def encode_section3(message: dict, context: dict):
"""Encodes section 3."""
buf = context['buf']
section3 = message['section3']
start = buf.tell()
buf.seek(start + 3)
buf.write(b'\x00') # Set to 0 per standard
number_of_subsets = shift_uint(section3['number_of_subsets'], 16, 0, 16)
buf.write(number_of_subsets)
flags_byte = 0
if section3['observed_flag']:
flags_byte |= 0x80
if section3['compressed_flag']:
flags_byte |= 0x40
buf.write(bytes([flags_byte]))
for descriptor in section3['descriptors']:
f, x, y = parse_ref(descriptor)
fx = bytearray(1)
fx = encode_uint(fx, f, 0, 2)
fx = encode_uint(fx, x, 2, 6)
buf.write(fx)
buf.write(bytes([y]))
end = buf.tell()
section_len = end - start
buf.seek(start)
section_len_b = shift_uint(section_len, 24, 0, 24)
buf.write(section_len_b)
buf.seek(end)
def get_sequence_description(fxy_str: str) -> pd.DataFrame:
"""Returns a sequence description used for encoding/decoding."""
summary = get_summary(fxy_str).copy(deep=True)
# Get rid of the old index because it references a combination of other tables
summary.reset_index(drop=True, inplace=True)
summary.rename(columns={
'BUFR_Scale': 'scale',
'BUFR_ReferenceValue': 'offset',
'FXY': 'fxy',
'Parent': 'parent',
'BUFR_DataWidth_Bits': 'bit_len',
},
inplace=True)
summary['type'] = np.nan
summary['text'] = summary['Title']
for i, row in summary.iterrows():
f, x, y = parse_ref(row['fxy'])
typename = ''
title = f'{row["Title"]} ({row["BUFR_Unit"]})'
if f == 2:
typename = 'operator'
elif row['BUFR_Unit'] == 'CCITT IA5':
typename = 'string'
elif row['BUFR_Unit'] == 'Replication':
typename = 'replication'
else:
typename = 'numeric'
summary.iloc[i, summary.columns.get_loc('type')] = typename
summary.iloc[i, summary.columns.get_loc('text')] = title
return summary
def table_d_lookup(f, x, y, parent=None):
"""Returns a data frame for a Table D."""
df = get_table_d(f, x, y)
sub_references = []
fxy_str = f'{f}{x:02d}{y:03d}'
if parent is None:
title = df.iloc[0]['Title_en']
subtitle = ''
sub_references.append((fxy_str, fxy_str, title, subtitle))
parent = fxy_str
for index, row in df.iterrows():
ref = row['FXY2']
title = row['ElementName_en']
subtitle = row['ElementDescription_en']
ref_f, ref_x, ref_y = parse_ref(ref)
if ref_f == 0:
sub_references.append((parent, ref, title, subtitle))
if ref_f == 2 and ref_x == 8:
sub_references.append((parent, ref, title, subtitle))
if ref_f == 1:
sub_references.append((parent, ref, title, subtitle))
if ref_f == 3:
sub_references.append((parent, ref, title, subtitle))
sub_refs = table_d_lookup(ref_f, ref_x, ref_y, ref)
sub_references.extend(sub_refs)
return sub_references
def encode_section4(message: dict, context: dict):
"""Encodes section 4."""
buf = context['buf']
write_buf = io.BytesIO()
start = buf.tell()
buf.seek(start + 3)
buf.write(b'\x00')
bit_offset = 0
sequence = message['section4'][:]
override_bitlength = None
for seq in sequence:
# Deal with operators
if seq['type'] == 'operator':
f, x, y = parse_ref(seq['fxy'])
if (f, x) == (2, 8):
if y > 0:
override_bitlength = y * 8
else:
override_bitlength = None
continue
if seq['bit_len'] < 1:
# Skip 0-length sections, they're for information purposes only
continue
if seq['type'] == 'numeric':
bitlen = seq['bit_len']
if override_bitlength:
bitlen = override_bitlength
value = float(seq['value'])
if seq['scale']:
value = value * math.pow(10, seq['scale'])
if seq['offset']:
value = value - seq['offset']
# The value should be ROUNDED to the nearest integer
value = int(np.round(value))
write_uint(write_buf, value, bit_offset, bitlen)
bit_offset += seq['bit_len']
elif seq['type'] == 'string':
bitlen = seq['bit_len']
if override_bitlength:
bitlen = override_bitlength
write_ascii(write_buf, seq['value'], bit_offset, bitlen)
bit_offset += seq['bit_len']
write_buf.seek(0)
buf.write(write_buf.read())
# Write section length
end = buf.tell()
section_len = end - start
section_len_b = shift_uint(section_len, 24, 0, 24)
buf.seek(start)
buf.write(section_len_b)
buf.seek(end)
def get_summary(fxy_str: str) -> pd.DataFrame:
"""Returns a summary table of the contents of the FXXYYY sequence."""
f, x, y = parse_ref(fxy_str)
references = table_d_lookup(f, x, y)
df = combine_references(references)
columns = [
'Parent',
'FXY',
'Title',
'Subtitle',
'BUFR_DataWidth_Bits',
'BUFR_Unit',
'BUFR_Scale',
'BUFR_ReferenceValue',
]
return df[columns]
def get_code_table_figure(fxy_str: str, code_figure: int) -> dict:
"""Returns the code table row for the given FXXYYY string."""
f, x, y = parse_ref(fxy_str)
filename = f'BUFRCREX_CodeFlag_en_{x:02d}.csv'
utf8_reader = codecs.getreader('utf-8')
with pkg_resources.resource_stream('bufrtools.tables',
f'data/{filename}') as f:
reader = csv.DictReader(utf8_reader(f))
for row in reader:
if fxy_str != row['FXY']:
continue
if '-' in row['CodeFigure']:
start, end = (int(i) for i in row['CodeFigure'].split('-'))
if start <= code_figure and code_figure <= end:
return row
else:
try:
val = int(row['CodeFigure'])
if val == code_figure:
return row
except ValueError:
pass
def get_code_table(fxy_str: str) -> pd.DataFrame:
"""Returns the code table for the given FXXYYY string."""
f, x, y = parse_ref(fxy_str)
filename = f'BUFRCREX_CodeFlag_en_{x:02d}.csv'
utf8_reader = codecs.getreader('utf-8')
with pkg_resources.resource_stream('bufrtools.tables',
f'data/{filename}') as f:
reader = csv.DictReader(utf8_reader(f))
rows = []
for row in reader:
if fxy_str != row['FXY']:
continue
if '-' in row['CodeFigure']:
start, end = (int(i) for i in row['CodeFigure'].split('-'))
for i in range(start, end + 1):
enumerated_row = copy.copy(row)
enumerated_row['CodeFigure'] = i
rows.append(enumerated_row)
else:
rows.append(row)
df = pd.DataFrame(rows)
df = df.astype({'CodeFigure': np.uint16})
return df
def combine_references(references) -> pd.DataFrame:
"""Returns a data frame for a generic reference, that will be flattened."""
frames = []
for parent, reference, title, subtitle in references:
f, x, y = parse_ref(reference)
if f == 2 and x == 8:
frames.append(
pd.DataFrame([{
'Parent': parent,
'FXY': f'{f}{x:02d}{y:03d}',
'ElementName_en':
f'Operator Change CCITT IA5 width to {8 * y}',
'BUFR_DataWidth_Bits': 0,
'BUFR_Unit': 'Operator',
'Title': title,
'Subtitle': subtitle,
}]))
elif f == 1:
if y == 0:
frames.append(
pd.DataFrame([{
'Parent': parent,
'FXY': f'{f}{x:02d}{y:03d}',
'ClassName_en': 'Replication',
'ElementName_en': f'Delayed Replication: {x}',
'BUFR_DataWidth_Bits': 0,
'BUFR_Unit': 'Replication',
'Title': title,
'Subtitle': subtitle,
}]))
else:
frames.append(
pd.DataFrame([{
'Parent': parent,
'FXY': f'{f}{x:02d}{y:03d}',
'ClassName_en': 'Replication',
'ElementName_en': f'Replication {x}',
'BUFR_DataWidth_Bits': 0,
'BUFR_Unit': 'Replication',
'Title': title,
'Subtitle': subtitle,
}]))
elif f == 3:
df = get_table_d(f, x, y)
name = df.iloc[0]['Title_en']
frames.append(
pd.DataFrame([{
'Parent': parent,
'FXY': f'{f}{x:02d}{y:03d}',
'ClassName_en': 'sequence',
'ElementName_en': f'{name}',
'BUFR_DataWidth_Bits': 0,
'BUFR_Unit': 'Sequence',
'Title': title,
'Subtitle': subtitle,
}]))
else:
df = get_table_b(f, x, y)
df['Parent'] = parent
df['Title'] = title
df['Subtitle'] = subtitle
frames.append(df)
return pd.concat(frames)