How Self-Calibration Powers Rich Text Editing in iOS Apps

In rich‑text scenarios such as comments or article titles, a piece of text may contain mentions, topics, links, and other special elements. Updating downstream services about each segment after every edit is impractical, and regex‑based detection cannot reliably distinguish user‑typed text from inserted special content.

The solution is a self‑calibration system: each special element carries a marker indicating its type and associated data, allowing a single pass over the text to retrieve all ranges, types, and data.

On iOS, NSMutableAttributedString provides the API to bind a dictionary of attributes to a range via -addAttributes:range: and related methods. By defining custom NSString keys (e.g., YYTextBackedStringAttributeName) you can store extra objects such as YYTextBackedString subclasses that hold model data for mentions, topics, links, etc.

YYKit declares YYTextBackedStringAttributeName and a YYTextBackedString class. Subclassing it (e.g., SVAtExtraTextBackedString, SVCommentSeekTextBackedString, SVCommentLinkTextBackedString) lets you attach specific model objects to the attributed string.

Parsing is performed by enumerating the YYTextBackedStringAttributeName over the whole string, extracting the custom objects and their ranges, as shown in the sample -parseAtExtraInfoModels method.

When static regex detection and self‑calibration are mixed, conflicts (e.g., overlapping mentions and topics) can be resolved by checking whether a range already contains a backed‑string object before applying the regex result.