def make_data(entries):
data_train = defaultdict(list)
for e in entries:
good_aggs = [
l['sorted_triples'] for l in e.lexes
if l['comment'] == 'good' and l['sorted_triples'] and sum(
len(x) for x in l['sorted_triples']) == len(e.triples)
]
if good_aggs:
all_aggs = partitions(e.triples)
for agg in all_aggs:
agg = tuple([tuple(x) for x in agg])
distance = calc_distance(agg, good_aggs)
data = (agg, distance)
data_train[len(e.triples)].append(data)
return data_train
def acc_majority_agg(train_sa_db, test_sa_db):
from more_itertools import partitions
from operator import itemgetter
counters = analyze_agg_patterns(train_sa_db)
tp = 0
for ts, aggs in test_sa_db:
n = len(ts)
all_partitions = list(partitions(ts))
all_patterns = [extract_agg_pattern(agg) for agg in all_partitions]
all_counts = [counters[n][pat] for pat in all_patterns]
choosen_agg = max(zip(all_partitions, all_counts),
key=itemgetter(1))[0]
choosen_agg = [tuple(agg_part) for agg_part in choosen_agg]
if choosen_agg in aggs:
tp += 1
return tp / len(test_sa_db)
def partition_mit(s: str) -> list:
# Copied from: https://stackoverflow.com/questions/4904430/find-all-list-permutations-of-splitting-a-string-in-python
if s:
for lst in mit.partitions(s):
yield ["".join(sublst) for sublst in lst]
else:
yield []
def select_sentence_aggregation(self, dp, n_triples):
sas = list(partitions(dp))
sas_scores = self.sa_scorer(sas, n_triples)
sas = sort_together([sas_scores, sas], reverse=True)[1]
return sas
def _span_partitions(self,
span: Span) -> Generator[List[Token], None, None]:
if len(span) == 1:
return span[0]
for partition in partitions(span):
yield [
spacy_utlis.make_token(nlp=self._nlp, word=''.join(element))
for element in partition
]
def select_sa(self, dp):
sas = list(partitions(dp))
sas_scores = self.sa_scorer(sas)
sas_scores, sas = sort_together([sas_scores, sas], reverse=True)
self.logger.debug('Sentence Aggregation: {}'.format('\n'.join(f'{score:.3f} -> {sa}' for score, sa in zip(sas_scores, sas))))
return [[tuple(sa_part) for sa_part in sa] for sa in sas[:self.max_sa]]
def is_split_number(what: int):
""" check if the number is split """
sqr = int(sqrt(what))
for comb in partitions(str(what)):
result = 0
for item in comb:
result += int(''.join(item))
if result == sqr:
return 1
return 0
def get_search_terms(current_line: str) -> Dict[int, Set[str]]:
# Remove unwanted characters
modified_line = current_line.translate(translation_table)
# Remove multiple spaces
modified_line = re.sub(r"\s\s+", " ", modified_line)
words = [word.rstrip(",") for word in modified_line.split(" ")]
results: Dict[int, Set[str]] = {}
for partition in partitions(words):
for subpartition in partition:
subpartition_word_count: int = len(subpartition)
if subpartition_word_count not in results.keys():
results[subpartition_word_count] = set()
result = " ".join(subpartition)
results[subpartition_word_count].update([result])
return results
def get_phrases(phrase_pronunciations: PhrasePronunciationList) -> PhraseList:
"""
Finally convert pronunciations of phrases back into english phrases consisting of words.
"""
phrases = []
print(
f'Looking over {len(phrase_pronunciations)} different pronunciations of sentence...')
for pronunciation in phrase_pronunciations:
for part in partitions(pronunciation):
try:
phrases.extend(part_to_phrases(part))
except KeyError:
pass
return phrases
def multiplicative_partitions(n, k=None):
factors = it.chain.from_iterable([p] * m
for p, m in sp.factorint(n).items())
# TODO: Try to avoid the filter_seen step by generating distinct partitions
# directly if possible.
#
# Otherwise, perhaps multiplicative partitioning should be implemented from
# scratch
# https://stackoverflow.com/questions/8558292/how-to-find-multiplicative-partitions-of-any-integer.
ps = mit.partitions(factors)
ps = it.takewhile(lambda p: len(p) <= k, ps) if k else ps
ps = map(lambda p: tuple(sorted(map(math.prod, p))), ps)
ps = filter_seen(ps)
ps = it.chain.from_iterable(map(mit.distinct_permutations, ps))
yield from ps
def acc_random(sa_db, seed=0):
from random import Random
from more_itertools import partitions
r = Random(seed)
tp = 0
for ts, aggs in sa_db:
all_partitions = list(partitions(ts))
random_partition = r.choice(all_partitions)
random_partition = [tuple(part) for part in random_partition]
if random_partition in aggs:
tp += 1
return tp / len(sa_db)
def _preprocess(z, stopwords=stopwords):
tokens = [tok for tok in word_tokenize(z) if tok not in stopwords]
parts = [[" ".join(tokens) for tokens in part]
for part in partitions(tokens)]
return parts
index = 0
for a, b, c in product(
letters, repeat=3): # These 3-letter combinations are called "roots"
index += 1
insert_word(0, 3, index, a + b + c, 1)
# Parse all books of the Tanach, insert all words/verses
for book in booklist:
parse_book(f"morphhb/wlc/{book}.xml", booklist[book])
# Populate formations table
print("inserting formations for", len(worddict), "items...")
for word, wordnum in worddict.items():
for part in partitions(word):
if len(part) != 1:
result = parse_partition(part)
if result:
insert_formation(wordnum, result)
if len(part) == 3:
result = parse_inside(part)
if result:
insert_formation(wordnum, result)
# Save the in-memory DB to a file
conn.commit()
db_file = sqlite3.connect("bible.db")
conn.backup(db_file)
def understanding_partition_size():
for x in range(1, 30):
deck = list(range(x))
print(x, 2**x, len(list(more_itertools.partitions(deck))))
