def test_newcol(tmpdir, db):
ds = StructureDataset.in_dir(str(tmpdir / 'd'))
# We rename the ID column of the ValueTable. Note that the propertyUrl
# remains the same:
ds['ValueTable', 'ID'].name = 'idx'
ds['ValueTable'].tableSchema.columns.extend([
Column(name='col1', datatype='anyURI'),
Column(name='col2', datatype='integer'),
Column(name='col3'),
])
ds.write(ValueTable=[{
'idx': '1',
'Language_ID': 'l',
'Parameter_ID': 'p',
'Value': 'v',
'Source': ['meier2015'],
'col2': 5,
'col1': anyURI().to_python('http://example.org')
}])
db.create()
with pytest.raises(IntegrityError): # A missing source is referenced!
db.load(ds)
ds.add_sources("@misc{meier2015,\ntitle={title}\n}")
db.load(ds)
assert db.fetchone("""\
select
s.title
from
SourceTable as s, ValueSource as vs, ValueTable as v
where
s.ID = vs.Source_ID and vs.Value_ID = v.id and v.id = 1""")[0] == 'title'
assert db.fetchone(
"select col1 from valuetable")[0] == 'http://example.org'
assert db.fetchone("select col2 from valuetable")[0] == 5
def test_db(tmpdir, dataset, mocker, capsys):
db = Database(str(tmpdir.join('lexibank.sqlite')))
db.load(dataset)
db.create(exists_ok=True)
with pytest.raises(ValueError):
db.create()
db.create(force=True)
db.load(dataset)
db.load_glottolog_data(dataset.glottolog)
db.load_concepticon_data(mocker.Mock(conceptsets={}))
for sql in db.sql:
db.fetchall(sql)
with db.connection() as conn:
db.fetchall('select * from dataset', conn=conn, verbose=True)
out, _ = capsys.readouterr()
assert 'select' in out
db.create(force=True)
db.load(dataset)
cldf_ds = dataset.cldf_reader()
cols = cldf_ds['FormTable'].tableSchema.columns
cols.append(Column(name='custom'))
cldf_ds.write_metadata()
db.load(dataset)
cols.pop()
cols.append(Column(name='custom', datatype='integer'))
cldf_ds.write_metadata()
with pytest.raises(ValueError):
db.load(dataset)
cols.pop()
cldf_ds.write_metadata()
db.load(dataset)
def make_column(spec):
if isinstance(spec, string_types):
if spec in TERMS.by_uri:
return TERMS.by_uri[spec].to_column()
return Column(name=spec, datatype='string')
if isinstance(spec, dict):
return Column.fromvalue(spec)
if isinstance(spec, Column):
return spec
raise TypeError(spec)
def make_column(spec):
if isinstance(spec, string_types):
if spec in TERMS.by_uri:
return TERMS.by_uri[spec].to_column()
return Column(name=spec, datatype='string')
if isinstance(spec, dict):
return Column.fromvalue(spec)
if isinstance(spec, Column):
return spec
raise TypeError(spec)
def to_column(self):
col = Column(
name=self.csvw_prop('name') or self.element.find(qname(RDFS, 'label')).text,
propertyUrl=self.element.attrib[qname(RDF, 'about')],
datatype=self.csvw_prop('datatype') or 'string')
for k in ['separator', 'null', 'valueUrl']:
v = self.csvw_prop(k)
if v:
setattr(col, k, v)
return col
def __str__(self):
# We overwrite the base class' method to fix the order of columns.
tg = TableGroup.fromvalue(self.MD)
for col in sorted(
self.column_labels, key=lambda t: (t == IPA_COLUMN, t.lower()), reverse=True):
if col != self.GRAPHEME_COL:
tg.tables[0].tableSchema.columns.append(
Column.fromvalue({"name": col, "null": self.NULL}))
return tg.tables[0].write(self.iteritems(), fname=None).decode('utf8').strip()
def __init__(self, dataset):
self._count = defaultdict(int)
self._cognate_count = defaultdict(int)
self.dataset = dataset
md = self.dataset.cldf_dir / MD_NAME
if not md.exists():
md = self.dataset.cldf_dir / ALT_MD_NAME
if not md.exists():
md = self.dataset.cldf_dir / MD_NAME
copy(Path(__file__).parent / MD_NAME, md)
self.wl = Wordlist.from_metadata(md)
default_cldf = Wordlist.from_metadata(
Path(__file__).parent / 'cldf-metadata.json')
self.objects = {}
self._obj_index = {}
for cls in [
self.dataset.lexeme_class,
self.dataset.language_class,
self.dataset.concept_class,
self.dataset.cognate_class,
]:
self.objects[cls.__cldf_table__()] = []
self._obj_index[cls.__cldf_table__()] = set()
cols = set(
col.header
for col in self.wl[cls.__cldf_table__()].tableSchema.columns)
properties = set(
col.propertyUrl.uri
for col in self.wl[cls.__cldf_table__()].tableSchema.columns
if col.propertyUrl)
for field in cls.fieldnames():
try:
col = default_cldf[cls.__cldf_table__(), field]
#
# We added Latitude and Longitude to the default metadata later, and want to
# make sure, existing datasets are upgraded silently.
#
if field in ['Latitude', 'Longitude'] \
and cls.__cldf_table__() == 'LanguageTable':
properties.add(col.propertyUrl.uri)
self.wl[cls.__cldf_table__(),
field].propertyUrl = col.propertyUrl
self.wl[cls.__cldf_table__(),
field].datatype = col.datatype
except KeyError:
col = Column(name=field, datatype="string")
if (col.propertyUrl and col.propertyUrl.uri not in properties) or \
((not col.propertyUrl) and (field not in cols)):
self.wl[cls.__cldf_table__()].tableSchema.columns.append(
col)
def test_update(tmpdir, db):
ds = Dictionary.in_dir(str(tmpdir / 'd1'))
ds.write(EntryTable=[], SenseTable=[])
ds2 = Dictionary.in_dir(str(tmpdir / 'd2'))
ds2.write(EntryTable=[], SenseTable=[])
db.create()
db.load(ds)
db.load(ds2)
ds.tables[0].tableSchema.columns.append(
Column(name='newcol', datatype='integer'))
db.load(ds)
ds.tables[0].tableSchema.columns[-1].datatype.base = 'string'
with pytest.raises(ValueError):
db.load(ds)
def __enter__(self):
super().__enter__()
default_cldf = Wordlist.from_metadata(
pathlib.Path(__file__).parent / MD_NAME)
self._obj_index = {}
for cls in [
self.dataset.lexeme_class,
self.dataset.language_class,
self.dataset.concept_class,
self.dataset.cognate_class,
]:
self.objects[cls.__cldf_table__()] = []
self._obj_index[cls.__cldf_table__()] = set()
cols = set(
col.header
for col in self.cldf[cls.__cldf_table__()].tableSchema.columns)
properties = set(
col.propertyUrl.uri
for col in self.cldf[cls.__cldf_table__()].tableSchema.columns
if col.propertyUrl)
for field in cls.fieldnames():
try:
col = default_cldf[cls.__cldf_table__(), field]
#
# We added Latitude and Longitude to the default metadata later, and want to
# make sure, existing datasets are upgraded silently.
#
if field in ['Latitude', 'Longitude'] \
and cls.__cldf_table__() == 'LanguageTable': # pragma: no cover
properties.add(col.propertyUrl.uri)
self.cldf[cls.__cldf_table__(),
field].propertyUrl = col.propertyUrl
self.cldf[cls.__cldf_table__(),
field].datatype = col.datatype
except KeyError:
col = Column(name=field, datatype="string")
if (col.propertyUrl and col.propertyUrl.uri not in properties) or \
((not col.propertyUrl) and (field not in cols)):
self.cldf[cls.__cldf_table__()].tableSchema.columns.append(
col)
return self
def create(self, req, filename=None, verbose=True, outfile=None):
cldf_cfg = req.registry.getUtility(ICldfConfig)
with TemporaryDirectory() as tmpd:
cls = getattr(dataset, cldf_cfg.module)
ds = cls.in_dir(tmpd)
ds.properties['dc:bibliographicCitation '] = text_citation(req, req.dataset)
ds.properties['dc:publisher'] = '%s, %s' % (
req.dataset.publisher_name, req.dataset.publisher_place)
ds.properties['dc:license'] = req.dataset.license
ds.properties['dc:issued'] = req.dataset.published.isoformat()
ds.properties['dc:title'] = req.dataset.name
ds.properties['dc:creator'] = req.dataset.formatted_editors()
ds.properties['dc:identifier'] = req.resource_url(req.dataset)
ds.properties['dcat:accessURL'] = req.route_url('download')
if DBSession.query(Sentence).count():
ds.add_component('ExampleTable')
if DBSession.query(DomainElement).count():
ds.add_component('CodeTable', {'name': 'Number', 'datatype': 'integer'})
ds.add_component('ParameterTable')
ds.add_component('LanguageTable')
ds.add_table('contributions.csv', 'ID', 'Name', 'Description', 'Contributors')
ds.add_columns(ds.primary_table, Column.fromvalue(
{
'name': 'Contribution_ID',
'datatype': 'string',
'valueUrl': url_template(req, 'contribution', 'contribution').uri,
}))
ds.add_foreign_key(
ds.primary_table, 'Contribution_ID', 'contributions.csv', 'ID')
ds['LanguageTable'].aboutUrl = url_template(req, 'language', 'ID')
ds['ParameterTable'].aboutUrl = url_template(req, 'parameter', 'ID')
ds[ds.primary_table].aboutUrl = url_template(req, 'value', 'ID')
cldf_cfg.custom_schema(req, ds)
for src in cldf_cfg.query(Source):
ds.sources.add(cldf_cfg.convert(Source, src, req))
fname = outfile or self.abspath(req)
transaction.abort()
tabledata = defaultdict(list)
for table, model in [
('ParameterTable', Parameter),
('CodeTable', DomainElement),
('LanguageTable', Language),
('ExampleTable', Sentence),
('contributions.csv', Contribution),
(ds.primary_table, Value),
]:
if verbose:
print('exporting {0} ...'.format(model))
transaction.begin()
for item in cldf_cfg.query(model):
tabledata[table].append(cldf_cfg.convert(model, item, req))
transaction.abort()
if verbose:
print('... done')
transaction.begin()
ds.write(**cldf_cfg.custom_tabledata(req, tabledata))
ds.validate()
shutil.make_archive(
fname.parent.joinpath(fname.stem).as_posix(), 'zip', tmpd.as_posix())
def test_make_column():
assert make_column(term_uri('source')).separator == ';'
assert make_column(Column('name')).datatype is None
with pytest.raises(TypeError):
make_column(5)
def create(self, req, filename=None, verbose=True, outfile=None):
cldf_cfg = req.registry.getUtility(ICldfConfig)
with TemporaryDirectory() as tmpd:
cls = getattr(dataset, cldf_cfg.module)
ds = cls.in_dir(tmpd)
ds.properties['dc:bibliographicCitation'] = text_citation(
req, req.dataset)
ds.properties['dc:publisher'] = '%s, %s' % (
req.dataset.publisher_name, req.dataset.publisher_place)
ds.properties['dc:license'] = req.dataset.license
ds.properties['dc:issued'] = req.dataset.published.isoformat()
ds.properties['dc:title'] = req.dataset.name
ds.properties['dc:creator'] = req.dataset.formatted_editors()
ds.properties['dc:identifier'] = req.resource_url(req.dataset)
ds.properties['dcat:accessURL'] = req.route_url('download')
if DBSession.query(Sentence).count():
ds.add_component('ExampleTable')
if DBSession.query(DomainElement).count():
ds.add_component('CodeTable', {
'name': 'Number',
'datatype': 'integer'
})
ds.add_component('ParameterTable')
ds.add_component('LanguageTable')
ds.add_table('contributions.csv', 'ID', 'Name', 'Description',
'Contributors')
ds.add_columns(
ds.primary_table,
Column.fromvalue({
'name':
'Contribution_ID',
'datatype':
'string',
'valueUrl':
url_template(req, 'contribution', 'contribution').uri,
}))
ds.add_foreign_key(ds.primary_table, 'Contribution_ID',
'contributions.csv', 'ID')
ds['LanguageTable'].aboutUrl = url_template(req, 'language', 'ID')
ds['ParameterTable'].aboutUrl = url_template(
req, 'parameter', 'ID')
ds[ds.primary_table].aboutUrl = url_template(req, 'value', 'ID')
cldf_cfg.custom_schema(req, ds)
for src in cldf_cfg.query(Source):
ds.sources.add(cldf_cfg.convert(Source, src, req))
fname = outfile or self.abspath(req)
transaction.abort()
tabledata = defaultdict(list)
for table, model in [
('ParameterTable', Parameter),
('CodeTable', DomainElement),
('LanguageTable', Language),
('ExampleTable', Sentence),
('contributions.csv', Contribution),
(ds.primary_table, Value),
]:
if verbose:
print('exporting {0} ...'.format(model))
transaction.begin()
for item in cldf_cfg.query(model):
tabledata[table].append(cldf_cfg.convert(model, item, req))
transaction.abort()
if verbose:
print('... done')
transaction.begin()
ds.write(**cldf_cfg.custom_tabledata(req, tabledata))
ds.validate()
shutil.make_archive(str(fname.parent / fname.stem), 'zip',
str(tmpd))
def add_metadata(fname: Path, logger: cli.logging.Logger = cli.logger):
if fname.name != "forms.csv":
cli.Exit.CLI_ARGUMENT_ERROR(
"A metadata-free Wordlist must be in a file called 'forms.csv'.")
default_wordlist = TableGroup.from_file(
pycldf.util.pkg_path("modules", "Wordlist-metadata.json"))
default_wordlist._fname = fname.with_name("Wordlist-metadata.json")
ds = pycldf.Wordlist(default_wordlist)
# `from_data` checks that the reqired columns of the FormTable are present,
# but it does not consolidate the columns further.
colnames = next(iterrows(fname))
understood_colnames = {
c.name
for c in ds[ds.primary_table].tableSchema.columns if c.name in colnames
}
more_columns = {
c.propertyUrl.uri: c
for c in ds[ds.primary_table].tableSchema.columns
if c.name not in understood_colnames
}
logger.info(
"CLDF freely understood the columns %s in your forms.csv.",
sorted(understood_colnames),
)
# Consider the columns that were not understood.
columns_without_metadata = set(colnames) - understood_colnames
for column_name in columns_without_metadata:
column: Column
# Maybe they are known CLDF properties?
if column_name in pycldf.terms.TERMS:
column = pycldf.TERMS[column_name].to_column()
# Maybe they are CLDF default column names?
elif column_name in DEFAULT_NAME_COLUMNS:
column = DEFAULT_NAME_COLUMNS[column_name]
# Maybe they are columns that Lexedata knows to handle?
elif column_name in LEXEDATA_COLUMNS:
column = LEXEDATA_COLUMNS[column_name]
# Maybe they are columns inherited from LingPy?
elif column_name.upper() in LINGPY_COLUMNS:
column = LINGPY_COLUMNS[column_name.upper()]
# Maybe they are some name we have seen before?
elif column_name in OTHER_KNOWN_COLUMNS:
column = OTHER_KNOWN_COLUMNS[column_name]
else:
# TODO: Maybe they look like they have a specific type?
...
# Otherwise, they are probably just text to be kept.
column = Column(
datatype=Datatype(base="string"),
default="",
null=[""],
name=column_name,
)
column.name = column_name
ds[ds.primary_table].tableSchema.columns.append(column)
summary = column.propertyUrl or column.datatype
logger.info(f"Column {column_name} seems to be a {summary} column.")
if column.propertyUrl:
to_be_replaced = more_columns.pop(column.propertyUrl.uri, None)
if to_be_replaced is not None:
ds[ds.primary_table].tableSchema.columns.remove(to_be_replaced)
for column in more_columns.values():
logger.info(
f"Also added column {column.name}, as expected for a FormTable.")
ds[ds.primary_table].tableSchema.columns.sort(
key=lambda k: colnames.index(k.name) if k.name in colnames else 1e10)
# TODO: Once lexedata is properly published, we can give a better URL.
ds.properties["dc:contributor"] = [
"https://github.com/Anaphory/lexedata/blob/master/src/lexedata/edit/add_metadata.py"
]
return ds
from csvw.dsv import iterrows
from csvw.metadata import Column, Datatype, TableGroup
from lexedata import cli
DEFAULT_NAME_COLUMNS = {
column.name: column
for column in (term.to_column() for term in pycldf.TERMS.values())
}
LEXEDATA_COLUMNS = {
"Status":
Column(
datatype=Datatype(base="string"),
default="",
lang="eng",
null=[""],
name="Value",
aboutUrl="...",
),
"Orthographic":
Column(
datatype=Datatype(base="string"),
default="",
null=[""],
name="Orthographic",
aboutUrl="...",
),
"Phonemic":
Column(
datatype=Datatype(base="string"),
default="",
