Elasticsearch attachment parsing usecase

Elastic Stack Elasticsearch
#1

Hi,

I am exploring ElasticSearch as a solution for one of our usecase. I wanted to know whether the current ElasticSearch capabilities could meet our requirements.
For our use case, we want to index excel files in our repository.
The excel files have a standard template. The data is present in 2 columns. Column A contains certain "questions" and Column B has the answers to those questions
For our use case, if the user searches for a question (say on Row 19 A, we want to be able to show him the answers in column B ( i.e Row 19 B).

From what I have read, I would have to use the ingest-attachment plugin to index the files. However once I have the files indexed, would I be able to map the contents of the files as "Key", "Value" pairs in the index to search on?

What would be the best way to go about this?

Thanks.

#2

No you can't. You'd better convert your document into CSV then import it with Logstash for example.

Here is an example: http://david.pilato.fr/blog/2015/04/28/exploring-capitaine-train-dataset/

#3

Thanks for the quick reply.
A follow-up question - The initial excel files are converted into html files using a tool we have. Would it be easier to index the html file directly in ElasticSearch via ingest-attachment?
Or is log-stash be the preferred approach to transform the data ( excel or html) before pushing to ElasticSearch?

#4

I believe you need structured data not unstructured.

So HTML won't help in my opinion.

#5

I may not have explained properly.
We currently already have a process where initial excel files are created from an excel template and converted to a html document.

I have an opportunity to index the file either from the excel doc or from the html doc that is generated; to get the desired output that I mentioned in the initial question.

Would it be easier to use logstash on the excel doc or the converted html document. Or is it the same effort/process using logstash irrespective of the file format.

#6

CSV is easier to parse IMO.

#7

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.