Page MenuHomePhabricator

Remove buildParserPHP regexes when phpeggy is fixed
Closed, ResolvedPublic

Description

Currently, in TexVC(PHP) there is a bug that makes it necessary to reformat regex conditions in some cases within the buildParserPHP builder when creating a new grammar Parser.php
Ideally, this is solved within phpeggy.

The reported issue in phpeggy is here:
https://github.com/MarcelBolten/phpeggy/issues/4

Here is a minimal example of the erroneous case:
https://gist.github.com/Hyper-Node/e103c1f3cf407a10dec9b137331ade7f

Edit TBD:
Remove the fixes in builder and check if all tests are still running again.
Keep this in mind when fixing the final unicode testcases.

As soon as the case is fixed, remove the regex reformatting from the buildparser.

Event Timeline

After correspondence with Marcel Bolton and rechecking the grammar without the modifications, all testcases have been running.

The issue was probably caused by an erroneous test case during development, such as :

	"\\text {-0-9a-zA-Z+*,=():/;?.!'` \x80-\u{00FF}} " .  // this is erronous \x80 notation
	"\\mbox {-0-9a-zA-Z+*,=():/;?.!'` \u{0080}-\u{00FF}} " . // this is ok notation
Physikerwelt changed the task status from Open to Stalled.Jul 29 2024, 2:44 PM
Physikerwelt changed the task status from Stalled to Open.Jul 29 2024, 2:47 PM

After correspondence with Marcel Bolton and rechecking the grammar without the modifications, all testcases have been running.

The issue was probably caused by an erroneous test case during development, such as :

	"\\text {-0-9a-zA-Z+*,=():/;?.!'` \x80-\u{00FF}} " .  // this is erronous \x80 notation
	"\\mbox {-0-9a-zA-Z+*,=():/;?.!'` \u{0080}-\u{00FF}} " . // this is ok notation

It seems there is nothing left to do here.