Introduction to Hadoop Mapper | Edureka Video Lecture | Cassandra: Learn and Master - Database Management

okay so let's look at the Hadoop mapper
what happens in the Hadoop memo
so for using the Cassandra base input
format the mapper is going to basically
get a byte buffer so by buffer is
basically a Java class it is basically
the buffer for collecting bytes and then
so the input value is basically a sorted
map of byte buffer and a column I column
is something as is an interface that
represents your column so icon is an
interface defined in the the Sandra API
that's basically an interface that
represents so the byte buffer is
basically the name of the column that
you want to vary an eye column is the
the interface that you want to track so
using the Cassandra byte buffer you -
this is a byte buffer util class that
Cassandra provides you can effectively
read and write my buffers in the Sangha
Cassandra basically uses the Java in i/o
library so a lot of use of my buffers is
done by Cassandra so if you are not that
familiar with the niño API this utility
that designer provides let's call is the
byte buffer util you can easily use this
to read and write information or convert
bite into industries and swing it back
in fights while using the sender so and
also importantly the map output should
be writable so if you look at the Hadoop
API your output he should be of the type
rightful so it should Troy it should
follow the writable interface so
Cassandra can also as you could use
those classes as well so that is
something that you need to take care of
while you're interacting with the Hadoop
that is bringing data from the sander
how that would look like in the next
life so this is typically how your map
would look like basically or you can
define your own class which will extend
the mapper class from the API and it
would take the three value as inputs and
the output key values so if you look at
this is basically the the byte array as
ASCII the value is basically the sort
map the byte array and the right column
information the output probably you
oversleep it as text and the output he
is going to be text output values will
be interactive so if this is your
typical class or the mapper
implementation if you want to bring here
from then for having called names column
name information can we also supply via
conflict so this can be done by setting
up the context so when you are sitting
the job information you can set this
information as part of the Constitution
and if you want to get it as part of the
map here's how here's how you do it so
basically and then in a set of metal
which gets invoked at the start of the
mapper you could just query in the
context get concentration get call name
so if Paulo name is the property this is
how I'm going to get the value of this
and if I get this value I mean basically
more saves the column name value then I
initiate the the incredible and the the
text classes for my Hadoop's
output and basically you break my map
method so I'm overriding the map method
for the mapper year so it's going to
take the byte array that the key this is
going to be the value and then context
information is basically passed to me
then you basically also have the IO
exception and
yeah cool beer so this is an important
thing that this is all written in Java
so people who are familiar with Java
will find easy people who are not
familiar with Java probably will find it
difficult to understand sport so I'm
going in pretty details so this should
be helpful so what you do is basically
all this column information is going to
be provided by your input format all
this data is pretty handy you what to do
is just no column start yet so if you
since it's a sorted map you just get it
and then the name or name the bytes
should give you all the I column
information so what it gives you is the
column class and then based the column
or the column base further gives you the
values out of whatever information is
being read from the Cassandra API so
what you can do is basically see I found
column and then columns but here would
give you all the I column information
and then the actual values can be
further derived from this pi column has
a value method which will give you the
value information for that particular
column so you already know the column
name here you have basically passed it
here so you already have the column name
the value can be dragged in this fashion
and then you say your own value is
eliminated inside the base maybe space
delimited or something you want to
basically tokenize what basically you
this example is for a work on exam so
supposing that this this value that
you're getting is stupid eyes in some
fashion
you can further use a string tokenizer
to tokenize it so in this case it's
default to space
elimination and you can also if it's
delimited
in some other way you can provide your
Eliminator
as a second argument to the string to
organize a class and that would tokenize
your
values effectively then what you do is
basically I trade to all your values and
then set up the the context you see what
the world wars is give the text that you
have defined here that the text class
and then the value for this now a
important thing that you do is sis all
your values are are going to be one this
is an effective optimization that you've
done you know that you just said the
private final static and this you just
say the value is 1 so you can just reuse
this object without the overhead of
creating new objects this is I would
read all the words and assigned of the
frequency to this and that's how your
memo would look like if you want to read
data from