def test__create_claim_json_wikibase_item(self):
expect = {
"mainsnak": {
"snaktype": "value",
"property": "P31",
"datatype": "wikibase-item",
"datavalue": {
"value": {
"entity-type": "item",
"numeric-id": 123
},
"type": "wikibase-entityid"
}
},
"type": "statement",
"rank": "normal"
}
compare(
expect,
ClaimFactory.create_claim_json(
SnakParameter(property_str="P31",
target_type="wikibase-item",
target="Q123")))
compare(
expect,
ClaimFactory.create_claim_json(
SnakParameter(property_str="P31",
target_type="wikibase-item",
target="123")))
def threshold_of_originality(self) -> JsonClaimDict:
qualifier_det = SnakParameter(self._DETERMINATION_METHOD, "wikibase-item",
self._THRESHOLD_OF_ORIGINALITY)
claim_parameter = SnakParameter(self._COPYRIGHT_STATUS, "wikibase-item", self._PUBLIC_DOMAIN)
claim = self.create_claim_json(claim_parameter,
qualifiers=[qualifier_det],
references=[[self._IMPORTED_FROM_WIKISOURCE]])
return claim
def published_95_years_ago(self) -> JsonClaimDict:
qualifier_jur = SnakParameter(self._APPLIES_TO_JURISDICTION, "wikibase-item",
self._UNITED_STATES_OF_AMERICA)
qualifier_det = SnakParameter(self._DETERMINATION_METHOD, "wikibase-item",
self._PUBLISHED_MORE_THAN_THAN_95_YEARS_AGO)
claim_parameter = SnakParameter(self._COPYRIGHT_STATUS, "wikibase-item", self._PUBLIC_DOMAIN)
claim = self.create_claim_json(claim_parameter,
qualifiers=[qualifier_jur, qualifier_det],
references=[[self._IMPORTED_FROM_WIKISOURCE]])
return claim
def xx_years_after_authors_death(self, years) -> JsonClaimDict:
qualifier_jur = SnakParameter(self._APPLIES_TO_JURISDICTION, "wikibase-item",
getattr(self, f"_COUNTRIES_WITH_{years}_YEARS_PMA_OR_SHORTER"))
qualifier_det = SnakParameter(self._DETERMINATION_METHOD, "wikibase-item",
getattr(self, f"_{years}_YEARS_AFTER_AUTHORS_DEATH"))
claim_parameter = SnakParameter(self._COPYRIGHT_STATUS, "wikibase-item", self._PUBLIC_DOMAIN)
claim = self.create_claim_json(claim_parameter,
qualifiers=[qualifier_jur, qualifier_det],
references=[[self._IMPORTED_FROM_WIKISOURCE]])
return claim
def test__create_claim_json_exception(self):
with self.assertRaises(BotException):
ClaimFactory.create_claim_json(
SnakParameter(property_str="P31",
target_type="tada",
target="123"))
with self.assertRaises(ValueError):
ClaimFactory.create_claim_json(
SnakParameter(property_str="P31",
target_type="time",
target="tada"))
def test__create_claim_json_with_qualifier(self):
expect = {
"mainsnak": {
"snaktype": "value",
"property": "P31",
"datatype": "string",
"datavalue": {
"value": "texttexttext",
"type": "string"
}
},
"type": "statement",
"rank": "normal",
"qualifiers": {
"P1234": [{
"snaktype": "value",
"property": "P1234",
"datatype": "string",
"datavalue": {
"value": "text",
"type": "string"
}
}],
"P5678": [{
"snaktype": "value",
"property": "P5678",
"datatype": "wikibase-item",
"datavalue": {
"value": {
"entity-type": "item",
"numeric-id": 123456
},
"type": "wikibase-entityid"
}
}]
},
"qualifiers-order": ["P1234", "P5678"]
}
main_parameter = SnakParameter(property_str="P31",
target_type="string",
target="texttexttext")
quali_snak_1 = SnakParameter(property_str="P1234",
target_type="string",
target="text")
quali_snak_2 = SnakParameter(property_str="P5678",
target_type="wikibase-item",
target="Q123456")
compare(
expect,
ClaimFactory.create_claim_json(
main_parameter, qualifiers=[quali_snak_1, quali_snak_2]))
def _get_claim_json(self) -> List[JsonClaimDict]:
return [
self.create_claim_json(
SnakParameter(property_str=self.get_property_string(),
target_type="wikibase-item",
target=id)) for id in self._get_author_list()
]
def test__create_claim_json_time_just_year(self):
expect = {
"mainsnak": {
"snaktype": "value",
"property": "P31",
"datatype": "time",
"datavalue": {
"value": {
"time": "+00000001234-01-01T00:00:00Z",
"precision": 9,
"after": 0,
"before": 0,
"timezone": 0,
"calendarmodel":
"http://www.wikidata.org/entity/Q1985727"
},
"type": "time"
}
},
"type": "statement",
"rank": "normal"
}
compare(
expect,
ClaimFactory.create_claim_json(
SnakParameter(property_str="P31",
target_type="time",
target="1234")))
def _get_claim_json(self) -> List[JsonClaimDict]:
wp_article = str(self.re_page.first_article["WIKIPEDIA"].value)
# if no wp_article is present, there is nothing to connect
if not wp_article:
return []
# handle the case of an implicit de:
if ":" not in wp_article:
wp_article = f"de:{wp_article}"
language, wp_page_str = wp_article.split(":")
wp_site: pywikibot.Site = pywikibot.Site(code=language,
fam="wikipedia")
wp_page: pywikibot.Page = pywikibot.Page(wp_site, wp_page_str)
# try to fetch the data item of the page, if not there is nothing to connect
try:
wp_data_item: pywikibot.ItemPage = wp_page.data_item()
except (pywikibot.exceptions.NoPageError,
pywikibot.exceptions.InvalidTitleError):
return []
# finally create the claim
snak = SnakParameter(property_str=self.get_property_string(),
target_type="wikibase-item",
target=wp_data_item.id)
return [
self.create_claim_json(
snak, references=[[self._IMPORTED_FROM_WIKISOURCE]])
]
def _get_claim_json(self) -> List[JsonClaimDict]:
neighbour_item = self._get_item_of_neighbour_lemma()
if neighbour_item:
snak_parameter = SnakParameter(
property_str=self.get_property_string(),
target_type="wikibase-item",
target=neighbour_item.id)
return [self.create_claim_json(snak_parameter)]
return []
def _get_claim_json(self) -> List[JsonClaimDict]:
start = str(self.re_page.first_article["SPALTE_START"].value)
end: str = ""
if self.re_page.first_article["SPALTE_END"].value not in ("", "OFF"):
end = str(self.re_page.first_article["SPALTE_END"].value)
columns = start
if end and start != end:
columns = f"{start}–{end}"
snak_parameter = SnakParameter(property_str=self.get_property_string(),
target_type="string",
target=columns)
return [self.create_claim_json(snak_parameter)]
def _get_claim_json(self) -> List[JsonClaimDict]:
article_type = get_article_type(self.re_page)
if article_type == "index":
target = self.INDEX
elif article_type == "prologue":
target = self.PROLOGUE
elif article_type == "crossref":
target = self.CROSS_REFERENCE_ITEM
else:
target = self.ENCYCLOPEDIC_ARTICLE_ITEM
snak_parameter = SnakParameter(property_str=self.get_property_string(),
target_type="wikibase-item",
target=target)
return [self.create_claim_json(snak_parameter)]
class ClaimFactory:
_authors = Authors()
_volumes = Volumes()
_IMPORTED_FROM_WIKISOURCE = SnakParameter(property_str="P143",
target_type="wikibase-item",
target="Q15522295")
def __init__(self, re_page: RePage, logger: WikiLogger):
self.wikisource: pywikibot.Site = pywikibot.Site(code="de", fam="wikisource", user="******")
self.wikipedia: pywikibot.Site = pywikibot.Site(code="de", fam="wikipedia", user="******")
self.wikidata: pywikibot.site.DataSite = self.wikisource.data_repository()
self.re_page = re_page
self.logger = logger
self._current_year = datetime.now().year
def _get_claim_json(self) -> List[JsonClaimDict]:
raise NotImplementedError
def execute_pre_action(self):
pass
def get_claims_to_update(self, data_item: pywikibot.ItemPage) -> ChangedClaimsDict:
"""
Every claim that is updated can possible add new claims, but can also remove existing claims at the item.
Which claims is removed or added depends of the specific implementation of the property factory. The standart
implementation will update all claims as expected by the factory (this include removing possible existing
claims).
:param: data_item: item where the specific property is going to be altered
:returns: A dictionary with claims to add and to remove is returned
"""
claim_list = [pywikibot.Claim.fromJSON(self.wikidata, claim_json)
for claim_json in self._get_claim_json()]
return self.get_diff_claims_for_replacement(claim_list, data_item)
@classmethod
def get_property_string(cls) -> str:
"""
Returns the property string for this class
"""
if regex_hit := re.search(r"^P\d{1,6}", cls.__name__):
return regex_hit.group(0)
return ""
def test__create_claim_json_string(self):
expect = {
"mainsnak": {
"snaktype": "value",
"property": "P31",
"datatype": "string",
"datavalue": {
"value": "texttexttext",
"type": "string"
}
},
"type": "statement",
"rank": "normal"
}
compare(
expect,
ClaimFactory.create_claim_json(
SnakParameter(property_str="P31",
target_type="string",
target="texttexttext")))
def _get_claim_json(self) -> List[JsonClaimDict]:
snak_parameter = SnakParameter(property_str=self.get_property_string(),
target_type="monolingualtext",
target=self.re_page.lemma_without_prefix)
return [self.create_claim_json(snak_parameter)]
def _get_claim_json(self) -> List[JsonClaimDict]:
snak = SnakParameter(property_str=self.get_property_string(),
target_type="time",
target=self._volume_of_first_article.year)
return [self.create_claim_json(snak)]
def test__create_claim_json_with_reference(self):
expect = {
"mainsnak": {
"snaktype": "value",
"property": "P1234",
"datatype": "string",
"datavalue": {
"value": "value",
"type": "string"
}
},
"type":
"statement",
"rank":
"normal",
"references": [{
"snaks": {
"P123": [{
"snaktype": "value",
"property": "P123",
"datatype": "string",
"datavalue": {
"value": "ref1",
"type": "string"
}
}],
"P234": [{
"snaktype": "value",
"property": "P234",
"datatype": "string",
"datavalue": {
"value": "ref2",
"type": "string"
}
}]
},
"snaks-order": ["P123", "P234"]
}, {
"snaks": {
"P345": [{
"snaktype": "value",
"property": "P345",
"datatype": "string",
"datavalue": {
"value": "ref3",
"type": "string"
}
}],
"P456": [{
"snaktype": "value",
"property": "P456",
"datatype": "string",
"datavalue": {
"value": "ref4",
"type": "string"
}
}]
},
"snaks-order": ["P345", "P456"]
}]
}
main_parameter = SnakParameter(property_str="P1234",
target_type="string",
target="value")
ref_snak_1 = SnakParameter(property_str="P123",
target_type="string",
target="ref1")
ref_snak_2 = SnakParameter(property_str="P234",
target_type="string",
target="ref2")
ref_snak_3 = SnakParameter(property_str="P345",
target_type="string",
target="ref3")
ref_snak_4 = SnakParameter(property_str="P456",
target_type="string",
target="ref4")
compare(
expect,
ClaimFactory.create_claim_json(
snak_parameter=main_parameter,
references=[[ref_snak_1, ref_snak_2], [ref_snak_3,
ref_snak_4]]))
def _get_claim_json(self) -> List[JsonClaimDict]:
snak_parameter = SnakParameter(
property_str=self.get_property_string(),
target_type="wikibase-item",
target=self._volume_of_first_article.data_item)
return [self.create_claim_json(snak_parameter)]
def _get_claim_json(self) -> List[JsonClaimDict]:
snak_parameter = SnakParameter(property_str=self.get_property_string(),
target_type="wikibase-item",
target=self.GERMAN)
return [self.create_claim_json(snak_parameter)]