Bug fix and enhancement with folder monitoring

src/data-collector.py

@@ -21,7 +21,7 @@ from threading import Thread, RLock
 import monitor
 import utils.agents.actor as actor
 from utils.agents.manager import Manager
-SYS_ARGS['host']='localhost'
+SYS_ARGS['host']='localhost'/
 ENDPOINT="http://:host/monitor".replace(":host",SYS_ARGS['host'])
 class Collector(Thread) :
 	def __init__(self):

src/monitor.py

@@ -29,7 +29,7 @@ class Analysis:
 		self.now = {"month":d.month,"year":d.year, "day":d.day,"hour":d.hour,"minute":d.minute}
 	def getNow(self):
 		d = datetime.datetime.now()
-		return {"month":d.month,"year":d.year, "day":d.day,"hour":d.hour,"minute":d.minute}
+		return {"month":d.month,"year":d.year, "day":d.day,"hour":d.hour,"minute":d.minute,"second":d.second}
 	def getName(self):
 		return self.__class__.__name__
 	def reboot(self,row,conf) :
@@ -167,7 +167,6 @@ class DetailProcess(Analysis):
 	
 	def init (self,names):
 		#Analysis.init(self)
-		
 		self.names = names;
 	def getName(self):
 		return "apps"
@@ -218,7 +217,7 @@ class DetailProcess(Analysis):
 		cmd = "ps -eo user,pid,pmem,pcpu,vsize,stat,command|grep -Ei \":app\"".replace(":app",name)
 		handler = subprocess.Popen(cmd,shell=True,stdout=subprocess.PIPE)
 		logs	= handler.communicate()[0].split('\n')
-		logs	= [row for row in logs if (row.strip() != '') and ('grep -Ei' in row )== False and (__file__ not in row)]
+		logs	= [row for row in logs if (row.strip() != '') and ('grep -Ei' in row )== False ]
 		
 		if len(logs) == 0:
 			return [dict(self.parse('',fields),**{'label':name}) ]
@@ -255,113 +254,43 @@ class DetailProcess(Analysis):
 class FileWatch(Analysis):
 	def __init__(self):
 		pass
-	def init(self,folders):
+	def init(self,folders):		
 		self.folders = folders;
+		self.cache = []
 	def getName(self):
 		return "folders"
-	def split(self,row):
-		
-		x = row.split(' ')
-		r = {}
-		months = ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec']
-		if x:
-			BYTES_TO_MB = 1000000
-			size = int(x[0])/BYTES_TO_MB
-			month	= months.index(x[1]) + 1
-			day	= int(x[2])
-			age = -1
-			hour=minute = 0
-			if ':' in x[3] :
-				hour,minute	= x[3].split(':')
-				now = datetime.datetime.today()
-				if month == now.month :
-					year	= now.year
-				else:
-					year = now.year - 1
-			else:
-				year = int(x[3])
-				hour = 0
-				minute = 0
-				
-			
-			
-			file_date = datetime.datetime(year,month,day,int(hour),int(minute))
-			# size = round(size,2)
-			#file_date = datetime.datetime(year,month,day,hour,minute)
-			now = datetime.datetime.now()
-			age = (now - file_date ).days
-			
-			return {"size":size,"age":age}	
-		return None
 
-	def evaluate(self,path):
-		cmd = "find  :path -print0|xargs -0 ls -ls |awk '{print $6,$7,$8,$9,$10}'".replace(":path",path)
-		handler = subprocess.Popen(cmd,shell=True,stdout=subprocess.PIPE)
-		ostream = handler.communicate()[0].split('\n')
-		ostream = [row for row in ostream if row.strip() != '']		
-		#return [self.split(stream) for stream in ostream if stream.strip() != '' and '.DS_Store' not in stream and 'total' not in stream]
-		#return [self.split(stream) for stream in ostream if path not in stream and not set(['','total','.DS_Store']) & set(stream.split(' '))]
-		return []
-	def toMB(self,size):
-		m = {'GB':1000,'TB':1000000}
-		v,u = size.split(' ')
-		return round(float(v)* m[u.upper()],2)
+	def evaluate(self,dir_path,r=[]):
+		"""
+			This function will recursively scan a folder and retrieve file sizes and age of the files.
+			The data will be returned as an array of {size,age,label} items
+		"""
+		for child in os.listdir(dir_path):
+			path = os.path.join(dir_path, child)
+			if os.path.isdir(path):				
+				self.evaluate(path,r)
+
+			else:
+				size 		= os.path.getsize(path)
+				file_date 	= os.path.getatime(path)
+				file_date	= datetime.datetime.fromtimestamp(file_date)
+				now 		= datetime.datetime.now()
+				age 		= (now - file_date ).days
+				r.append({"label":path,"size":size,"age":age,"date":self.getNow()})
+		return r
 
 	def reboot(self,rows,limit) :
-		return np.sum([ int(self.toMB(item['size']) > self.toMB(limit)) for item in rows]) > 0
+		return np.sum([ 1 for item in rows if rows['size'] > limit ]) > 0
 	def composite(self):
-		d = [] #-- vector of details (age,size)
-		
-		now = datetime.datetime.today()
-		for folder in self.folders:
-			if os.path.exists(folder):
-				xo_raw = self.evaluate(folder)
-				xo = np.array(ML.Extract(['size','age'],xo_raw))
-				if len(xo) == 0:
-					continue
-				name = re.findall("([a-z,A-Z,0-9]+)",folder)
-				name = folder.split(os.sep)
-				if len(name) == 1:
-					name = [folder]
-				else:
-					i = len(name) -1
-					name = [name[i-1]+' '+name[i]]
-
-				name = name[0]
-				size = round(np.sum(xo[:,0]),2)
-				if size > 1000 :
-					size = round(size/1000,2)
-					units = ' GB'
-				elif size > 1000000:
-					size = round(size/1000000,2)
-					units = ' TB'
-				else:
-					size = size 
-					units = ' MB'
-				size = str(size)+ units
-				age = round(np.mean(xo[:,1]),2)
-				if age > 30 and age <= 365 :
-					age = round(age/30,2)
-					units = ' Months'
-				elif age > 365 :
-					age = round(age/365,2) 
-					units = ' Years'
-				else:
-					age = age
-					units = ' Days'
-				age = str(age)+units
-				N = len(xo[:,1])
-				xo = {"label":folder} #,"details":xo_raw,"summary":{"size":size,"age":age,"count":len(xo[:,1])}}
-				xo = dict(xo,**{"size":size,"age":age,"count":N})
-				xo["name"] = name
-				xo['day'] = now.day
-				xo['month'] = now.month
-				xo['year'] = now.year
-				xo['date'] = time.mktime(now.timetuple())
-				
-				d.append(xo)
-		
-		return d
+		out = []
+		for folder in self.folders :
+			
+			r = self.evaluate(folder,[])
+			file_count 	= len(r)
+			age 		= {"mean":np.mean([item['age'] for item in r] ),"var":np.mean([item['age'] for item in r])}
+			size 		= {"mean":np.mean([item['size'] for item in r] ),"var":np.mean([item['size'] for item in r])}
+			out.append({"label":folder,"stats":{"age":age,"size":size,"file_count":file_count},"logs":r})
+		return out
 
 
 # class Monitor (Thread):

src/utils/agents/actor.py

@@ -193,8 +193,8 @@ class Folders(Actor):
             @pre : isValid
         """
         folder = item['label']
-	name = folder.split(os.sep)
-	name = name[len(name)-1]
+	    name = folder.split(os.sep)
+	    name = name[len(name)-1]
         signature='-'.join([name,str(item['date']),str(item['count']),'files'])
         tarball=os.sep.join([folder,'..',signature])
         shutil.make_archive(tarball,'tar',folder)

test/TestServerMonitor.py

@@ -4,7 +4,7 @@ from monitor import Env, DetailProcess, ProcessCounter, Sandbox, FileWatch
 import monitor
 import os
 import json
-from utils.workers import Top, Learner
+# from utils.workers import Top, Learner
 #from multiprocessing import Lock
 from threading import Lock
 path = os.environ['MONITOR_CONFIG_PATH']
@@ -29,8 +29,8 @@ class TestMonitorServer(unittest.TestCase):
 		p = DetailProcess()
 		p.init(['kate','rabbitmq-server','python','apache2','firefox'])
 		r = p.composite()
-		for row in r:
-			print row['label'],row['status'],row['proc_count']
+		#for row in r:
+		#	print row['label'],row['status'], sum([1 for item in r if item['label']==row['label']])
 		self.assertTrue(r)
 	
 	def test_ProcessCount(self):
@@ -46,22 +46,13 @@ class TestMonitorServer(unittest.TestCase):
 		p	= Sandbox()
 		p.init({"sandbox":sandbox_path,"requirements":requirements_path})
 		p.composite()
-	def test_StartTop(self):		
-		lock = Lock()
-		p = Top(CONFIG,lock)		
-		#p.start()
-		
-		#p.join()
 	def test_StartLearner(self):
-		lock = Lock()
-		p = Learner(CONFIG,lock)
-		p.start()
+		pass
 	def test_FileWatch(self):
-		conf =CONFIG['monitor']['folder']
-		path =os.environ['FILE_PATH']
 		fw  = FileWatch()
-		fw.init([path])
-		print fw.composite()
+		fw.init(CONFIG['folders'])
+		#r =  fw.evaluate('/Users/steve/git/resume')
+		fw.composite()
 		
 if __name__ == '__main__' :
 	unittest.main()