def main(args):
with open(args.data_file('2.3', 'obsolete_refs.json')) as fp:
obsolete = json.load(fp)
with transaction.manager:
provider = Provider.get('glottolog20121')
for ref in provider.refs:
if ref.id in obsolete:
Config.add_replacement(ref, None, session=DBSession, model=Source)
DBSession.delete(ref)
else:
assert len(ref.providers) > 1
DBSession.flush()
DBSession.delete(provider)
def __init__(self, req, *args, **kw):
if 'cq' in kw:
self.complexquery = get_params(kw)
elif 'cq' in req.params:
self.complexquery = get_params(req.params)
else:
self.complexquery = None
if 'provider' in kw:
self.provider = kw['provider']
elif 'provider' in req.params:
self.provider = Provider.get(req.params['provider'], default=None)
else:
self.provider = None
super(Refs, self).__init__(req, *args, **kw)
if self.language:
self.language_sources = [s.pk for s in self.language.sources]
def update_providers(args, verbose=False):
filepath = args.data_dir.joinpath('references', 'bibtex', 'BIBFILES.ini')
p = RawConfigParser()
with io.open(filepath, encoding='utf-8-sig') as fp:
p.readfp(fp)
provider_map = get_map(Provider)
for section in p.sections():
sectname = section[:-4] if section.endswith('.bib') else section
id_ = slug(sectname)
attrs = {
'name': p.get(section, 'title'),
'description': p.get(section, 'description'),
'abbr': p.get(section, 'abbr'),
}
if id_ in provider_map:
provider = provider_map[id_]
for a in list(attrs):
before, after = getattr(provider, a), attrs[a]
if before == after:
del attrs[a]
else:
setattr(provider, a, after)
attrs[a] = (before, after)
if attrs:
args.log.info('updating provider %s %s' %
(slug(id_), sorted(attrs)))
if verbose:
for a, (before, after) in attrs.items():
before, after = (' '.join(_.split())
for _ in (before, after))
if before != after:
args.log.info('%s\n%r\n%r' % (a, before, after))
else:
args.log.info('adding provider %s' % slug(id_))
DBSession.add(Provider(id=id_, **attrs))
def update(args):
author = 'ISO 639-3 Registration Authority'
pid = 'iso6393'
dtid = 'overview'
dt = Doctype.get(dtid)
provider = Provider.get(pid, default=None)
if provider is None:
provider = Provider(
id=pid,
abbr=pid,
name=author,
description="Change requests submitted to the ISO 639-3 registration authority.")
iid = max(int(DBSession.execute(
"select max(cast(id as integer)) from source").fetchone()[0]), 500000)
pk = int(DBSession.execute("select max(pk) from source").fetchone()[0])
for crno, affected in args.json['changerequests'].items():
year, serial = crno.split('-')
title = 'Change Request Number %s' % crno
ref = Ref.get(title, key='title', default=None)
if not ref:
iid += 1
pk += 1
ref = Ref(
pk=pk,
id=str(iid),
name='%s %s' % (author, year),
bibtex_type=EntryType.misc,
number=crno,
description=title,
year=year,
year_int=int(year),
title=title,
author=author,
address='Dallas',
publisher='SIL International',
url='http://www.sil.org/iso639-3/cr_files/%s.pdf' % crno,
doctypes_str=dtid,
providers_str=pid,
language_note=', '.join('%(Language Name)s [%(Affected Identifier)s]' % spec for spec in affected),
jsondata=dict(hhtype=dtid, src=pid))
ref.doctypes.append(dt)
ref.providers.append(provider)
for spec in affected:
lang = Languoid.get(spec['Affected Identifier'], key='hid', default=None)
if lang and lang not in ref.languages:
ref.languages.append(lang)
DBSession.add(ref)
transaction.commit()
transaction.begin()
matched = 0
near = 0
max_identifier_pk = DBSession.query(
Identifier.pk).order_by(desc(Identifier.pk)).first()[0]
families = []
for family in DBSession.query(Languoid)\
.filter(Languoid.level == LanguoidLevel.family)\
.filter(Language.active == True)\
.all():
isoleafs = set()
for row in DBSession.query(TreeClosureTable.child_pk, Languoid.hid)\
.filter(family.pk == TreeClosureTable.parent_pk)\
.filter(Languoid.pk == TreeClosureTable.child_pk)\
.filter(Languoid.hid != None)\
.filter(Languoid.level == LanguoidLevel.language)\
.filter(Languoid.status == LanguoidStatus.established)\
.all():
if len(row[1]) == 3:
isoleafs.add(row[1])
families.append((family, isoleafs))
families = sorted(families, key=lambda p: len(p[1]))
for mid, leafs in args.json['macrolanguages'].items():
leafs = set(leafs)
found = False
for family, isoleafs in families:
if leafs == isoleafs:
if mid not in [c.name for c in family.identifiers
if c.type == IdentifierType.iso.value]:
family.codes.append(Identifier(
id=str(max_identifier_pk + 1),
name=mid,
type=IdentifierType.iso.value))
max_identifier_pk += 1
matched += 1
found = True
break
elif leafs.issubset(isoleafs):
print '~~~', family.name, '-->', mid, 'distance:', len(leafs), len(isoleafs)
near += 1
found = True
break
if not found:
print '---', mid, leafs
print 'matched', matched, 'of', len(args.json['macrolanguages']), 'macrolangs'
print near
def update(args):
author = 'ISO 639-3 Registration Authority'
pid = 'iso6393'
dtid = 'overview'
dt = Doctype.get(dtid)
provider = Provider.get(pid, default=None)
if provider is None:
provider = Provider(
id=pid,
abbr=pid,
name=author,
description=
"Change requests submitted to the ISO 639-3 registration authority."
)
iid = max(
int(
DBSession.execute(
"select max(cast(id as integer)) from source").fetchone()[0]),
500000)
pk = int(DBSession.execute("select max(pk) from source").fetchone()[0])
for crno, affected in args.json['changerequests'].items():
year, serial = crno.split('-')
title = 'Change Request Number %s' % crno
ref = Ref.get(title, key='title', default=None)
if not ref:
iid += 1
pk += 1
ref = Ref(pk=pk,
id=str(iid),
name='%s %s' % (author, year),
bibtex_type=EntryType.misc,
number=crno,
description=title,
year=year,
year_int=int(year),
title=title,
author=author,
address='Dallas',
publisher='SIL International',
url='http://www.sil.org/iso639-3/cr_files/%s.pdf' % crno,
language_note=', '.join(
'%(Language Name)s [%(Affected Identifier)s]' % spec
for spec in affected),
jsondata=dict(hhtype=dtid, src=pid))
ref.doctypes.append(dt)
ref.providers.append(provider)
for spec in affected:
lang = Languoid.get(spec['Affected Identifier'],
key='hid',
default=None)
if lang and lang not in ref.languages:
ref.languages.append(lang)
DBSession.add(ref)
transaction.commit()
transaction.begin()
matched = 0
near = 0
max_identifier_pk = DBSession.query(Identifier.pk).order_by(
desc(Identifier.pk)).first()[0]
families = []
for family in DBSession.query(Languoid)\
.filter(Languoid.level == LanguoidLevel.family)\
.filter(Language.active == True)\
.all():
isoleafs = set()
for row in DBSession.query(TreeClosureTable.child_pk, Languoid.hid)\
.filter(family.pk == TreeClosureTable.parent_pk)\
.filter(Languoid.pk == TreeClosureTable.child_pk)\
.filter(Languoid.hid != None)\
.filter(Languoid.level == LanguoidLevel.language)\
.filter(Languoid.status == LanguoidStatus.established)\
.all():
if len(row[1]) == 3:
isoleafs.add(row[1])
families.append((family, isoleafs))
families = sorted(families, key=lambda p: len(p[1]))
for mid, leafs in args.json['macrolanguages'].items():
leafs = set(leafs)
found = False
for family, isoleafs in families:
if leafs == isoleafs:
if mid not in [
c.name for c in family.identifiers
if c.type == IdentifierType.iso.value
]:
family.codes.append(
Identifier(id=str(max_identifier_pk + 1),
name=mid,
type=IdentifierType.iso.value))
max_identifier_pk += 1
matched += 1
found = True
break
elif leafs.issubset(isoleafs):
print '~~~', family.name, '-->', mid, 'distance:', len(
leafs), len(isoleafs)
near += 1
found = True
break
if not found:
print '---', mid, leafs
print 'matched', matched, 'of', len(
args.json['macrolanguages']), 'macrolangs'
print near
