Skip to main content
Extract with Regex Block Screenshot

​
Overview

The Extract With Regex Block allows you to extract one or more strings from an input string using a regular expression. It’s useful for parsing responses from Language Models (LLMs), extracting specific values from files, or any scenario where you need to match patterns in text.

​
Key Features

  • Extract multiple strings using capture groups
  • Configurable error handling for failed matches
  • Support for multiline mode
  • Flexible output options including all matches and success/failure flags

​
Inputs

​
string
required
The string to match the regex against. Non-string inputs will be coerced to strings.
​
string
The regular expression to use for matching. Only available when “Use Regex Input” is enabled in settings.

​
Outputs

​
string
One output port per capture group in the regex, containing the matched content for that capture group.
​
string[]
An array of all matched strings from the first capture group.
​
boolean
True if the regex matched the input string, false otherwise.
​
boolean
True if the regex did not match the input string, false otherwise.

​
Editor Settings

​
boolean
default:false
If enabled, the block will error if the regex doesn’t match the input string.
​
boolean
default:false
Enables multiline mode for the regex, affecting ^ and $ behavior.
​
string
default:"([a-zA-Z]+)"
The regular expression to use for matching.
​
boolean
default:false
When enabled, allows the regex to be provided via an input port instead of being set in the settings.

​
Example: Extract a command from an LLM response

  1. Create a Chat Block with a System Prompt:
  2. Add an Extract With Regex Block and connect the Chat Block’s output to its input.
  3. Set the Regex to !([a-zA-Z]+).
  4. Connect the Output 1 port to a Text Block to display the matched command.

​
Error Handling

  • The block will error if the input string is not provided or if the regex is invalid.
  • If “Error on failed” is enabled, the block will error when the regex doesn’t match the input.

​
FAQ

For special characters like newlines, you can include them literally in the regex editor. They will be preserved in the regular expression.

​
See Also