Meziantou.Framework

Limitations

edit

URL patterns

  • Regular expressions are .NET regular expressions. The URL Pattern standard defines the regular expression groups with the JavaScript syntax. This library runs them with the .NET engine, so the syntax that differs between the two does not behave the way it does in a browser; for instance, the set operations of the JavaScript v flag ([\d&&[0-1]]) are not supported.
  • Regular expressions have no timeout. A pattern with a regular expression that backtracks a lot can take a long time to match some URLs. Do not create patterns from text that users supply, or check that it has no regular expression group (HasRegExpGroups).
  • No URL type. The library parses URLs to match them, but returns no parsed URL: use System.Uri for that. The values in a match result are in the normalized form of the URL standard, which can differ from what System.Uri returns for the same URL, for instance on percent-encoding.
  • No pattern generation. A pattern cannot build a URL from the values of its groups.

Query strings

  • Strings in, strings out. The helpers work on URL strings and do not validate them: they split the string at the first ? and the first #.
  • Names are case-sensitive, and parameters are written back in the name=value form, so a parameter written without = (?debug) gains one (?debug=) when the query string is written again.

Public Suffix List

  • A snapshot. The list is the one of the package version you use: a suffix added to the list after that version is unknown until you update the package. See Version of the list.
  • Domain names only. IP addresses are not domain names: TryGetDomainInfo returns false for a Uri whose host is one.

Internationalized host names and invariant globalization

The host names that contain non-ASCII characters are converted to their ASCII form with IdnMapping, in URL patterns and in the Public Suffix List. IdnMapping relies on ICU: in an application that runs with InvariantGlobalization enabled, the conversion does not apply the whole mapping of the standard, so some non-ASCII host names are converted differently from a browser. ASCII host names are not affected.