def setIndexMaps(self, valuesR, valuesG=None, valuesB=None):
indexMapR = IndexMap(len(valuesR[0]), valuesR)
self._indexMaps.append(indexMapR)
self.addElement(indexMapR)
# Either one or three indexMaps
if (valuesG != None and valuesB != None):
indexMapG = IndexMap(len(valuesG[0]), valuesG)
self._indexMaps.append(indexMapG)
self.addElement(indexMapG)
indexMapB = IndexMap(len(valuesB[0]), valuesB)
self._indexMaps.append(indexMapB)
self.addElement(indexMapB)
def readChild(self, element):
child = None
if element.tag == 'Array':
child = Array()
child.read(element)
integers = bitDepthIsInteger(self.getAttribute('outBitDepth'))
child.setValuesAreIntegers(integers)
self._array = child
elif element.tag == 'IndexMap':
child = IndexMap()
child.read(element)
self._indexMaps.append(child)
return child
def readChild(self, element):
child = None
elementType = self.getElementType(element.tag)
if elementType == 'Array':
rawHalfs = not (self.getAttribute('rawHalfs') in [None, False])
child = Array(rawHalfs=rawHalfs)
child.read(element)
integers = bitDepthIsInteger(self.getAttribute('outBitDepth'))
child.setValuesAreIntegers(integers)
self._array = child
elif elementType == 'IndexMap':
child = IndexMap()
child.read(element)
self._indexMaps.append(child)
return child
try:
f_name = self.files.get(i)
except KeyError as err_key:
print('Can\'t find key: ', err_key, ' , ignored.')
continue
file_path = self.files.dir_path + '/' + f_name
pdf_reader = PdfFileReader(file_path)
pdf_writer.appendPagesFromReader(pdf_reader)
print(' ', f_name, ' is merged.')
self.merged += 1
output_file = self.files.dir_path + '/' + self.output
with open(output_file, 'wb') as outf:
pdf_writer.write(outf)
print('Finished merging ', self.merged, ' files.')
end_time = time.time()
print('Merging completed in ', end_time - start_time, ' s')
if __name__ == '__main__':
files = IndexMap()
files.set_dir('/path/to/dir_of_PDF_files')
RE_LIST = [
'\d+'
]
files.set_patterns(RE_LIST)
files.construct_map()
merger = PdfMerger(files, 'output.pdf')
merger.work()
import numpy as np
from IndexMap import IndexMap
from PageRank import PageRank
from StronglyConnectedComponent import SCC
nan = float("nan")
if __name__ == '__main__':
N = 7115
E = 103689
A_t = np.zeros((N, N), dtype=np.float)
A = np.zeros((N, N), dtype=np.float)
index_map = IndexMap()
print "Loading ",
with open("Wiki-Vote.txt", mode='r') as data_file:
for line in data_file:
if line[0] == '#':
continue
line = line.replace('\n', '')
ij_list = line.split('\t')
i = index_map.getIndex(int(ij_list[0]))
j = index_map.getIndex(int(ij_list[1]))
A_t[j, i] = 1
A[i, j] = 1
print "is finished."
print " [1] : SCC .............................................. "
scc = SCC(A)
A_temp = scc.removeDeadEnds()
print "> PageRank started..."
