Text Extractor
Text Extractor reads a text and pulls out the data in it: names of people, companies, places, dates and amounts, plus emails, web addresses, phones, IBANs, VAT numbers and other fixed formats. You can also add your own patterns, such as the format of your order numbers. It does not use a language model or need a credential, and it works in English, Spanish, French, German and Italian.
When to use it
- Pulling the order number and email out of each customer message.
- Finding the dates and amounts in a contract.
- Processing hundreds of texts at once, such as reviews or tickets.
Configuration
| Field | What it does |
|---|---|
| Language (required) | The texts’ language: English (default), Spanish, French, German or Italian. |
| Text (required) | A single text to analyse, for example an email body. Fill in this field or Data. |
| Data | A list of records, one per text, to analyse many at once. |
| Text Column | With Data, the field of each record that holds the text. Default: text. |
Advanced options
| Field | What it does |
|---|---|
| Entity Types | Which data types are kept, such as PERSON, ORG (organisations), GPE (countries and cities), DATE or MONEY. Empty, all of them. |
| Custom Patterns | Your own patterns, as name and pattern pairs, for example {"ORDER": "ORD-\d{6}"}. The name becomes the type of the data found. |
| Include Built-in Patterns | Also looks for emails, web addresses, phones, IBANs, EU VAT numbers, cards and other fixed formats. On by default. |
Output
The Result output holds the data found in each text, with its type.
In tool mode, an agent can use the component by itself. See Tool Mode.
Example
An online shop wants to know which order each customer email mentions.
- Connect the email body to Text and choose the right language in Language.
- In Custom Patterns, add
{"ORDER": "ORD-\d{6}"}. - Use Result to look the order up in your database.
Good practices
- Choose the language carefully: recognising names and places depends on it. If your texts mix languages, use one component per language.
Related components
- Gemini OCR: reads the text of scanned documents.