Add support for like() function (#130)

* add like support

* Update README.md

* Update SplToCatalystTest.scala

Author: chirag-s-db

README.md

@@ -19,7 +19,7 @@ There's basic support for the most used commands like `addtotals`, `bin`, `colle
 `streamstats`, `table`, `where`.
 
 There's also basic support for functions like `auto()`, `cidr_match()`, `coalesce()`, `count()`, 
-`ctime()`, `earliest()`, `if()`, `isnotnull()`, `latest()`, `len()`, `lower()`, `max()`, 
+`ctime()`, `earliest()`, `if()`, `isnotnull()`, `latest()`, `len()`, `like()`, `lower()`, `max()`, 
 `memk()`, `min()`, `mvappend()`, `mvcount()`, `mvfilter()`, `mvindex()`, `none()`, 
 `null()`, `num()`, `replace()`, `rmcomma()`, `rmunit()`, `round()`, `strftime()`, 
 `substr()`, `sum()`, `term()`, `values()`. 

src/main/scala/com/databricks/labs/transpiler/spl/catalyst/SplToCatalyst.scala

@@ -173,6 +173,30 @@ object SplToCatalyst extends Logging {
       determineMax(ctx, call)
     case "len" =>
       Length(attrOrExpr(ctx, call.args.head))
+    case "like" =>
+      val field = attrOrExpr(ctx, call.args.head)
+      val pattern = attrOrExpr(ctx, call.args(1))
+      pattern match {
+        case Literal(patternLiteral: UTF8String, StringType) =>
+          val patternString = patternLiteral.toString
+          // If the pattern is a simple LIKE (%foo%) pattern, we can convert it into a CONTAINS
+          // expression.
+          // For this to be safe, the pattern must start with %, end with % (unescaped), and contain
+          // exactly 2 instances of the wildcard character %. Note that this approach is
+          // conservative, as there may exist cases like %foo\%bar% that can be safely converted
+          // (as the wildcard in the middle of the string is escaped).
+          if (patternString.length > 2 &&
+              patternString.charAt(0) == '%' &&
+              patternString.charAt(patternString.length - 1) == '%' &&
+              patternString.charAt(patternString.length - 2) != '\\' &&
+              patternString.count(_ == '%') == 2) {
+            Contains(field,
+              Literal(patternString.substring(1, patternString.length - 1)))
+          } else {
+            Like(field, pattern, '\\')
+          }
+        case _ => Like(field, pattern, '\\')
+      }
     case "substr" =>
       val str = attrOrExpr(ctx, call.args.head)
       val pos = expression(ctx, call.args(1))

src/test/scala/com/databricks/labs/transpiler/spl/catalyst/SplToCatalystTest.scala

@@ -1096,6 +1096,39 @@ class SplToCatalystTest extends AnyFunSuite with PlanTestBase {
         )
     }
 
+    test("simple LIKE converted to CONTAINS") {
+        check(ast.SearchCommand(
+            ast.Call("like", Seq(ast.Field("a"), ast.StrValue("%foo%")))),
+            (_, tree) =>
+                Filter(
+                    Contains(
+                        UnresolvedAttribute("a"),
+                        Literal.create("foo")),
+                    tree)
+        )
+    }
+
+    test("complex LIKE not converted to CONTAINS") {
+        check(ast.SearchCommand(
+            ast.Call("like", Seq(ast.Field("a"), ast.StrValue("%foo%bar%")))),
+            (_, tree) =>
+                Filter(
+                    Like(
+                        UnresolvedAttribute("a"),
+                        Literal.create("%foo%bar%"), '\\'),
+                    tree)
+        )
+        check(ast.SearchCommand(
+            ast.Call("like", Seq(ast.Field("a"), ast.StrValue("%foo\\%")))),
+            (_, tree) =>
+                Filter(
+                    Like(
+                        UnresolvedAttribute("a"),
+                        Literal.create("%foo\\%"), '\\'),
+                    tree)
+        )
+    }
+
     test("eventstats max(colA) AS maxA by colC") {
         check(ast.EventStatsCommand(
             allNum = false,