Support columns containing backquotes

SemyonSinchenko · Apr 2, 2025 · Aug 17, 2020 · Feb 28, 2025 · Feb 28, 2025 · Feb 28, 2025
commit dca14279778503adb9270b24d29be6e54a7a294f
diff --git a/src/main/scala/org/graphframes/GraphFrame.scala b/src/main/scala/org/graphframes/GraphFrame.scala
@@ -882,7 +882,8 @@ object GraphFrame extends Serializable with Logging {
    * Note: This can be replaced with org.apache.spark.sql.catalyst.util.QuotingUtils.quoteIfNeeded
    * once support for Spark 3 has been dropped
    */
-  private[graphframes] def quote(column: String): String = s"`$column`"
+  private[graphframes] def quote(column: String): String =
+    s"`${column.replace("`", "``")}`"
 
   /**
    * Helper for column names containing a dot. Quotes the given column name with backticks to

diff --git a/src/test/scala/org/graphframes/GraphFrameSuite.scala b/src/test/scala/org/graphframes/GraphFrameSuite.scala
@@ -93,6 +93,19 @@ class GraphFrameSuite extends SparkFunSuite with GraphFrameTestSparkContext {
     g.pageRank.maxIter(10).run()
   }
 
+  test("construction from DataFrames with backquote in column names") {
+    val g = GraphFrame(
+      vertices.withColumnRenamed("name", "a `name`"),
+      edges.withColumnRenamed("action", "the `action`"))
+    g.vertices.collect().foreach { case Row(id: Long, name: String) =>
+      assert(localVertices(id) === name)
+    }
+    g.edges.collect().foreach { case Row(src: Long, dst: Long, action: String) =>
+      assert(localEdges((src, dst)) === action)
+    }
+    g.pageRank.maxIter(10).run()
+  }
+
   test("construction from edge DataFrame") {
     val g = GraphFrame.fromEdges(edges)
     assert(g.vertices.columns === Array("id"))
@@ -312,6 +325,20 @@ class GraphFrameSuite extends SparkFunSuite with GraphFrameTestSparkContext {
             nullable = false))))
   }
 
+  test("nestAsCol with backquote in column names") {
+    val df = vertices.withColumnRenamed("name", "a `name`")
+    val col = nestAsCol(df, "attr")
+    assert(
+      df.select(col).schema === StructType(
+        Seq(
+          StructField(
+            "attr",
+            StructType(Seq(
+              StructField("id", LongType, nullable = false),
+              StructField("a `name`", StringType, nullable = true))),
+            nullable = false))))
+  }
+
   test("skewed long ID assignments") {
     val spark = this.spark
     import spark.implicits._

diff --git a/src/test/scala/org/graphframes/lib/ShortestPathsSuite.scala b/src/test/scala/org/graphframes/lib/ShortestPathsSuite.scala
@@ -106,4 +106,11 @@ class ShortestPathsSuite extends SparkFunSuite with GraphFrameTestSparkContext {
     doTest(Some(vertices))
   }
 
+  test("Test vertices with backquote in column name") {
+    val verticeSeq =
+      Seq((1L, "one"), (2L, "two"), (3L, "three"), (4L, "four"), (5L, "five"), (6L, "six"))
+    val vertices = sqlContext.createDataFrame(verticeSeq).toDF("id", "a `name`")
+    doTest(Some(vertices))
+  }
+
 }
diff --git a/src/test/scala/org/graphframes/lib/TriangleCountSuite.scala b/src/test/scala/org/graphframes/lib/TriangleCountSuite.scala
@@ -99,6 +99,20 @@ class TriangleCountSuite extends SparkFunSuite with GraphFrameTestSparkContext {
       .foreach { case Row(vid: Long, count: Long, _) => assert(count === 1) }
   }
 
+  test("Count with backquote in column name") {
+    val edges = sqlContext.createDataFrame(Array(0L -> 1L, 1L -> 2L, 2L -> 0L)).toDF("src", "dst")
+    val vertices = sqlContext
+      .createDataFrame(Seq((0L, "a"), (1L, "b"), (2L, "c")))
+      .toDF("id", "a `column`")
+    val g = GraphFrame(vertices, edges)
+    val v2 = g.triangleCount.run()
+    TestUtils.testSchemaInvariants(g, v2)
+    TestUtils.checkColumnType(v2.schema, "count", DataTypes.LongType)
+    v2.select("id", "count", quote("a `column`"))
+      .collect()
+      .foreach { case Row(vid: Long, count: Long, _) => assert(count === 1) }
+  }
+
   test("no triangle") {
     val edges = spark.createDataFrame(Array(0L -> 1L, 1L -> 2L)).toDF("src", "dst")
     val g = GraphFrame.fromEdges(edges)