[
https://issues.apache.org/jira/browse/MIME4J-283?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16746393#comment-16746393
]
ASF GitHub Bot commented on MIME4J-283:
---------------------------------------
GitHub user hirthwork opened a pull request:
https://github.com/apache/james-mime4j/pull/26
MIME4J-283 DecoderUtil performance fix
DecoderUtil currently uses the following regex pattern for rfc2047-encoded
words:
`"(.*?)=\\?(.+?)\\?(\\w)\\?(.*?)\\?="`
First capturing group `(.*?)` is a very expensive regular expression
causing next pattern node evaluation on every input character. Because of this
decoding of 4 KB input (`To:` field with 40-80 recipients) takes up to 200ms on
modern CPUs.
At the same time, this capturing group used only to store separator text
between encoded words. Proposed patch reuses existing `tailIndex` for separator
text extraction and same input decoding now takes only 1-2ms.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/hirthwork/james-mime4j master
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/james-mime4j/pull/26.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #26
----
commit 028b10572f48f9e4818571c240e76db9ac6acadb
Author: Dmitry Potapov <dpotapov@...>
Date: 2019-01-18T15:17:43Z
MIME4J-283 DecoderUtil performance fix
----
> DecoderUtil performance fix
> ---------------------------
>
> Key: MIME4J-283
> URL: https://issues.apache.org/jira/browse/MIME4J-283
> Project: James Mime4j
> Issue Type: Improvement
> Components: parser (core)
> Affects Versions: master, 0.8.2
> Reporter: Dmitry Potapov
> Priority: Minor
> Attachments: patch
>
>
> DecoderUtil currently uses the following regex pattern for rfc2047-encoded
> words:
> {code:java}
> "(.*?)=\\?(.+?)\\?(\\w)\\?(.*?)\\?="
> {code}
> First capturing group {{(.*?)}} is a very expensive regular expression
> causing next pattern node evaluation on every input character. Because of
> this decoding of 4 KB input ({{To:}} field with 40-80 recipients) takes up to
> 200ms on modern CPUs.
> At the same time, this capturing group used only to store separator text
> between encoded words. Proposed patch reuses existing {{tailIndex}} for
> separator text extraction and same input decoding now takes only 1-2ms.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
