Better support for stringr in Snowflake

* Added support for str_starts() and str_ends() by using Snowflake's REGEXP_INSTR() function * Refactored str_detect() to use Snowflake's CONTAINS() function instead of hacking with REGEXP() which anchors the start and end by default
tidyverse · Nov 10, 2023 · f67a525 · f67a525
1 parent 388a6ee
commit f67a525
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 18 deletions.
diff --git a/R/backend-snowflake.R b/R/backend-snowflake.R
@@ -38,23 +38,32 @@ sql_translation.Snowflake <- function(con) {
       str_locate = function(string, pattern) {
         sql_expr(POSITION(!!pattern, !!string))
       },
-      # REGEXP on Snowflaake "implicitly anchors a pattern at both ends", which
-      # str_detect does not.  Left- and right-pad `pattern` with .* to get
-      # str_detect-like behavior
       str_detect = function(string, pattern, negate = FALSE) {
-        sql_str_pattern_switch(
-          string = string,
-          pattern = {{ pattern }},
-          negate = negate,
-          f_fixed = sql_str_detect_fixed_instr("detect"),
-          f_regex = function(string, pattern, negate = FALSE) {
-            if (isTRUE(negate)) {
-              sql_expr(!(((!!string)) %REGEXP% (".*" || (!!pattern) || ".*")))
-            } else {
-              sql_expr(((!!string)) %REGEXP% (".*" || (!!pattern) || ".*"))
-            }
-          }
-        )
+        con <- sql_current_con()
+
+        if (negate) {
+          translate_sql(!CONTAINS(!!string, !!pattern), con = con)
+        } else {
+          translate_sql(CONTAINS(!!string, !!pattern), con = con)
+        }
+      },
+      str_starts = function(string, pattern, negate = FALSE) {
+        con <- sql_current_con()
+
+        if (negate) {
+          translate_sql(REGEXP_INSTR(!!string, !!pattern) != 1L, con = con)
+        } else {
+          translate_sql(REGEXP_INSTR(!!string, !!pattern) == 1L, con = con)
+        }
+      },
+      str_ends = function(string, pattern, negate = FALSE) {
+        con <- sql_current_con()
+
+        if (negate) {
+          translate_sql(REGEXP_INSTR(!!string, !!pattern, 1L, 1L, 1L) != LENGTH(!!string) + 1L, con = con)
+        } else {
+          translate_sql(REGEXP_INSTR(!!string, !!pattern, 1L, 1L, 1L) == LENGTH(!!string) + 1L, con = con)
+        }
       },
       # On Snowflake, REGEXP_REPLACE is used like this:
       # REGEXP_REPLACE( <subject> , <pattern> [ , <replacement> ,

diff --git a/tests/testthat/test-backend-snowflake.R b/tests/testthat/test-backend-snowflake.R
@@ -25,15 +25,19 @@ test_that("custom stringr functions translated correctly", {
   local_con(simulate_snowflake())
 
   expect_equal(test_translate_sql(str_locate(x, y)), sql("POSITION(`y`, `x`)"))
-  expect_equal(test_translate_sql(str_detect(x, y)), sql("(`x`) REGEXP ('.*' || `y` || '.*')"))
-  expect_equal(test_translate_sql(str_detect(x, y, negate = TRUE)), sql("!((`x`) REGEXP ('.*' || `y` || '.*'))"))
+  expect_equal(test_translate_sql(str_detect(x, y)), sql("CONTAINS(`x`, `y`)"))
+  expect_equal(test_translate_sql(str_detect(x, y, negate = TRUE)), sql("NOT(CONTAINS(`x`, `y`))"))
   expect_equal(test_translate_sql(str_replace(x, y, z)), sql("REGEXP_REPLACE(`x`, `y`, `z`, 1.0, 1.0)"))
   expect_equal(test_translate_sql(str_replace(x, "\\d", z)), sql("REGEXP_REPLACE(`x`, '\\\\d', `z`, 1.0, 1.0)"))
   expect_equal(test_translate_sql(str_replace_all(x, y, z)), sql("REGEXP_REPLACE(`x`, `y`, `z`)"))
   expect_equal(test_translate_sql(str_squish(x)), sql("REGEXP_REPLACE(TRIM(`x`), '\\\\s+', ' ')"))
   expect_equal(test_translate_sql(str_remove(x, y)), sql("REGEXP_REPLACE(`x`, `y`, '', 1.0, 1.0)"))
   expect_equal(test_translate_sql(str_remove_all(x, y)), sql("REGEXP_REPLACE(`x`, `y`)"))
   expect_equal(test_translate_sql(str_trim(x)), sql("TRIM(`x`)"))
+  expect_equal(test_translate_sql(str_starts(x, y)), sql("REGEXP_INSTR(`x`, `y`) = 1"))
+  expect_equal(test_translate_sql(str_starts(x, y, negate = TRUE)), sql("REGEXP_INSTR(`x`, `y`) != 1"))
+  expect_equal(test_translate_sql(str_ends(x, y)), sql("REGEXP_INSTR(`x`, `y`, 1, 1, 1) = (LENGTH(`x`) + 1)"))
+  expect_equal(test_translate_sql(str_ends(x, y, negate = TRUE)), sql("REGEXP_INSTR(`x`, `y`, 1, 1, 1) != (LENGTH(`x`) + 1)"))
 })
 
 test_that("aggregates are translated correctly", {