Xml parser plugin for Embulk

Embulk parser plugin for parsing xml data by XPath perfectly!

Features

• namespace awareness
• nullable columns
• complex json array columns (with restrictions)

Overview

• Plugin type: parser
• Guess supported: no

Configuration

• type: specify this plugin as "xpath2" (string, required)
• root: root element to start fetching each entries (string, required)
• schema: specify the attribute of table and data type (required)
• namespaces: specify namespaces (required)
• stop_on_invalid_record: stop bulk load transaction if a invalid record is found (boolean, default is false)

Example

parser:
  type: xpath2
  root: '/ns1:root/ns2:entry'
  schema:
    - { path: 'ns2:id', name: id, type: long }
    - { path: 'ns2:title', name: title, type: string }
    - { path: 'ns2:meta/ns2:author', name: author, type: string }
    - { path: 'ns2:date', name: date, type: timestamp, format: '%Y%m%d' }
    - { path: 'ns2:ratings/ns2:rating[@by="subscribers"]', name: ratings, type: json }
  namespaces: {ns1: 'http://example.com/ns1/', ns2: 'http://example.com/ns2/'}

Then you can fetch entries from the following xml:

<?xml version="1.0"?>
<ns1:root
  xmlns:ns1="http://example.com/ns1/"
  xmlns:ns2="http://example.com/ns2/">
  <ns2:entry>
    <ns2:id>1</ns2:id>
    <ns2:title>Hello!</ns2:title>
    <ns2:meta>
      <ns2:author>maji-KY</ns2:author>
    </ns2:meta>
    <ns2:date>20010101</ns2:date>
    <ns2:ratings>
      <ns2:rating by="subscribers">1</ns2:rating>
      <ns2:rating by="subscribers">2</ns2:rating>
      <ns2:rating>3</ns2:rating>
    </ns2:ratings>
  </ns2:entry>
</ns1:root>

complex json array column

Usage

parser:
  type: xpath2
  root: '/ns1:root/ns2:entry'
  schema:
    - { path: 'ns2:id', name: id, type: long }
    - path: 'ns2:list'
      name: list
      type: json
      structure: # adding structure key to enabling complex json array column
        - path: 'ns2:list'
          name: list
          type: array
        - path: 'ns2:list/ns2:elements'
          name: elements
          type: array
        - path: 'ns2:list/ns2:elements/ns2:name'
          name: elementName
          type: string
        - path: 'ns2:list/ns2:elements/ns2:value'
          name: elementValue
          type: long
        - path: 'ns2:list/ns2:elements/ns2:active'
          name: elementActive
          type: boolean
  namespaces: {ns1: 'http://example.com/ns1/', ns2: 'http://example.com/ns2/'}

Structure configuration

• path: specify path from the XPath of the column (string, required)
• name: json key name (string)
• type: json data type (One of array, string, long, boolean., required)

Then you can fetch entries from the following xml:

<?xml version="1.0"?>
<ns1:root
        xmlns:ns1="http://example.com/ns1/"
        xmlns:ns2="http://example.com/ns2/">
    <ns2:entry>
        <ns2:id>1</ns2:id>
        <ns2:list>
            <ns2:elements>
                <ns2:name>foo1</ns2:name>
                <ns2:value>1</ns2:value>
                <ns2:active>true</ns2:active>
            </ns2:elements>
            <ns2:elements>
                <ns2:name>foo2</ns2:name>
                <ns2:value>2</ns2:value>
                <ns2:active>false</ns2:active>
            </ns2:elements>
        </ns2:list>
        <ns2:list>
            <ns2:elements>
                <ns2:name>bar1</ns2:name>
                <ns2:value>3</ns2:value>
                <ns2:active>true</ns2:active>
            </ns2:elements>
        </ns2:list>
    </ns2:entry>
</ns1:root>

result of list column:

{
  "list": [
    {
      "elements": [
        {
          "elementActive": true,
          "elementName": "foo1",
          "elementValue": 1
        },
        {
          "elementActive": false,
          "elementName": "foo2",
          "elementValue": 2
        }
      ]
    },
    {
      "elements": [
        {
          "elementActive": true,
          "elementName": "bar1",
          "elementValue": 3
        }
      ]
    }
  ]
}

Build

$ ./gradlew gem

Benchmark

$ sbt benchmark/jmh:run