Develop Reference

How to access SparkContext on executors to save DataFrame to Cassandra?

How can I use SparkContext (to create SparkSession or Cassandra Sessions) on executors?
If I pass it as a parameter to the foreach or foreachPartition, then it will have a null value. Shall I create a new SparkContext in each executor?
What I'm trying to do is as follows:
Read a dump directory with millions of XML files:
dumpFiles = Directory.listFiles(dumpDirectory)
dumpFilesRDD = sparkContext.parallize(dumpFiles, numOfSlices)
dumpFilesRDD.foreachPartition(dumpFilePath->parse(dumpFilePath))
In parse(), every XML file is validated, parsed and inserted into several tables using Spark SQL. Only valid XML files will present objects of same type that can be saved. Portion of the data needs to be replaced by other keys before being inserted into one of the tables.
In order to do that, SparkContext is needed in the function parse to use sparkContext.sql().

If I'd rephrase your question, what you want is to:
Read a directory with millions of XML files
Parse them
Insert them into a database
That's a typical Extract, Transform and Load (ETL) process that terribly easy in Spark SQL.
Loading XML files can be done using a separate package spark-xml:
spark-xml A library for parsing and querying XML data with Apache Spark, for Spark SQL and DataFrames. The structure and test tools are mostly copied from CSV Data Source for Spark.
You can "install" the package using --packages command-line option:
$SPARK_HOME/bin/spark-shell --packages com.databricks:spark-xml_2.11:0.4.1
Quoting spark-xml's Scala API (with some changes to use SparkSession instead):
// Step 1. Loading XML files
val path = "the/path/to/millions/files/*.xml"
val spark: SparkSession = ???
val files = spark.read
.format("com.databricks.spark.xml")
.option("rowTag", "book")
.load(path)
That makes the first requirement almost no-brainer. You've got your million XML files taken care by Spark SQL.
Step 2 is about parsing the lines (from the XML files) and marking rows to be saved to appropriate tables.
// Step 2. Transform them (using parse)
def parse(line: String) = ???
val parseDF = files.map { line => parse(line) }
Your parse function could return something (as the main result) and the table that something should be saved to.
With the table markers, you split the parseDF into DataFrames per table.
val table1DF = parseDF.filter($"table" === "table1")
And so on (per table).
// Step 3. Insert into DB
table1DF.write.option(...).jdbc(...)
That's just a sketch of what you may really be after, but that's the general pattern to follow. Decompose your pipeline into digestable chunks and tackle one chunk at a time.

It is important to keep in mind that in Spark we are not supposed to program in terms of executors.
In Spark programming model, your driver program is mostly a self-contained program where certain sections will be automatically converted to a physical execution plan. Ultimately a bunch of tasks distributed across worker/executors.
When you need to execute something for each partition, you can use something like mapPartitions(). Refer Spark : DB connection per Spark RDD partition and do mapPartition for further details. Pay attention to how the dbConnection object is enclosed in the function body.
It is not clear what you mean by a parameter. If it is just data (not a DB connection or similar), I think you need to use a boradcast variable.

Read newly appended lines to csv file using python or java

Jmeter creates csv file at starting of the test, after that Jmeter appends incremental results (new lines) to csv file till test is done. Below is the format
1459239209060,152,Client token ,200,OK,data
1459239209074,136,Client token ,200,OK,data
1459239209217,17,/mydata,200,OK,data 1
1459239209219,70,/mydata,200,OK,data 1
1459239209235,14,/mydata,200,OK,data 1
So I want to read only newly appended values for every time (gap of 1 sec/2 sec /3sec). So is there any way to do this.

Run your JMeter script via Taurus framework, it fully supports JMeter .jmx files and reports interim statistics on average response times in console:
Taurus is an open-source tool so you can check i.e. https://github.com/Blazemeter/taurus/blob/master/bzt/modules/console.py for implementation details if you need just that bit.

Merging PDFs with Sejda fails with stream output

Using Sejda 1.0.0.RELEASE, I basically followed the tutorial for splitting a PDF but tried merging instead (org.sejda.impl.itext5.MergeTask, MergeParameters, ...). All works great with the FileTaskOutput:
parameters.setOutput(new FileTaskOutput(new File("/some/path/merged.pdf")));
However I am unable to change this to StreamTaskOutput correctly:
OutputStream os = new FileOutputStream("/some/path/merged.pdf");
parameters.setOutput(new StreamTaskOutput(os));
parameters.setOutputName("merged.pdf");
No error is reported, but the resulting file cannot be read by Preview.app and is approximately 31 kB smaller (out of the ~1.2 MB total result) than the file saved above.
My first idea was: stream is not being closed properly! So I added os.close(); to the end of CompletionListener, still the same problem.
Remarks:
The reason I need to use StreamTaskOutput is that this merge logic will live in a web app, and the merged PDF will be sent directly over HTTP. I could store the temporary file and serve that one, but that is a hack.
Due to licencing issues, I cannot use the iText 5 version of the task.
Edit
Turns out, the reason is that StreamTaskOutput zips the result into a ZIP file! OutputWriterHelper.copyToStream() is the culprit. If I rename merged.pdf to merged.zip, it's a valid ZIP file containing a perfectly valid merged.pdf file!
Could anyone (dear authors of the library) comment on why this is happening?

The idea is that when a task consumes a MultipleOutputTaskParameters producing multiple output documents, the StreamTaskOutput has to group them to be able to write all of them to a stream output. Unfortunately Sejda currently applies the same logic to SingleOutputTaskParameters, hence your issue. We can fix this in Sejda 2.0 because it makes more sense to directly stream the out document in case of SingleOutputTaskParameters. For Sejda 1.x I'm not sure how to address this remaining compatible with the existing behaviour.

Reading N-Quads in Jena

I'm trying to read an N-Quads file with Jena, but all I get is an empty model. The file I'm trying to read is taken from the example in N-Quads documentation:
<http://example.org/#spiderman> <http://www.perceive.net/schemas/relationship/enemyOf> <http://example.org/#green-goblin> <http://example.org/graphs/spiderman> .
(I saved it as a file named file.nq).
The way I'm loading the model is using the RDFDataMgr. But it didn't work with Model.read either.
RDFDataMgr.loadModel("file.nq", Lang.NQUADS)
yields an empty model.
What am I missing? Doesn't Jena support N-Quads out-of-the-box?

Yes, Jena supports N-Quads. Try loadDataset.
N-Quads is for multiple graphs and you have read it into one graph. What you get is just the default graph triples, in this case, none.
There is a warning emitted:
WARN riot :: Only triples or default graph data expected : named graph data ignored
If you didn't get that then (1) you are running an old copy (2) you have turned logging off (3) the file is empty.

Trailing null (\x00) characters when writing text to Accumulo

I am trying to write the name of a file into Accumulo. I am using accumulo-core-1.43.
For some reason, certain files seem to be written into Accumulo with trailing \x00 characters at the end of the name. The upload is coming through a Java servlet (using the jquery file upload plugin). In the servlet, I check the name of the file with a System.out.println and it looks normal, and I even tried unescaping the string with
org.apache.commons.lang.StringEscapeUtils.unescapeJava(...);
The actual writing to accumulo looks like this:
Mutation mut = new Mutation(new Text(checkSum));
Value val = new Value(new Text(filename).getBytes());
long timestamp = System.currentTimeMillis();
mut.put(new Text(colFam), new Text(EMPTY_BYTES), timestamp, val);
but nothing unusual showed up there (perhaps \x00 isn't escaped)? But then if I do a scan on my table in accumulo, there will be one or more \x00 in the file name.
The problem this seems to cause is that I return that string within XML when I retrieve a list of files (where it shows up) and pass that back to the browser, the the XSL that is supposed to render the information in the XML no longer works when there's these extra characters (not sure why that is the case either).
In chrome, for the response on these calls, I see that there's three red dots after the file name, and when I hover over it, \u0 pops up (which I think is a different representation of 0/null?).
Anyway, I'm just trying to figure out why this happens, or at the very least, how I can filter out \x00 characters before returning the file in Java. any ideas?

You are likely incorrectly using the Hadoop Text class -- this is not an error with Accumulo. Specifically, you make the mistake in your above example:
Value val = new Value(new Text(filename).getBytes());
You must adhere to the length of provided by the Text class. See the Text javadoc for more information. If you're using Hadoop-2.2.0, you can use the provided copyBytes method on Text. If you're on older version of Hadoop where this method doesn't yet exist, you can use something like the ByteBuffer class or the System.arraycopy method to get a copy of the byte[] with the proper limits enforced.