Commit graph

5 commits

Author SHA1 Message Date
Ori Livneh 907de9a26c Increase AbuseFilterTokenizer cache TTL from 10m to 1d
AbuseFilterTokenizer is referentially transparent -- the mapping of input to
outputs does not vary on nonlocal state. So the cache TTL can be much longer.

Change-Id: I9e6ec4347dbb940c3d73538d550a0f045706264c
2015-10-24 10:30:24 -07:00
Ori Livneh ecbd159ce5 Track tokenizer cache hits / misses
Change-Id: I65d4c6064c37e9957b6f0aca4d3032f26bdf9bde
2015-10-22 13:30:16 -07:00
Ori Livneh f3472dfdd0 Use isset() to check array element exists rather than relying on @ operator
Change-Id: I0ecdcdd1426b2e76a326bc50b6ea0ca1cbad3d22
2015-10-21 23:58:00 +00:00
Aaron Schulz 7251c178fb Small doc fixes for AbuseFilterTokenizer
Change-Id: Ic85fa1c772ff7023883ff84449471a310109ba07
2015-08-25 15:51:30 -07:00
Ori Livneh bab9832415 Move rule tokenization to new AbuseFilterTokenizer class
* Move AbuseFilterParser::nextToken() and the various AbuseFilterParser
  properties that accompanied it to a new class, AbuseFilterTokenizer.
* Tokenize rules eagerly and cache the result in APC.

Change-Id: I15f5b5b65e8c4ec4fba3000d7c9fd78b98967d1d
2015-08-25 14:00:10 -07:00