Python/Elasticsearch Bulk problem

[Ivan_Devic (Ivan Dević)]

Hi all,

I would like to perform bulk request of next action. I want to create document if it does not exist, but I would like to update it, if it does exist. Within the update, I append new values (new_a and new_b) to arrays which are inside a single doc (a and b). For a single document, next part of code in Python seems to work flawlessly, when I invoke update function with following body:

{"script":{"inline": "ctx._source.a.addAll(params.new_a); ctx._source.b.addAll(params.new_b)",
           "params": {"a": new_a, "b": new_b}},
 "upsert": {"a": new_a, "b": new_b}}

However, when I try to take the present logic and expand it into Bulk request, I kind of hit the wall. Thank you in advance for help

EDIT: Preview purposes only

[spinscale (Alexander Reelsen)]

Hey,

can you maybe explain in more detail, what exactly fails and how your code looks like? are you calling the bulk endpoint yourself are you using the bulk helper?

Thanks!

--Alex

[Ivan_Devic (Ivan Dević)]

Hi Alex,

so what I do is I loop over my data in a python loop and I yield every map (dictionary) that I need to update. What I send to helpers.bulk() function from a Python Elasticsearch library is next scheme

{"_index":  index_name,
 "op_type": "update",
 "_type": doc_type
 "_id": doc_id
 "_source": {"source": updateSchema}}

in which updateSchema is a map from my original post. The scripts executes, but rather then storing results from scripts, it stores them as literal strings. Additionally, I tried the variant with body explained on next link, but again, scripts are not being executed but stored as strings.

EDIT:

Figured it out, op_type has to be sent with a starting "_" to be interpreted properly