def test_log_or_raise(mocker):
from clldutils.misc import log_or_raise
with pytest.raises(ValueError):
log_or_raise('')
log = mocker.Mock()
log_or_raise('test', log=log)
assert log.warn.called
def check(self, clts=None, log=None, ipa_col=IPA_COLUMN):
"""
Check a profile for consistency, logging problems.
For each grapheme, raise:
- a warning if there are duplicate entries
- an error if there are inconsistencies
- an error if the mapping has invalid BIPA
"""
mapping = collections.defaultdict(list)
if self.fname:
# We read the profile from disk because segments.Profile already skips duplicate
# graphemes, which we want to investigate more closely.
for spec in dsv.reader(self.fname, dicts=True, delimiter='\t'):
mapping[spec[self.GRAPHEME_COL]].append(spec[ipa_col])
for grapheme in mapping:
# check mapping consistency
if len(mapping[grapheme]) >= 2:
if len(set(mapping[grapheme])) == 1:
log_or_raise(
"Duplicate, redundant entry or entries for grapheme [{}]."
.format(grapheme),
log=log,
level='warning')
else:
log_or_raise(
"Inconsist entries for grapheme [{}]: multiple mappings {}."
.format(grapheme, str(mapping[grapheme])),
log=log,
level='error')
# check BIPA consistency
if clts:
for value in mapping[grapheme]:
if value:
# check for unknown sounds
unknown = [
isinstance(clts.bipa[segment],
pyclts.models.UnknownSound)
for segment in ipa2tokens(value)
if segment and segment != 'NULL'
]
if any(unknown):
log_or_raise(
"Mapping [{}] ({}) -> [{}] ({}) includes an unknown sound."
.format(grapheme,
unicode2codepointstr(grapheme), value,
unicode2codepointstr(value)),
log=log,
level='error')
def validate(self, log=None, validators=None):
validators = validators or []
validators.extend(VALIDATORS)
success = True
default_tg = TableGroup.from_file(
pkg_path('modules', '{0}{1}'.format(self.module, MD_SUFFIX)))
for default_table in default_tg.tables:
dtable_uri = default_table.common_props['dc:conformsTo']
try:
table = self[dtable_uri]
except KeyError:
log_or_raise('{0} requires {1}'.format(self.module, dtable_uri), log=log)
success = False
table = None
if table:
default_cols = {
c.propertyUrl.uri for c in default_table.tableSchema.columns
if c.required or c.common_props.get('dc:isRequiredBy')}
cols = {
c.propertyUrl.uri for c in table.tableSchema.columns
if c.propertyUrl}
table_uri = table.common_props['dc:conformsTo']
for col in default_cols - cols:
log_or_raise('{0} requires column {1}'.format(table_uri, col), log=log)
success = False
for table in self.tables:
type_uri = table.common_props.get('dc:conformsTo')
if type_uri:
try:
TERMS.is_cldf_uri(type_uri)
except ValueError:
success = False
log_or_raise('invalid CLDF URI: {0}'.format(type_uri), log=log)
# FIXME: check whether table.common_props['dc:conformsTo'] is in validators!
validators_ = []
for col in table.tableSchema.columns:
if col.propertyUrl:
col_uri = col.propertyUrl.uri
try:
TERMS.is_cldf_uri(col_uri)
except ValueError:
success = False
log_or_raise('invalid CLDF URI: {0}'.format(col_uri), log=log)
for table_, col_, v_ in validators:
if (not table_ or table is self.get(table_)) and col is self.get((table, col_)):
validators_.append((col, v_))
fname = Path(table.url.resolve(table._parent.base))
if fname.exists():
for fname, lineno, row in table.iterdicts(log=log, with_metadata=True):
for col, validate in validators_:
try:
validate(self, table, col, row)
except ValueError as e:
log_or_raise(
'{0}:{1}:{2} {3}'.format(fname.name, lineno, col.name, e),
log=log)
success = False
if not table.check_primary_key(log=log):
success = False
if not self.tablegroup.check_referential_integrity(log=log):
success = False
return success
def validate(self, log=None, validators=None):
validators = validators or []
validators.extend(VALIDATORS)
success = True
default_tg = TableGroup.from_file(
pkg_path('modules', '{0}{1}'.format(self.module, MD_SUFFIX)))
for default_table in default_tg.tables:
dtable_uri = default_table.common_props['dc:conformsTo']
try:
table = self[dtable_uri]
except KeyError:
log_or_raise('{0} requires {1}'.format(self.module, dtable_uri), log=log)
success = False
table = None
if table:
default_cols = {
c.propertyUrl.uri for c in default_table.tableSchema.columns
if c.required or c.common_props.get('dc:isRequiredBy')}
cols = {
c.propertyUrl.uri for c in table.tableSchema.columns
if c.propertyUrl}
table_uri = table.common_props['dc:conformsTo']
for col in default_cols - cols:
log_or_raise('{0} requires column {1}'.format(table_uri, col), log=log)
success = False
for table in self.tables:
type_uri = table.common_props.get('dc:conformsTo')
if type_uri:
try:
TERMS.is_cldf_uri(type_uri)
except ValueError:
success = False
log_or_raise('invalid CLDF URI: {0}'.format(type_uri), log=log)
# FIXME: check whether table.common_props['dc:conformsTo'] is in validators!
validators_ = []
for col in table.tableSchema.columns:
if col.propertyUrl:
col_uri = col.propertyUrl.uri
try:
TERMS.is_cldf_uri(col_uri)
except ValueError:
success = False
log_or_raise('invalid CLDF URI: {0}'.format(col_uri), log=log)
for table_, col_, v_ in validators:
if (not table_ or table is self.get(table_)) and col is self.get((table, col_)):
validators_.append((col, v_))
fname = Path(table.url.resolve(table._parent.base))
if fname.exists():
for fname, lineno, row in table.iterdicts(log=log, with_metadata=True):
for col, validate in validators_:
try:
validate(self, table, col, row)
except ValueError as e:
log_or_raise(
'{0}:{1}:{2} {3}'.format(fname.name, lineno, col.name, e),
log=log)
success = False
if not table.check_primary_key(log=log):
success = False
else:
log_or_raise('{0} does not exist'.format(fname), log=log)
success = False
if not self.tablegroup.check_referential_integrity(log=log):
success = False
return success