Implement str_like() closes #280

NAMESPACE

@@ -20,6 +20,7 @@ export(str_glue)
 export(str_glue_data)
 export(str_interp)
 export(str_length)
+export(str_like)
 export(str_locate)
 export(str_locate_all)
 export(str_match)

NEWS.md

@@ -9,6 +9,9 @@
 
 * New `str_split_n()` function to extract only a single piece from a string
   (#278, @bfgray3).
+
+* New `str_like()` function which allows the use of SQL wildcards
+  (#280, @rjpat).
 
 # stringr 1.4.0
 

R/detect.r

@@ -104,3 +104,41 @@ str_ends <- function(string, pattern, negate = FALSE) {
          }
   )
 }
+
+#' Detect the presence of a pattern in the string using SQL LIKE convention.
+#'
+#' @description
+#' Follows the structure of the SQL `LIKE` operator:
+#' * Must match the entire string
+#' * `_` matches a single character (like `.`)
+#' * \verb{\%} matches any number of characters (`.*`)
+#' * \verb{\\\%} and `\_` match literal \verb{\%} and `_`
+#' * The match is case insentistive by default
+#'
+#' @inheritParams str_detect
+#' @param pattern A charcter vector containing a SQL "like" pattern.
+#'   See above for details.
+#' @param ignore_case Ignore case of matches?  Defaults to `TRUE` to match
+#'   the SQL `LIKE` operator.
+#' @return A logical vector.
+#' @export
+#' @examples
+#' fruit <- c("apple", "banana", "pear", "pinapple")
+#' str_like(fruit, "app")
+#' str_like(fruit, "app%")
+#' str_like(fruit, "ba_ana")
+#' str_like(fruit, "%APPLE")
+str_like <- function(string, pattern, ignore_case = TRUE) {
+  if (!is.character(pattern) || is.object(pattern)) {
+    stop("`pattern` must be a character vector", call. = FALSE)
+  }
+
+  pattern <- regex(like_to_regex(pattern), ignore_case = ignore_case)
+  stri_detect_regex(string, pattern, opts_regex = opts(pattern))
+}
+
+like_to_regex <- function(pattern) {
+  converted <- stri_replace_all_regex(pattern, "(?<!\\\\|\\[)%(?!\\])", "\\.\\*")
+  converted <- stri_replace_all_regex(converted, "(?<!\\\\|\\[)_(?!\\])", "\\.")
+  paste0("^", converted, "$")
+}

tests/testthat/test-detect.r

@@ -49,3 +49,20 @@ test_that("str_ends works", {
   # Special typing of patterns.
   expect_true(str_ends("ab", fixed("B", ignore_case = TRUE)))
 })
+
+
+# str_like ----------------------------------------------------------------
+
+test_that("like_to_regex generates expected regexps",{
+  expect_equal(like_to_regex("ab%"), "^ab.*$")
+  expect_equal(like_to_regex("ab_"), "^ab.$")
+
+  # escaping
+  expect_equal(like_to_regex("ab\\%"), "^ab\\%$")
+  expect_equal(like_to_regex("ab[%]"), "^ab[%]$")
+})
+
+test_that("str_like works", {
+  expect_true(str_like("abc", "ab%"))
+  expect_error(str_like("abc", regex("x")), "character vector")
+})