  investor_name,portfolio_name  ,  isin  ,  market_value,currency \nx,y,z,1,a", lead_and_lag_whitespace)
+  expect_setequal(names(determine_headers(lead_and_lag_whitespace)), proper_names)
+})
+
+test_that("column names with leading and lagging whitespace are properly determined (tab-padded)", {
+  proper_names <- c("investor_name", "portfolio_name", "isin", "market_value", "currency")
+  lead_and_lag_whitespace <- withr::local_tempfile()
+  writeLines("\tinvestor_name,portfolio_name\t,\tisin\t,\tmarket_value,currency \nx,y,z,1,a", lead_and_lag_whitespace)
+  expect_setequal(names(determine_headers(lead_and_lag_whitespace)), proper_names)
+})
+
+test_that("column names with leading and lagging whitespace are properly determined (double-tab-padded)", {
+  proper_names <- c("investor_name", "portfolio_name", "isin", "market_value", "currency")
+  lead_and_lag_whitespace <- withr::local_tempfile()
+  writeLines("\t\tinvestor_name,portfolio_name\t\t,\t\tisin\t\t,\t\tmarket_value,currency \nx,y,z,1,a", lead_and_lag_whitespace)
+  expect_setequal(names(determine_headers(lead_and_lag_whitespace)), proper_names)
+})
+
+test_that("column names with leading and lagging whitespace are properly determined (space-and-tab-padded)", {
+  proper_names <- c("investor_name", "portfolio_name", "isin", "market_value", "currency")
+  lead_and_lag_whitespace <- withr::local_tempfile()
+  writeLines(" \tinvestor_name,portfolio_name\t ,\t isin\t\t, \tmarket_value,currency \nx,y,z,1,a", lead_and_lag_whitespace)
   expect_setequal(names(determine_headers(lead_and_lag_whitespace)), proper_names)
 })