MAY"> MUST"> MUST NOT"> OPTIONAL"> RECOMMENDED"> REQUIRED"> SHALL"> SHALL NOT"> SHOULD"> SHOULD NOT"> ]> greenbytes GmbH

Hafenweg 16 MuensterNW48155 Germany julian.reschke@greenbytes.de http://greenbytes.de/tech/webdav/

The parsing of Uniform Resource Identifiers (URIs, RFC 3986) and Internationalized Resource Identifiers (IRIs, RFC 3987) is defined in terms of Augmented Backus-Naur Form (ABNF). The ABNF grammars are defined in terms of valid identifiers, and thus technically do not address how to handle invalid ones. The URI specification however includes a note how to use Regular Expressions for parsing, and this note applies to invalid identifiers as well. This document introduces terminology referring to potentially invalid identifiers, and demonstrates how the rules in the URI specification can be applied to them. Distribution of this document is unlimited. Although this is not a work item of the IRI Working Group, comments should be sent to the IRI mailing list at public-iri@w3.org, which may be joined by sending a message with subject "subscribe" to public-iri-request@w3.org. Discussions of the IRI Working Group are archived at . XML versions, latest edits, and the issues list for this document are available from . Umbrella issue for editorial fixes/enhancements. Expand for IRIs. Re-state the parsing algorithm as a procedural algorithm, maybe in JS? We can turn the regular expression into a concise ABNF grammar if that helps, we might also adapt it if that is found to be necessary, but I do not see a reason why hundreds of lines of prose code or JavaScript code would help (and anyone who would like to have that anyway can easily de- rive it from the expression or from a grammar). Define pre-processing steps for extraction of candidate references from content (WS stripping)? Define post-processing steps, such as query component rewriting based on document encoding.

The parsing of Uniform Resource Identifiers (URIs, ) and Internationalized Resource Identifiers (IRIs, ) is defined in terms of Augmented Backus-Naur Form (ABNF). The ABNF grammars are defined in terms of valid identifiers, and thus technically do not address how to handle invalid ones. The URI specification however includes a note how to use Regular Expressions for parsing, and this note applies to invalid identifiers as well. This document introduces terminology referring to potentially invalid identifiers, and demonstrates how the rules in the URI specification can be applied to them.

In addition to the terms defined in the URI specification, namely the Syntax Components (see ), this document defines: Candidate URI Reference A string that may or may not be a valid URI-reference according to . Candidate Scheme Component A string that may or may not be a valid URI scheme component according to . Candidate Authority Component A string that may or may not be a valid URI authority component according to . Candidate Path Component A string that may or may not be a valid URI path component according to . Candidate Query Component A string that may or may not be a valid URI query component according to . Candidate Fragment Component A string that may or may not be a valid URI fragment component according to .

The regular expression given in will parse any input string into a Candidate Scheme Component, a Candidate Authority Component, a Candidate Path Component, a Candidate Query Component, and a Candidate Fragment Component. Note that of these five components, all components except for the Path Component can be undefined. If each of the defined components is valid according to the related URI component definition, the input was a valid URI reference.

In section 3.2 you have "The result will be a valid URI Reference if and only if the components used by the algorithm were valid themselves." I have some doubts about "only if", consider for instance removing dot segments, which might remove a malformed part, if I recall correctly. defines Reference Resolution based on the five components. This algorithm works both for components obtained from valid and invalid references. The result will be a valid URI Reference if and only if the components used by the algorithm were valid themselves.

TBD

There are no IANA Considerations related to this specification.

World Wide Web Consortium

timbl@w3.org

Day Software

fielding@gbiv.com

Adobe Systems Incorporated

LMM@acm.org

shows results for the parsing/resolution processing described above, based on a test implementation written in XSLT 2.0.

Added issue combine-valid.