Support Questions

Find answers, ask questions, and share your expertise
Announcements
Celebrating as our community reaches 100,000 members! Thank you!

How to parse w/ fixed width instead of char delimited contents?

avatar
Contributor

I am trying to parse data from file contents that are generated by fixed width instead of by a delimiter. As a simplified example, the value for data attribute 1 is in position 1-2, for attribute 2 is in position 3-6, and attribute 3 is in position 7-8 in each line. Then, the file contents should be transformed as below.

Before

AABBBBCC

DDEEEEFF

After

AA;BBBB;CC

DD;EEEE;FF

I assume there may be a way to capture substrings per line? Please assist.

1 ACCEPTED SOLUTION

avatar
Expert Contributor

Kausha,

You can use ReplaceText to do this. In your example above, you can use a Replacement Strategy of "Regex Replace".

Set Evaluation Mode to "Line-by-Line"

The Search Value would then be:

(.{2})(.{4})(.{2})

And the Replacement Value would be:

$1;$2;$3

Does that help?

View solution in original post

19 REPLIES 19

avatar

This is a bit tricky and will require a bit of regex magic. You will want to capture each length of data attribute (2, 4 and 2 respectively) into capture groups then use those capture groups to replace the content.

You'll use the ReplaceText processor with a search value of "(.{2})(.{4})(.{2})" and a replacement value of "$1;$2;$3" and configure it to evaluate line by line. This will go through the contents of the flowfile line by line and replace the contents like you want.

Comment below if you run into any problems!

avatar
Contributor

I have posted to @mpayne also: "Yes. This works well, but is there a way to store the values as attributes?" Ultimately, I want to use the AttributesToJSON processor.

avatar

Ah so you also want to extract the text to be attributes of the flowfile. Is the structure of the contents only ever two lines and do you want to create JSON using both of those lines or split them into separate flowfiles?

avatar
Contributor

The number of rows for the flow files will vary. Each line of data will represent a record/item. Also, I want the data output in its original file, not in separate flowfiles.

avatar

What's your end goal JSON schema look like? What identifiers are each of the values going to use (for every line)? As a note, AttributesToJSON only creates flat JSON objects (no nested fields).

avatar
Contributor

The output, based on this example would be JSON array

[{"field1":"AA","field2":"BBBB","field3":"CC"},{"field1":"DD","field2":"EEEE","field3":"FF"}]

avatar

Assuming you are okay with using Hive for this, you would just create a table with one column (column name something like row) and then load the whole file into that table. Run a query to then split the columns and insert in another table.

Here are more details and code snippet. https://martin.atlassian.net/wiki/pages/viewpage.action?pageId=21299205

avatar

didnt realize the question was about nifi.. my bad.

avatar
Guru

I agree with @jpercivall and @mpayne ReplaceText is the best way to go. I created a quick workflow that you can reference. This was assuming the input of AABBBBCC as you suggested. You can change the GetFile path and PutFile path, and the regex in ReplaceText to test with your real data.fixedwidthexample.xml

avatar
Contributor

I have attempted, unsuccessfully, using the ReplaceText processor. The method works if I have a small/set number of lines in my file.

Do you have any guidance on the ReplaceTextWithMapping processor and how the mapping file should be formatted?