Refactor class Power to use pandas dataframes

monitoring/db/scripts/get_detailed_power.py

@@ -34,7 +34,7 @@ def parse_jobid(s):
 class Power:
 	def __init__(self, nodes):
 		self.nodes = nodes
-		self.epochs = OrderedDict()
+		self.epochs = None
 		self.first_ts = None
 		self.last_ts = None
 		self.warnings = ""
@@ -83,20 +83,25 @@ class Power:
 
 	@classmethod
 	def from_db(cls, db, jobid, interval, hawk_ai):
-		all_list = db.db_to_list(jobid, interval, hawk_ai)
-		if not all_list:
-			raise RuntimeError
+		df = db.db_to_pf(jobid, interval, hawk_ai)
+		power = cls.from_pandas(df, {'time': 'time', 'name': 'node', 'avg': 'power'})
 
-		power = cls.from_list(all_list)
+		return power
 
-		power.df = db.db_to_pf(jobid, interval, hawk_ai)
-		# power.df.to_pickle("pickle_df")
-		power.by_node = power.df.groupby('name')
-		power.by_epoch = power.df.groupby('time')
+	@classmethod
+	def from_pandas(cls, dataframe, columns):
+		dataframe.rename(columns=columns, inplace=True)
+		by_node = dataframe.groupby('node')
+		nodes = list(by_node.groups.keys())
+
+		power = cls(nodes)
+		power.epochs = dataframe.groupby('time')
+		power.dataframe = dataframe
+		power.by_node = power.dataframe.groupby('node')
 		power.energy = power.summarize_energy()
 
-		return power 
- 
+		return power
+
 	def to_file(self, jobid, header=""):
 		"""Dumps power data to file. Returns filename is succesfull and None if unsucessfull."""
 		fname = self.filename(jobid)
@@ -133,7 +138,7 @@ class Power:
 
 	def body(self):
 		_body = ""
-		for epoch in self.epochs.items():
+		for epoch in self.epochs:
 			_body += self.pretty_print(self.summarize_epoch(epoch))
 		return _body
 	
@@ -141,8 +146,8 @@ class Power:
 		return ts, ""
 
 	@staticmethod
-	def summarize_values(val):
-		values = np.asarray(val)
+	def summarize_values(df):
+		values = df['power'].values
 		head = values[0]
 		min, max = values.min(), values.max()
 		avg, stddev = values.mean(), values.std()
@@ -153,7 +158,10 @@ class Power:
 		ts, values = epoch
 		return self.summarize_time(ts) \
 				+ self.summarize_values(values) \
-				+ tuple(values)
+				 + self.all_values(values)
+
+	def all_values(self, values):
+		return tuple(values['power'].tolist())
 
 	def energy_total(self):
 		energy = None
@@ -165,7 +173,7 @@ class Power:
 	def energy_node(group):
 		"""Left-sided Riemann sum is enough, as time is lower bound of bucket"""
 		delta_t = group["time"].diff().shift(-1)/1000.  # in seconds
-		pow =  group["avg"]
+		pow =  group['power']
 		return (delta_t * pow).iloc[:-1].sum()	
 
 	def summarize_energy(self):