Support searching for numeric literals with digit separators in GitHub Code Search

[tristaninf]

Select Topic Area

Question

Body

Hello GitHub Team,

I am writing to request an enhancement to the GitHub Code Search functionality. Specifically, I would like to see support for searching numeric literals with digit separators (underscores) in addition to the standard numeric format.

Current Limitation:

Currently, when searching for a numeric value such as 999999, it is not possible to find matches that include digit separators like 999_999. This feature would be extremely helpful for developers who utilize digit separators for better readability in their code, as it would allow us to locate all relevant instances of a numeric value regardless of formatting.

Feature Request:

I propose that GitHub Code Search be enhanced to:

1. Recognize numeric literals with digit separators (e.g., 999_999) as equivalent to their standard numeric counterparts (e.g., 999999).
2. Allow for searching both formats simultaneously using a consistent search pattern or an enhanced regular expression feature.

Benefits:

• Improved code search accuracy and efficiency for repositories that utilize digit separators for readability.
• Easier maintenance and code review processes by ensuring all instances of a numeric value can be identified, regardless of formatting.
• Enhanced developer experience by aligning search capabilities with modern coding standards and practices.

Example:

A search for 999999 should also return results for 999_999, 99_99_99, etc. Additionally, it would be beneficial if regular expressions could be used to specify patterns that match numeric values with optional digit separators.

Thank you for considering this enhancement. I believe it would significantly improve the usability of GitHub Code Search for many developers.

[look]

Thanks for the suggestion. I recommend using a regular expression for this use case.

For example, to search for variations of 1,000,000 you could use: /1[_,0]{6,}/. This isn't exactly what you want: it will match 100_000 since it has 6 characters after the 1 just like 1000000 does. However, it will narrow the search space considerably. If there's a format regularly used in your codebase, you could provide a stricter regular expression.