file separati vengono isolati in cartelle differenti

This commit is contained in:
2019-12-02 20:12:00 +01:00
parent 8628c3dbfb
commit 01dd92e4da
2 changed files with 15 additions and 8 deletions

View File

@@ -57,6 +57,9 @@ class bananaSPLITTER(threading.Thread):
except IOError as e:
self.log.critical("Impossibile aprire il file: {}! [{}]".format(self.fileName,e))
raise BaseException("OpenFile")
except UnicodeDecodeError as ee:
self.log.critical("Il file [{}] contiene caratteri non compatibili con la codifica scelta! [{}]"
.format(self.fileParams['name'],ee))
pass
def remEmptyLines(self):
@@ -205,6 +208,8 @@ class bananaSPLITTER(threading.Thread):
pass
def saveSeparate(self):
os.mkdir(self.paths['OUTworkPath']+slugify(self.fileName))
self.paths['OUTworkPath']=self.paths['OUTworkPath']+slugify(self.fileName)
self.log.info("Salvo gli articoli in file separati...")
self.log.debug("Persorso: {0}".format(self.paths['OUTworkPath'].format('nomeFile')))
for ff in self.fileList:

View File

@@ -4,6 +4,7 @@ Created on 1 dic 2019
@author: Emanuele Trabattoni
'''
import os
from glob import glob
from libsplit import bananaSPLITTER
from libconfload import bananaCONF
from libfancylogger import fancyLogger
@@ -15,14 +16,15 @@ confl.open()
confl.use("testEN.json")
splconf = confl.getParams("splitter")
splconf["name"] = splconf["paths"]["INworkPath"]+"GUARDIAN 1989.txt"
splitter = bananaSPLITTER(fileParams=splconf, logger=logger)
splitter.openFile()
splitter.remEmptyLines()
splitter.splitFile()
splitter.removeDuplicates()
splitter.saveBody()
splitter.saveSeparate()
for f in glob(splconf["paths"]["INworkPath"]+"*.txt"):
splconf["name"] = f
splitter = bananaSPLITTER(fileParams=splconf, logger=logger)
splitter.openFile()
splitter.remEmptyLines()
splitter.splitFile()
splitter.removeDuplicates()
splitter.saveBody()
splitter.saveSeparate()