[SE-0351] Revise regex builder proposal #1634

rxwei · 2022-04-25T08:55:10Z

Revision:

Capture takes throwing closures. (Throwing customization hooks swift-experimental-string-processing#261)
Rename Output associated type to RegexOutput. (Rename RegexComponent.Output swift-experimental-string-processing#281)
Define primary associated type for RegexComponent.

Clarification:

Clarify that RegexComponent and Regex: RegexComponent will be in the stdlib, not in RegexBuilder.
Make detailed design driven by example and move API definition to a collapsible. This is so that the complex result builder machinery won't obscure the API design.
Add alternative considered section about unifying Capture and TryCapture.

milseman

Quick review of the first 2/3rds of this PR. Sorry it's not more in-depth, but I wanted to get the feedback out early

proposals/0351-regex-builder.md

milseman · 2022-04-25T13:13:01Z

proposals/0351-regex-builder.md

 ```

-By conforming standard library types to `RegexComponent`, we allow them to be used inside the regex builder DSL as a match target.
+Note:
+- `RegexComponent` and `Regex`'s conformance to `RegexComponent` are available without importing `RegexBuilder`. All other types and conformances introduced in this proposal are in the `RegexBuilder` module.


Is this where we landed, with the String conformance not being in the stdlib proper (i.e. Swift)? I thought we were leaning the other way but I'll have to crawl over conversation threads to see.

Either way, let's be very explicit what we mean by "other types". Perhaps pointing out String but also b, c, and d.

There's also a very real possibility of CharacterClass getting pulled into the Swift module at some point, perhaps even during the Unicode review.

Is this where we landed, with the String conformance not being in the stdlib proper (i.e. Swift)?

Yes.

Either way, let's be very explicit what we mean by "other types".

"All other types" seems pretty clear to me. The sentence below it already says explicitly:

By conforming standard library types to RegexComponent, we allow them to .... These conformances are available in the RegexBuilder module.

proposals/0351-regex-builder.md

Azoy

I think this reads really well now (including Michael's comments). I'm super excited for this!

proposals/0351-regex-builder.md

milseman · 2022-04-26T01:41:37Z

Argh, Github comment threads are so hard to see over time. The remaining point is:

A regex is considered invalid when it contains a use of reference without it ever being captured in the regex. When this occurs in the regex builder DSL, a runtime error will be reported. Similarly, the use of a reference in a Regex.Match.subscript(_:) must have been captured in the regex that produced the match.

Is the intent to trap if that branch of an alternation wasn't taken, or to trap if the reference wasn't used with the regex?

edit: to clarify, "ever being captured" could be confused as meaning it was set to a value instead of nil. I.e. captured at run-time instead of compile-time (noting that compilation time is not Swift compilation time but regex compilation time).

Maybe we can define "regex compilation time" as a term and use that to clarify?

milseman

Temporarily marking this as requesting changes concerning the trapping behavior of reference subscript and some other concerns. Don't want this to be merged accidentally.

Revision: - Capture takes throwing closures. (swiftlang/swift-experimental-string-processing#261) - Rename `Output` associated type to `RegexOutput`. (swiftlang/swift-experimental-string-processing#281) - Define primary associated type for `RegexComponent`. Clarification: - Clarify that `RegexComponent` and `Regex: RegexComponent` will be in the stdlib, not in `RegexBuilder`. - Make detailed design driven by example and move API definition to a collapsible. This is so that the complex result builder machinery won't obscure the API design. - Add alternative considered section about unifying `Capture` and `TryCapture`.

proposals/0351-regex-builder.md

milseman · 2022-05-03T13:20:25Z

proposals/0351-regex-builder.md

+  /// An anchor that matches at the first position of a match in the input
+  /// string.
+  ///
+  /// This anchor is equivalent to `\y` in regex syntax.


Suggested change

/// This anchor is equivalent to `\y` in regex syntax.

/// This anchor is equivalent to `\G` in regex syntax.

IIRC

\y is correct according to @natecook1000

Add a section for `mapOutput` method

proposals/0351-regex-builder.md

milseman · 2022-05-04T01:39:43Z

proposals/0351-regex-builder.md

+  }
+}.map { _, c1, c2, c3 in
+  SemanticVersion(major: c1, minor: c2, patch: c3 ?? 0)
+}


This is such a compelling use

proposals/0351-regex-builder.md

benrimmington · 2022-06-12T14:50:17Z

In the "Future directions" section, two instances of CustomMatchingRegexComponent could be updated to CustomConsumingRegexComponent.

At the end of the document, [Regex Syntax] is a broken link, and [String Processing Algorithms] could also be updated to:

[Regex Syntax]:                 https://github.com/apple/swift-evolution/blob/main/proposals/0355-regex-syntax-run-time-construction.md
[String Processing Algorithms]: https://github.com/apple/swift-evolution/blob/main/proposals/0357-regex-string-processing-algorithms.md

`buildEither` was removed from the regex builder DSL proposal. See swiftlang/swift-evolution#1634.

Also fix broken links.

rxwei requested review from milseman, natecook1000 and Azoy April 25, 2022 08:55

rxwei force-pushed the 0351-dsl-update branch 6 times, most recently from e02cfbf to f73b130 Compare April 25, 2022 09:05

milseman reviewed Apr 25, 2022

View reviewed changes

natecook1000 reviewed Apr 25, 2022

View reviewed changes

Azoy approved these changes Apr 25, 2022

View reviewed changes

proposals/0351-regex-builder.md Outdated Show resolved Hide resolved

rxwei force-pushed the 0351-dsl-update branch from f73b130 to f2dfbab Compare April 26, 2022 01:46

milseman suggested changes Apr 26, 2022

View reviewed changes

rxwei force-pushed the 0351-dsl-update branch from 23379d7 to 1198ce2 Compare April 27, 2022 04:59

rxwei force-pushed the 0351-dsl-update branch from 1198ce2 to 5478972 Compare April 27, 2022 05:38

natecook1000 reviewed Apr 30, 2022

View reviewed changes

proposals/0351-regex-builder.md Outdated Show resolved Hide resolved

Add section about anchors and lookaheads

77e6942

milseman reviewed May 3, 2022

View reviewed changes

natecook1000 and others added 3 commits May 3, 2022 17:54

Add section re mapOutput

160182b

Move mapOutput to Regex instead of RegexComponent

5ad7d0e

Merge pull request #2 from natecook1000/0351-dsl-update-mapoutput

a7db2ce

Add a section for `mapOutput` method

milseman reviewed May 4, 2022

View reviewed changes

milseman self-requested a review May 4, 2022 01:40

milseman approved these changes May 4, 2022

View reviewed changes

milseman reviewed May 4, 2022

View reviewed changes

proposals/0351-regex-builder.md Outdated Show resolved Hide resolved

Address comments.

d2b03bf

milseman reviewed May 5, 2022

View reviewed changes

proposals/0351-regex-builder.md Outdated Show resolved Hide resolved

natecook1000 and others added 3 commits May 28, 2022 18:55

Minor fix in RegexComponent declaration

711805b

Capture transform shouldn't be restricted to taking a Substring

0184c7a

Fix syntax highlighting

190418e

rxwei added a commit to rxwei/swift-experimental-string-processing that referenced this pull request Jun 14, 2022

Remove buildEither.

894cb39

`buildEither` was removed from the regex builder DSL proposal. See swiftlang/swift-evolution#1634.

rxwei mentioned this pull request Jun 14, 2022

Remove buildEither. swiftlang/swift-experimental-string-processing#482

Merged

stephentyrone and others added 3 commits June 28, 2022 16:02

Update 0351-regex-builder.md

8b17522

Revise RegexComponentBuilder.buildPartialBlock(first:).

a744060

Also fix broken links.

Update 0351-regex-builder.md

3161496

airspeedswift merged commit 164ca7f into swiftlang:main Jul 6, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[SE-0351] Revise regex builder proposal #1634

[SE-0351] Revise regex builder proposal #1634

rxwei commented Apr 25, 2022

milseman left a comment

milseman Apr 25, 2022

rxwei Apr 26, 2022 •

edited

Loading

Azoy left a comment

milseman commented Apr 26, 2022 •

edited

Loading

milseman left a comment

milseman May 3, 2022

rxwei May 4, 2022

milseman May 4, 2022

benrimmington commented Jun 12, 2022

	/// This anchor is equivalent to `\y` in regex syntax.
	/// This anchor is equivalent to `\G` in regex syntax.

[SE-0351] Revise regex builder proposal #1634

[SE-0351] Revise regex builder proposal #1634

Conversation

rxwei commented Apr 25, 2022

milseman left a comment

Choose a reason for hiding this comment

milseman Apr 25, 2022

Choose a reason for hiding this comment

rxwei Apr 26, 2022 • edited Loading

Choose a reason for hiding this comment

Azoy left a comment

Choose a reason for hiding this comment

milseman commented Apr 26, 2022 • edited Loading

milseman left a comment

Choose a reason for hiding this comment

milseman May 3, 2022

Choose a reason for hiding this comment

rxwei May 4, 2022

Choose a reason for hiding this comment

milseman May 4, 2022

Choose a reason for hiding this comment

benrimmington commented Jun 12, 2022

rxwei Apr 26, 2022 •

edited

Loading

milseman commented Apr 26, 2022 •

edited

Loading