def annif2jskos(cls, annif_suggestion: List[dict],
annif_project_id: str) -> _ConceptScheme:
""" Transform an Annif suggestion into a JSKOS concept scheme.
:param annif_suggestion: the output of calling AnnifClient.suggest
:param annif_project_id: Annif API project identifier
"""
# get project details:
annif = AnnifClient()
project = annif.get_project(annif_project_id)
language = project.get("language")
name = project.get("name")
# setup concept scheme based on project details:
scheme = _ConceptScheme(pref_label=_LanguageMap({language: name}))
# make concept from result and add to concept scheme:
for result in annif_suggestion:
label = result.get("label")
uri = result.get("uri")
concept = _Concept(uri=uri,
pref_label=_LanguageMap({language: label}))
scheme.concepts.append(concept)
return scheme
def _get_keywords_from_annif(payload):
annif = AnnifClient()
try:
results = annif.analyze(project_id='yso-en', text=payload, limit=5)
except:
_logger.exception('Could not connect to annif')
raise Exception('Could not connect to annif')
keywords = [{
'label': row['label'],
'score': row['score']
} for row in results]
for row in results:
_logger.info(row)
_logger.info(keywords)
return keywords
def _set_input(self, text: str, **kwargs) -> _ConceptScheme:
""" Use words suggested by Annif on the basis of text to set JSKOS Concept Scheme.
:param text: input text
:param **kwargs: required or optional Annif parameters
"""
annif = AnnifClient()
annif_suggestion = annif.suggest(project_id=kwargs.get("project_id"),
text=text,
limit=kwargs.get("limit"),
threshold=kwargs.get("threshold"))
scheme = _Utility.annif2jskos(annif_suggestion, kwargs.get("project_id"))
return scheme
from tqdm import tqdm
if not os.path.exists('skr.xml'):
print('Downloading SKR data')
call(
"curl 'https://apurahat.skr.fi/myonnot/xml_apurahat?vuosi=2018&mkr=KR&ala=Tiede' -H 'User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:62.0) Gecko/20100101 Firefox/62.0' -H 'Accept: */*' -H 'Accept-Language: en-US,en;q=0.5' --compressed -H 'Referer: https://apurahat.skr.fi/myonnot' -H 'Connection: keep-alive' -o skr.xml",
shell=True)
with open('./skr.xml', 'r') as fd:
data = xmltodict.parse(fd.read())
Apuraha = namedtuple(
'Apuraha',
['saaja', 'otsikko', 'puhdas_otsikko', 'ala', 'rahasto', 'summa', 'annif'])
client = AnnifClient()
memory = joblib.Memory('./cache', verbose=0)
def clean_otsikko(otsikko):
i = otsikko.find('käsittelevään')
if i != -1:
otsikko = otsikko[:i]
i = otsikko.find('käsittelevän')
if i != -1:
otsikko = otsikko[:i]
return otsikko.strip()
from finna_client import FinnaClient
finna = FinnaClient()
results = finna.search("bicycle", limit=5)
for rec in results['records']:
print(rec['title'])
from annif_client import AnnifClient
annif = AnnifClient()
results = annif.analyze(project_id='yso-en',
text="The quick brown fox",
limit=5)
for result in results:
print(result)
def client():
return AnnifClient()
def test_create_client_api_base():
client = AnnifClient('http://localhost:5000/v1/')
assert isinstance(client, AnnifClient)
assert client.api_base == 'http://localhost:5000/v1/'
def test_create_client_default():
client = AnnifClient()
assert isinstance(client, AnnifClient)
assert client.api_base == API_BASE