docteste

querykmers_tpmiso.py

@@ -6,18 +6,37 @@ class SimpleBloomFilter:
 		self.num_hashes = num_hashes
 		self.bit_array = [0] * size
 
-	def _hashes(self, item):
+	def _hashes(self, item): 
+		"""
+		Donnes des valeurs de hachage pour un élément donnée
+		
+		>>> bf = SimpleBloomFilter(10,2)
+		>>> len(bf._hasches("test"))==2
+		True
+		"""
 		hash_values = []
 		for i in range(self.num_hashes):
-			hash_func = hashlib.sha256((str(i) + item).encode()).hexdigest()
+			hash_func = hashlib.sha256((str(i) + item).encode()).hexdigest() 
 			hash_values.append(int(hash_func, 16) % self.size)
 		return hash_values
 
 	def add(self, item):
+		"""
+		Ajoute un élément dans le Filtre Bloom
+		"""
 		for pos in self._hashes(item):
 			self.bit_array[pos] = 1
 
 	def contains(self, item):
+		"""
+		Verifie si un élemnt est présent dans le filtre de Bloom
+		>>> bf = SimpleBloomFilter(10, 2)
+        >>> bf.add("test")
+        >>> bf.contains("test")
+        True
+        >>> bf.contains("not_in")
+        False
+		"""
 		return all(self.bit_array[pos] for pos in self._hashes(item))
 
 	def merge(self, other):
@@ -70,6 +89,17 @@ class Structure:
 		return nodes[0] if nodes else None  
 
 	def query(self, kmer):
+		"""
+		>>> datasets = ["D1", "D2"]
+        >>> kmers_dict = {"D1": ["ACGT"], "D2": ["TGCA"]}
+        >>> s = Structure(datasets, kmers_dict, 10, 1)
+        >>> s.query("ACGT")
+        ['D1']
+        >>> s.query("TGCA")
+        ['D2']
+        >>> s.query("GCTA")
+        []
+        """
 		results = []
 		self._query_recursive(self.root, kmer, results)
 		return results