def __init__(self, configPath, defaultDict={}):
if os.path.isdir(configPath):
raise "config file dict can only be created with file."
self.configPath = configPath
try:
f = open(self.configPath, "r")
self.config = json.load(f)
f.close()
except IOError:
print "no config file found:", self.configPath
self.config = defaultDict
def __init__(self, srcRoot, storageRoot, stateStoragePath = 'd:/state.txt'):
print "src root is:", srcRoot
self.srcRoot = transform.transformDirToInternal(srcRoot)
self.storageRoot = transform.transformDirToInternal(storageRoot)
self.stateStoragePath = stateStoragePath
try:
f = open(self.stateStoragePath,'r')
self.config = json.load(f)
f.close()
except IOError:
self.config = {}
def readEncryptedZipLog(self, encryptedZipLogPath):
zipLogPath = os.path.join(self.decryptionWorkingDir,
os.path.basename(encryptedZipLogPath).replace('.enclog', '.log'))
self.decCopier.copy(encryptedZipLogPath, zipLogPath)
try:
f = open(zipLogPath,'r')
res = json.load(f)
f.close()
return res
except IOError:
return {}
def processItem(self, job, item):
monitoringFullPath = transform.transformDirToInternal(item['monitoringPath'])
archiveId = gZipFolderCollectionPrefix + monitoringFullPath
if not self.collectionInDbForMonitoringPath.has_key(monitoringFullPath):
self.collectionInDbForMonitoringPath[monitoringFullPath] = collectionDatabase.collectionOnMongoDbBase(archiveId, self.dbInst.getCollectionDb())
objUuid = self.dbInst.addVirtualObj({"monitoringPath": monitoringFullPath, "zippedInfoCollectionId": archiveId});
idInCol = objUuid
self.zippedInfoCollectionList.addObj(idInCol, objUuid)
#Save the item in the archive collection: zippedInfoColllection://D:/tmp/
fullPath = transform.transformDirToInternal(item["fullPath"])
relativePath = transform.getRelativePathFromFull(fullPath, monitoringFullPath)
if not os.path.exists(fullPath):
job.delete()
return False#No job release, job was deleted.
#################################################################
# Start process the
#################################################################
if not self.collectionInDbForMonitoringPath[monitoringFullPath].exists(relativePath):
#This item is not in the collection, so we need to extract info from this item
newObj = self.dbInst.getFsObjFromFullPath(fullPath)
self.collectionInDbForMonitoringPath[monitoringFullPath].addObj(relativePath, newObj["uuid"])
zipFilePath = transform.transformDirToInternal(
fileTools.getTimestampWithFreeName(self.workingDir, gInfoFileDecryptedExt, gInfoFilePrefix))
self.decCopier.copy(fullPath, zipFilePath)
for i in zippedInfo(self.workingDir).enumItems(zipFilePath):
print '--------------------------------------------------'
print i
fp = open(i, 'r')
loadedFileInfo = json.load(fp)
print loadedFileInfo
for i in zippedInfo(self.workingDir).enumZippedFiles(zipFilePath):
fp = open(i, 'r')
print 'data file extracted:', i
'''
else:
#This item is not in the collection, so we need to extract info from this item
newObj = self.dbInst.getFsObjFromFullPath(fullPath)
self.collectionInDbForMonitoringPath[monitoringFullPath].addObj(relativePath, newObj["uuid"])
zipFilePath = transform.transformDirToInternal(
fileTools.getTimestampWithFreeName(self.workingDir, gInfoFileDecryptedExt, gInfoFilePrefix))
self.decCopier.copy(fullPath, zipFilePath)
for i in zippedInfo(self.workingDir).enumItems(zipFilePath):
print '--------------------------------------------------'
print i
fp = open(i, 'r')
loadedFileInfo = json.load(fp)
print loadedFileInfo
for i in zippedInfo(self.workingDir).enumZippedFiles(zipFilePath):
fp = open(i, 'r')
print 'data file extracted:', i
'''
return True#Release job
def __init__(self, srcRoot, storageRoot, stateStoragePath = 'd:/state.txt'):
#print 'src root:',srcRoot
self.srcRoot = transform.transformDirToInternal(srcRoot)
self.storageRoot = transform.transformDirToInternal(storageRoot)
#print self.srcRoot
#print self.storageRoot
self.stateStoragePath = stateStoragePath
try:
f = open(self.stateStoragePath,'r')
self.config = json.load(f)
f.close()
except IOError:
self.config = {}
#print 'storage root:', self.storageRoot
self.curArchivedSize = 0
self.curArchive = None
def processJob(self, job, item):
monitoringFullPath = transform.transformDirToInternal(item['monitoringPath'])
archiveId = "zippedInfoColllection://" + monitoringFullPath
if not self.collectionInDbForMonitoringPath.has_key(monitoringFullPath):
self.collectionInDbForMonitoringPath[monitoringFullPath] = collectionDatabase.collectionOnMongoDbBase(archiveId, self.dbInst.getCollectionDb())
#Save the item in the archive collection: zippedInfoColllection://D:/tmp/
fullPath = transform.transformDirToInternal(item["fullPath"])
relativePath = transform.getRelativePathFromFull(fullPath, monitoringFullPath)
if not self.collectionInDbForMonitoringPath[monitoringFullPath].exists(relativePath):
#This item is not in the collection, so we need to extract info from this item
newObj = self.dbInst.getFsObjFromFullPath(fullPath)
self.collectionInDbForMonitoringPath[monitoringFullPath].addObj(relativePath, newObj["uuid"])
for i in zippedInfo(self.workingDir).enumItems(fullPath):
fp = open(i, 'r')
loadedFileInfo = json.load(fp)
print loadedFileInfo
return True
def importDb(input, password = None):
dbSysInst = dbSys.dbSysSmart()
sf = open(input, 'r')
package = json.load(sf)
en = enc.encryptorBase64Out(password)
l = package["add"]
if (password is None) and package.has_key("encPass"):
print 'need password for importing'
return
if (not (password is None)):
if unicode(str(md5.new(password).hexdigest())) != package["encPass"]:
print 'pass not match:', unicode(str(md5.new(password+package["time-duration"]).hexdigest())), package["encPass"]
return
else:
l = listEncryptor.dictListDecryptor(l, en)
l = listEncryptor.dictListEncryptor(l)
dbSysInst.importDb(l)
def decryptBackup(sourceFullPath, target, password = None):
sf = open(sourceFullPath, 'r')
l = json.load(sf)
if l.has_key("encPass"):
if password is None:
print 'need password'
return
else:
en = enc.encryptorBase64Out(password)
if str(md5.new(password+l["time-duration"]).hexdigest()) != l["encPass"]:
print 'pass not match:', str(md5.new(password+l["time-duration"]).hexdigest()), l["encPass"]
return
res = dbExporterV3.dictListDecryptor(l["add"], en)
else:
res = l["add"]
l["add"] = res
del l["encPass"]
s = json.dumps(l, sort_keys=True, indent=4)
f = open(target,'w')
f.write(s)
f.close()
def importDb(input, username, targetPasswd, password = None):
dbSysInst = dbSys.dbSysSmart()
sf = open(input, 'r')
package = json.load(sf)
en = enc.encryptorBase64Out(password)
l = package["add"]
if (password is None) and package.has_key("encPass"):
print 'need password for importing'
return
if (not (password is None)):
if unicode(str(md5.new(password).hexdigest())) != package["encPass"]:
print 'pass not match:', unicode(str(md5.new(password+package["time-duration"]).hexdigest())), package["encPass"]
return
else:
l = listEncryptor.dictListDecryptor(l, en)
sysUser = service.ufsUser(u'system.user', u'system.pass')
if userMan.userManager().verifyPasswd(username, targetPasswd, dbSys.dbSysSmart(sysUser).getDb("passwdDb")):
print 'importing'
l = listEncryptor.dictListEncryptor(l, xorEnc.encryptorTxtOut(targetPasswd))
#print l
l = listEncryptor.setUser(l, username)
#print l
dbSysInst.importDb(l)
def subClassProcessItem(self, processingObj):
encZipFileFullPath = processingObj["fullPath"]
encryptedZipLogPath = encZipFileFullPath
ncl(encryptedZipLogPath)
zipLogPath = os.path.join(self.decryptionWorkingDir,
os.path.basename(encryptedZipLogPath).replace(gTargetEncLog, gZipLogExt))
cl(zipLogPath)
self.decCopier.copy(encryptedZipLogPath, zipLogPath)
#Extract log file from zip
logZip = zipClass.ZFile(zipLogPath, 'r')
#logPath = zipLogPath.replace(gZipLogExt, gLogExt)
logLoaded = False
for logFileName in logZip.list():
logZip.extract(logFileName, self.decryptionWorkingDir)
logPath = os.path.join(self.decryptionWorkingDir, logFileName)
#cl(logPath)
#Read log file
try:
f = open(logPath,'r')
newLog = json.load(f)
f.close()
except IOError:
newLog = None
raise 'log not read'
if not (newLog is None):
#########################
#Log loaded, update collection
#########################
logLoaded = True
print 'extracting info from log', encZipFileFullPath
for i in newLog:
relaPath = transform.formatRelativePath(i)
ncl(newLog[i])
#Remove the uuid in log file
if newLog[i].has_key("uuid"):
newLog[i]["originalUuid"] = newLog[i]["uuid"]
del newLog[i]["uuid"]
#################
#Get collection
#################
if self.logCollection.exists(relaPath):
itemUuid = self.logCollection.getObjUuid(relaPath)
ncl("returned uuid:", itemUuid)
item = self.objDb.getObjFromUuid(itemUuid)
#Conflict, check if update needed
ncl(newLog[i]["timestamp"])
ncl(item["timestamp"])
if newLog[i]["timestamp"] > item["timestamp"]:
#The new item is newer, replace the old one
ncl('updating duplicated item to 1st one:', newLog[i]["timestamp"], item["timestamp"])
objUuid = self.objDb.addVirtualObj(newLog[i])
self.logCollection.updateObjUuid(relaPath, objUuid)
else:
ncl("no update, ignore")
else:
#Add object to obj objDb
objUuid = self.objDb.addVirtualObj(newLog[i])
#Add obj to collection
self.logCollection.addObj(relaPath, objUuid)
ncl("added new item", relaPath, newLog[i])
'''
encZipFileFullPath = transform.transformDirToInternal(encZipFileFullPath)
zipFileFullPath = self.getZipFile(encZipFileFullPath.replace(gTargetEncLog, gTargetEnc))
extractedItemInfo = self.getItemState(relaPath)
extractedItemFullPath = os.path.join(self.workingDir, i)
extractedItem = extractedZipStorageItem(self.workingDir,
extractedItemFullPath, extractedItemInfo, zipFileFullPath, None, relaPath)
'''
yield newLog[i]
'''
def subClassProcessItem(self, processingObj):
#Check if the log file and data file are both OK
encZipFileFullPath = processingObj["fullPath"]
if not (re.search('\.enclog$', encZipFileFullPath) is None):
#An encrypted zip file, go on
if os.path.exists(encZipFileFullPath.replace('.enclog', '.enc')):
#Log and Data are both OK
encryptedZipLogPath = encZipFileFullPath
ncl(encryptedZipLogPath)
zipLogPath = os.path.join(self.decryptionWorkingDir,
os.path.basename(encryptedZipLogPath).replace('.enclog', '.log'))
ncl(zipLogPath)
self.decCopier.copy(encryptedZipLogPath, zipLogPath)
try:
f = open(zipLogPath,'r')
newLog = json.load(f)
f.close()
except IOError:
newLog = None
if not (newLog is None):
#########################
#Log loaded, update collection
#########################
for i in newLog:
relaPath = transform.formatRelativePath(i)
ncl(newLog[i])
#################
#Get collection
#################
if self.logCollection.exists(relaPath):
itemUuid = self.logCollection.getObjUuid(relaPath)
ncl("returned uuid:", itemUuid)
item = self.db.getObjFromUuid(itemUuid)
#Conflict, check if update needed
ncl(newLog[i]["timestamp"])
ncl(item["timestamp"])
if newLog[i]["timestamp"] > item["timestamp"]:
#The new item is newer, replace the old one
ncl('updating duplicated item to 1st one:', newLog[i]["timestamp"], item["timestamp"])
objUuid = self.db.addVirtualObj(newLog[i])
self.logCollection.updateObjUuid(relaPath, objUuid)
else:
ncl("no update, ignore")
else:
#Add object to obj db
objUuid = self.db.addVirtualObj(newLog[i])
#Add obj to collection
self.logCollection.addObj(relaPath, objUuid)
ncl("added new item", relaPath, newLog[i])
#################################
#Process data
#################################
encZipFileFullPath = transform.transformDirToInternal(encZipFileFullPath)
ncl(encZipFileFullPath)
zipFileFullPath = self.getZipFile(encZipFileFullPath.replace(".enclog", ".enc"))
#For all element in the zip file
#Enumerate all files in the decrypted zip file
zf = zipClass.ZFile(zipFileFullPath, 'r')
#Generate a log file if it does not exist
for i in zf.list():
#zf.extract(i, self.workingDir)
extractedItemFullPath = os.path.join(self.workingDir, i)
relaPath = transform.formatRelativePath(i)
extractedItemInfo = self.getItemState(relaPath)
extractedItem = encZipStorage.extractedZipStorageItem(self.workingDir,
extractedItemFullPath, extractedItemInfo, zf, relaPath)
###########################
#Store the file
###########################
self.targetCollection.store(extractedItem)
##########################
#Everything goes OK
#Quit
##########################
return
else:
cl('Load log file failed',encZipFileFullPath)
else:
#Data file not exist, push it back
ncl('Data file not exist, push it back: ',encZipFileFullPath)
else:
ncl('not a encrypted zip file: ',encZipFileFullPath)
return
############
#Item not processed, push it back
############
processorBase.cacheCollectionProcessorBase.subClassProcessItem(self, processingObj)
def subClassProcessItem(self, processingObj):
encZipFileFullPath = processingObj["fullPath"]
encryptedZipLogPath = encZipFileFullPath
ncl(encryptedZipLogPath)
zipLogPath = os.path.join(
self.decryptionWorkingDir, os.path.basename(encryptedZipLogPath).replace(gTargetEncLog, gZipLogExt)
)
cl(zipLogPath)
self.decCopier.copy(encryptedZipLogPath, zipLogPath)
# Extract log file from zip
logZip = zipClass.ZFile(zipLogPath, "r")
# logPath = zipLogPath.replace(gZipLogExt, gLogExt)
logLoaded = False
for logFileName in logZip.list():
logZip.extract(logFileName, self.decryptionWorkingDir)
logPath = os.path.join(self.decryptionWorkingDir, logFileName)
# cl(logPath)
# Read log file
try:
f = open(logPath, "r")
newLog = json.load(f)
f.close()
except IOError:
newLog = None
raise "log not read"
if not (newLog is None):
#########################
# Log loaded, update collection
#########################
logLoaded = True
print "extracting info from log", encZipFileFullPath
for i in newLog:
relaPath = transform.formatRelativePath(i)
ncl(newLog[i])
# Remove the uuid in log file
if newLog[i].has_key("uuid"):
newLog[i]["originalUuid"] = newLog[i]["uuid"]
del newLog[i]["uuid"]
#################
# Get collection
#################
if self.logCollection.exists(relaPath):
itemUuid = self.logCollection.getObjUuid(relaPath)
ncl("returned uuid:", itemUuid)
item = self.objDb.getObjFromUuid(itemUuid)
# Conflict, check if update needed
ncl(newLog[i]["timestamp"])
ncl(item["timestamp"])
if newLog[i]["timestamp"] > item["timestamp"]:
# The new item is newer, replace the old one
ncl("updating duplicated item to 1st one:", newLog[i]["timestamp"], item["timestamp"])
objUuid = self.objDb.addVirtualObj(newLog[i])
self.logCollection.updateObjUuid(relaPath, objUuid)
else:
ncl("no update, ignore")
else:
# Add object to obj objDb
objUuid = self.objDb.addVirtualObj(newLog[i])
# Add obj to collection
self.logCollection.addObj(relaPath, objUuid)
ncl("added new item", relaPath, newLog[i])
if logLoaded:
#################################
# Process data
#################################
print "extracting info from log complete", encZipFileFullPath
encZipFileFullPath = transform.transformDirToInternal(encZipFileFullPath)
ncl(encZipFileFullPath)
zipFileFullPath = self.getZipFile(encZipFileFullPath.replace(gTargetEncLog, gTargetEnc))
# For all element in the zip file
# Enumerate all files in the decrypted zip file
zf = zipClass.ZFile(zipFileFullPath, "r")
# Generate a log file if it does not exist
for i in zf.list():
# zf.extract(i, self.workingDir)
extractedItemFullPath = os.path.join(self.workingDir, i)
relaPath = transform.formatRelativePath(i)
ncl(relaPath)
extractedItemInfo = self.getItemState(relaPath)
extractedItem = extractedZipStorageItem(
self.workingDir, extractedItemFullPath, extractedItemInfo, zipFileFullPath, zf, relaPath
)
"""
###########################
#Store the file
###########################
#self.targetCollection.store(extractedItem)
"""
###########################
# Returning object
###########################
yield extractedItem
##########################
# Everything goes OK
# Quit
##########################
return
else:
cl("Load log file failed", encZipFileFullPath)
