Github user thunterdb commented on a diff in the pull request:
https://github.com/apache/spark/pull/19439#discussion_r147661078
--- Diff: mllib/src/main/scala/org/apache/spark/ml/image/HadoopUtils.scala
---
@@ -0,0 +1,120 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.ml.image
+
+import scala.language.existentials
+import scala.util.Random
+
+import org.apache.commons.io.FilenameUtils
+import org.apache.hadoop.conf.{Configuration, Configured}
+import org.apache.hadoop.fs.{Path, PathFilter}
+import org.apache.hadoop.mapreduce.lib.input.FileInputFormat
+
+import org.apache.spark.sql.SparkSession
+
+private object RecursiveFlag {
+
+ /**
+ * Sets a value of spark recursive flag.
+ * If value is a None, it unsets the flag.
+ *
+ * @param value value to set
+ * @param spark existing spark session
+ * @return previous value of this flag
+ */
+ def setRecursiveFlag(value: Option[String], spark: SparkSession):
Option[String] = {
+ val flagName = FileInputFormat.INPUT_DIR_RECURSIVE
+ val hadoopConf = spark.sparkContext.hadoopConfiguration
+ val old = Option(hadoopConf.get(flagName))
+
+ value match {
+ case Some(v) => hadoopConf.set(flagName, v)
+ case None => hadoopConf.unset(flagName)
+ }
+
+ old
+ }
+}
+
+/**
+ * Filter that allows loading a fraction of HDFS files.
+ */
+private class SamplePathFilter extends Configured with PathFilter {
+ val random = new Random()
+
+ // Ratio of files to be read from disk
+ var sampleRatio: Double = 1
+
+ override def setConf(conf: Configuration): Unit = {
+ if (conf != null) {
+ sampleRatio = conf.getDouble(SamplePathFilter.ratioParam, 1)
+ }
+ }
+
+ override def accept(path: Path): Boolean = {
+ // Note: checking fileSystem.isDirectory is very slow here, so we use
basic rules instead
+ !SamplePathFilter.isFile(path) || random.nextDouble() < sampleRatio
--- End diff --
I would prefer that we do not use a seed and that the result is
deterministic, based for example on some hash of the file name, to make it more
robust to future code changes. That being said, there is no fundamental issues
with the current implementation and other developers may have differing
opinions, so the current implementation is fine as far as I am concerned.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
