Improve Recovery of Unterminated Regular Expressions #58289
Conversation
typescript-bot
added
the
For Uncommitted Bug
|
This PR doesn't have any linked issues. Please open an issue that references this PR. From there we can discuss and prioritise. |
| @@ -2389,7 +2390,8 @@ export function createScanner(languageVersion: ScriptTarget, skipTrivia: boolean | |||
| function reScanSlashToken(): SyntaxKind { | |||
| if (token === SyntaxKind.SlashToken || token === SyntaxKind.SlashEqualsToken) { | |||
| // Quickly get to the end of regex such that we know the flags | |||
| let p = tokenStart + 1; | |||
There was a problem hiding this comment.
This is eliminated for readability
There was a problem hiding this comment.
I know this is bad. Are there any solutions that does not affect pressing enter?
There was a problem hiding this comment.
Does this now become division?
| @@ -49,5 +49,5 @@ file.tsx(11,20): error TS1161: Unterminated regular expression literal. | |||
| !!! error TS2304: Cannot find name 'data'. | |||
| ~ | |||
| !!! error TS1005: ';' expected. | |||
| ~~ | |||
There was a problem hiding this comment.
Should we include angle brackets (in addition to parens and braces) to the logic?
There was a problem hiding this comment.
I should have named it regExpScanningRecovery.ts (which is analogous to jsonParserRecovery.ts, I forgot the fact that the recovery part will still remain in the scanner even after the regex worker is moved out)
| "\t\v\r\n", | ||
| "\u3000\u2028", | ||
| ]; | ||
| it("stops parsing unterminated regexes at correct position", () => { |
There was a problem hiding this comment.
I didn’t name each test case; it’s unrealistic to give each of the test cases a meaningful name. Perhaps they should be named by their indices?
This is now recovered.
The change is no longer necessary since it’s moved to `checker.ts` in microsoft#58295. The partially reverts microsoft@1a5228d.
|
Comments marked “Outdated” are not actually outdated, they’re just due to the merge |
|
Please inform me if you would like me to move the last commit to #58320 or a new PR. |
Per #55600 (comment), to not let
scanRegularExpressionWorkeraffect the end position of the regex, I separated the logic for unterminated regexes out from the well-terminated ones. This should make movingscanRegularExpressionWorkeraway in the future much easier. Note that unterminated regexes are no longer parsed.