Preperation for merging multiple conllus, also added ATE files

This commit is contained in:
Kikimanox
2022-10-18 10:07:01 +02:00
parent be84cdae1b
commit 5833dfafc8
25 changed files with 95006 additions and 21 deletions
@@ -1,6 +1,14 @@
import connexion
import json
from pathlib import Path
from swagger_server.models.izlusci_body import IzlusciBody # noqa: E501
from swagger_server.classla import cl_utils
import requests
# ATEapi_endpoint = "http://localhost:5000/predict"
ATEapi_endpoint = "http://ate-api:5000/predict"
def get_candidates(body): # noqa: E501
@@ -16,6 +24,45 @@ def get_candidates(body): # noqa: E501
if connexion.request.is_json:
body = IzlusciBody.from_dict(connexion.request.get_json()) # noqa: E501
# Todo: What is "conllus" in body input anyway??
# Todo: Kaj je s prepovedanimi besedami?
# file_ids = [10000, 10001, 10002, 10003]
# big_conllu = cl_utils.multipla_conllus_to_one_from_file_ids(file_ids)
# txt = Path('../ATEapi/temp_1.conllu').read_text('utf-8') # LOCAL ONLY
# # Todo: does this close the file after the request?
# files = [
# ('file', ('temp_1.conllu', open('../ATEapi/temp_1.conllu', 'rb'),
# 'application/octet-stream'))
# ]
# res = requests.post(ATEapi_endpoint, files=files)
# # res.status_code
# # res.text
#
# return res.text, res.status_code
#
# example_data = ""
# with open("C:\\Users\\Kiki\\Desktop\\Untitled-2.json", "r", encoding='utf-8') as f:
# example_data = json.loads(f.read())
#
# d = 0
#
# ret = {'terminoloski_kandidati': [
# {
# 'POSoznake': tk['msd'],
# 'kandidat': tk['terms'], # more to bit lemma al terms?
# 'kanonicnaoblika': tk['canonical'],
# 'ranking': tk['ranking'],
# 'podporneutezi': [
# 6.0274563, # ????????
# 6.0274563 # ??????
# ],
# 'pogostostpojavljanja': [101, 71] # ???????
# }
# for tk in example_data
# ]}
data = {
"terminoloski_kandidati": [
{
@@ -97,7 +97,7 @@ def try_do_jobs_classla():
[ex.submit(execute_classla_job, job) for job in unfinished_jobs]
except Exception as e:
print(f"Exception in {__name__}: {e}")
print(f"Exception in try_do_jobs_classla")
finally:
time.sleep(3)
@@ -113,7 +113,7 @@ def try_do_jobs_doc2text():
with cf.ThreadPoolExecutor(max_workers=DOC2TEXT_CONCURANCE_LIMIT) as ex:
[ex.submit(execute_doc2text_job, job) for job in unfinished_jobs]
except Exception as e:
print(f"Exception in {__name__}: {e}")
print(f"Exception in try_do_jobs_doc2text")
finally:
time.sleep(3)