Github user srowen commented on a diff in the pull request:

    https://github.com/apache/spark/pull/20611#discussion_r202363783
  
    --- Diff: 
sql/core/src/main/scala/org/apache/spark/sql/execution/command/tables.scala ---
    @@ -303,94 +303,44 @@ case class LoadDataCommand(
               s"partitioned, but a partition spec was provided.")
           }
         }
    -
    -    val loadPath =
    +    val loadPath = {
           if (isLocal) {
    -        val uri = Utils.resolveURI(path)
    -        val file = new File(uri.getPath)
    -        val exists = if (file.getAbsolutePath.contains("*")) {
    -          val fileSystem = FileSystems.getDefault
    -          val dir = file.getParentFile.getAbsolutePath
    -          if (dir.contains("*")) {
    -            throw new AnalysisException(
    -              s"LOAD DATA input path allows only filename wildcard: $path")
    -          }
    -
    -          // Note that special characters such as "*" on Windows are not 
allowed as a path.
    -          // Calling `WindowsFileSystem.getPath` throws an exception if 
there are in the path.
    -          val dirPath = fileSystem.getPath(dir)
    -          val pathPattern = new File(dirPath.toAbsolutePath.toString, 
file.getName).toURI.getPath
    -          val safePathPattern = if (Utils.isWindows) {
    -            // On Windows, the pattern should not start with slashes for 
absolute file paths.
    -            pathPattern.stripPrefix("/")
    -          } else {
    -            pathPattern
    -          }
    -          val files = new File(dir).listFiles()
    -          if (files == null) {
    -            false
    -          } else {
    -            val matcher = fileSystem.getPathMatcher("glob:" + 
safePathPattern)
    -            files.exists(f => 
matcher.matches(fileSystem.getPath(f.getAbsolutePath)))
    -          }
    -        } else {
    -          new File(file.getAbsolutePath).exists()
    -        }
    -        if (!exists) {
    -          throw new AnalysisException(s"LOAD DATA input path does not 
exist: $path")
    -        }
    -        uri
    +        val localFS = FileContext.getLocalFSFileContext()
    +        localFS.makeQualified(new Path(path))
           } else {
    -        val uri = new URI(path)
    -        val hdfsUri = if (uri.getScheme() != null && uri.getAuthority() != 
null) {
    -          uri
    -        } else {
    -          // Follow Hive's behavior:
    -          // If no schema or authority is provided with non-local inpath,
    -          // we will use hadoop configuration "fs.defaultFS".
    -          val defaultFSConf = 
sparkSession.sessionState.newHadoopConf().get("fs.defaultFS")
    -          val defaultFS = if (defaultFSConf == null) {
    -            new URI("")
    -          } else {
    -            new URI(defaultFSConf)
    -          }
    -
    -          val scheme = if (uri.getScheme() != null) {
    --- End diff --
    
    Hm. I was trying to understand where this logic went. I see that's sort of 
in the call to `makeQualified`. I couldn't find the docs for that method 
overload though because it's actually "LimitedPrivate" in Hadoop. I think we 
shouldn't call this method? can we instead just restore this logic?


---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org

Reply via email to