Extract vs. Split

Extract and split transformations do not do the same thing:

• A split transformation separates a single column into one or more separate columns based on one or more values in the source column that identify where the data should be split. These delimiters can be determined by the application or specified by the user when defining the transformation.

• An extract transformation matches literal or pattern values from a source column and stores it in a separate column. 

  NOTE: The source column is untouched by extract transformations.

Extract methods

In the Transformer page, you can use the following methods to extract values: 

Extract text or patterns

A primary use of extraction is to remove literal or patterned values of text from a column of values. Suppose your dataset included a column of LinkedIn updates. You can use one of the following methods to extract keywords from these values. 

Extract single values

The following example transformation extracts the word #bigdata from the column msg_LinkedIn:

Notes:

• The option parameter identifies that the pattern to match is a custom one specified by the user.
• The Number of matches to extract parameter defaults to 1, meaning that the transformation extracts a maximum of one value from each cell. This value can be set from 1-50. 

Extract values by example

You can generate a new column of values extracted from a source column by entering example values to match with source values. Values with similar patterns may also be matched based on your entered example value.

For more information on transformation by example, see Overview of TBE.

Constrain matching

Within the extract transformation, you can specify literals or patterns before or after which the match is found. This method can be used to remove parts of each cell value from erroneously matching on the literal or pattern that is desired.

The following example extracts the second three-digit element of a phone number, skipping the area code:

Extract single patterns

You can also do pattern-based extractions using  or regular expressions.

• Regular expressions are a standards-based method of describing patterns of characters for matching purposes. Regular expressions are very powerful but can be difficult to use.  
• A  is a proprietary method of describing patterns, which is much simpler to use than regular expressions.
• For more information on both types of patterns, see Text Matching.

The following example extracts all words that begin with # in the msg_LinkedIn column:

Notes:

• The Text to extract parameter has changed:

  Element	Description
  Two back-ticks (`)	Indicate that the expression between them represents a .
  \#	The slash indicates that the character right after it should be interpreted as a character only; it should not be interpreted as any special character in the pattern.
  {alphanum-underscore}	This element is used to indicate a single alphanumeric or underscore character.
  +	Adding the plus sign after the above character signifies that the pattern can match on a sequence of alphanumeric or underscore characters of one or more length.

• The Number of matches to extract parameter has been increased to grab up to 50 hashtags.

Advanced options

Extract multiple values

In your pattern expressions, you can use the vertical pipe character (|) to define multiple patterns to find. The following example extracts any value from the myDate column that ends in 7 pr in 8:

You can use the vertical pipe in both  and regular expressions.

Extract first or last characters

You can extract the first or last set of characters from a column into a new column. In the following example, the first five characters from the ProductName column are extracted into a new product identifier column:

You can change the Option value to Last characters to extract from the right side of the column value.

Extract and remove

If you need to remove the characters that you extracted, you can use the following transformation. In this case, the first five characters, which were extracted in the previous transformation, are removed:

Extract by positions

You can extract values between specified index positions within a set of column values. In the following example, the text between the fifth and tenth characters in a column are extracted to a new column.

Extract by Data Type

You can perform extractions that are specific to a data type or based on failures of the data to match a specified data type.

Extract date values

You can use functions to extract values from Datetime columns.  The example below extracts the year value from the myDate column:

The following functions can be used to extract values from a Datetime column, as long as the values are present in the formatted date:

You can also reformat the whole Datetime column using the DATEFORMAT function. The following reformats the column to show only the two-digit year:

Extract numeric values

You can extract numerical data from text values. In the following example, the first number is extracted from the address column, which would correspond to extracting the street number for the address:

Empty values in this new column might indicate a formatting problem with the address.

Extract components of a URL

URL components

Using functions, you can extract specific elements of a valid URL. The following transformation pulls the domain values from the myURL column:

In some cases, the function may not return values. For example, the SUBDOMAIN function returns empty values if there is no sub-domain part of the URL.

The following functions can be used to extract values from a set of URLs:

Query parameters

You can extract query parameter values from an URL. The following example extracts the store_id value from the storeURL field value:

Extract object values

If your data includes sets of arrays, you can extract array elements into columns for each key, with the values written to each key column.

Suppose your restaurant dataset includes a set of characteristics in the restFeatures column in the following JSON format:

{
  "Credit": "Y",
  "Accessible": "Y",
  "Restrooms": "Y",
  "EatIn": "Y",
  "ToGo": "N",
  "AlcoholBeer": "Y",
  "AlcoholHard": "N",
  "TotalTables": "10",
  "TotalTableSeats": "36",
  "Counter": "Y",
  "CounterSeats": "8"
}

You can use the following transformation to extract the values from TotalTableSeats and CounterSeats into separate columns:

After the above is executed, you can perform a simple sum of the TotalTableSeats and CounterSeats columns to determine the total number of seats in the restaurant.

Extract array values

In some cases, your data may contain arrays of repeated key-value pairs, where each pair would exist on a separate line. Suppose you have a column called, Events, which contains date and time information about the musician described in the same row of data. The Events column might look like the following:

[{"Date":"2018-06-15","Time":"19:00"},{"Date":"2018-06-17","Time":"19:00"},{"Date":"2018-06-19","Time":"20:00"},{"Date":"2018-06-20","Time":"20:00"}]

The following transformation creates a separate row for each entry in the Events column, populating the other fields in the new rows with the data from the original row:

Extract Values into a List

You can also extract sets of values into an array list of values. 

Extract matches into array

Using , you can extract the values of the column to form a new column of arrays. The following example shows the usage of {any} pattern to extract the cell values and form a new array column.

Transformation:

Results:

Extract hashtags