[abp2blocklist] Make rules case-sensitive whereever possible

[sebastian]

Background

By default the regular expression given by trigger.url-filter matches case-insensitive. That's fine as it is the default behavior of Adblock Plus filters as well.

However, when the block list is compiled to a finite state machine by Safari, case-insensitive rules compile slower, result in larger code, and are slightly slower at runtime, as outlined in an ​official blog post.

While Adblock Plus filters are supposed to match case-insensitive (by default), it doesn't matter much for filters that can only possibly match the protocol + host part, as that part is always lowercase anyway.

Moreover, there is the $match_case filter option which is currently ignored when converting Adblock Plus filters to WebKit blocking rules.

What to change

Set trigger.url-filter-is-case-sensitive to true for rules generated from filters that either specify the $match_case option, or filters that can potentially only match the protocol + host part. These are:

• Element hiding filters, as those can only be limited to a domain in the first place
• Blocking filters and exception rules ...
  • with implied begin of host (starting with ||) or protocol delimiter (://)
  • that don't match any further alpha characters after the next occurrence of any of these:
    • / (begin of path)
    • ? (begin of query)
    • * (wildcard)
    • ^ (boundary)

[kzar]

​https://hg.adblockplus.org/abp2blocklist/rev/6d4363ef65e5