Skip to content

CSV source : a step further

unsplash drawers

Ever wondered how Fast2 could help you dive into archives and come out with burried content and metadata? It actually is quite simple, as long as you have the right tools in your hands.

The CSVSource task has been designed to receive a CSV file as input.

Basic usage

With little to no configuration, each line represents one document with different values matching the column header. From a Fast2 standpoint, a CSV with following content :

header1,header2
value1_A,value2_A
value1_B,value2_B

will generate 2 documents :

  • The first document will have 2 data, header1: value1_A and header2: value2_A
  • The second document will have 2 data, header1: value1_B and header2: value2_B

Although the default before resolved the data names from the 1st row (column headers), these names can be overwritten by the user, or even enriched.

Change data names

As for the first option (overwriting data names), the configuration needs to focus on the "New column names to set".

Enter each new header on a new line, making sure your input covers all the columns found in the CSV file.

CSV source task configuration for new data names

With such a setting, Fast2 will map the data retrieved from the CSV directly under those new data names.

Example

Let's consider processing a CSV file with the following content:

header1,header2
value1 ,value2

With the default settings, the document in Fast2 would have such dataset:

{
    "header1": "value1",
    "header2": "value2"
}

If the CSVSource task is configured as shown below,

Parameterized CSV source task configuration for new data names

the created document will only have a dataset looking like:

{
    "new header A": "value1",
    "new header B": "value2"
}

Fast2 will keep no trace of the old header names, generating a document with a dataset populated from the CSV file alongside new data names.

Of course this data name mapping could have been handled by an additional task, such as Drools or JSTransform (just to name a few).

But this CSV task here combines these 2 steps (of parsing and mapping) into a single one, lowering room for error and freeing the document of unnecessary information you'd not even have used.

Create extra columns based on existing data

This feature requires the configuration of the extracolumns option.

Syntaxe Enter one line per new data you intend to create.

Syntax goes as follows :

[variable]=[function]:[param1]:[param2]:[param3]:[param4]...

Rules

  1. The separator is the character : (semi-colon).
  2. Parameters have to striclty match the format $<data_name>. A data with the name "key" will be accessed under $key.
  3. Parameters can use other params
param1=stringLength:$keyA
param2=substring:$param1:3:5

Supported functions

Function Description
stringLength length of param1
substring substring of param1, from param2, during param3 characters
concat concatenation of all params
ifeq if param1 equals param2, then takes value of param3, otherwise param4

Example

We consider the following CSV content as input :

header1,header2,header3,header4
value1 ,value2 ,value3 ,this-is-the-value4

In the following examples, a new data with the name 'var_name' will be created with the value depending on the chosen option.

How to write it Value of the 'var_name' data
var_name=stringLength:$header1 Length (as integer) of the data 'value1' under 'header1', so 6.
var_name=substring:$header1:0:4 First 4 characters of the value under key header1, so valu.
var_name=substring:$header4:4:7 The 7 consecutive characters starting from the 5th one (counting from 0). So is-the-.
var_name=concat:OK/:$header1:/:$header2:_:$header3:.pdf Output : ok/value1/value2_value3.pdf
var_name=concat:KO/:$header1:.pdf Output : KO/value1.pdf
var_name=ifeq:$header1:18:$header2:$header3 Value of 'header1' is not equal to '18', so the output is value3.