[SPARK-43935][SQL][PYTHON][CONNECT] Add xpath_* functions to Scala and Python

connector/connect/client/jvm/src/main/scala/org/apache/spark/sql/functions.scala

@@ -4282,6 +4282,93 @@ object functions {
   def array_except(col1: Column, col2: Column): Column =
     Column.fn("array_except", col1, col2)
 
+  /**
+   * Returns a string array of values within the nodes of xml that match the XPath expression.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath(xml: Column, path: Column): Column =
+    Column.fn("xpath", xml, path)
+
+  /**
+   * Returns true if the XPath expression evaluates to true, or if a matching node is found.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_boolean(xml: Column, path: Column): Column =
+    Column.fn("xpath_boolean", xml, path)
+
+  /**
+   * Returns a double value, the value zero if no match is found, or NaN if a match is found but
+   * the value is non-numeric.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_double(xml: Column, path: Column): Column =
+    Column.fn("xpath_double", xml, path)
+
+  /**
+   * Returns a double value, the value zero if no match is found, or NaN if a match is found but
+   * the value is non-numeric.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_number(xml: Column, path: Column): Column =
+    Column.fn("xpath_number", xml, path)
+
+  /**
+   * Returns a float value, the value zero if no match is found, or NaN if a match is found but
+   * the value is non-numeric.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_float(xml: Column, path: Column): Column =
+    Column.fn("xpath_float", xml, path)
+
+  /**
+   * Returns an integer value, or the value zero if no match is found, or a match is found but the
+   * value is non-numeric.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_int(xml: Column, path: Column): Column =
+    Column.fn("xpath_int", xml, path)
+
+  /**
+   * Returns a long integer value, or the value zero if no match is found, or a match is found but
+   * the value is non-numeric.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_long(xml: Column, path: Column): Column =
+    Column.fn("xpath_long", xml, path)
+
+  /**
+   * Returns a short integer value, or the value zero if no match is found, or a match is found
+   * but the value is non-numeric.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_short(xml: Column, path: Column): Column =
+    Column.fn("xpath_short", xml, path)
+
+  /**
+   * Returns the text contents of the first xml node that matches the XPath expression.
+   *
+   * @group "xml_funcs"
+   * @since 3.5.0
+   */
+  def xpath_string(xml: Column, path: Column): Column =
+    Column.fn("xpath_string", xml, path)
+
   private def newLambdaVariable(name: String): proto.Expression.UnresolvedNamedLambdaVariable = {
     proto.Expression.UnresolvedNamedLambdaVariable
       .newBuilder()

connector/connect/client/jvm/src/test/scala/org/apache/spark/sql/PlanGenerationTestSuite.scala

@@ -1655,6 +1655,42 @@ class PlanGenerationTestSuite
     fn.to_date(fn.col("s"), "yyyy-MM-dd")
   }
 
+  temporalFunctionTest("xpath") {
+    fn.xpath(fn.col("s"), lit("a/b/text()"))
+  }
+
+  temporalFunctionTest("xpath_boolean") {
+    fn.xpath_boolean(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_double") {
+    fn.xpath_double(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_number") {
+    fn.xpath_number(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_float") {
+    fn.xpath_float(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_int") {
+    fn.xpath_int(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_long") {
+    fn.xpath_long(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_short") {
+    fn.xpath_short(fn.col("s"), lit("a/b"))
+  }
+
+  temporalFunctionTest("xpath_string") {
+    fn.xpath_string(fn.col("s"), lit("a/b"))
+  }
+
   temporalFunctionTest("unix_date") {
     fn.unix_date(fn.to_date(fn.col("s"), "yyyy-MM-dd"))
   }

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath.explain

@@ -0,0 +1,2 @@
+Project [xpath(s#0, a/b/text()) AS xpath(s, a/b/text())#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_boolean.explain

@@ -0,0 +1,2 @@
+Project [xpath_boolean(s#0, a/b) AS xpath_boolean(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_double.explain

@@ -0,0 +1,2 @@
+Project [xpath_double(s#0, a/b) AS xpath_double(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_float.explain

@@ -0,0 +1,2 @@
+Project [xpath_float(s#0, a/b) AS xpath_float(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_int.explain

@@ -0,0 +1,2 @@
+Project [xpath_int(s#0, a/b) AS xpath_int(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_long.explain

@@ -0,0 +1,2 @@
+Project [xpath_long(s#0, a/b) AS xpath_long(s, a/b)#0L]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_number.explain

@@ -0,0 +1,2 @@
+Project [xpath_number(s#0, a/b) AS xpath_number(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_short.explain

@@ -0,0 +1,2 @@
+Project [xpath_short(s#0, a/b) AS xpath_short(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/explain-results/function_xpath_string.explain

@@ -0,0 +1,2 @@
+Project [xpath_string(s#0, a/b) AS xpath_string(s, a/b)#0]
++- LocalRelation <empty>, [d#0, t#0, s#0, x#0L, wt#0]

connector/connect/common/src/test/resources/query-tests/queries/function_xpath.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b/text()"
+          }
+        }]
+      }
+    }]
+  }
+}

connector/connect/common/src/test/resources/query-tests/queries/function_xpath_boolean.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath_boolean",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b"
+          }
+        }]
+      }
+    }]
+  }
+}

connector/connect/common/src/test/resources/query-tests/queries/function_xpath_double.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath_double",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b"
+          }
+        }]
+      }
+    }]
+  }
+}

connector/connect/common/src/test/resources/query-tests/queries/function_xpath_float.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath_float",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b"
+          }
+        }]
+      }
+    }]
+  }
+}

connector/connect/common/src/test/resources/query-tests/queries/function_xpath_int.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath_int",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b"
+          }
+        }]
+      }
+    }]
+  }
+}

connector/connect/common/src/test/resources/query-tests/queries/function_xpath_long.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath_long",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b"
+          }
+        }]
+      }
+    }]
+  }
+}

connector/connect/common/src/test/resources/query-tests/queries/function_xpath_number.json

@@ -0,0 +1,29 @@
+{
+  "common": {
+    "planId": "1"
+  },
+  "project": {
+    "input": {
+      "common": {
+        "planId": "0"
+      },
+      "localRelation": {
+        "schema": "struct\u003cd:date,t:timestamp,s:string,x:bigint,wt:struct\u003cstart:timestamp,end:timestamp\u003e\u003e"
+      }
+    },
+    "expressions": [{
+      "unresolvedFunction": {
+        "functionName": "xpath_number",
+        "arguments": [{
+          "unresolvedAttribute": {
+            "unparsedIdentifier": "s"
+          }
+        }, {
+          "literal": {
+            "string": "a/b"
+          }
+        }]
+      }
+    }]
+  }
+}