Skip to content

Commit 7ddf874

Browse files
feragferag
committed
Creating EvaluatorBase
1 parent 670375b commit 7ddf874

File tree

3 files changed

+108
-224
lines changed

3 files changed

+108
-224
lines changed

‎api/evaluator.py‎

Lines changed: 72 additions & 194 deletions
Original file line numberDiff line numberDiff line change
@@ -6,6 +6,7 @@
66
import sys
77
import urllib
88
import xml.etree.ElementTree as ET
9+
from abc import ABC, abstractmethod
910
from functools import wraps
1011

1112
import idutils
@@ -63,53 +64,7 @@ def wrapper(plugin, **kwargs):
6364
return wrapper
6465

6566

66-
class ConfigTerms(property):
67-
def __init__(self, term_id):
68-
self.term_id = term_id
69-
70-
def __call__(self, wrapped_func):
71-
@wraps(wrapped_func)
72-
def wrapper(plugin, **kwargs):
73-
metadata = plugin.metadata
74-
has_metadata = True
75-
76-
term_list = ast.literal_eval(plugin.config[plugin.name][self.term_id])
77-
# Get values in config for the given term
78-
if not term_list:
79-
msg = (
80-
"Cannot find any value for term <%s> in configuration"
81-
% self.term_id
82-
)
83-
has_metadata = False
84-
else:
85-
# Get metadata associated with the term ID
86-
term_metadata = pd.DataFrame(
87-
term_list, columns=["element", "qualifier"]
88-
)
89-
term_metadata = ut.check_metadata_terms_with_values(
90-
metadata, term_metadata
91-
)
92-
if term_metadata.empty:
93-
msg = (
94-
"No access information can be found in the metadata for: %s. Please double-check the value/s provided for '%s' configuration parameter"
95-
% (term_list, self.term_id)
96-
)
97-
has_metadata = False
98-
99-
if not has_metadata:
100-
logger.warning(msg)
101-
return (0, [{"message": msg, "points": 0}])
102-
103-
# Update kwargs with collected metadata for the required terms
104-
kwargs.update(
105-
{self.term_id: {"list": term_list, "metadata": term_metadata}}
106-
)
107-
return wrapped_func(plugin, **kwargs)
108-
109-
return wrapper
110-
111-
112-
class Evaluator(object):
67+
class EvaluatorBase(ABC):
11368
"""A class used to define FAIR indicators tests. It contains all the references to all the tests
11469
11570
...
@@ -126,115 +81,79 @@ class Evaluator(object):
12681
lang : Language
12782
"""
12883

129-
def __init__(self, item_id, oai_base=None, lang="en", plugin=None, config=None):
84+
def __init__(self, item_id, oai_base=None, lang="en", config=None, name=None):
13085
self.item_id = item_id
13186
self.oai_base = oai_base
13287
self.metadata = None
133-
self.access_protocols = []
13488
self.cvs = []
135-
self.config = config
136-
# configuration terms
137-
self.terms_access_metadata = pd.DataFrame()
138-
self.terms_license_metadata = pd.DataFrame()
139-
140-
logger.debug("OAI_BASE IN evaluator: %s" % oai_base)
141-
if oai_base is not None and oai_base != "" and self.metadata is None:
142-
metadataFormats = ut.oai_metadataFormats(oai_base)
143-
dc_prefix = ""
144-
for e in metadataFormats:
145-
if metadataFormats[e] == "http://www.openarchives.org/OAI/2.0/oai_dc/":
146-
dc_prefix = e
147-
logger.debug("DC_PREFIX: %s" % dc_prefix)
148-
149-
try:
150-
id_type = idutils.detect_identifier_schemes(self.item_id)[0]
151-
except Exception as e:
152-
id_type = "internal"
153-
154-
logger.debug("Trying to get metadata")
155-
try:
156-
item_metadata = ut.oai_get_metadata(
157-
ut.oai_check_record_url(oai_base, dc_prefix, self.item_id)
158-
).find(".//{http://www.openarchives.org/OAI/2.0/}metadata")
159-
except Exception as e:
160-
logger.error("Problem getting metadata: %s" % e)
161-
item_metadata = ET.fromstring("<metadata></metadata>")
162-
data = []
163-
for tags in item_metadata.findall(".//"):
164-
metadata_schema = tags.tag[0 : tags.tag.rfind("}") + 1]
165-
element = tags.tag[tags.tag.rfind("}") + 1 : len(tags.tag)]
166-
text_value = tags.text
167-
qualifier = None
168-
data.append([metadata_schema, element, text_value, qualifier])
169-
self.metadata = pd.DataFrame(
170-
data, columns=["metadata_schema", "element", "text_value", "qualifier"]
171-
)
172-
173-
if self.metadata is not None:
174-
if len(self.metadata) > 0:
175-
self.access_protocols = ["http", "oai-pmh"]
17689

17790
# Config attributes
178-
self.name = plugin
179-
if self.name == None:
180-
self.name = "oai-pmh"
181-
try:
182-
self.identifier_term = ast.literal_eval(
183-
self.config[self.name]["identifier_term"]
184-
)
185-
self.terms_quali_generic = ast.literal_eval(
186-
self.config[self.name]["terms_quali_generic"]
187-
)
188-
self.terms_quali_disciplinar = ast.literal_eval(
189-
self.config[self.name]["terms_quali_disciplinar"]
190-
)
191-
self.terms_access = ast.literal_eval(self.config[self.name]["terms_access"])
192-
self.terms_cv = ast.literal_eval(self.config[self.name]["terms_cv"])
193-
self.supported_data_formats = ast.literal_eval(
194-
self.config[self.name]["supported_data_formats"]
195-
)
196-
self.terms_qualified_references = ast.literal_eval(
197-
self.config[self.name]["terms_qualified_references"]
198-
)
199-
self.terms_relations = ast.literal_eval(
200-
self.config[self.name]["terms_relations"]
201-
)
202-
self.terms_license = ast.literal_eval(
203-
self.config[self.name]["terms_license"]
204-
)
205-
self.metadata_quality = 100 # Value for metadata quality
206-
self.terms_access_protocols = ast.literal_eval(
207-
self.config[self.name]["terms_access_protocols"]
208-
)
209-
self.metadata_standard = ast.literal_eval(
210-
self.config[self.name]["metadata_standard"]
211-
)
212-
self.fairsharing_username = ast.literal_eval(
213-
self.config["fairsharing"]["username"]
214-
)
91+
self.identifier_term = ast.literal_eval(
92+
self.config[self.name]["identifier_term"]
93+
)
94+
self.terms_quali_generic = ast.literal_eval(
95+
self.config[self.name]["terms_quali_generic"]
96+
)
97+
self.terms_quali_disciplinar = ast.literal_eval(
98+
self.config[self.name]["terms_quali_disciplinar"]
99+
)
100+
self.terms_cv = ast.literal_eval(self.config[self.name]["terms_cv"])
101+
self.supported_data_formats = ast.literal_eval(
102+
self.config[self.name]["supported_data_formats"]
103+
)
104+
self.terms_qualified_references = ast.literal_eval(
105+
self.config[self.name]["terms_qualified_references"]
106+
)
107+
self.terms_relations = ast.literal_eval(
108+
self.config[self.name]["terms_relations"]
109+
)
110+
self.metadata_access_manual = ast.literal_eval(
111+
self.config[self.name]["metadata_access_manual"]
112+
)
113+
self.data_access_manual = ast.literal_eval(
114+
self.config[self.name]["data_access_manual"]
115+
)
116+
self.terms_access_protocols = ast.literal_eval(
117+
self.config[self.name]["terms_access_protocols"]
118+
)
215119

216-
self.fairsharing_password = ast.literal_eval(
217-
self.config["fairsharing"]["password"]
218-
)
219-
self.fairsharing_metadata_path = ast.literal_eval(
220-
self.config["fairsharing"]["metadata_path"]
221-
)
222-
self.fairsharing_formats_path = ast.literal_eval(
223-
self.config["fairsharing"]["formats_path"]
224-
)
225-
self.internet_media_types_path = ast.literal_eval(
226-
self.config["internet media types"]["path"]
227-
)
228-
self.metadata_schemas = ast.literal_eval(
229-
self.config[self.name]["metadata_schemas"]
230-
)
231-
except Exception as e:
232-
logger.error("Problem loading plugin config: %s" % e)
120+
# self.vocabularies = ast.literal_eval(self.config[self.name]["vocabularies"])
233121

234-
# Translations
235-
self.lang = lang
236-
logger.debug("El idioma es: %s" % self.lang)
237-
logger.debug("METAdata: %s" % self.metadata)
122+
self.dict_vocabularies = ast.literal_eval(
123+
self.config[self.name]["dict_vocabularies"]
124+
)
125+
126+
self.vocabularies = list(self.dict_vocabularies.keys())
127+
self.metadata_standard = ast.literal_eval(
128+
self.config[self.name]["metadata_standard"]
129+
)
130+
131+
self.metadata_authentication = ast.literal_eval(
132+
self.config[self.name]["metadata_authentication"]
133+
)
134+
self.metadata_persistence = ast.literal_eval(
135+
self.config[self.name]["metadata_persistence"]
136+
)
137+
self.terms_vocabularies = ast.literal_eval(
138+
self.config[self.name]["terms_vocabularies"]
139+
)
140+
141+
self.fairsharing_username = ast.literal_eval(
142+
self.config["fairsharing"]["username"]
143+
)
144+
145+
self.fairsharing_password = ast.literal_eval(
146+
self.config["fairsharing"]["password"]
147+
)
148+
self.fairsharing_metadata_path = ast.literal_eval(
149+
self.config["fairsharing"]["metadata_path"]
150+
)
151+
self.fairsharing_formats_path = ast.literal_eval(
152+
self.config["fairsharing"]["formats_path"]
153+
)
154+
self.internet_media_types_path = ast.literal_eval(
155+
self.config["internet media types"]["path"]
156+
)
238157
global _
239158
_ = self.translation()
240159

@@ -267,6 +186,11 @@ def eval_persistency(self, id_list, data_or_metadata="(meta)data"):
267186

268187
return (points, msg_list)
269188

189+
@abstractmethod
190+
def get_metadata(self):
191+
"""Method to be implemented by plugins."""
192+
raise NotImplementedError("Derived class mus implement get_metadata method")
193+
270194
def eval_uniqueness(self, id_list, data_or_metadata="(meta)data"):
271195
points = 0
272196
msg_list = []
@@ -2066,49 +1990,3 @@ def check_standard_license(self, license_id_or_url):
20661990
% _url
20671991
)
20681992
return license_name
2069-
2070-
2071-
class ConfigTerms(property):
2072-
def __init__(self, term_id):
2073-
self.term_id = term_id
2074-
2075-
def __call__(self, wrapped_func):
2076-
@wraps(wrapped_func)
2077-
def wrapper(plugin, **kwargs):
2078-
metadata = plugin.metadata
2079-
has_metadata = True
2080-
2081-
term_list = ast.literal_eval(plugin.config[plugin.name][self.term_id])
2082-
# Get values in config for the given term
2083-
if not term_list:
2084-
msg = (
2085-
"Cannot find any value for term <%s> in configuration"
2086-
% self.term_id
2087-
)
2088-
has_metadata = False
2089-
else:
2090-
# Get metadata associated with the term ID
2091-
term_metadata = pd.DataFrame(
2092-
term_list, columns=["element", "qualifier"]
2093-
)
2094-
term_metadata = ut.check_metadata_terms_with_values(
2095-
metadata, term_metadata
2096-
)
2097-
if term_metadata.empty:
2098-
msg = (
2099-
"No access information can be found in the metadata for: %s. Please double-check the value/s provided for '%s' configuration parameter"
2100-
% (term_list, self.term_id)
2101-
)
2102-
has_metadata = False
2103-
2104-
if not has_metadata:
2105-
logger.warning(msg)
2106-
return (0, msg)
2107-
2108-
# Update kwargs with collected metadata for the required terms
2109-
kwargs.update(
2110-
{self.term_id: {"list": term_list, "metadata": term_metadata}}
2111-
)
2112-
return wrapped_func(plugin, **kwargs)
2113-
2114-
return wrapper

‎plugins/gbif/config.ini‎

Lines changed: 25 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
[Generic]
22
doi_url = https://doi.org/
3-
api_config = /FAIR_eva/fair-api.yaml
3+
api_config = fair-api.yaml
44
endpoint= https://api.gbif.org/v1
55
[local]
66
only_local = false
@@ -47,6 +47,30 @@ terms_access = [['access', ''], ['rights', '']]
4747
# Accepted access protocols
4848
terms_access_protocols =['http','https','ftp']
4949

50+
# Manual metadata access
51+
metadata_access_manual = ['TODO']
52+
53+
# Manual data access
54+
data_access_manual = ['TODO']
55+
56+
#Policy of metadata persistence
57+
metadata_persistence = []
58+
59+
#Authentication for EPOS
60+
metadata_authentication = []
61+
62+
#terms that use vocabularies and vocabularies used
63+
dict_vocabularies= {'ROR': 'https://ror.org/', 'PIC': 'https://ec.europa.eu/info/funding-tenders/opportunities/portal/screen/how-to-participate/participant-register', 'imtypes': 'https://www.iana.org/assignments/media-types/media-types.xhtml', 'TRL': 'TRL', 'temporal': 'https://www.iso.org/iso-8601-date-and-time-format.html', 'Rolecode': 'Rolecode', 'spdx': 'https://spdx.org/licenses/', 'ORCID': 'https://orcid.org/'}
64+
65+
terms_vocabularies=[['identifiers','relatedDataProducts'],
66+
['',''],
67+
['availableFormats',''],
68+
['',''],
69+
['temporalCoverage','relatedDataProducts'],#no temporal metatdata
70+
['',''],
71+
['license',''],
72+
['contactPoints','relatedDataProducts']]
73+
5074
# Metadata terms wich includes controlled vocabularies. More controlled vocabularies can be imlpemented in plugins
5175
terms_cv = [['coverage', 'spatial'], ['subject', 'lcsh']]
5276

0 commit comments

Comments
 (0)