step back on classifier for now

This commit is contained in:
Kyle Drake 2016-01-15 20:20:16 -08:00
parent 3ef075b24b
commit a37586bad4
5 changed files with 8 additions and 11 deletions

View file

@ -33,7 +33,6 @@ gem 'dnsruby'
gem 'base32'
gem 'coveralls', require: false
gem 'sanitize'
gem 'linnaeus', git: 'https://github.com/neocities/linnaeus.git', branch: 'soften_deps_and_perf'
gem 'will_paginate'
platform :mri, :rbx do

View file

@ -133,6 +133,3 @@ $country_codes = {}
CSV.foreach("./files/country_codes.csv") do |row|
$country_codes[row.last] = row.first
end
$classifier = Linnaeus::Classifier.new redis_db: 1
$trainer = Linnaeus::Trainer.new redis_db: 1

View file

@ -1247,17 +1247,17 @@ class Site < Sequel::Model
def classify(path)
return nil unless classification_allowed? path
$classifier.classify process_for_classification(path)
#$classifier.classify process_for_classification(path)
end
def classification_scores(path)
return nil unless classification_allowed? path
$classifier.classification_scores process_for_classification(path)
#$classifier.classification_scores process_for_classification(path)
end
def train(path, category='ham')
return nil unless classification_allowed? path
$trainer.train(category, process_for_classification(path))
# $trainer.train(category, process_for_classification(path))
site_file = site_files_dataset.where(path: path).first
site_file.classifier = category
site_file.save_changes validate: false
@ -1265,7 +1265,7 @@ class Site < Sequel::Model
def untrain(path, category='ham')
return nil unless classification_allowed? path
$trainer.untrain(category, process_for_classification(path))
# $trainer.untrain(category, process_for_classification(path))
site_file = site_files_dataset.where(path: path).first
site_file.classifier = category
site_file.save_changes validate: false

View file

@ -1,5 +1,4 @@
require 'sanitize'
require 'linnaeus'
class SiteFile < Sequel::Model
CLASSIFIER_LIMIT = 1_000_000.freeze

View file

@ -332,9 +332,10 @@ describe 'site_files' do
describe 'classification' do
before do
$trainer.instance_variable_get('@db').redis.flushall
puts "TODO FINISH CLASSIFIER"
#$trainer.instance_variable_get('@db').redis.flushall
end
=begin
it 'trains files' do
upload 'files[]' => Rack::Test::UploadedFile.new('./tests/files/classifier/ham.html', 'text/html')
upload 'files[]' => Rack::Test::UploadedFile.new('./tests/files/classifier/spam.html', 'text/html')
@ -348,6 +349,7 @@ describe 'site_files' do
@site.classify('spam.html').must_equal 'spam'
@site.classify('phishing.html').must_equal 'phishing'
end
=end
end
end
end