Turn Messy Web Data Into Clean, Structured Datasets

Clean Data Before It Leaves Your Machine

The Dataset Builder transforms raw scraped values into standardized fields before export. Instead of exporting messy, page-specific values and then cleaning them in a spreadsheet, you define the structure you want up front. RTILA X maps scraped elements into those fields, applies any transformations you configure, and produces a dataset that is ready to use.

This approach reduces manual work after the automation runs. Your Google Sheet, database, or CSV does not need another layer of processing before it becomes useful. The cleaning happens inside the automation where it can be repeated consistently on every run.

Property Types & Selectors

The dataset builder supports multiple ways to capture data from a page. You can use CSS selectors, XPath expressions, and direct text selectors. The right choice depends on the structure of the site and how stable the markup is.

You are not locked into one selector type for the entire dataset. Different columns can use different selector strategies, which lets you mix simple CSS for stable elements with XPath for complex or deeply nested content. The builder also handles text normalization, so whitespace and formatting differences across pages do not pollute the final export.

Data Transformations

Each dataset column can include transformation rules. You can strip unwanted characters, convert numbers from strings, reformat dates, split combined values into separate fields, and apply conditional replacements. These transformations execute automatically as rows are collected, so the exported dataset reflects your desired format immediately.

Common examples include extracting a numeric price from a string like $1,299.00, removing HTML tags from captured content, and converting a date string such as 12 Aug 2024 into a consistent ISO format. These small fixes save significant downstream work.

File Downloads

Some extraction tasks produce more than structured rows. RTILA X can automatically download files during scraping, including PDFs, images, spreadsheets, and archives. You can then associate those downloaded files with rows in your dataset, rename them based on extracted values, and store them in project folders.

For example, a product listing scraper could capture the product name, price, and description while also downloading the product image for each row. The resulting dataset includes both structured values and references to the downloaded files, creating a complete package for further analysis.

Nested Datasets

Many pages contain repeating blocks with multiple fields inside each block. A product card may include the product title, price, image URL, and rating. The dataset builder understands this repeating structure and lets you create nested fields within a single row rather than flattening everything into one long list.

Nested datasets are useful when you need to preserve relationships between values. They export cleanly to formats like JSON and can be expanded into separate rows when exporting to spreadsheet formats.

Preview and Validation

Before committing to a full run, the builder shows a live preview of the data being collected. You can see how each column will be populated, whether selectors are capturing the right values, and which rows may contain missing data.

Validation options allow you to flag missing required fields, skip rows that do not match a condition, or add fallback values. This makes the automation more robust when pages vary slightly or when optional elements are missing.

Less Manual Cleanup

Deduplication and column-level rules mean your spreadsheets arrive ready to use. The combination of well-defined columns, transformations, and validation removes most of the cleanup that normally follows raw scraping.