def _query_join_relation(self, query: BaseQuery,
root_relation: str) -> BaseQuery:
"""
Helper function that applies necessary joins for dotted columns on a
SQLAlchemy query object
:param query: SQLAlchemy query object
:param root_relation: The root part of a dotted column, so the root relation
:return: Transformed SQLAlchemy Query
"""
relations = self.get_related_model_and_join(root_relation)
for relation in relations:
model_relation, relation_join = relation
# Support multiple joins for the same table
if self.is_model_already_joined(query, model_relation):
# Since the join already exists apply a new aliased one
model_relation = aliased(model_relation)
# The binary expression needs to be inverted
relation_pk = self.get_pk(model_relation)
relation_join = BinaryExpression(relation_join.left,
relation_pk,
relation_join.operator)
query = query.join(model_relation, relation_join, isouter=True)
return query
def query_articles_by_tag(self, tag: str, query_obj: BaseQuery = None):
if query_obj is not None:
query_obj = query_obj.join(Article.tags).filter(Tag.name == tag)
else:
query_obj = Article.query.join(
Article.tags).filter(Tag.name == tag)
return query_obj
def apply_filter(query: BaseQuery) -> BaseQuery:
for param, value in request.args.items():
if param not in {"fields", "sort", "page", "limit"}:
if param == "username":
query = query.filter(
GithubUserInfo.username.like(f"%{value}%"))
continue
if param == "language":
langs_list = value.split(" ")
for lang in langs_list:
query = query.join(GithubUserInfo.repositories).filter(
GithubRepositories.languages.contains(lang))
continue
if param.startswith("number_of_repositories"):
if param.endswith("[gt]"):
users_ids_list = []
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
if len(user.repositories) > int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
elif param.endswith("[gte]"):
users_ids_list = []
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
if len(user.repositories) >= int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
elif param.endswith("[lt]"):
users_ids_list = []
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
if len(user.repositories) < int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
elif param.endswith("[lte]"):
users_ids_list = []
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
if len(user.repositories) <= int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
else:
users_ids_list = []
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
if len(user.repositories) == int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
if param.startswith("date"):
pass
## last_commit_date scraping functionality to be added to database
## try:
## value = datetime.strptime(value, "%d-%m-%Y").date()
## if param.endswith("[gt]"):
## query = query.filter(GithubUserInfo.date > value)
## continue
## elif param.endswith("[gte]"):
## query = query.filter(GithubUserInfo.date >= value)
## continue
## elif param.endswith("[lt]"):
## query = query.filter(GithubUserInfo.date < value)
## continue
## elif param.endswith("[lte]"):
## query = query.filter(GithubUserInfo.date <= value)
## continue
## else:
## query = query.filter(GithubUserInfo.date == value)
## continue
## except ValueError:
## continue
if param.startswith("stars"):
if param.endswith("[gt]"):
users_ids_list = []
users_stars_dict = {}
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
users_stars_dict[user] = 0
for repo in user.repositories:
users_stars_dict[user] = (
users_stars_dict[user] + repo.stars)
if users_stars_dict[user] > int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
elif param.endswith("[gte]"):
users_ids_list = []
users_stars_dict = {}
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
users_stars_dict[user] = 0
for repo in user.repositories:
users_stars_dict[user] = (
users_stars_dict[user] + repo.stars)
if users_stars_dict[user] >= int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
elif param.endswith("[lt]"):
users_ids_list = []
users_stars_dict = {}
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
users_stars_dict[user] = 0
for repo in user.repositories:
users_stars_dict[user] = (
users_stars_dict[user] + repo.stars)
if users_stars_dict[user] < int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
elif param.endswith("[lte]"):
users_ids_list = []
users_stars_dict = {}
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
users_stars_dict[user] = 0
for repo in user.repositories:
users_stars_dict[user] = (
users_stars_dict[user] + repo.stars)
if users_stars_dict[user] <= int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
else:
users_ids_list = []
users_stars_dict = {}
subquery = (query.join(
GithubUserInfo.repositories).distinct(
GithubUserInfo.id).all())
for user in subquery:
users_stars_dict[user] = 0
for repo in user.repositories:
users_stars_dict[user] = (
users_stars_dict[user] + repo.stars)
if users_stars_dict[user] == int(value):
users_ids_list.append(user.id)
query = query.filter(
GithubUserInfo.id.in_(users_ids_list))
continue
return query.all()
return query
def gather_muts_and_sites(self,
mutations: BaseQuery,
sites: BaseQuery,
show_progress=True,
occurrences_in: List[MutationSource] = None,
intersection=None) -> MotifsData:
"""If occurrences_in is provided, the count of mutations will
represent number of occurrences of mutations in provided
sources, instead of number of distinct substitutions.
"""
if intersection:
accepted_sites = sites.join(Mutation.affected_sites).filter(
and_(*[Mutation.in_sources(source)
for source in intersection])).all()
else:
accepted_sites = sites.all()
mutations_affecting_sites = mutations.filter(
Mutation.affected_sites.any(Site.types.contains(self.site_type)))
muts_around_sites_with_motif = defaultdict(dict)
muts_breaking_sites_motif = defaultdict(dict)
sites_with_broken_motif = defaultdict(set)
sites_with_motif = select_sites_with_motifs(accepted_sites,
self.site_specific_motifs)
if occurrences_in:
def mutation_count(mut: Mutation):
return sum([
mut.sources_map[source.name].get_value()
if source.name in mut.sources_map else 0
for source in occurrences_in
])
else:
def mutation_count(mut):
return 1
is_affected = self.breaking_modes[self.mode]
if show_progress:
ptm_muts = mutations_affecting_sites.count()
mutations_affecting_sites = tqdm(mutations_affecting_sites,
total=ptm_muts)
for mutation in mutations_affecting_sites:
sites = mutation.affected_sites
for site in sites:
if site not in accepted_sites:
continue
for motif_name, motif in self.site_specific_motifs.items():
if site in sites_with_motif[motif_name]:
count = mutation_count(mutation)
muts_around_sites_with_motif[motif_name][
mutation] = count
mutated_sequence = mutate_sequence(site,
mutation,
offset=7)
if is_affected(mutated_sequence, motif):
sites_with_broken_motif[motif_name].add(site)
muts_breaking_sites_motif[motif_name][
mutation] = count
return MotifsData(
sites_with_motif=sites_with_motif,
sites_with_broken_motif=sites_with_broken_motif,
muts_around_sites_with_motif=muts_around_sites_with_motif,
muts_breaking_sites_motif=muts_breaking_sites_motif)
