def resize_toolbar_bitmap_each(path,
percent,
method=ResizeMethod.Bicubic,
out_path=None):
bmp = Bitmap.fromFileEx(path)
bmps = bmp.splitHorizontal()
width, height = bmps[0].size
width = round(width * percent / 100)
height = round(height * percent / 100)
size = (width, height)
print(
f'resize toolbar bitmap {percent} {method.name}: {bmps[0].size} => {size}'
)
resized = []
for bmp in bmps:
bmp = bmp.resize(size, method=method)
resized.append(bmp)
bmp = Bitmap.concatHorizontal(resized)
if not out_path:
name, ext = os.path.splitext(path)
out_path = f"{name}_{height}_{percent}_{method.name}{ext}"
save_bitmap(bmp, out_path)
def concat_images(horizontal, paths, out_path):
if horizontal:
print('concat horizontal:', ', '.join(paths), '=>', out_path)
else:
print('concat vertical:', ', '.join(paths), '=>', out_path)
bmps = []
for path in paths:
bmp = Bitmap.fromFileEx(path)
bmps.append(bmp)
if horizontal:
bmp = Bitmap.concatHorizontal(bmps)
else:
bmp = Bitmap.concatVertical(bmps)
save_bitmap(bmp, out_path)
def second_approach_algorithm(self, input_logs, min_sup_rel,
pass_next_level):
if min_sup_rel == None:
min_sup_rel = self.min_sup
#STEP 0: SCAN THE DATABASE TO STORE THE FIRST BIT POSITION OF EACH SEQUENCE AND CALCULATE THE TOTAL NUMBER OF BIT FOR EACH BITMAP
bit_index = 0
for (idx, row) in input_logs.iterrows():
self.sequence_size.append(bit_index)
bit_index += len(
row.loc['index_in_rules_plus_one']
) # I want in bit_index the size of the sequence to know the start id of the new sequence
self.last_bit_index = bit_index - 1 # last item of the last seq
self.min_sup = math.ceil(
min_sup_rel * len(self.sequence_size)) #absolute minimum support
if self.min_sup == 0:
self.min_sup = 1
#STEP1: SCAN THE DATABASE TO CREATE THE BITMAP VERTICAL DATABASE REPRESENTATION
sid = tid = 0 # sequence id and itemset id
for (idx, row) in input_logs.iterrows():
for itemset in row.loc['index_in_rules_plus_one']:
if itemset in self.vertical_db:
self.vertical_db[itemset].register_bit(
sid, tid, self.sequence_size)
else:
self.vertical_db[itemset] = Bitmap(self.last_bit_index)
self.vertical_db[itemset].register_bit(
sid, tid, self.sequence_size)
tid += 1
sid += 1
tid = 0
# STEP2: REMOVE INFREQUENT ITEMS FROM THE DATABASE BECAUSE THEY WILL NOT APPEAR IN ANY FREQUENT SEQUENTIAL PATTERNS
for key in list(self.vertical_db.keys()):
# if the cardinality of this bitmap is lower than the itemset is removed
if self.vertical_db[key].get_support() < self.min_sup:
del self.vertical_db[key]
#STEP3: BREADTH FIRST SEARCH
prefixs = pd.DataFrame(columns=['prefix', 'bitmap'])
for key in self.vertical_db.keys():
# We create a prefix with that item
prefix = Prefix()
prefix.add_itemset(Itemset())
prefix.itemsets[0].add_item(key)
prefixs = prefixs.append(
{
'prefix': prefix,
'bitmap': self.vertical_db[key]
},
ignore_index=True)
self.bfs_search(prefixs, 2, pass_next_level)
self.ranking_results()
def convert_image(path, out_path=None):
if not out_path:
name, ext = os.path.splitext(path)
if ext.lower() == 'bmp':
out_path = name + '-converted' + '.bmp'
else:
out_path = name + '.bmp'
print('convert image:', path, '=>', out_path)
bmp = Bitmap.fromFileEx(path)
#bmp.resolution = (96, 96)
save_bitmap(bmp, out_path)
def spam_algorithm(self, input_logs, min_sup_rel):
#STEP 0: SCAN THE DATABASE TO STORE THE FIRST BIT POSITION OF EACH SEQUENCE AND CALCULATE THE TOTAL NUMBER OF BIT FOR EACH BITMAP
bit_index = 0
for (idx, row) in input_logs.iterrows():
self.sequence_size.append(bit_index)
bit_index += len(
row.loc['index_in_rules_plus_one']
) # I want in bit_index the size of the sequence to know the start id of the new sequence
self.last_bit_index = bit_index - 1
self.min_sup = math.ceil(
min_sup_rel * len(self.sequence_size)) #absolute minimum support
if self.min_sup == 0:
self.min_sup = 1
#STEP1: SCAN THE DATABASE TO CREATE THE BITMAP VERTICAL DATABASE REPRESENTATION
sid = tid = 0 # sequence id and itemset id
for (idx, row) in input_logs.iterrows():
for itemset in row.loc['index_in_rules_plus_one']:
if itemset in self.vertical_db:
self.vertical_db[itemset].register_bit(
sid, tid, self.sequence_size)
else:
self.vertical_db[itemset] = Bitmap(self.last_bit_index)
self.vertical_db[itemset].register_bit(
sid, tid, self.sequence_size)
tid += 1
sid += 1
tid = 0
# STEP2: REMOVE INFREQUENT ITEMS FROM THE DATABASE BECAUSE THEY WILL NOT APPEAR IN ANY FREQUENT SEQUENTIAL PATTERNS
frequent_items = list()
for key in list(self.vertical_db.keys()):
# if the cardinality of this bitmap is lower than the itemset is removed
if self.vertical_db[key].get_support() < self.min_sup:
del self.vertical_db[key]
else:
# otherwise, we save this item as a frequent sequential pattern of size 1
if self.min_pattern_length <= 1 and self.max_pattern_length >= 1:
self.save_pattern(key)
frequent_items.append(key)
# STEP3: WE PERFORM THE RECURSIVE DEPTH FIRST SEARCH TO FIND LONGER SEQUENTIAL PATTERNS RECURSIVELY
if self.max_pattern_length > 1:
# for each frequent item
for key in list(self.vertical_db.keys()):
# We create a prefix with that item
prefix = Prefix()
prefix.add_itemset(Itemset())
prefix.itemsets[0].add_item(key)
# We call the depth first search method with that prefix list of frequent items to try to find sequential patterns by appending some of these items
self.dfs_pruning(prefix, self.vertical_db[key], frequent_items,
frequent_items, key, 2)
def dump_bitmap(path):
print('dump bitmap:', path)
bmp = Bitmap.fromFile(path)
print(bmp.fileHeader)
print(bmp.infoHeader)
data_path = path + '.data'
print('write:', data_path, len(bmp.data))
with open(data_path, 'wb') as fd:
fd.write(bmp.data)
dump_path = os.path.splitext(path)[0] + '-dump.bmp'
print('write:', dump_path)
bmp.save(dump_path)
def flip_image(horizontal, path, out_path=None):
if not out_path:
name, ext = os.path.splitext(path)
out_path = name + '-flip' + ext
if horizontal:
print('flip horizontal:', path, '=>', out_path)
else:
print('flip vertical:', path, '=>', out_path)
bmp = Bitmap.fromFileEx(path)
if horizontal:
bmp = bmp.flipHorizontal()
else:
bmp = bmp.flipVertical()
save_bitmap(bmp, out_path)
def resize_toolbar_bitmap_whole(path, percent, method=ResizeMethod.Bicubic, out_path=None):
bmp = Bitmap.fromFileEx(path)
width, height = bmp.size
width = round(width * percent/100)
height = round(height * percent/100)
size = (width, height)
print(f'resize toolbar bitmap {percent} {method.name}: {bmp.size} => {size}')
bmp = bmp.resize(size, method=method)
if not out_path:
name, ext = os.path.splitext(path)
out_path = f"{name}_{height}_{percent}_{method.name}{ext}"
save_bitmap(bmp, out_path)
def split_image(horizontal, path, dims=None, out_path=None, ext=None):
name, old_ext = os.path.splitext(path)
if not out_path:
out_path = name + '-split'
if not ext:
ext = old_ext
if isinstance(dims, str):
dims = _parse_split_dims(dims)
if horizontal:
print('split horizontal:', path, dims, '=>', out_path)
else:
print('split vertical:', path, dims, '=>', out_path)
bmp = Bitmap.fromFileEx(path)
if horizontal:
bmps = bmp.splitHorizontal(dims)
else:
bmps = bmp.splitVertical(dims)
save_bitmap_list(bmps, out_path, ext)
H.writeSpeakersDictionary('./dictionaries/EvenOdd/speakersEvenOdd.pkl')
print('\n', len(dates), '\n')
H.writeDebatesDictionary('./dictionaries/EvenOdd/debatesEvenOdd.pkl')
comm.send(data1, dest=1, tag=1)
comm.send(data, dest=2, tag=1)
else:
data = comm.recv(source=0, tag=1)
debates = readDebatesDictionary(
'./dictionaries/EvenOdd/debatesEvenOdd.pkl')
speakers = readSpeakersDictionary(
'./dictionaries/EvenOdd/speakersEvenOdd.pkl')
myBitmap = Bitmap(len(debates), len(speakers))
for x in data:
debateNameofInterest = x.get('name') + ': ' + x.find(
'{http://docs.oasis-open.org/legaldocml/ns/akn/3.0}heading').text
for y in x.iter(
'{http://docs.oasis-open.org/legaldocml/ns/akn/3.0}speech'):
my1 = debates.get(debateNameofInterest)
my2 = speakers.get(y.get('by'))
myBitmap.createBitmap(my1 - 1, my2 - 1)
array = (myBitmap.array)
comm.send(array, dest=0)
if rank == 0:
even = comm.recv(source=1)