Archives of Support Questions (Read Only)

shailuk · ‎09-21-2017

39448-screen-shot-2017-09-20-at-113720-pm.png

As you can see the 3 mb got transferred in screenshot - 1. And in the screenshot -2 I am specifying my Path where I want it stored. I basically want the CSV file written in the untitled text file. Am I doing it right.? I very new to NIFI. Please Help..:

@Wynner

39449-screen-shot-2017-09-20-at-114113-pm.png

Shu_ashu · ‎09-21-2017

@Shailesh Nookala
Sure, if you are thinking to insert the data to Sql Server we can do that by using NiFi, You don’t have to download the file at all.

You can do that couple of ways by using ExtractText Processor and extract all the contents of csv file into attributes and prepare insert statement by using ReplaceText Processor and push the data to Sql server.

Another method is preparing json document for input CSV file and extracting all the attributes of flow file and prepare record to insert data to SQL server.
I’m using Json method which is easy to extract attributes of flowfiles.

Example:-

Let’s consider you are having the csv file having just 2 records in it.
1,test,1000

2,test1,2000

As I’m using NiFi 1.1 there is no direct processor which can convert the incoming csv file to json.

we need to follow

InferAvroSchema-->  ConvertCSVToAvro --> ConvertAvroToJSON

the output would be in Json format with the same contents of our input CSV file.
But in new versions of NiFi(I think from 1.3) there are processors to convert CSV to Json directly using ConvertRecord Processor.
InferAvroSchema:-

In this processor I’m giving the incoming csv file definition (or) you can read header definition from incoming flowfile by changing Get CSV Header Definition From Data property to true NiFi can read definition from 1 line of file.
as I know my incoming file having id,name,salary I have given this definition and keep Schema Output Destination to flowfile-attribute we can make use of this schema in ConvertCSVToAvro processor.

40470-infer-avro-schema.png

ConvertCSVToAvro:-

change the Record Schema property to

${inferred.avro.schema}

Which can infer the schema from flowfile attribute.

40471-csvtoavro.png

ConvertAvroToJSON:-

drag and drop ConvertAvroToJSON processor and leave the properties to default.
Output:-
[{"id": 1, "name": "test", "salary": 1000},{"id": 2, "name": "test1", "salary": 2000}]

SplitJson:-

If you are having more than 1 record in your csv file use splitjson processor because if ConvertAvroToJSON processors find more than one record in the flowfile it will make an array and keep all the records inside the array.
As we need only 1 record in the flowfile before inserting data to Sqlserver.

If your flowfile having array then json path expression property should be

$.*

as it will dynamically splits the array of records to one record for one flowfile.

Input:-

[{"id": 1, "name": "test","salary": 1000},{"id": 2, "name":"test1", "salary": 2000}]

Output:-

flowfile1:-
{"id": 1, "name": "test", "salary":
1000}flowfile2:-
{"id": 2, "name": "test1", "salary":
2000}

As Splijson splits the array into individual records.

EvaluateJsonPath configs:-

in this processor extracting all the contents of flowfile to attributes by changing the list of properties

Destination as flowfile-attribute

extracting contents of json file by adding new properties by clicking + symbol on right corner.

id as $.id

name as $.name

salary as $.salary

ReplaceText Configs:-

We are preparing insert statement in this processor and change

Replacement Strategy property as AlwaysReplace

Use insert statement and give destination table name and use the extracted attributes to replace the contents of values dynamically.

insert into sqlserver_tablename (id,name,salary) values (${id},${name},${salary})

output:-

flowfile1 content:-
insert into sqlserver_tablename (id,name,salary) values(1,test,1000)
flowfile2 content:-
insert into sqlserver_tablename (id,name,salary) values(2,test1,2000)

40472-replacetext-sqlstmt.png
PutSQL Configs:-
use putsql processor and make a connection pool to sql server enable and use the connection pool.
so Putsql processor will executes all the insert statements that we are having contents of the flowfile and all the data will get inserted to sql server.

FlowScreenshot:-

40475-insert-sqlserver-flow.png

View solution in original post

shailuk · ‎09-24-2017

@Yash

Hi..I changed the relationship in all my processors. I get this error in all of them. Please take a look.

I am getting the relationship error in all my processors.

39503-sc-1.png

This is what I did in the bridge that connects "InferAvroSchema" and "convert CSV to AVRO".(screenshot - 2)

39504-sc-2.png

This is my "inferavro" configuration

.

39505-sc-3.png

When I select success in "Automatically terminate relationships" it gets selected and after I connect it to the next processor it gets un-selected automatically.

Please take a look.

Shu_ashu · ‎09-24-2017

@Shailesh Nookala
as you have given Inferavroschema processor Success relation to convertAvrotoJson processor then you need to

Except success relation
Auto terminate the following relations failure,original,unsupported content i.e click on the check boxes for these 3 relations.
Make sure you haven't selected success relation.
Click on Apply button

40500-autoterminate.png

why we haven't Autoterminated Success Relation?

Reason is we have connected success relation to convertavrotojson processor.

Once you are done with this do the same steps for all the other processors auto terminate all the relations Except the relation that you have connected to the next processor.

For the final processor(PutSQL) Auto Terminate success and reconnect failure and retry to the same processor.

shailuk · ‎10-08-2017

@Shu

Can you please help me fix this error. It says Incorrect syntax near','. Not able to figure out where I am making this error.

40732-screen-shot-2017-10-yjmzo.jpg

This is my CSV file.

My insert statement is

insert into [test_shailesh].[dbo].[nifiTest](policyID,statecode,county) values (${policyID},${statecode},${county})

shailuk · ‎10-08-2017

@Shu

This is the output of my EVALUATEJSONPATH Processor -

40741-screen-shot-2017-10-08-at-105336-am.png

Guess this is fine.

This is the output of my REPLACETEXT processor

40742-screen-shot-2017-10-08-at-110240-am.png

Don't really much about this.

This is the output of my PUTSQL processor

40743-screen-shot-2017-10-08-at-105933-am.png

If you see the values section is blank with just two commas. Can you please tell me what is going wrong.

My insert statement in REPLACETEXT processor

insert into [test_shailesh].[dbo].[nifiTest](policyID,statecode,county) values (${policyID},${statecode},${county})

my overall flow -

40744-screen-shot-2017-10-08-at-110346-am.jpg

shailuk · ‎10-08-2017

@Shu

I am not able to attach my CSV file. My CSV file has these two records.

policyID,statecode,county

498960,498960,498960

498960,498960,792148

name of my CSV file - FL_insurance_sample.csv

I created a table in my sql server with these three columns names all being varchar data type.

Shu_ashu · ‎10-08-2017

@Shailesh Nookala, i tested out with your csv file its working fine, can you attach all of your processors configs screenshots please.

shailuk · ‎10-09-2017

@Shu

Sure man. Here are the screenshots of all my processors -

1) GetFTP processor -

39714-capture.png

(contd)

39715-ss-2.png

2) InferAROSchema -

39716-inferavroschema-3.png

3) ConvertCSVtoAVRO

39717-csvtoavro-4.png

4) AVROToJSON -

39718-avrotojson-5.png

(I'll post the rest it in the next answer since I allowed to post only 5 attachments)

shailuk · ‎10-09-2017

@Shu

It worked man! Thank you so very much:)

LoL. One last thing, What if the my table has more number of columns lets say 15-20 columns. Do we still have to hard code all the column names in the processors.?

eg -

In the insert statement (in replaceText processor) where we do

insert into table_name value (${column1},${column2}................)

Is there a way that nifi can do this dynamically? I mean if there are 20-25 columns in my table then it would be a pain to mention all the column names in the insert statement. Can I use regex or something.?

Cloudera Community

Archives of Support Questions (Read Only)

I used GetFTP processor to get a CSV file from filezilla and transferred it to a PUTFILE. it worked but when I open the directory where I wanted the file to save in laptop. It's not there.