Load classifer db into sample data hash

This commit is contained in:
Joshua Peek
2012-07-23 13:13:52 -05:00
parent 97ae7c1a11
commit 0c9a947f39
5 changed files with 52 additions and 63 deletions

View File

@@ -18,7 +18,7 @@ class TestClassifier < Test::Unit::TestCase
end
def test_instance_freshness
serialized = Linguist::MD5.hexdigest(Classifier.instance.to_hash)
serialized = Linguist::MD5.hexdigest(Sample::DATA)
latest = Linguist::MD5.hexdigest(Linguist::Sample.classifier.to_hash)
# Just warn, it shouldn't scare people off by breaking the build.
@@ -55,16 +55,16 @@ class TestClassifier < Test::Unit::TestCase
end
def test_instance_classify_empty
results = Classifier.instance.classify("")
results = Classifier.new(Sample::DATA).classify("")
assert results.first[1] < 0.5, results.first.inspect
end
def test_instance_classify_nil
assert_equal [], Classifier.instance.classify(nil)
assert_equal [], Classifier.new(Sample::DATA).classify(nil)
end
def test_verify
data = Classifier.instance.to_hash
data = Sample::DATA
assert_equal data['languages_total'], data['languages'].inject(0) { |n, (_, c)| n += c }
assert_equal data['tokens_total'], data['language_tokens'].inject(0) { |n, (_, c)| n += c }
@@ -80,7 +80,7 @@ class TestClassifier < Test::Unit::TestCase
languages = Language.all.select { |l| l.extensions.include?(extname) }.map(&:name)
next unless languages.length > 1
results = Classifier.instance.classify(File.read(sample[:path]), languages)
results = Classifier.new(Sample::DATA).classify(File.read(sample[:path]), languages)
assert_equal language.name, results.first[0], "#{sample[:path]}\n#{results.inspect}"
end
end