Parsing

Audience: Low-code Engineers

Skill Prerequisites: Actions, Tokens

The Parsing actions read values out of text and turn text from one format into another. They read JSON and XML into tokens, build JSON, find and replace text with regular expressions, convert Markdown to HTML and clean HTML. Every action reads its input from tokens and writes its result to tokens. None of them create files.

Choosing an action

Action	What it does	Use it to
Parse JSON Into Tokens	Turns the properties of a JSON object or array into tokens, such as [Response:customer.email].	Read values from an API response or from JSON stored in the database.
Parse XML Into JSON Tokens	Converts XML to JSON, then turns it into tokens the same way, for example [Order:order.@id].	Read values from a SOAP or other XML response.
Create JSON Object	Builds a JSON object from name, value and type rows, escaping string values for you.	Build the JSON body of an API request from form values, lists or objects.
Regex	Gets the first match of one or more patterns in a text, each in its own token.	Pull an order number or tracking code out of an email or document.
Regex - Multiple Results	Gets every match of one pattern, joined into CSV, JSON or XML text.	Collect all the email addresses, links or hashtags in a text.
Regex Replace	Replaces every match of a pattern, with one input and pattern per row.	Keep only the digits of a phone number, or reformat a date.
Markdown to Html	Converts Markdown, including tables and task lists, to HTML.	Show a reply from an AI model, or user notes written in Markdown, as formatted HTML.
Sanitize Html	Removes tags, attributes and scripts from HTML, and repairs broken HTML.	Clean HTML from users, an API or an AI model before you show or store it.

Licensing

Regex, Regex - Multiple Results, Regex Replace, Markdown to Html and Sanitize Html require the Scripting feature to be licensed. If it isn't, they fail with an error.

Text, tokens and lists

• Results are text. Parse JSON Into Tokens and Parse XML Into JSON Tokens expand nested objects with dots, but a JSON array stays as JSON text in one token. Regex - Multiple Results also returns text, not a list.
• Looping. To run actions for each item, load the JSON text with Create List from JSON and use Execute Actions for each List Entry.
• Building JSON safely. Tokens typed inside JSON or XML aren't escaped, so a quote or & in a value can break it. Build JSON with Create JSON Object, using a String row for each value. To turn a list into a top-level array, use List to JSON.
• Sending and receiving. These actions are often paired with Server Request: Create JSON Object builds the request body, and the parse actions read the response.

Regular expressions

The three regex actions share these rules:

• They use .NET regular expression syntax.
• Matching ignores case. Start the pattern with (?-i) to match case.
• There are no multiline or singleline settings. Use (?m) or (?s) at the start of the pattern.
• When a pattern has groups, Regex and Regex - Multiple Results return the first group instead of the whole match.
• Only Regex Replace can replace tokens in the pattern, with its Apply tokens on pattern option.
• There's no match timeout. Keep patterns simple, because nested repetition such as (a+)+$ can run for a very long time on long inputs.

HTML from untrusted sources

Markdown to Html copies raw HTML in the Markdown, including <script> tags, to the output. Display Message inserts tokens as HTML without encoding. So when the Markdown or HTML comes from users or an AI model, run Sanitize Html on it before you show or store it. Keep in mind that Sanitize Html's default Disable scripts option also removes every href and src, so links and images lose their targets. A regex can't reliably remove unsafe HTML, so don't use Regex Replace for this.

For other ways to work with lists and JSON, see Lists Objects. For text read from PDF and Word files, see the PDF Generator and Word add-ons.

Revised 10/02/2026