diff --git a/languedoc.py b/languedoc.py
--- a/languedoc.py
+++ b/languedoc.py
@@ -1,9 +1,8 @@
 import os
-import re
 import random
 import itertools
 
-from shared import identify, extract_ngram_freqs, TOP_NGRAM_COUNT
+from shared import preprocess, identify, extract_ngram_freqs, rank_ngram_freqs, Sample
 
 random.seed(19181028)
 
@@ -11,11 +10,6 @@ CROSSVALIDATION_SOURCE_COUNT = 5
 TEST_LENS = [8, 16, 32, 64]
 
 
-def preprocess(text):
-	text = re.sub(r"[\W\d_]+", " ", " "+text+" ")
-	return text.lower()
-
-
 def merge_ngram_freqs(freqs):
 	n = len(freqs)
 	res = dict()
@@ -28,78 +22,28 @@ def merge_ngram_freqs(freqs):
 	return res
 
 
-class Sample:
-	def __init__(self, language="??", text=""):
-		self.language = language
-		self.frequencies = dict()
-		self._ranked_ngrams = dict()
-
-		if text:
-			self._extract(text)
-
-	def _extract(self, text):
-		for k in range(1, 4):
-			self.frequencies.update(extract_ngram_freqs(text, k))
-
-	@staticmethod
-	def merge(samples):
-		assert len({x.language for x in samples}) == 1
-
-		res = Sample(samples[0].language)
-		res.frequencies = merge_ngram_freqs([x.frequencies for x in samples])
-
-		return res
-
-	@property
-	def ranked_ngrams(self):
-		if not self._ranked_ngrams:
-			ordered_ngrams = sorted(self.frequencies.items(), key=lambda kv: -kv[1])[:TOP_NGRAM_COUNT]
-			self._ranked_ngrams = dict(zip([key for (key, freq) in ordered_ngrams], itertools.count(0)))
-
-		return self._ranked_ngrams
-
-	def compare(self, other):
-		"""take k most common
-		use frequencies x order
-		use letter, digrams, trigrams
-		use absolute x square"""
-		res = sum(abs(v-other.ranked_ngrams.get(k, len(other.ranked_ngrams))) for (k, v) in self.ranked_ngrams.items()) + \
-					sum(abs(v-self.ranked_ngrams.get(k, len(self.ranked_ngrams))) for (k, v) in other.ranked_ngrams.items())
-
-		return res
-
-	def print_overview(self):
-		print(f"Sample({self.language}):")
-
-		for freqs in self.frequencies:
-			x = [
-				(k, round(v, 3))
-				for (k, v) in sorted(freqs.items(), key=lambda kv: -kv[1])
-			]
-			print("  ", x[:8], "...", x[-8:])
-
-		print()
-
-
 class SampleSet:
 	def __init__(self, language):
 		self.language = language
 		self.texts = []
-		self.samples = []
+		self.frequencies = []
 
 	def add(self, text):
 		self.texts.append(text)
-		self.samples.append(Sample(self.language, text))
+		self.frequencies.append(extract_ngram_freqs(text))
 
 	def create_model(self):
-		return Sample.merge(self.samples)
+		merged_frequencies = merge_ngram_freqs(self.frequencies)
+		res = Sample(self.language, rank_ngram_freqs(merged_frequencies))
+		return res
 
 	def generate_tests(self, n):
-		for (i, (text, sample)) in enumerate(itertools.cycle(zip(self.texts, self.samples))):
+		for (i, (text, freqs)) in enumerate(itertools.cycle(zip(self.texts, self.frequencies))):
 			if i >= n:
 				break
 
-			yield (text, Sample.merge([x for x in self.samples if x is not sample]))
+			ranked_ngrams = rank_ngram_freqs(merge_ngram_freqs([f for f in self.frequencies if f is not freqs]))
+			yield (text, Sample(self.language, ranked_ngrams))
 
 
 def cross_validate(sample_sets):