Skip to main content

Text Matching

This section enables you to work with text matching in Dataprep by Trifacta application.

Match Types

Dataprep by Trifacta supports the following types of text matching clauses:

  • String literals match specified strings exactly. Written using single quotes ('...') or double quotes ("...").

  • Regular expressions enable pattern-based matching. Regular expressions are written using forward slashes (/.../). The syntax is based on RE2 and PCRE regular expressions.

    Note

    Regular expressions are considered a developer-level capability and can have significant consequences if they are improperly specified. Unless you are comfortable with regular expressions, you should use Wrangle instead.

  • are custom selectors for patterns in your data and provide a simpler and more readable alternative to regular expressions. They are written using backticks (`...`).

  • Column names are simple text strings in Wrangle. If the column name contains a space, it must be bracketed in curly braces: {my Column Name}.

Dataprep by Trifacta Patterns Syntax

The following tables contain syntax information about Wrangle :

Note

You can use Wrangle as parameters in your flow. You assign the pattern to a variable, which can be used in your recipe steps. For more information, see Manage Parameters Dialog.

Tip

After using Wrangle , regular expressions, or string literals in a recipe step, you can reuse them in your transformations where applicable.

Character patterns

These patterns apply to single characters and strings of characters

Pattern

Description

%

Match any character, exactly once

%?

Match any character, zero or one times

%*

Match any character, zero or more times

%+

match any character, one or more times

%{3}

Match any character, exactly three times

%{3,5}

Match any character, 3, 4, or 5 times

#

Digit character [0-9]

{any}

Match any character, exactly once

{alpha}

Alpha character [A-Za-z_]

{upper}

Uppercase alpha character [A-Z_]

{lower}

Lowercase alpha character [a-z_]

{digit}

Digit character [0-9]

{delim}

Single delimiter character e.g :, ,, |, /, -, ., \s

{delim-ws}

Single delimiter and all the whitespace around it

{alpha-numeric}

Match a single alphanumeric character

{alphanum-underscore}

Match a single alphanumeric character or underscore character

{at-username}

Match @username values

{hashtag}

Match #hashtag values

{hex}

Match hexadecimal number (e.g. 2FA3)

Position patterns

These patterns describe positions relative to the entire string.

Pattern

Description

{start}

Match the start of the line

{end}

Match the end of the line

Type patterns

These patterns can be used to match strings that fit a particular data type, except for Datetime patterns.

Pattern

Description

{phone}

Match a valid U.S. phone number.

{email}

Match a valid email address.

{url}

Match a valid URL.

{ip-address}

Match a valid IP address.

{hex-ip-address}

Match a valid hexadecimal IP address (e.g. 0x0CA40012)

{bool}

Match a valid Boolean value.

{street}

Match a U.S.-formatted street address (e.g. 123 Main Street)

{occupancy}

Match a valid U.S.-formatted occupancy address value (e.g. Apt 2D)

{city}

Match a city name within U.S.-formatted address value

{state}

Match a valid U.S. state value (e.g. California).

{state-abbrev}

Match a valid two-letter U.S. state abbreviation value (e.g. CA)

{zip}

Match a valid five-digit zip code

Datetime patterns

Pattern

Description

{month}

Match full name of month (e.g. January)

{month-abbrev}

Match short name of month (e.g. Jan)

{time}

Match time value in HOUR:MINUTE:SECOND format (e.g. 11:59:23)

{period}

Match time period of the day: AM/PM

{dayofweek}

Match long name for day of the week (e.g. Sunday).

{dayofweek-abbrev}

Match short name for day of the week (e.g. Sun).

{utcoffset}

Match a valid UTC offset value (e.g. -0500, +0400, Z)

Note

You can use the Datetime data type formatting tokens as part of your Wrangle to build a variety of matching patterns for date and time values.

Grouping patterns

Pattern

Description

{[...]}

character class matches characters in brackets

{![...]}

negated class matches characters not in brackets

(...)

grouping, including captures

#, %, ?, *, +, {, }, (, ), \, ’, \n, \t

escaped characters or pattern modifiers Use a double backslash (\\) to denote an escaped string literal. For more information, see Escaping Strings in Transformations.

|

logical OR

  • Logical AND is the implied operator when you concatenate text matching patterns.

  • Logical NOT is managed using negated classes.

See also Capture Group References.

Patterns Examples

Basic

Match first three characters:

`{start}%{3}`

Match last four letters (numeric or other character types do not match):

`{alpha}{4}{end}`

Match first word:

`{start}{alpha}+`

Matches date values in general YYYY*MM*dd format:

`{yyyy}{delim}{MM}{delim}{dd}`

Matches time values in 12-hour format:

`{h}{delim}{mm}{delim}{s}`

In transformations

The following transformation masks credit card number patterns, except for the last four digits:

Transformation Name

Replace text or patterns

Parameter: Columns

myCreditCardNumbers

Parameter: Find

`{start}{digit}{4}{any}{digit}{4}{any}{digit}{4}{any}({digit}{4}){end}`

Parameter: Replace with

XXXX-XXXX-XXXX-$1

Notes:

  • The inclusion of the {start} and {end} tokens assures that the matches are made only when the pattern is found across the entire value in a cell.

  • The parenthesis in the Find value identify the capture group, which is referenced in the Replace With value as $1. See Capture Group References.

The above transformation matches values based on the structure of the data, instead of the data type.

  • Some values that follow this pattern are not valid credit card numbers, so it's meaningful to check against the data type.

  • If for some reason, you have values that are not credit card numbers yet follow the credit card pattern, those values will be masked as well by this transformation.

So to be safe, you might try the following set of transformations to ensure that you are matching on credit card values.

Step 1: If the number in your source column is valid, write it to a new column.

Transformation Name

New formula

Parameter: Formula type

Single row formula

Parameter: Formula

IFVALID(myCreditCardNumbers,'Creditcard'),$col,'')

Parameter: New column name

myCreditCardNumbersMasked

Notes:

  • The IFVALID function tests to see if a set of values is valid for a specified data type, 'Creditcard' in this case. For more information on the strings that you can use to test against data type, see Valid Data Type Strings.

  • The $col is a reference to the value in the column where the evaluation is being performed. For more information, see Source Metadata References.

Step 2: The myCreditCardNumbersMasked column now contains values that are valid credit card numbers from your source column. You can now apply the masking step.

Transformation Name

Replace text or patterns

Parameter: Columns

myCreditCardNumbersMasked

Parameter: Find

`{start}{digit}{4}{any}{digit}{4}{any}{digit}{4}{any}({digit}{4}){end}`

Parameter: Replace with

XXXX-XXXX-XXXX-$1

Step 3: If needed, you can move the masked values back to the source column.

Transformation Name

Edit column with formula

Parameter: Columns

myCreditCardNumbers

Parameter: Formula

IF(myCreditCardNumbersMasked<>'',myCreditCardNumbersMasked,'')

The myCreditCardNumbers column now contains only valid credit card numbers that have been asked. The application is likely to infer the data type of the column as String.

Delete the myCreditCardNumbersMasked column.

In this section: