json-py-es

Alexander Liu

To import raw JSON data files to ElasticSearch in one line of commands

Very fast -- 4 to 10 times faster when processing big data.

Installation

pip install jsonpyes

Notice: Before using pip to install jsonpyes, firstly you need to install python-pip on your system. ( Supports Python2.6, Python2.7 )

jsonpyes

Instructions:

There are 3 proccesses of importing raw JSON data to ElasticSearch
1. Only validating raw JSON data
2. Without validating ,just import data to ElasticSearch
3. After validating successfully, then import data to ElasticSearch

A valid JSON file here refers to a JSON file stacked with many lines of data

file valid_data.json and its content

{"key1": "valueA", "key2": {"sub_key1": "value2A", "sub_key2": ["Good", "Morning"]}}
{"key1": "valueB", "key2": {"sub_key1": "value2B", "sub_key2": ["Good", "Afternoon"]}}
...
{"key1": "valueC", "key2": {"sub_key1": "value2C", "sub_key2": ["Good", "Evening"]}}

Functions included

1. Validating JSON format data

jsonpyes --data raw_data.json --check

If the json data file is valid:

If the json data file is invalid:

2. Only importing without validating

jsonpyes --data raw_data.json --bulk http://localhost:9200 --import --index myindex2 --type mytype2

Notice: If the raw JSON data file is invalid, jsonpyes will not import it.

Or enable multi-threads jsonpyes --data raw_data.json --bulk http://localhost:9200 --import --index myindex2 --type mytype2 --thread 8

jsonpyes supports multi-threads when importing data to elasticsearch

Multi-threads comparison

No multi-threads
With 8 threads and jsonpyes cuts files into pieces, then destributes to workers fairly

As you can see these two containers have same docs loaded, if we use --thread 8 it could be several times faster, usually 5 to 10 times faster. That really depends on your computer/server resources. This was tested on a 4GB RAM / 2.4Ghz intel i5 Linux x64 laptop system.

And it works.