def _old2new(self, row):
if not self._allow_reading_old_format:
raise IOError('Please convert to new format. ' +
'Use: python -m ase.db.convert ' + self.filename)
extra = decode(row[25])
return row[:-1] + (encode(
extra['keywords']), encode(
extra['key_value_pairs']), encode(extra['data']), 42)
def _old2new(self, row):
if not self._allow_reading_old_format:
raise IOError('Please convert to new format. ' +
'Use: python -m ase.db.convert ' + self.filename)
extra = decode(row[25])
return row[:-1] + (encode(extra['keywords']),
encode(extra['key_value_pairs']),
encode(extra['data']),
42)
def _old2new(self, row):
if not self._allow_reading_old_format:
raise IOError('Please convert to new format. ' +
'Use: python -m ase.db.convert ' + self.filename)
if len(row) == 26:
extra = decode(row[25])
return row[:-1] + (encode(
extra['keywords']), encode(
extra['key_value_pairs']), encode(extra['data']), 42)
else:
keywords = decode(row[-4])
kvp = decode(row[-3])
kvp.update(dict((keyword, 1) for keyword in keywords))
return row[:-4] + (encode(kvp), ) + row[-2:]
def _old2new(self, row):
if not self._allow_reading_old_format:
raise IOError('Please convert to new format. ' +
'Use: python -m ase.db.convert ' + self.filename)
if len(row) == 26:
extra = decode(row[25])
return row[:-1] + (encode(extra['keywords']),
encode(extra['key_value_pairs']),
encode(extra['data']),
42)
else:
keywords = decode(row[-4])
kvp = decode(row[-3])
kvp.update(dict((keyword, 1) for keyword in keywords))
return row[:-4] + (encode(kvp),) + row[-2:]
def sync(self):
"""Write data dictionary.
Write bool, int, float, complex and str data, shapes and
dtypes for ndarrays and class names for other objects."""
assert self.shape[0] == 0
i = self.fd.tell()
s = encode(self.data).encode()
writeint(self.fd, len(s))
self.fd.write(s)
n = len(self.offsets)
if self.nitems >= n:
offsets = np.zeros(n * N1, np.int64)
offsets[:n] = self.offsets
self.itemoffsets = align(self.fd)
offsets.tofile(self.fd)
writeint(self.fd, self.itemoffsets, 24)
self.offsets = offsets
self.offsets[self.nitems] = i
writeint(self.fd, i, self.itemoffsets + self.nitems * 8)
self.nitems += 1
writeint(self.fd, self.nitems, 16)
self.fd.flush()
self.fd.seek(0, 2) # end of file
self.data = {}
def sync(self):
"""Write data dictionary.
Write bool, int, float, complex and str data, shapes and
dtypes for ndarrays and class names for other objects."""
assert self.shape[0] == 0
i = self.fd.tell()
s = encode(self.data).encode()
writeint(self.fd, len(s))
self.fd.write(s)
n = len(self.offsets)
if self.nitems >= n:
offsets = np.zeros(n * N1, np.int64)
offsets[:n] = self.offsets
self.itemoffsets = align(self.fd)
offsets.tofile(self.fd)
writeint(self.fd, self.itemoffsets, 24)
self.offsets = offsets
self.offsets[self.nitems] = i
writeint(self.fd, i, self.itemoffsets + self.nitems * 8)
self.nitems += 1
writeint(self.fd, self.nitems, 16)
self.fd.flush()
self.fd.seek(0, 2) # end of file
self.data = {}
def _write(self, atoms, keywords, key_value_pairs, data):
Database._write(self, atoms, keywords, key_value_pairs, data)
con = self._connect()
self._initialize(con)
cur = con.cursor()
id = None
if isinstance(atoms, dict):
dct = atoms
unique_id = dct['unique_id']
cur.execute('SELECT id FROM systems WHERE unique_id=?',
(unique_id,))
rows = cur.fetchall()
if rows:
id = rows[0][0]
self._delete(cur, [id], ['keywords', 'text_key_values',
'number_key_values'])
dct['mtime'] = now()
else:
dct = self.collect_data(atoms)
if 'constraints' in dct:
constraints = encode(dct['constraints'])
else:
constraints = None
numbers = dct.get('numbers')
row = (dct['unique_id'],
dct['ctime'],
dct['mtime'],
dct['user'],
blob(numbers),
blob(dct.get('positions')),
blob(dct.get('cell')),
int(np.dot(dct.get('pbc'), [1, 2, 4])),
blob(dct.get('initial_magmoms')),
blob(dct.get('initial_charges')),
blob(dct.get('masses')),
blob(dct.get('tags')),
blob(dct.get('momenta')),
constraints)
if 'calculator' in dct:
row += (dct['calculator'],
encode(dct['calculator_parameters']))
else:
row += (None, None)
magmom = dct.get('magmom')
if magmom is not None:
# magmom can be one or three numbers (non-collinear case)
magmom = np.array(magmom)
row += (dct.get('energy'),
dct.get('free_energy'),
blob(dct.get('forces')),
blob(dct.get('stress')),
blob(dct.get('dipole')),
blob(dct.get('magmoms')),
blob(magmom),
blob(dct.get('charges')),
encode(keywords),
encode(key_value_pairs),
encode(data),
len(numbers))
if id is None:
q = self.default + ', ' + ', '.join('?' * len(row))
cur.execute('INSERT INTO systems VALUES ({0})'.format(q),
row)
else:
q = ', '.join(line.split()[0].lstrip() + '=?'
for line in init_statements[0].splitlines()[2:])
cur.execute('UPDATE systems SET {0} WHERE id=?'.format(q),
row + (id,))
if id is None:
id = self.get_last_id(cur)
if len(numbers) > 0:
count = np.bincount(numbers)
unique_numbers = count.nonzero()[0]
species = [(int(Z), int(count[Z]), id) for Z in unique_numbers]
cur.executemany('INSERT INTO species VALUES (?, ?, ?)',
species)
text_key_values = []
number_key_values = []
for key, value in key_value_pairs.items():
if isinstance(value, (float, int)):
number_key_values.append([key, float(value), id])
else:
assert isinstance(value, (str, unicode))
text_key_values.append([key, value, id])
cur.executemany('INSERT INTO text_key_values VALUES (?, ?, ?)',
text_key_values)
cur.executemany('INSERT INTO number_key_values VALUES (?, ?, ?)',
number_key_values)
cur.executemany('INSERT INTO keywords VALUES (?, ?)',
[(keyword, id) for keyword in keywords])
# Insert keys in keywords table also so that it is easy to query
# for the existance of keys:
cur.executemany('INSERT INTO keywords VALUES (?, ?)',
[(key, id) for key in key_value_pairs])
con.commit()
con.close()
return id