def execute_server_listener(self):
print('Waiting for connections')
while True:
connection, address = self.server.accept()
server_log('Connection %s established to %s' % (connection, address))
# todo: open thread for new server loop (possibly new object)
start_new_thread(Server, (connection, address, self.password_hash))
def validate_relative_path(self, relative_path):
"""
Validates a path returning an error message and the simplified path
:param relative_path:
:return: error (in case there is an error), simplified path in case there is no error
"""
if not relative_path:
current_path = ''
return '', current_path
else:
if relative_path.startswith('/'):
complete_path = self.root_folder_abs_directory + relative_path
elif relative_path.startswith('~') and self.abs_root_folder:
complete_path = path.expanduser('~') + relative_path[1:]
else:
complete_path = path.join(
self.root_folder_abs_directory,
path.join(self.current_path, relative_path))
complete_path = path.normpath(path.realpath(complete_path))
server_log("Trying to access: %s" % complete_path)
if not complete_path.startswith(self.root_folder_abs_directory):
return 'Cannot go outside root folder', ''
else:
if not path.exists(complete_path):
return 'Directory not found', ''
elif not path.isdir(complete_path):
return ERROR_NOT_A_DIRECTORY, ''
else:
current_path = path.relpath(complete_path,
self.root_folder_abs_directory)
if current_path == '.':
current_path = ''
return '', current_path
def text_to_vector(text):
"""
Parses input plain text and returns a numpy vector based on the
pre-loaded `word_model_wv`.
Args:
* text: plain text string
"""
tokens = list(
map(
lambda x: x.lemma_,
filter(
lambda tok: tok.lemma_ in word_model_wv and tok.lemma_ not in
stop_words,
nlp(text),
),
))
num_tokens = len(tokens)
# Generate error message if no valid tokens are found
if num_tokens == 0:
message = "Valid tokens not found in user input"
server_log(f"{message}\n")
abort(400, message=message)
# Calculate the mean of all token vectors and convert it to a vector
# whose shape is accepted by KNN search function.
vec = np.zeros(word_model_wv.vector_size)
for tok in tokens:
vec += word_model_wv[tok]
vec = (vec / num_tokens)[np.newaxis, :]
return vec
def bind(self, host, port):
try:
self.server_socket = socket.socket(socket.AF_INET,
socket.SOCK_STREAM)
utils.server_log("Binding socket to port: " + str(port))
self.server_socket.bind((host, port))
except socket.error as e:
utils.server_log("Socket error: " + str(e))
sys.exit()
def cd_command(self, args_list):
if not args_list:
self.connection.sendall('%sNo path sent' % INVALID)
else:
path = args_list[0]
error = self.file_manager.resolve_path(path)
if not error:
self.pwd_command([])
else:
server_log('Access denied')
self.connection.sendall(('%s%s' % (INVALID, error)).encode())
def test_paper_knn(data_row):
"""Confirm the closest neighbor of an input data row."""
pmc, np_vec = get_np_vec(data_row)
closest = get_paper_knn(np_vec)[0]
server_log(
f"Closest neighbor: {closest['pmcid']}; distance: {closest['distance']}"
)
assert closest['pmcid'] == pmc
assert closest['distance'] < 1e-2 # ensure that distance is small enough
server_log(f"{pmc} confirmed\n")
def get_doi_neighbors(user_doi):
"""
Find the closest papers and journals given an input paper's DOI.
Arguments:
* user_doi: biorxiv DOI
"""
server_log(f"Received user DOI ({user_doi})")
content, paper_metadata, xml_found = get_doi_content(user_doi)
file_type = 'XML' if xml_found else 'PDF'
server_log(f"Downloaded {file_type} content of {user_doi}")
query_vec = parse_content(content, is_xml=xml_found)
server_log(f"Start searching {user_doi}")
paper_knn = get_paper_knn(query_vec)
journal_knn = get_journal_knn(query_vec)
coordinates = get_coordinates(query_vec)
server_log(f"Finished searching {user_doi}\n")
return {
"paper_neighbors": paper_knn,
"journal_neighbors": journal_knn,
"coordinates": coordinates,
"paper_info": paper_metadata,
"xml_found": xml_found
}
def server_loop(self):
print('Waiting for user authentication')
self.state = USER_AUTH
while True:
server_log('Awaiting message')
received_byte_list = self.get_byte_list()
server_log('Decoding received message')
self.data_received = self.decode(received_byte_list)
if not self.data_received:
server_log('Connection lost')
return
server_log('Executing message')
self.function_switcher[self.state]()
server_log('Concluded')
def get_neighbors(user_doi):
"""
Find the closest papers and journals given an input paper's DOI.
Arguments:
- user_doi: biorxiv DOI
"""
server_log(f"Received user DOI ({user_doi})")
content, paper_metadata = get_doi_content(user_doi)
server_log(f"Downloaded PDF content of {user_doi}")
query_vec = parse_content(content)
server_log(f"Start searching {user_doi}")
paper_knn = get_paper_knn(query_vec)
journal_knn = get_journal_knn(query_vec)
coordinates = get_coordinates(query_vec)
server_log(f"Finished searching {user_doi}\n")
return {
"paper_neighbors": paper_knn,
"journal_neighbors": journal_knn,
"coordinates": coordinates,
"paper_info": paper_metadata
}
def listen(self, max_connections):
try:
self.server_socket.listen(max_connections)
while True:
client, addr = self.server_socket.accept()
self.connection_counter += 1
utils.server_log("Connection has been established [" +
addr[0] + ":" + str(addr[1]) + "]")
client.settimeout(60)
threading.Thread(target=self.handle_client,
name="Connection Thread " +
str(self.connection_counter),
args=(client, addr)).start()
except socket.error as e:
utils.server_log("Socket error: " + str(e))
sys.exit()
def ping_biorxiv_or_medrxiv(doi, server="biorxiv"):
"""
This function pings biorxiv or medrxiv to see if doi exists
within their repository
Args:
doi - a doi that grabs the most current version of a preprint
"""
api_url = f"https://api.biorxiv.org/details/{server}/{doi}"
try:
response = requests.get(api_url)
except Exception as e:
message = f"Cannot connect to {api_url}"
server_log(f"{message}: {e}\n")
abort(404, message=message)
if response.status_code != 200:
message = f"Invalid response from {api_url}"
server_log(f"{message}\n")
abort(response.status_code, message=message)
try:
content = response.json()
except Exception as e:
message = f"Cannot convert response from {api_url} to json format"
server_log(f"{message}: {e}\n")
abort(404, message=message)
if len(content["collection"]) < 1:
return None
return content
def write_file(self):
server_log('writing file: %s' % self.item_name)
error = self.file_manager.write_file(self.simplified_abs_path, self.item_name, self.data_received)
if error:
server_log('File write failed: ' + error)
else:
server_log('File write successfully')
self.state = READING
def get_doi_content(user_doi):
"""
This function is designed to render the paper-journal
network for the user given a biorxiv doi
Args:
user_doi - a biorxiv doi that grabs the most current version of a preprint
"""
# Try pinging biorxiv server first
content = ping_biorxiv_or_medrxiv(user_doi, server="biorxiv")
pdf_url = f"http://biorxiv.org/content"
# If no match found try medrxiv
if content is None:
content = ping_biorxiv_or_medrxiv(user_doi, server="medrxiv")
pdf_url = f"http://medrxiv.org/content"
# If no match at all then raise the red flag
if content is None:
message = f"Cannot find document {user_doi} in either biorxiv or medrxiv."
server_log(f"{message}\n")
abort(404, message=message)
latest_paper = content['collection'][-1]
paper_metadata = {
"title": latest_paper['title'],
"authors": latest_paper['authors'],
"doi": latest_paper['doi'],
"accepted_date": latest_paper['date'],
"publisher": "Cold Spring Harbor Laboratory"
}
# Grab latest version of PDF file
pdf_url = f"{pdf_url}/{user_doi}v{latest_paper['version']}.full.pdf"
try:
response = requests.get(pdf_url)
except Exception as e:
message = f"Cannot connect to {pdf_url}"
server_log(f"{message}: {e}\n")
abort(404, message=message)
if response.status_code != 200:
message = f"Invalid response from {pdf_url}"
server_log(f"{message}\n")
abort(response.status_code, message=message)
return response.content, paper_metadata
def get_text_neighbors(user_text):
"""
Find the closest papers and journals given an input plain text.
Arguments:
* user_text: user's input plain text
"""
server_log("Received user's plain text")
query_vec = text_to_vector(user_text)
server_log(f"Start plain text searching")
paper_knn = get_paper_knn(query_vec)
journal_knn = get_journal_knn(query_vec)
coordinates = get_coordinates(query_vec)
server_log(f"Finished plain text searching\n")
return {
"paper_neighbors": paper_knn,
"journal_neighbors": journal_knn,
"coordinates": coordinates,
}
def main():
utils.server_log("Snowflake started", constants.time_datestring)
rpc = RPCServers(6994, 6993)
rpc.start()
def get_doi_content(user_doi):
"""
This function is designed to render the paper-journal
network for the user given a biorxiv doi
Args:
user_doi - a biorxiv doi that grabs the most current version of a preprint
"""
# Try pinging biorxiv server first
content = ping_biorxiv_or_medrxiv(user_doi, server="biorxiv")
doc_url = f"http://biorxiv.org/content"
# If no match found try medrxiv
if content is None:
content = ping_biorxiv_or_medrxiv(user_doi, server="medrxiv")
doc_url = f"http://medrxiv.org/content"
# If no match at all then raise the red flag
if content is None:
message = f"Cannot find document {user_doi} in either biorxiv or medrxiv."
server_log(f"{message}\n")
abort(404, message=message)
latest_paper = content["collection"][-1]
paper_metadata = {
"title": latest_paper["title"],
"authors": latest_paper["authors"],
"doi": latest_paper["doi"],
"accepted_date": latest_paper["date"],
"publisher": "Cold Spring Harbor Laboratory",
}
# Grab latest version of the XML file if available
accepted_date = latest_paper["date"].replace("-", "/")
file_url = f"{doc_url}/early/{accepted_date}/{str(Path(user_doi).stem)}.source.xml"
xml_found = False
try:
response = requests.get(file_url)
if response.status_code == 200:
xml_found = True
except Exception as e:
message = f"Cannot connect to {file_url}"
server_log(f"{message}: {e}\n")
# If xml not found then use PDF version
if not xml_found:
# Grab latest version of PDF file
file_url = f"{doc_url}/{user_doi}v{latest_paper['version']}.full.pdf"
try:
response = requests.get(file_url)
except Exception as e:
message = f"Cannot connect to {file_url}"
server_log(f"{message}: {e}\n")
abort(404, message=message)
if response.status_code != 200:
message = f"Invalid response from {file_url}"
server_log(f"{message}\n")
abort(response.status_code, message=message)
return response.content, paper_metadata, xml_found
for pool in poolscol.find():
pools.append(pool)
pools_json = dumps(pools)
client.send(
protocol.response(
len(pools_json.encode("utf-8"))).encode(
"utf-8"))
client.send(pools_json.encode("utf-8"))
client.close()
except:
client.close()
if __name__ == "__main__":
if "CGRU_LOCATION" in os.environ:
utils.server_log("CGRU_LOCATION=" + os.environ['CGRU_LOCATION'])
else:
utils.server_log("CGRU_LOCATION is not set!")
sys.exit()
# Loads Mongo DB config
mongodb_config = utils.get_mongodb_config()
# Loads pool server config
Config.check()
Config.load()
# Pool server setup
poolServer = PoolServer()
poolServer.mongodb_host = mongodb_config["host"]
poolServer.mongodb_port = str(mongodb_config["port"])
