uso di multiprocessing al posto d threading per una concorrenza vera

serve? da decidere

bananaSPLIT/libbananasplit/libsplit.py

@@ -3,13 +3,12 @@ Created on 2 nov 2019
 
 @author: Emanuele Trabattoni
 '''
-from libfancylogger import fancyLogger
 from slugify.slugify import slugify
-import threading, time, parse, re, copy, os
+import time, parse, re, copy, os, multiprocessing
 
-class bananaSPLITTER(threading.Thread):
+class bananaSPLITTER(multiprocessing.Process):
 	def __init__(self, fileParams=None, logger=None):
-		threading.Thread.__init__(self)
+		multiprocessing.Process.__init__(self)
 		self.fileParams = fileParams
 		self.log = logger
 		self.rawFile = None
@@ -44,7 +43,7 @@ class bananaSPLITTER(threading.Thread):
 				for idx, ff in enumerate(self.fileList):
 					ff['duplicate']=False
 					self.fileList[idx]=ff
-				print('Salto il controllo dei duplicati..')
+				self.log.warn('Salto il controllo dei duplicati..')
 			if self.settings['saveSeparateFiles']:
 				self.saveSeparate()
 			if self.settings['saveBodyFile']:
@@ -128,11 +127,11 @@ class bananaSPLITTER(threading.Thread):
 										newsPaperName = self.settings['nameNotFoundStr']
 								except:
 									self.log.warn("E' successo qualcosa mentre stavo cercando il nome della pubblicazione,\
-									controlla i file di uscita! \n\t[{}]".format(prevLine.strip()))
+									controlla i file di uscita! [{}]".format(prevLine.strip()))
 							else:
 								newsPaperName = self.settings['nameNotFoundStr']
 						except:
-							self.log.warn("Ho trovato una riga ambigua.. potrebbe essere una data ma non so: \n\t[{}]". format(l.strip('\r\n')))
+							self.log.warn("Ho trovato una riga ambigua.. potrebbe essere una data ma non so: [{}]". format(l.strip('\r\n')))
 							pass
 					elif lineWords[0] in self.docStruct['headWords']:
 						#cambio stato e inizializzo un nuovo documento da riempire
@@ -237,17 +236,17 @@ class bananaSPLITTER(threading.Thread):
 	
 	def saveBody(self):
 		self.log.info('Salvo gli articoli in un singolo file vicino agli originali...')
+		self.log.debug('Persorso: {0}'.format(self.outPath))
 		os.chdir(self.outPath)
-		print ('Persorso: {0}'.format(self.outPath))
 		try:
 			fName=slugify(self.fileName)
 			fName='BODYFILE_{0}.txt'.format(fName[:self.settings['maxTitleLen']])
 			fileContent = os.linesep.join([cc['content'] for cc in self.fileList])
-			out=open(self.paths['OUTworkPath']+'{0}'.format(fName),'wb')
+			out=open('{0}'.format(fName),'wb')
 			out.write(fileContent.encode(self.settings['encoding']))
 			out.close()
 		except IOError as e:
-			print("OOPS! Qualcosa e\' andato storto, non riesco a scrivere il file: {}".format(e))
+			self.log.error("Qualcosa e\' andato storto, non riesco a scrivere il file: {}".format(e))
 		pass
 
 	

bananaSPLIT/libbananasplit/testEN.json

@@ -67,7 +67,7 @@
 			"saveSeparateFiles": true,
 			"saveBodyFile": true,
 			"saveBodyNumber": true,
-			"delLF": true,
+			"delLF": false,
 			"delWordBreak": true,
 			"delChars": [
 				"'",

bananaSPLIT/libtestmain.py

@@ -10,25 +10,24 @@ from libsplit import bananaSPLITTER
 from libconfload import bananaCONF
 from libfancylogger import fancyLogger
 
-print("CWD-> "+os.getcwd())
-logger = fancyLogger(fileLog = False)
-confl = bananaCONF(workdir=r"./libbananasplit", logger=logger)
-confl.open()
-
-confl.use("testEN.json")
-splconf = confl.getParams("splitter")
-splist = []
-os.chdir(splconf["paths"]["INworkPath"])
-for f in glob("*.txt"):
-	splconf["name"] = f
-	logger.info("-"*80)
-	splitter = bananaSPLITTER(fileParams=deepcopy(splconf), logger=logger)
-	splist.append(splitter)
-	splitter.start()
-	splitter.join()
-	del splitter
-
-logger.info("\n"+"="*50+"\n\tFINITO!!!\n"+"="*50)
+if __name__ == "__main__":
+	print("CWD-> "+os.getcwd())
+	logger = fancyLogger(fileLog = False)
+	confl = bananaCONF(workdir=r"./libbananasplit", logger=logger)
+	confl.open()
+	
+	confl.use("testEN.json")
+	splconf = confl.getParams("splitter")
+	splist = []
+	os.chdir(splconf["paths"]["INworkPath"])
+	for f in glob("*.txt"):
+		splconf["name"] = f
+		logger.info("-"*80)
+		splitter = bananaSPLITTER(fileParams=deepcopy(splconf), logger=logger)
+		splist.append(splitter)
+		splitter.start()
+		
+	logger.info("\n"+"="*50+"\n\tFINITO!!!\n"+"="*50)