The least complicated way to import a CSV into Elasticsearch

import.sh

#!/bin/bash

# The simplest possible way to import a CSV into Elasticsearch without having to use a UI or depending on an Elasticsearch client library.
#
# Set $INDEX (the index name to import to) and $URL (the _bulk endpoint URL) environment variables.
#
# Requires `bash`, `awk`, `curl`, and [`mlr`](https://miller.readthedocs.io/en/6.12.0/) (for converting CSV to JSON.)
#
# Usage: sh ./import.sh < filename.csv

INDEX=entities-github_project_item-default
URL="http://elastic:changeme@localhost:9200/$INDEX/_bulk"

# Convert to ndjson
mlr --icsv --ojsonl cat  | \
# Add the index operation above each line
awk -v i="$INDEX" \
    '{ printf "{ \"index\": \{ \"_index\": \"%s\" \} \}\n%s\n", i, $0 }' | \
# Post to _bulk API
curl -H  "Content-Type: application/json; charset=UTF-8" -XPOST "$URL" --data-binary @- 

One line version:

mlr --icsv --ojsonl cat | awk -v i="$INDEX" \ '{ printf "{ \"index\": \{ \"_index\": \"%s\" \} \}\n%s\n", i, $0 }' | curl -H  "Content-Type: application/json; charset=UTF-8" -XPOST "$URL" --data-binary @-