def get_symbols_in_sublist(self, subdir):
'''
@summary: Returns all the symbols belonging to that subdir of the data store.
@param subdir: Specifies which subdir you want.
@return: A list of symbols belonging to that subdir
'''
pathtolook = self.rootdir + self.midPath + subdir
stocksAtThisPath = cached_listdir(pathtolook)
#Next, throw away everything that is not a .pkl And these are our stocks!
try:
stocksAtThisPath = [
x for x in stocksAtThisPath
if (str(x).find(str(self.fileExtensionToRemove)) > -1)
]
#Now, we remove the .pkl to get the name of the stock
stocksAtThisPath = [
(x.partition(str(self.fileExtensionToRemove))[0])
for x in stocksAtThisPath
]
except:
print("error: no path to " + subdir)
stocksAtThisPath = list()
return stocksAtThisPath
def getPathOfFile(self, symbol_name, bDelisted=False):
'''
@summary: Since a given pkl file can exist in any of the folders- we need to look for it in each one until we find it. Thats what this function does.
@return: Complete path to the pkl file including the file name and extension
'''
if not bDelisted:
for path1 in self.folderList:
if (os.path.exists(str(path1) + str(symbol_name + ".pkl"))):
# Yay! We found it!
return (str(str(path1) + str(symbol_name) + ".pkl"))
#if ends
elif (os.path.exists(str(path1) + str(symbol_name + ".csv"))):
# Yay! We found it!
return (str(str(path1) + str(symbol_name) + ".csv"))
#for ends
else:
''' Special case for delisted securities '''
lsPaths = []
for sPath in self.folderList:
if re.search('Delisted Securities', sPath) == None:
continue
for sFile in cached_listdir(sPath):
if not re.match('%s-\d*.pkl' % symbol_name, sFile) == None:
lsPaths.append(sPath + sFile)
lsPaths.sort()
return lsPaths
print("Did not find path to " + str(symbol_name) +
". Looks like this file is missing")
def get_all_symbols(self):
'''
@summary: Returns a list of all the symbols located at any of the paths for this source. @see: {__init__}
@attention: This will discard all files that are not of type pkl. ie. Only the files with an extension pkl will be reported.
'''
listOfStocks = list()
#Path does not exist
if (len(self.folderList) == 0):
raise ValueError("DataAccess source not set")
for path in self.folderList:
stocksAtThisPath = list()
#print str(path)
stocksAtThisPath = cached_listdir(str(path))
#Next, throw away everything that is not a .pkl And these are our stocks!
stocksAtThisPath = [
x for x in stocksAtThisPath
if (str(x).find(str(self.fileExtensionToRemove)) > -1)
]
#Now, we remove the .pkl to get the name of the stock
stocksAtThisPath = [
(x.partition(str(self.fileExtensionToRemove))[0])
for x in stocksAtThisPath
]
listOfStocks.extend(stocksAtThisPath)
#for stock in stocksAtThisPath:
#listOfStocks.append(stock)
return listOfStocks
def getStocks(listOfPaths):
listOfStocks = list()
#Path does not exist
print("Reading in all stock names...")
fileExtensionToRemove = ".h5"
for path in listOfPaths:
stocksAtThisPath = list()
stocksAtThisPath = cached_listdir(str(path))
#Next, throw away everything that is not a .h5 And these are our stocks!
stocksAtThisPath = [
x for x in stocksAtThisPath
if (str(x).find(str(fileExtensionToRemove)) > -1)
]
#Now, we remove the .h5 to get the name of the stock
stocksAtThisPath = [(x.partition(str(fileExtensionToRemove))[0])
for x in stocksAtThisPath]
for stock in stocksAtThisPath:
listOfStocks.append(stock)
return listOfStocks