def categoryLink(self, m):
uri = 'http://lagen.nu/concept/%s' % util.ucfirst(
m.group(1)).replace(' ', '_')
if len(m.groups()) == 2:
# lcwl = "Labeled Category WikiLink"
return '<a class="lcwl" href="%s">%s</a>' % (uri, m.group(2))
else:
# cwl = "Category wikilink"
return '<a class="cwl" href="%s">%s</a>' % (uri, m.group(1))
def capitalizedLink(self, m):
if m.group(1).startswith('SFS/'):
uri = 'http://rinfo.lagrummet.se/publ/%s' % m.group(1).lower()
else:
uri = 'http://lagen.nu/concept/%s' % util.ucfirst(
m.group(1)).replace(' ', '_')
if len(m.groups()) == 3:
# lwl = "Labeled WikiLink"
return '<a class="lwl" href="%s">%s%s</a>' % (uri, m.group(2), m.group(3))
else:
return '<a class="wl" href="%s">%s%s</a>' % (uri, m.group(1), m.group(2))
def infer_metadata(self, resource, basefile):
# remove the bogus dcterms:issued thing that we only added to
# aid URI generation. NB: This is removed in the superclass'
# postprocess_doc as well, because for this lagen.nu-derived
# class it needs to be done at this point, but for use of the
# superclass directly, it needs to be done at some point.
for o in resource.objects(DCTERMS.issued):
if not o.datatype:
resource.remove(DCTERMS.issued, o)
sameas_uri = self.sameas_minter.space.coin_uri(resource)
resource.add(OWL.sameAs, URIRef(sameas_uri))
resource.graph.add((URIRef(self.canonical_uri(basefile, True)),
OWL.sameAs, resource.identifier))
# then find each rpubl:konsolideringsunderlag, and create
# owl:sameas for them as well
for subresource in resource.objects(RPUBL.konsolideringsunderlag):
# sometimes there'll be a rpubl:konsolideringsunderlag to
# a resource URI but no actual data about that
# resource. This seems to happen if SFST is updated but
# SFSR is not. In those cases we can't generate a
# owl:sameAs URI since we have no other data about the
# resource.
if subresource.value(RDF.type):
uri = self.sameas_minter.space.coin_uri(subresource)
subresource.add(OWL.sameAs, URIRef(uri))
desc = Describer(resource.graph, resource.identifier)
de = DocumentEntry(self.store.documententry_path(basefile))
if de.orig_updated:
desc.value(RINFOEX.senastHamtad, de.orig_updated)
if de.orig_checked:
desc.value(RINFOEX.senastKontrollerad, de.orig_checked)
rooturi = URIRef(desc.getrel(RPUBL.konsoliderar))
v = self.commondata.value(rooturi, DCTERMS.alternate, any=True)
if v:
desc.value(DCTERMS.alternate, v)
v = self.commondata.value(rooturi, RDFS.label, any=True)
if v:
# don't include labels if they're essentially the same as
# dcterms:title (legalref needs it to be able to parse
# refs to laws that typically don't include SFS numbers,
# so that's why they're in sfs.ttl
basetitle = str(resource.value(DCTERMS.title)).rsplit(" (")[0]
if not v.startswith(basetitle.lower()):
desc.value(RDFS.label, util.ucfirst(v))
def infer_metadata(self, resource, basefile):
# remove the bogus dcterms:issued thing that we only added to
# aid URI generation. NB: This is removed in the superclass'
# postprocess_doc as well, because for this lagen.nu-derived
# class it needs to be done at this point, but for use of the
# superclass directly, it needs to be done at some point.
for o in resource.objects(DCTERMS.issued):
if not o.datatype:
resource.remove(DCTERMS.issued, o)
sameas_uri = self.sameas_minter.space.coin_uri(resource)
resource.add(OWL.sameAs, URIRef(sameas_uri))
resource.graph.add((URIRef(self.canonical_uri(basefile, True)),
OWL.sameAs, resource.identifier))
# then find each rpubl:konsolideringsunderlag, and create
# owl:sameas for them as well
for subresource in resource.objects(RPUBL.konsolideringsunderlag):
# sometimes there'll be a rpubl:konsolideringsunderlag to
# a resource URI but no actual data about that
# resource. This seems to happen if SFST is updated but
# SFSR is not. In those cases we can't generate a
# owl:sameAs URI since we have no other data about the
# resource.
if subresource.value(RDF.type):
uri = self.sameas_minter.space.coin_uri(subresource)
subresource.add(OWL.sameAs, URIRef(uri))
desc = Describer(resource.graph, resource.identifier)
de = DocumentEntry(self.store.documententry_path(basefile))
if de.orig_updated:
desc.value(RINFOEX.senastHamtad, de.orig_updated)
if de.orig_checked:
desc.value(RINFOEX.senastKontrollerad, de.orig_checked)
rooturi = URIRef(desc.getrel(RPUBL.konsoliderar))
v = self.commondata.value(rooturi, DCTERMS.alternate, any=True)
if v:
desc.value(DCTERMS.alternate, v)
v = self.commondata.value(rooturi, RDFS.label, any=True)
if v:
# don't include labels if they're essentially the same as
# dcterms:title (legalref needs it to be able to parse
# refs to laws that typically don't include SFS numbers,
# so that's why they're in sfs.ttl
basetitle = str(resource.value(DCTERMS.title)).rsplit(" (")[0]
if not v.startswith(basetitle.lower()):
desc.value(RDFS.label, util.ucfirst(v))
def prop_sanitize_identifier(identifier):
if not identifier:
return identifier # allow infer_identifier to do it's magic later
if identifier.startswith("prop"):
identifier = util.ucfirst(identifier)
if identifier.startswith("PROP"):
identifier = identifier.replace("PROP", "Prop")
if identifier.startswith("Prop "):
identifier = identifier.replace("Prop ", "Prop. ")
if re.match("Prop\.\d{4}", identifier): # missing space
identifier = identifier.replace("Prop.", "Prop. ")
if "\xa0" in identifier: # Non-breakable space
identifier = identifier.replace("\xa0", " ")
if not identifier.startswith("Prop. "):
identifier = "Prop. " + identifier
# identify and correct the not-uncommon "2009/2010:87" pattern (should be 2009/10:87)
m = re.search(r"(\d{4})/(\d{4}):(\d+)$", identifier)
if m and m.group(2) != "2000" and int(m.group(1)) == int(m.group(2)) - 1:
identifier = identifier.replace(m.group(2), m.group(2)[-2:])
if not re.match(r"^Prop\. (19|20)\d{2}(|/\d{2}|/2000):(|B ?|U ?)[1-9]\d{0,2}$", identifier):
raise ValueError("Irregular identifier %s" % identifier)
return Literal(identifier)
def authorLink(self, m):
uri = 'http://wiki.lagen.nu/index.php/%s' % util.ucfirst(
m.group(1)).replace(' ', '_')
return '<a class="awl" href="%s">%s</a>' % (uri, m.group(2))
def hiddenLink(self, m):
uri = 'http://lagen.nu/concept/%s' % util.ucfirst(
m.group(1)).replace(' ', '_')
return '<a class="hcwl" rel="dct:subject" href="%s"/>' % uri
def sokord_uri(value):
return self.config.url + "concept/%s" % util.ucfirst(value).replace(' ', '_')
def parse_from_soup(self, soup):
# Step 1: Find out basic metadata
rubrik = soup.first("title").string
beslutsdatum = soup.first(
"meta", {'name': 'SG_Beslutsdatum'})['content']
beslutsdatum = datetime.strptime(beslutsdatum, "%Y-%m-%d").date()
diarienummer = soup.first(
"meta", {'name': 'SG_Dokumentbet'})['content']
arendetyp = soup.first("meta", {'name': 'Subject'})['content']
# the keywords for a documents is contained in a metatag
# formatted like:
# <meta name="Keywords" content="hets_mot_folkgrupp\nmeddelarfrihet\åklagare">
#
# Transform this into an array like:
# [u'http://lagen.nu/concept/Hets_mot_folkgrupp',
# u'http://lagen.nu/concept/Meddelarfrihet',
# u'http://lagen.nu/concept/Åklagare']
nyckelord = soup.first("meta", {'name': 'Keywords'})['content']
begrepp = ['http://lagen.nu/concept/%s' % util.ucfirst(
x).strip().replace(" ", "_") for x in nyckelord.split("\n")]
# Step 2: Using the metadata, construct the canonical URI for this document
uri = LegalURI.construct({'type': LegalRef.MYNDIGHETSBESLUT,
'myndighet': 'jk',
'dnr': diarienummer})
# self.log.debug("URI: %s" % uri)
# Step 3: Create a RDF graph of all our metadata (so far)
g = Graph()
g.bind('dct', self.ns['dct'])
g.bind('rinfo', self.ns['rinfo'])
g.bind('rinfoex', self.ns['rinfoex'])
g.bind('xsd', util.ns['xsd'])
g.add((
URIRef(uri), self.ns['dct']['title'], Literal(rubrik, lang="sv")))
g.add((URIRef(uri), self.ns['rinfo']['beslutsdatum'],
Literal(beslutsdatum, lang="sv")))
g.add((URIRef(uri), self.ns['rinfo']['diarienummer'],
Literal(diarienummer, lang="sv")))
g.add((URIRef(uri), self.ns['rinfoex']['arendetyp'],
Literal(arendetyp, lang="sv")))
for s in begrepp:
g.add((URIRef(uri), self.ns['dct']['subject'], URIRef(s)))
g.add((URIRef(uri), self.ns['dct']['identifier'], Literal(
"JK %s" % diarienummer, lang="sv")))
g.add((URIRef(uri), RDF.type, self.rdf_type))
# Step 4: Process the actual text of the document
self.parser = LegalRef(LegalRef.LAGRUM,
LegalRef.KORTLAGRUM,
LegalRef.RATTSFALL,
LegalRef.FORARBETEN)
# newer documents have a semantic structure with h1 and h2
# elements. Older have elements like <p class="Rubrik_1">. Try
# to determine which one we're dealing with?
tag = soup.find('a', {'name': "Start"})
if tag:
# self.log.debug("Using new-style document structure")
elements = tag.parent.findAllNext()
else:
# self.log.debug("Using old-style document structure")
elements = soup.findAll("p")
# self.log.debug("Found %d elements" % len(elements))
from collections import deque
elements = deque(elements)
body = self.make_sektion(elements, "Referat av beslut")
# Step 5: Combine the metadata and the document, and return it
doc = {'meta': g,
'body': body,
'lang': 'sv',
'uri': uri}
return doc
def keyword_uri(self, keyword):
baseuri = "https://lagen.nu/begrepp/"
return baseuri + util.ucfirst(keyword).replace(' ', '_').replace('"', "%22").replace("»", "//")