def validate_copy(self, data, columns, pfx=""):
"""Validate tab-separated fields"""
ok, _unicode = skytools.safe_utf8_decode(data)
if ok:
return data
# log error
vals = data.split('\t')
for i, v in enumerate(vals):
ok, tmp = skytools.safe_utf8_decode(v)
if not ok:
self.show_error(columns[i], v, pfx, tmp)
# return safe data
return _unicode.encode('utf8')
def validate_copy(self, data, columns, pfx=""):
"""Validate tab-separated fields"""
ok, _unicode = skytools.safe_utf8_decode(data)
if ok:
return data
# log error
vals = data.split('\t')
for i, v in enumerate(vals):
ok, tmp = skytools.safe_utf8_decode(v)
if not ok:
self.show_error(columns[i], v, pfx, tmp)
# return safe data
return _unicode.encode('utf8')
def validate_string(self, value, pfx=""):
"""validate string"""
ok, u = skytools.safe_utf8_decode(value)
if ok:
return value
_pfx = pfx and (pfx+': ') or ""
self.log.info('%sFixed invalid UTF8 in string <%s>', _pfx, value)
return u.encode('utf8')
def validate_string(self, value, pfx=""):
"""validate string"""
ok, u = skytools.safe_utf8_decode(value)
if ok:
return value
_pfx = pfx and (pfx + ': ') or ""
self.log.info('%sFixed invalid UTF8 in string <%s>', _pfx, value)
return u.encode('utf8')
def validate_dict(self, data, pfx=""):
"""validates data in dict"""
for k, v in data.items():
if v:
ok, u = skytools.safe_utf8_decode(v)
if not ok:
self.show_error(k, v, pfx, u)
data[k] = u.encode('utf8')
return data
def validate_dict(self, data, pfx=""):
"""validates data in dict"""
for k, v in data.items():
if v:
ok, u = skytools.safe_utf8_decode(v)
if not ok:
self.show_error(k, v, pfx, u)
data[k] = u.encode('utf8')
return data
