From 3184d538d9edebc4f9fdbc5331847159bfbe2cdd Mon Sep 17 00:00:00 2001 From: dakkar Date: Fri, 2 Nov 2018 14:50:46 +0000 Subject: let's use Perl 6! minimal email parser, for now --- bayes | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+) create mode 100644 bayes diff --git a/bayes b/bayes new file mode 100644 index 0000000..37352ce --- /dev/null +++ b/bayes @@ -0,0 +1,55 @@ +#!/usr/bin/env perl6 +use v6.d.PREVIEW; + +grammar Message { + regex TOP { + + + + } + + token newline { [\x0d\x0a] | [\x0a\x0d] | \x0a | \x0d } + token separator { + [\x0a\x0d\x0a\x0d] | [\x0d\x0a\x0d\x0a] | \x0a ** 2 | \x0d ** 2 + } + token body { .* } + regex headers { +
+ % + } + regex header { + \: \h* + || + } + token name { + <-[:\s]>+ + } + regex value { + + % [ \h+] + } + token line { \N* } + token junk { \N+ } +} + +class Message-actions { + method TOP($/) { + make %( headers => $/.made, body => $/.Str ); + } + method headers($/) { + make %( |$/
ยป.made ); + } + method header($/) { + make $/ ?? () !! ( $/.Str => $/.made ); + } + method value($/) { + make $/.join(' ') + } +} + +sub MAIN(*@files) { + say "Starting";my $start = now; + my @messages = @files.race(:degree(10) :batch(100)).map({ + my $email = $_.IO.slurp(:enc); + my $match = Message.parse($email,:actions(Message-actions.new)); + }); + say "Took { now - $start } for { +@messages } messages"; +} -- cgit v1.2.3