# -*- coding: utf-8 -*-
""" ASAM MDF version 3 file format module """
from __future__ import division, print_function
import logging
import os
import sys
import time
import xml.etree.ElementTree as ET
from collections import defaultdict
from copy import deepcopy
from functools import reduce
from itertools import product
from math import ceil
from struct import unpack
from tempfile import TemporaryFile
from numpy import (
arange,
array,
array_equal,
column_stack,
concatenate,
dtype,
flip,
float64,
interp,
packbits,
roll,
uint8,
union1d,
unpackbits,
zeros,
)
from numpy.core.records import fromarrays, fromstring
from pandas import DataFrame
from .signal import Signal
from . import v2_v3_constants as v23c
from .conversion_utils import conversion_transfer
from .utils import (
CHANNEL_COUNT,
CONVERT_LOW,
CONVERT_MINIMUM,
ChannelsDB,
MdfException,
SignalSource,
as_non_byte_sized_signed_int,
fix_dtype_fields,
fmt_to_datatype_v3,
get_fmt_v3,
get_min_max,
get_unique_name,
get_text_v3,
validate_memory_argument,
validate_version_argument,
count_channel_groups,
is_file_like,
)
from .v2_v3_blocks import (
Channel,
ChannelConversion,
ChannelDependency,
ChannelExtension,
ChannelGroup,
DataBlock,
DataGroup,
FileIdentificationBlock,
HeaderBlock,
ProgramBlock,
TextBlock,
TriggerBlock,
)
from .version import __version__
PYVERSION = sys.version_info[0]
if PYVERSION == 2:
# pylint: disable=W0622
from .utils import bytes
# pylint: enable=W0622
logger = logging.getLogger('asammdf')
__all__ = ['MDF3', ]
def write_cc(conversion, defined_texts, blocks=None, address=None, stream=None):
if conversion:
if stream:
tell = stream.tell
write = stream.write
for key, item in conversion.referenced_blocks.items():
if isinstance(item, TextBlock):
text = item['text']
if text in defined_texts:
conversion[key] = defined_texts[text]
else:
if stream:
address = tell()
conversion[key] = address
defined_texts[text] = address
item.address = address
if stream:
write(bytes(item))
else:
address += item['block_len']
blocks.append(item)
elif isinstance(item, ChannelConversion):
if stream:
write_cc(item, defined_texts, blocks, stream=stream)
address = tell()
item.address = address
conversion[key] = address
write(bytes(item))
else:
item.address = address
conversion[key] = address
address += item['block_len']
blocks.append(item)
address = write_cc(item, defined_texts, blocks, address)
return address
[docs]class MDF3(object):
"""The *header* attibute is a *HeaderBlock*.
The *groups* attribute is a list of dicts, each one with the following keys
* ``data_group`` - DataGroup object
* ``channel_group`` - ChannelGroup object
* ``channels`` - list of Channel objects (when *memory* is *full* or *low*)
or addresses (when *memory* is *minimum*) with the same order as found in
the mdf file
* ``channel_dependencies`` - list of *ChannelArrayBlock* in case of channel
arrays; list of Channel objects (when *memory* is *full* or *low*) or
addresses (when *memory* is *minimum*) in case of structure channel
composition
* ``data_block`` - DataBlock object when *memory* is *full* else address of
data block
* ``data_location``- integer code for data location (original file,
temporary file or memory)
* ``data_block_addr`` - list of raw samples starting addresses, for *low*
and *minimum* memory options
* ``data_block_type`` - list of codes for data block type
* ``data_block_size`` - list of raw samples block size
* ``sorted`` - sorted indicator flag
* ``record_size`` - dict that maps record ID's to record sizes in bytes
* ``size`` - total size of data block for the current group
* ``trigger`` - *Trigger* object for current group
Parameters
----------
name : string
mdf file name (if provided it must be a real file name) or
file-like object
memory : str
memory optimization option; default `full`
* if *full* the data group binary data block will be memorised in RAM
* if *low* the channel data is read from disk on request, and the
metadata is memorized into RAM
* if *minimum* only minimal data is memorized into RAM
version : string
mdf file version ('2.00', '2.10', '2.14', '3.00', '3.10', '3.20' or
'3.30'); default '3.30'
Attributes
----------
attachments : list
list of file attachments
channels_db : dict
used for fast channel access by name; for each name key the value is a
list of (group index, channel index) tuples
groups : list
list of data group dicts
header : HeaderBlock
mdf file header
identification : FileIdentificationBlock
mdf file start block
masters_db : dict
used for fast master channel access; for each group index key the value
is the master channel index
memory : str
memory optimization option
name : string
mdf file name
version : str
mdf version
"""
_terminate = False
def __init__(self, name=None, memory='full', version='3.30', **kwargs):
memory = validate_memory_argument(memory)
self.groups = []
self.header = None
self.identification = None
self.name = name
self.memory = memory
self.channels_db = ChannelsDB(version=3)
self.masters_db = {}
self.version = version
self._master_channel_cache = {}
self._master_channel_metadata = {}
# used for appending to MDF created with memory=False
self._tempfile = TemporaryFile()
self._tempfile.write(b'\0')
self._file = None
self._read_fragment_size = 0
self._write_fragment_size = 8 * 2 ** 20
self._single_bit_uint_as_bool = False
self._callback = kwargs.get('callback', None)
if name:
if is_file_like(name):
self._file = name
self.name = 'From_FileLike.mf4'
self._from_filelike = True
else:
self._file = open(self.name, 'rb')
self._from_filelike = False
self._read()
else:
self._from_filelike = False
version = validate_version_argument(version, hint=3)
self.identification = FileIdentificationBlock(version=version)
self.version = version
self.header = HeaderBlock(version=self.version)
def _load_group_data(self, group):
""" get group's data block bytes"""
offset = 0
if self.memory == 'full':
yield group['data_block']['data'], offset
else:
channel_group = group['channel_group']
if group['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
# go to the first data block of the current data group
stream = self._file
else:
stream = self._tempfile
# go to the first data block of the current data group
if group['sorted']:
samples_size = channel_group['samples_byte_nr']
if not samples_size:
yield b'', 0
else:
if self._read_fragment_size:
split_size = self._read_fragment_size // samples_size
split_size *= samples_size
else:
channels_nr = len(group['channels'])
if self.memory == 'minimum':
y_axis = CONVERT_MINIMUM
else:
y_axis = CONVERT_LOW
split_size = interp(
channels_nr,
CHANNEL_COUNT,
y_axis,
)
split_size = int(split_size)
split_size = split_size // samples_size
split_size *= samples_size
if split_size == 0:
split_size = samples_size
blocks = zip(
group['data_block_addr'],
group['data_block_size'],
)
if PYVERSION == 2:
blocks = iter(blocks)
cur_size = 0
data = []
while True:
try:
address, size = next(blocks)
current_address = address
except StopIteration:
break
stream.seek(address)
while size >= split_size - cur_size:
stream.seek(current_address)
if data:
data.append(stream.read(split_size - cur_size))
yield b''.join(data), offset
current_address += split_size - cur_size
else:
yield stream.read(split_size), offset
current_address += split_size
offset += split_size
size -= split_size - cur_size
data = []
cur_size = 0
if size:
stream.seek(current_address)
data.append(stream.read(size))
cur_size += size
offset += size
if data:
yield b''.join(data), offset
elif not offset:
yield b'', 0
else:
record_id = group['channel_group']['record_id']
if PYVERSION == 2:
record_id = chr(record_id)
cg_size = group['record_size']
if group['data_group']['record_id_len'] <= 2:
record_id_nr = group['data_group']['record_id_len']
else:
record_id_nr = 0
cg_data = []
blocks = zip(
group['data_block_addr'],
group['data_block_size'],
)
if PYVERSION == 2:
blocks = iter(blocks)
for address, size in blocks:
stream.seek(address)
data = stream.read(size)
i = 0
while i < size:
rec_id = data[i]
# skip record id
i += 1
rec_size = cg_size[rec_id]
if rec_id == record_id:
rec_data = data[i: i + rec_size]
cg_data.append(rec_data)
# consider the second record ID if it exists
if record_id_nr == 2:
i += rec_size + 1
else:
i += rec_size
cg_data = b''.join(cg_data)
size = len(cg_data)
yield cg_data, offset
offset += size
def _prepare_record(self, group):
""" compute record dtype and parents dict for this group
Parameters
----------
group : dict
MDF group dict
Returns
-------
parents, dtypes : dict, numpy.dtype
mapping of channels to records fields, records fiels dtype
"""
memory = self.memory
if group['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
grp = group
record_size = grp['channel_group']['samples_byte_nr'] << 3
next_byte_aligned_position = 0
types = []
current_parent = ""
parent_start_offset = 0
parents = {}
group_channels = set()
if memory != 'minimum':
channels = grp['channels']
else:
channels = [
Channel(address=ch_addr, stream=stream, load_metadata=False)
for ch_addr in grp['channels']
]
# the channels are first sorted ascending (see __lt__ method of Channel
# class): a channel with lower start offset is smaller, when two
# channels havethe same start offset the one with higer bit size is
# considered smaller. The reason is that when the numpy record is built
# and there are overlapping channels, the parent fields mustbe bigger
# (bit size) than the embedded channels. For each channel the parent
# dict will have a (parent name, bit offset) pair: the channel value is
# computed using the values from the parent field, and the bit offset,
# which is the channel's bit offset within the parent bytes.
# This means all parents will have themselves as parent, and bit offset
# of 0. Gaps in the records are also considered. Non standard integers
# size is adjusted to the first higher standard integer size (eq. uint
# of 28bits will be adjusted to 32bits)
sortedchannels = sorted(enumerate(channels), key=lambda i: i[1])
for original_index, new_ch in sortedchannels:
# skip channels with channel dependencies from the numpy record
if new_ch['ch_depend_addr']:
continue
start_offset = new_ch['start_offset']
try:
additional_byte_offset = new_ch['aditional_byte_offset']
start_offset += 8 * additional_byte_offset
except KeyError:
pass
bit_offset = start_offset % 8
data_type = new_ch['data_type']
bit_count = new_ch['bit_count']
name = new_ch.name
# handle multiple occurance of same channel name
name = get_unique_name(group_channels, name)
group_channels.add(name)
if start_offset >= next_byte_aligned_position:
parent_start_offset = (start_offset // 8) * 8
# check if there are byte gaps in the record
gap = (parent_start_offset - next_byte_aligned_position) // 8
if gap:
types.append(('', 'a{}'.format(gap)))
# adjust size to 1, 2, 4 or 8 bytes for nonstandard integers
size = bit_offset + bit_count
if data_type == v23c.DATA_TYPE_STRING:
next_byte_aligned_position = parent_start_offset + size
if next_byte_aligned_position <= record_size:
dtype_pair = (name, get_fmt_v3(data_type, size))
types.append(dtype_pair)
parents[original_index] = name, bit_offset
else:
next_byte_aligned_position = parent_start_offset
elif data_type == v23c.DATA_TYPE_BYTEARRAY:
next_byte_aligned_position = parent_start_offset + size
if next_byte_aligned_position <= record_size:
dtype_pair = (name, get_fmt_v3(data_type, size))
types.append(dtype_pair)
parents[original_index] = name, bit_offset
else:
next_byte_aligned_position = parent_start_offset
else:
if size > 32:
next_byte_aligned_position = parent_start_offset + 64
elif size > 16:
next_byte_aligned_position = parent_start_offset + 32
elif size > 8:
next_byte_aligned_position = parent_start_offset + 16
else:
next_byte_aligned_position = parent_start_offset + 8
if next_byte_aligned_position <= record_size:
dtype_pair = (name, get_fmt_v3(data_type, size))
types.append(dtype_pair)
parents[original_index] = name, bit_offset
else:
next_byte_aligned_position = parent_start_offset
current_parent = name
else:
max_overlapping = next_byte_aligned_position - start_offset
if max_overlapping >= bit_count:
parents[original_index] = (
current_parent,
start_offset - parent_start_offset,
)
if next_byte_aligned_position > record_size:
break
gap = (record_size - next_byte_aligned_position) >> 3
if gap:
dtype_pair = ('', 'a{}'.format(gap))
types.append(dtype_pair)
if PYVERSION == 2:
types = fix_dtype_fields(types, 'latin-1')
parents_ = {}
for key, (name, offset) in parents.items():
if isinstance(name, unicode):
parents_[key] = name.encode('latin-1'), offset
else:
parents_[key] = name, offset
parents = parents_
return parents, dtype(types)
def _get_not_byte_aligned_data(self, data, group, ch_nr):
big_endian_types = (
v23c.DATA_TYPE_UNSIGNED_MOTOROLA,
v23c.DATA_TYPE_FLOAT_MOTOROLA,
v23c.DATA_TYPE_DOUBLE_MOTOROLA,
v23c.DATA_TYPE_SIGNED_MOTOROLA,
)
record_size = group['channel_group']['samples_byte_nr']
if self.memory != 'minimum':
channel = group['channels'][ch_nr]
else:
if group['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream=self._file
else:
stream=self._tempfile,
channel = Channel(
address=group['channels'][ch_nr],
stream=stream,
load_metadata=False,
)
bit_offset = channel['start_offset'] % 8
byte_offset = channel['start_offset'] // 8
bit_count = channel['bit_count']
byte_count = bit_offset + bit_count
if byte_count % 8:
byte_count = (byte_count >> 3) + 1
else:
byte_count >>= 3
types = [
('', 'a{}'.format(byte_offset)),
('vals', '({},)u1'.format(byte_count)),
('', 'a{}'.format(record_size - byte_count - byte_offset)),
]
vals = fromstring(data, dtype=dtype(types))
vals = vals['vals']
if channel['data_type'] not in big_endian_types:
vals = flip(vals, 1)
vals = unpackbits(vals)
vals = roll(vals, bit_offset)
vals = vals.reshape((len(vals) // 8, 8))
vals = packbits(vals)
vals = vals.reshape((len(vals) // byte_count, byte_count))
if bit_count < 64:
mask = 2 ** bit_count - 1
masks = []
while mask:
masks.append(mask & 0xFF)
mask >>= 8
for i in range(byte_count - len(masks)):
masks.append(0)
masks = masks[::-1]
for i, mask in enumerate(masks):
vals[:, i] &= mask
if channel['data_type'] not in big_endian_types:
vals = flip(vals, 1)
if bit_count <= 8:
size = 1
elif bit_count <= 16:
size = 2
elif bit_count <= 32:
size = 4
elif bit_count <= 64:
size = 8
else:
size = bit_count // 8
if size > byte_count:
extra_bytes = size - byte_count
extra = zeros((len(vals), extra_bytes), dtype=uint8)
types = [
('vals', vals.dtype, vals.shape[1:]),
('', extra.dtype, extra.shape[1:]),
]
vals = fromarrays([vals, extra], dtype=dtype(types))
vals = vals.tostring()
fmt = get_fmt_v3(channel['data_type'], bit_count)
if size <= byte_count:
if channel['data_type'] in big_endian_types:
types = [
('', 'a{}'.format(byte_count - size)),
('vals', fmt),
]
else:
types = [
('vals', fmt),
('', 'a{}'.format(byte_count - size)),
]
else:
types = [('vals', fmt), ]
vals = fromstring(vals, dtype=dtype(types))['vals']
if channel['data_type'] in v23c.SIGNED_INT:
return as_non_byte_sized_signed_int(vals, bit_count)
else:
return vals
def _validate_channel_selection(self, name=None, group=None, index=None):
"""Gets channel comment.
Channel can be specified in two ways:
* using the first positional argument *name*
* if there are multiple occurrences for this channel then the
*group* and *index* arguments can be used to select a specific
group.
* if there are multiple occurrences for this channel and either the
*group* or *index* arguments is None then a warning is issued
* using the group number (keyword argument *group*) and the channel
number (keyword argument *index*). Use *info* method for group and
channel numbers
If the *raster* keyword argument is not *None* the output is
interpolated accordingly.
Parameters
----------
name : string
name of channel
group : int
0-based group index
index : int
0-based channel index
Returns
-------
group_index, channel_index : (int, int)
selected channel's group and channel index
"""
if name is None:
if group is None or index is None:
message = (
'Invalid arguments for channel selection: '
'must give "name" or, "group" and "index"'
)
raise MdfException(message)
else:
gp_nr, ch_nr = group, index
if gp_nr > len(self.groups) - 1:
raise MdfException('Group index out of range')
if index > len(self.groups[gp_nr]['channels']) - 1:
raise MdfException('Channel index out of range')
else:
if name not in self.channels_db:
raise MdfException('Channel "{}" not found'.format(name))
else:
if group is None:
gp_nr, ch_nr = self.channels_db[name][0]
if len(self.channels_db[name]) > 1:
message = (
'Multiple occurances for channel "{}". '
'Using first occurance from data group {}. '
'Provide both "group" and "index" arguments'
' to select another data group'
)
message = message.format(name, gp_nr)
logger.warning(message)
else:
for gp_nr, ch_nr in self.channels_db[name]:
if gp_nr == group:
if index is None:
break
elif index == ch_nr:
break
else:
if index is None:
message = 'Channel "{}" not found in group {}'
message = message.format(name, group)
else:
message = (
'Channel "{}" not found in group {} '
'at index {}'
)
message = message.format(name, group, index)
raise MdfException(message)
return gp_nr, ch_nr
def _read(self):
stream = self._file
memory = self.memory
cg_count = count_channel_groups(stream, 3)
if self._callback:
self._callback(0, cg_count)
current_cg_index = 0
# performance optimization
read = stream.read
seek = stream.seek
dg_cntr = 0
seek(0)
self.identification = FileIdentificationBlock(
stream=stream,
)
self.header = HeaderBlock(stream=stream)
self.version = (
self.identification['version_str']
.decode('latin-1')
.strip(' \n\t\0')
)
# this will hold mapping from channel address to Channel object
# needed for linking dependency blocks to referenced channels after
# the file is loaded
ch_map = {}
ce_map = {}
cc_map = {}
# go to first date group
dg_addr = self.header['first_dg_addr']
# read each data group sequentially
while dg_addr:
data_group = DataGroup(
address=dg_addr,
stream=stream,
)
record_id_nr = data_group['record_id_len']
cg_nr = data_group['cg_nr']
cg_addr = data_group['first_cg_addr']
data_addr = data_group['data_block_addr']
# read trigger information if available
trigger_addr = data_group['trigger_addr']
if trigger_addr:
trigger = TriggerBlock(
address=trigger_addr,
stream=stream,
)
else:
trigger = None
new_groups = []
for i in range(cg_nr):
new_groups.append({})
grp = new_groups[-1]
grp['channels'] = []
grp['data_block'] = None
grp['trigger'] = trigger
grp['channel_dependencies'] = []
if record_id_nr:
grp['sorted'] = False
else:
grp['sorted'] = True
kargs = {
'first_cg_addr': cg_addr,
'data_block_addr': data_addr,
}
if self.version >= '3.20':
kargs['block_len'] = v23c.DG_POST_320_BLOCK_SIZE
else:
kargs['block_len'] = v23c.DG_PRE_320_BLOCK_SIZE
kargs['record_id_len'] = record_id_nr
grp['data_group'] = DataGroup(**kargs)
# read each channel group sequentially
grp['channel_group'] = ChannelGroup(
address=cg_addr,
stream=stream,
)
# go to first channel of the current channel group
ch_addr = grp['channel_group']['first_ch_addr']
ch_cntr = 0
grp_chs = grp['channels']
while ch_addr:
# read channel block and create channel object
load_metadata = memory != 'minimum'
new_ch = Channel(
address=ch_addr,
stream=stream,
load_metadata=load_metadata,
)
# check if it has channel dependencies
if new_ch['ch_depend_addr']:
dep = ChannelDependency(
address=new_ch['ch_depend_addr'],
stream=stream,
)
grp['channel_dependencies'].append(dep)
else:
grp['channel_dependencies'].append(None)
# update channel map
ch_map[ch_addr] = (ch_cntr, dg_cntr)
for name in (new_ch.name, new_ch.display_name):
self.channels_db.add(name, dg_cntr, ch_cntr)
if new_ch['channel_type'] == v23c.CHANNEL_TYPE_MASTER:
self.masters_db[dg_cntr] = ch_cntr
# go to next channel of the current channel group
ch_cntr += 1
if memory != 'minimum':
grp_chs.append(new_ch)
else:
grp_chs.append(ch_addr)
ch_addr = new_ch['next_ch_addr']
cg_addr = grp['channel_group']['next_cg_addr']
dg_cntr += 1
current_cg_index += 1
if self._callback:
self._callback(current_cg_index, cg_count)
if self._terminate:
self.close()
return
# store channel groups record sizes dict and data block size in
# each new group data belong to the initial unsorted group, and
# add the key 'sorted' with the value False to use a flag;
# this is used later if memory=False
cg_size = {}
total_size = 0
for grp in new_groups:
record_id = grp['channel_group']['record_id']
if PYVERSION == 2:
record_id = chr(record_id)
cycles_nr = grp['channel_group']['cycles_nr']
record_size = grp['channel_group']['samples_byte_nr']
cg_size[record_id] = record_size
record_size += record_id_nr
total_size += record_size * cycles_nr
grp['record_size'] = cg_size
grp['size'] = total_size
if memory == 'full':
# read data block of the current data group
dat_addr = data_group['data_block_addr']
if dat_addr:
seek(dat_addr)
data = read(total_size)
else:
data = b''
if record_id_nr == 0:
grp = new_groups[0]
grp['data_location'] = v23c.LOCATION_MEMORY
grp['data_block'] = DataBlock(data=data)
else:
# agregate data for each record ID in the cg_data dict
cg_data = defaultdict(list)
i = 0
size = len(data)
while i < size:
rec_id = data[i]
# skip record id
i += 1
rec_size = cg_size[rec_id]
rec_data = data[i: i + rec_size]
cg_data[rec_id].append(rec_data)
# possibly skip 2nd record id
if record_id_nr == 2:
i += rec_size + 1
else:
i += rec_size
for grp in new_groups:
grp['data_location'] = v23c.LOCATION_MEMORY
record_id = grp['channel_group']['record_id']
if PYVERSION == 2:
record_id = chr(record_id)
data = cg_data[record_id]
data = b''.join(data)
grp['channel_group']['record_id'] = 1
grp['data_block'] = DataBlock(data=data)
else:
for grp in new_groups:
grp['data_location'] = v23c.LOCATION_ORIGINAL_FILE
grp['data_group']['data_block_addr'] = data_group['data_block_addr']
grp['data_block_addr'] = [data_group['data_block_addr'], ]
grp['data_block_size'] = [total_size, ]
self.groups.extend(new_groups)
# go to next data group
dg_addr = data_group['next_dg_addr']
# finally update the channel depency references
for grp in self.groups:
for dep in grp['channel_dependencies']:
if dep:
for i in range(dep['sd_nr']):
ref_channel_addr = dep['ch_{}'.format(i)]
channel = ch_map[ref_channel_addr]
dep.referenced_channels.append(channel)
[docs] def add_trigger(self,
group,
timestamp,
pre_time=0,
post_time=0,
comment=''):
""" add trigger to data group
Parameters
----------
group : int
group index
timestamp : float
trigger time
pre_time : float
trigger pre time; default 0
post_time : float
trigger post time; default 0
comment : str
trigger comment
"""
comment_template = """<EVcomment>
<TX>{}</TX>
</EVcomment>"""
try:
group = self.groups[group]
except IndexError:
return
trigger = group['trigger']
if comment:
try:
comment = ET.fromstring(comment)
comment = comment.find('.//TX').text
except:
pass
if trigger:
count = trigger['trigger_events_nr']
trigger['trigger_events_nr'] += 1
trigger['block_len'] += 24
trigger['trigger_{}_time'.format(count)] = timestamp
trigger['trigger_{}_pretime'.format(count)] = pre_time
trigger['trigger_{}_posttime'.format(count)] = post_time
if comment:
if trigger.comment is None:
comment = '{}. {}'.format(
count + 1,
comment,
)
comment = comment_template.format(comment)
trigger.comment = comment
else:
current_comment = trigger.comment
try:
current_comment = ET.fromstring(current_comment)
current_comment = current_comment.find('.//TX').text
except:
raise
comment = '{}\n{}. {}'.format(
current_comment,
count + 1,
comment,
)
comment = comment_template.format(comment)
trigger.comment = comment
else:
trigger = TriggerBlock(
trigger_event_nr=1,
trigger_0_time=timestamp,
trigger_0_pretime=pre_time,
trigger_0_posttime=post_time,
)
if comment:
comment = '1. {}'.format(
comment,
)
comment = comment_template.format(comment)
trigger.comment = comment
group['trigger'] = trigger
[docs] def append(
self,
signals,
acquisition_info='Python',
common_timebase=False,
units=None,
):
"""Appends a new data group.
For channel dependencies type Signals, the *samples* attribute must be
a numpy.recarray
Parameters
----------
signals : list | Signal | pandas.DataFrame
list of *Signal* objects, or a single *Signal* object, or a pandas
*DataFrame* object
acquisition_info : str
acquisition information; default 'Python'
common_timebase : bool
flag to hint that the signals have the same timebase
units : dict
will contain the signal units mapped to the singal names when
appending a pandas DataFrame
Examples
--------
>>> # case 1 conversion type None
>>> s1 = np.array([1, 2, 3, 4, 5])
>>> s2 = np.array([-1, -2, -3, -4, -5])
>>> s3 = np.array([0.1, 0.04, 0.09, 0.16, 0.25])
>>> t = np.array([0.001, 0.002, 0.003, 0.004, 0.005])
>>> names = ['Positive', 'Negative', 'Float']
>>> units = ['+', '-', '.f']
>>> info = {}
>>> s1 = Signal(samples=s1, timstamps=t, unit='+', name='Positive')
>>> s2 = Signal(samples=s2, timstamps=t, unit='-', name='Negative')
>>> s3 = Signal(samples=s3, timstamps=t, unit='flts', name='Floats')
>>> mdf = MDF3('new.mdf')
>>> mdf.append([s1, s2, s3], 'created by asammdf v1.1.0')
>>> # case 2: VTAB conversions from channels inside another file
>>> mdf1 = MDF3('in.mdf')
>>> ch1 = mdf1.get("Channel1_VTAB")
>>> ch2 = mdf1.get("Channel2_VTABR")
>>> sigs = [ch1, ch2]
>>> mdf2 = MDF3('out.mdf')
>>> mdf2.append(sigs, 'created by asammdf v1.1.0')
>>> df = pd.DataFrame.from_dict({'s1': np.array([1, 2, 3, 4, 5]), 's2': np.array([-1, -2, -3, -4, -5])})
>>> units = {'s1': 'V', 's2': 'A'}
>>> mdf2.append(df, units=units)
"""
if isinstance(signals, Signal):
signals = [signals, ]
elif isinstance(signals, DataFrame):
self._append_dataframe(signals, acquisition_info, units=units)
return
version = self.version
# check if the signals have a common timebase
# if not interpolate the signals using the union of all timbases
if signals:
timestamps = signals[0].timestamps
if not common_timebase:
for signal in signals[1:]:
if not array_equal(signal.timestamps, timestamps):
different = True
break
else:
different = False
if different:
times = [s.timestamps for s in signals]
timestamps = reduce(union1d, times).flatten().astype(float64)
signals = [s.interp(timestamps) for s in signals]
times = None
else:
timestamps = array([])
if self.version < '3.10':
if timestamps.dtype.byteorder == '>':
timestamps = timestamps.byteswap().newbyteorder()
for signal in signals:
if signal.samples.dtype.byteorder == '>':
signal.samples = signal.samples.byteswap().newbyteorder()
if self.version >= '3.00':
channel_size = v23c.CN_DISPLAYNAME_BLOCK_SIZE
elif self.version >= '2.10':
channel_size = v23c.CN_LONGNAME_BLOCK_SIZE
else:
channel_size = v23c.CN_SHORT_BLOCK_SIZE
memory = self.memory
file = self._tempfile
write = file.write
tell = file.tell
kargs = {
'module_nr': 0,
'module_address': 0,
'type': v23c.SOURCE_ECU,
'description': b'Channel inserted by Python Script',
}
ce_block = ChannelExtension(**kargs)
canopen_time_fields = (
'ms',
'days',
)
canopen_date_fields = (
'ms',
'min',
'hour',
'day',
'month',
'year',
'summer_time',
'day_of_week',
)
defined_texts, cc_map, si_map = {}, {}, {}
dg_cntr = len(self.groups)
gp = {}
gp['channels'] = gp_channels = []
gp['channel_dependencies'] = gp_dep = []
gp['signal_types'] = gp_sig_types = []
gp['string_dtypes'] = []
self.groups.append(gp)
cycles_nr = len(timestamps)
fields = []
types = []
parents = {}
ch_cntr = 0
offset = 0
field_names = set()
if signals:
master_metadata = signals[0].master_metadata
else:
master_metadata = None
if master_metadata:
time_name = master_metadata[0]
else:
time_name = 'time'
if signals:
# conversion for time channel
kargs = {
'conversion_type': v23c.CONVERSION_TYPE_NONE,
'unit': b's',
'min_phy_value': timestamps[0] if cycles_nr else 0,
'max_phy_value': timestamps[-1] if cycles_nr else 0,
}
conversion = ChannelConversion(**kargs)
conversion.unit = 's'
source = ce_block
# time channel
t_type, t_size = fmt_to_datatype_v3(
timestamps.dtype,
timestamps.shape,
)
kargs = {
'short_name': time_name.encode('latin-1'),
'channel_type': v23c.CHANNEL_TYPE_MASTER,
'data_type': t_type,
'start_offset': 0,
'min_raw_value': timestamps[0] if cycles_nr else 0,
'max_raw_value': timestamps[-1] if cycles_nr else 0,
'bit_count': t_size,
'block_len': channel_size,
'version': version,
}
channel = Channel(**kargs)
channel.name = name = time_name
channel.conversion = conversion
channel.source = source
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
self.channels_db.add(name, dg_cntr, ch_cntr)
self.masters_db[dg_cntr] = 0
# data group record parents
parents[ch_cntr] = name, 0
# time channel doesn't have channel dependencies
gp_dep.append(None)
fields.append(timestamps)
types.append((name, timestamps.dtype))
field_names.add(name)
offset += t_size
ch_cntr += 1
gp_sig_types.append(0)
for signal in signals:
sig = signal
names = sig.samples.dtype.names
name = signal.name
if names is None:
sig_type = v23c.SIGNAL_TYPE_SCALAR
else:
if names in (canopen_time_fields, canopen_date_fields):
sig_type = v23c.SIGNAL_TYPE_CANOPEN
elif names[0] != sig.name:
sig_type = v23c.SIGNAL_TYPE_STRUCTURE_COMPOSITION
else:
sig_type = v23c.SIGNAL_TYPE_ARRAY
gp_sig_types.append(sig_type)
# conversions for channel
conversion = conversion_transfer(signal.conversion)
conversion.unit = unit = signal.unit
israw = signal.raw
if not israw and not unit:
conversion = None
min_val, max_val = get_min_max(signal.samples)
if sig_type == v23c.SIGNAL_TYPE_SCALAR:
# source for channel
if signal.source:
source = signal.source
if source.source_type != 2:
kargs = {
'type': v23c.SOURCE_ECU,
'description': source.name.encode('latin-1'),
'ECU_identification': source.path.encode('latin-1'),
}
else:
kargs = {
'type': v23c.SOURCE_VECTOR,
'message_name': source.name.encode('latin-1'),
'sender_name': source.path.encode('latin-1'),
}
new_source = ChannelExtension(**kargs)
else:
new_source = ce_block
# compute additional byte offset for large records size
if offset > v23c.MAX_UINT16:
additional_byte_offset = ceil((offset - v23c.MAX_UINT16) / 8)
start_bit_offset = offset - additional_byte_offset * 8
else:
start_bit_offset = offset
additional_byte_offset = 0
s_type, s_size = fmt_to_datatype_v3(
signal.samples.dtype,
signal.samples.shape,
)
name = signal.name
comment = signal.comment
display_name = signal.display_name
if signal.samples.dtype.kind == 'u' and signal.bit_count <= 4:
s_size_ = signal.bit_count
else:
s_size_ = s_size
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size_,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'version': version,
}
if s_size < 8:
s_size = 8
channel = Channel(**kargs)
channel.name = signal.name
channel.comment = signal.comment
channel.source = new_source
channel.conversion = conversion
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
offset += s_size
self.channels_db.add(name, dg_cntr, ch_cntr)
self.channels_db.add(display_name, dg_cntr, ch_cntr)
# update the parents as well
field_name = get_unique_name(field_names, name)
parents[ch_cntr] = field_name, 0
if signal.samples.dtype.kind == 'S':
gp['string_dtypes'].append(signal.samples.dtype)
fields.append(signal.samples)
if s_type != v23c.DATA_TYPE_BYTEARRAY:
types.append((field_name, signal.samples.dtype))
else:
types.append((field_name, signal.samples.dtype, signal.samples.shape[1:]))
field_names.add(field_name)
ch_cntr += 1
# simple channels don't have channel dependencies
gp_dep.append(None)
# second, add the composed signals
elif sig_type in (
v23c.SIGNAL_TYPE_CANOPEN,
v23c.SIGNAL_TYPE_STRUCTURE_COMPOSITION):
new_dg_cntr = len(self.groups)
new_gp = {}
new_gp['channels'] = new_gp_channels = []
new_gp['channel_dependencies'] = new_gp_dep = []
self.groups.append(new_gp)
new_fields = []
new_types = []
new_parents = {}
new_ch_cntr = 0
new_offset = 0
new_field_names = set()
# conversion for time channel
kargs = {
'conversion_type': v23c.CONVERSION_TYPE_NONE,
'unit': b's',
'min_phy_value': timestamps[0] if cycles_nr else 0,
'max_phy_value': timestamps[-1] if cycles_nr else 0,
}
conversion = ChannelConversion(**kargs)
conversion.unit = 's'
source = ce_block
# time channel
t_type, t_size = fmt_to_datatype_v3(
timestamps.dtype,
timestamps.shape,
)
kargs = {
'short_name': time_name.encode('latin-1'),
'channel_type': v23c.CHANNEL_TYPE_MASTER,
'data_type': t_type,
'start_offset': 0,
'min_raw_value': timestamps[0] if cycles_nr else 0,
'max_raw_value': timestamps[-1] if cycles_nr else 0,
'bit_count': t_size,
'block_len': channel_size,
'version': version,
}
channel = Channel(**kargs)
channel.name = name = time_name
channel.source = source
channel.conversion = conversion
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
self.channels_db.add(name, new_dg_cntr, new_ch_cntr)
self.masters_db[new_dg_cntr] = 0
# data group record parents
new_parents[new_ch_cntr] = name, 0
# time channel doesn't have channel dependencies
new_gp_dep.append(None)
new_fields.append(timestamps)
new_types.append((name, timestamps.dtype))
new_field_names.add(name)
new_offset += t_size
new_ch_cntr += 1
names = signal.samples.dtype.names
if names == (
'ms',
'days'):
channel_group_comment = 'From mdf v4 CANopen Time channel'
elif names == (
'ms',
'min',
'hour',
'day',
'month',
'year',
'summer_time',
'day_of_week'):
channel_group_comment = 'From mdf v4 CANopen Date channel'
else:
channel_group_comment = 'From mdf v4 structure channel composition'
for name in names:
samples = signal.samples[name]
# conversions for channel
min_val, max_val = get_min_max(samples)
kargs = {
'conversion_type': v23c.CONVERSION_TYPE_NONE,
'unit': signal.unit.encode('latin-1'),
'min_phy_value': min_val if min_val <= max_val else 0,
'max_phy_value': max_val if min_val <= max_val else 0,
}
conversion = ChannelConversion(**kargs)
# source for channel
if signal.source:
source = signal.source
if source.source_type != 2:
kargs = {
'type': v23c.SOURCE_ECU,
'description': source.name.encode('latin-1'),
'ECU_identification': source.path.encode('latin-1'),
}
else:
kargs = {
'type': v23c.SOURCE_VECTOR,
'message_name': source.name.encode('latin-1'),
'sender_name': source.path.encode('latin-1'),
}
source = ChannelExtension(**kargs)
else:
source = ce_block
# compute additional byte offset for large records size
if new_offset > v23c.MAX_UINT16:
additional_byte_offset = (new_offset - v23c.MAX_UINT16) >> 3
start_bit_offset = new_offset - additional_byte_offset << 3
else:
start_bit_offset = new_offset
additional_byte_offset = 0
s_type, s_size = fmt_to_datatype_v3(
samples.dtype,
samples.shape,
)
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'version': version,
}
if s_size < 8:
s_size = 8
channel = Channel(**kargs)
channel.name = name
channel.source = source
channel.conversion = conversion
if memory != 'minimum':
new_gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
new_gp_channels.append(channel.address)
new_offset += s_size
self.channels_db.add(name, new_dg_cntr, new_ch_cntr)
# update the parents as well
field_name = get_unique_name(new_field_names, name)
new_parents[new_ch_cntr] = field_name, 0
new_fields.append(samples)
new_types.append((field_name, samples.dtype))
new_field_names.add(field_name)
new_ch_cntr += 1
# simple channels don't have channel dependencies
new_gp_dep.append(None)
# channel group
kargs = {
'cycles_nr': cycles_nr,
'samples_byte_nr': new_offset >> 3,
'ch_nr': new_ch_cntr,
}
new_gp['channel_group'] = ChannelGroup(**kargs)
new_gp['channel_group'].comment = channel_group_comment
new_gp['size'] = cycles_nr * (new_offset >> 3)
# data group
if self.version >= '3.20':
block_len = v23c.DG_POST_320_BLOCK_SIZE
else:
block_len = v23c.DG_PRE_320_BLOCK_SIZE
new_gp['data_group'] = DataGroup(block_len=block_len)
# data block
if PYVERSION == 2:
new_types = fix_dtype_fields(new_types)
new_types = dtype(new_types)
new_gp['types'] = new_types
new_gp['parents'] = new_parents
new_gp['sorted'] = True
samples = fromarrays(new_fields, dtype=new_types)
try:
block = samples.tostring()
if memory == 'full':
new_gp['data_location'] = v23c.LOCATION_MEMORY
kargs = {'data': block}
new_gp['data_block'] = DataBlock(**kargs)
else:
new_gp['data_location'] = v23c.LOCATION_TEMPORARY_FILE
if cycles_nr:
data_address = tell()
new_gp['data_group']['data_block_addr'] = data_address
self._tempfile.write(block)
else:
new_gp['data_group']['data_block_addr'] = 0
except MemoryError:
if memory == 'full':
raise
else:
new_gp['data_location'] = v23c.LOCATION_TEMPORARY_FILE
data_address = tell()
new_gp['data_group']['data_block_addr'] = data_address
for sample in samples:
self._tempfile.write(sample.tostring())
# data group trigger
new_gp['trigger'] = None
else:
names = signal.samples.dtype.names
name = signal.name
component_names = []
component_samples = []
if names:
samples = signal.samples[names[0]]
else:
samples = signal.samples
shape = samples.shape[1:]
dims = [list(range(size)) for size in shape]
for indexes in product(*dims):
subarray = samples
for idx in indexes:
subarray = subarray[:, idx]
component_samples.append(subarray)
indexes = ''.join('[{}]'.format(idx) for idx in indexes)
component_name = '{}{}'.format(name, indexes)
component_names.append(component_name)
# add channel dependency block for composed parent channel
sd_nr = len(component_samples)
kargs = {'sd_nr': sd_nr}
for i, dim in enumerate(shape[::-1]):
kargs['dim_{}'.format(i)] = dim
parent_dep = ChannelDependency(**kargs)
gp_dep.append(parent_dep)
# source for channel
if signal.source:
source = signal.source
if source.source_type != 2:
kargs = {
'type': v23c.SOURCE_ECU,
'description': source.name.encode('latin-1'),
'ECU_identification': source.path.encode('latin-1'),
}
else:
kargs = {
'type': v23c.SOURCE_VECTOR,
'message_name': source.name.encode('latin-1'),
'sender_name': source.path.encode('latin-1'),
}
source = ChannelExtension(**kargs)
else:
source = ce_block
min_val, max_val = get_min_max(samples)
s_type, s_size = fmt_to_datatype_v3(
samples.dtype,
(),
True,
)
# compute additional byte offset for large records size
if offset > v23c.MAX_UINT16:
additional_byte_offset = (offset - v23c.MAX_UINT16) >> 3
start_bit_offset = offset - additional_byte_offset << 3
else:
start_bit_offset = offset
additional_byte_offset = 0
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'version': version,
}
if s_size <8:
s_size = 8
channel = Channel(**kargs)
channel.comment = signal.comment
channel.display_name = signal.display_name
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
self.channels_db.add(name, dg_cntr, ch_cntr)
ch_cntr += 1
for i, (name, samples) in enumerate(
zip(component_names, component_samples)):
if i < sd_nr:
dep_pair = ch_cntr, dg_cntr
parent_dep.referenced_channels.append(dep_pair)
description = b'\0'
else:
description = '{} - axis {}'.format(signal.name, name)
description = description.encode('latin-1')
min_val, max_val = get_min_max(samples)
s_type, s_size = fmt_to_datatype_v3(
samples.dtype,
(),
)
shape = samples.shape[1:]
# source for channel
if signal.source:
source = signal.source
if source.source_type != 2:
kargs = {
'type': v23c.SOURCE_ECU,
'description': source.name.encode('latin-1'),
'ECU_identification': source.path.encode('latin-1'),
}
else:
kargs = {
'type': v23c.SOURCE_VECTOR,
'message_name': source.name.encode('latin-1'),
'sender_name': source.path.encode('latin-1'),
}
source = ChannelExtension(**kargs)
else:
source = ce_block
# compute additional byte offset for large records size
if offset > v23c.MAX_UINT16:
additional_byte_offset = (offset - v23c.MAX_UINT16) >> 3
start_bit_offset = offset - additional_byte_offset << 3
else:
start_bit_offset = offset
additional_byte_offset = 0
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'description': description,
'version': version,
}
if s_size < 8:
s_size = 8
channel = Channel(**kargs)
channel.name = name
channel.source = source
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
size = s_size
for dim in shape:
size *= dim
offset += size
self.channels_db.add(name, dg_cntr, ch_cntr)
# update the parents as well
field_name = get_unique_name(field_names, name)
parents[ch_cntr] = field_name, 0
fields.append(samples)
types.append((field_name, samples.dtype, shape))
field_names.add(field_name)
gp_dep.append(None)
ch_cntr += 1
for name in names[1:]:
samples = signal.samples[name]
component_names = []
component_samples = []
shape = samples.shape[1:]
dims = [list(range(size)) for size in shape]
for indexes in product(*dims):
subarray = samples
for idx in indexes:
subarray = subarray[:, idx]
component_samples.append(subarray)
indexes = ''.join('[{}]'.format(idx) for idx in indexes)
component_name = '{}{}'.format(name, indexes)
component_names.append(component_name)
# add channel dependency block for composed parent channel
sd_nr = len(component_samples)
kargs = {'sd_nr': sd_nr}
for i, dim in enumerate(shape[::-1]):
kargs['dim_{}'.format(i)] = dim
parent_dep = ChannelDependency(**kargs)
gp_dep.append(parent_dep)
# source for channel
if signal.source:
source = signal.source
if source.source_type != 2:
kargs = {
'type': v23c.SOURCE_ECU,
'description': source.name.encode('latin-1'),
'ECU_identification': source.path.encode('latin-1'),
}
else:
kargs = {
'type': v23c.SOURCE_VECTOR,
'message_name': source.name.encode('latin-1'),
'sender_name': source.path.encode('latin-1'),
}
source = ChannelExtension(**kargs)
else:
source = ce_block
min_val, max_val = get_min_max(samples)
s_type, s_size = fmt_to_datatype_v3(
samples.dtype,
(),
)
# compute additional byte offset for large records size
if offset > v23c.MAX_UINT16:
additional_byte_offset = (offset - v23c.MAX_UINT16) >> 3
start_bit_offset = offset - additional_byte_offset << 3
else:
start_bit_offset = offset
additional_byte_offset = 0
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'version': version,
}
if s_size < 8:
s_size = 8
channel = Channel(**kargs)
channel.name = name
channel.comment = signal.comment
channel.source = source
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
self.channels_db.add(name, dg_cntr, ch_cntr)
ch_cntr += 1
for i, (name, samples) in enumerate(
zip(component_names, component_samples)):
if i < sd_nr:
dep_pair = ch_cntr, dg_cntr
parent_dep.referenced_channels.append(dep_pair)
description = b'\0'
else:
description = '{} - axis {}'.format(signal.name, name)
description = description.encode('latin-1')
min_val, max_val = get_min_max(samples)
s_type, s_size = fmt_to_datatype_v3(
samples.dtype,
(),
)
shape = samples.shape[1:]
# source for channel
if signal.source:
source = signal.source
if source.source_type != 2:
kargs = {
'type': v23c.SOURCE_ECU,
'description': source.name.encode('latin-1'),
'ECU_identification': source.path.encode('latin-1'),
}
else:
kargs = {
'type': v23c.SOURCE_VECTOR,
'message_name': source.name.encode('latin-1'),
'sender_name': source.path.encode('latin-1'),
}
source = ChannelExtension(**kargs)
else:
source = ce_block
# compute additional byte offset for large records size
if offset > v23c.MAX_UINT16:
additional_byte_offset = (offset - v23c.MAX_UINT16) >> 3
start_bit_offset = offset - additional_byte_offset << 3
else:
start_bit_offset = offset
additional_byte_offset = 0
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'description': description,
'version': version,
}
if s_size < 8:
s_size = 8
channel = Channel(**kargs)
channel.name = name
channel.source = source
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
size = s_size
for dim in shape:
size *= dim
offset += size
self.channels_db.add(name, dg_cntr, ch_cntr)
# update the parents as well
field_name = get_unique_name(field_names, name)
parents[ch_cntr] = field_name, 0
fields.append(samples)
types.append((field_name, samples.dtype, shape))
field_names.add(field_name)
gp_dep.append(None)
ch_cntr += 1
# channel group
kargs = {
'cycles_nr': cycles_nr,
'samples_byte_nr': offset >> 3,
'ch_nr': ch_cntr,
}
if self.version >= '3.30':
kargs['block_len'] = v23c.CG_POST_330_BLOCK_SIZE
else:
kargs['block_len'] = v23c.CG_PRE_330_BLOCK_SIZE
gp['channel_group'] = ChannelGroup(**kargs)
gp['channel_group'].comment = acquisition_info
gp['size'] = cycles_nr * (offset >> 3)
# data group
if self.version >= '3.20':
block_len = v23c.DG_POST_320_BLOCK_SIZE
else:
block_len = v23c.DG_PRE_320_BLOCK_SIZE
gp['data_group'] = DataGroup(block_len=block_len)
# data block
if PYVERSION == 2:
types = fix_dtype_fields(types, 'latin-1')
types = dtype(types)
gp['types'] = types
gp['parents'] = parents
gp['sorted'] = True
if signals:
samples = fromarrays(fields, dtype=types)
else:
samples = array([])
block = samples.tostring()
if memory == 'full':
gp['data_location'] = v23c.LOCATION_MEMORY
kargs = {'data': block}
gp['data_block'] = DataBlock(**kargs)
else:
gp['data_location'] = v23c.LOCATION_TEMPORARY_FILE
if cycles_nr:
data_address = tell()
gp['data_group']['data_block_addr'] = data_address
gp['data_block_addr'] = [data_address, ]
gp['data_block_size'] = [len(block), ]
self._tempfile.write(block)
else:
gp['data_group']['data_block_addr'] = 0
gp['data_block_addr'] = [0, ]
gp['data_block_size'] = [0, ]
# data group trigger
gp['trigger'] = None
def _append_dataframe(self, df, source_info='', units=None):
"""
Appends a new data group from a Pandas data frame.
"""
units = units or {}
t = df.index
index_name = df.index.name
time_name = index_name or 'time'
time_unit = 's'
version = self.version
timestamps = t
# if self.version < '3.10':
# if timestamps.dtype.byteorder == '>':
# timestamps = timestamps.byteswap().newbyteorder()
# for signal in signals:
# if signal.samples.dtype.byteorder == '>':
# signal.samples = signal.samples.byteswap().newbyteorder()
if self.version >= '3.00':
channel_size = v23c.CN_DISPLAYNAME_BLOCK_SIZE
elif self.version >= '2.10':
channel_size = v23c.CN_LONGNAME_BLOCK_SIZE
else:
channel_size = v23c.CN_SHORT_BLOCK_SIZE
memory = self.memory
file = self._tempfile
write = file.write
tell = file.tell
kargs = {
'module_nr': 0,
'module_address': 0,
'type': v23c.SOURCE_ECU,
'description': b'Channel inserted by Python Script',
}
ce_block = ChannelExtension(**kargs)
defined_texts, cc_map, si_map = {}, {}, {}
dg_cntr = len(self.groups)
gp = {}
gp['channels'] = gp_channels = []
gp['channel_dependencies'] = gp_dep = []
gp['signal_types'] = gp_sig_types = []
gp['string_dtypes'] = []
self.groups.append(gp)
cycles_nr = len(timestamps)
fields = []
types = []
parents = {}
ch_cntr = 0
offset = 0
field_names = set()
if df.shape[0]:
# conversion for time channel
kargs = {
'conversion_type': v23c.CONVERSION_TYPE_NONE,
'unit': b's',
'min_phy_value': timestamps[0] if cycles_nr else 0,
'max_phy_value': timestamps[-1] if cycles_nr else 0,
}
conversion = ChannelConversion(**kargs)
conversion.unit = 's'
source = ce_block
# time channel
t_type, t_size = fmt_to_datatype_v3(
timestamps.dtype,
timestamps.shape,
)
kargs = {
'short_name': time_name.encode('latin-1'),
'channel_type': v23c.CHANNEL_TYPE_MASTER,
'data_type': t_type,
'start_offset': 0,
'min_raw_value': timestamps[0] if cycles_nr else 0,
'max_raw_value': timestamps[-1] if cycles_nr else 0,
'bit_count': t_size,
'block_len': channel_size,
'version': version,
}
channel = Channel(**kargs)
channel.name = name = time_name
channel.conversion = conversion
channel.source = source
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
self.channels_db.add(name, dg_cntr, ch_cntr)
self.masters_db[dg_cntr] = 0
# data group record parents
parents[ch_cntr] = name, 0
# time channel doesn't have channel dependencies
gp_dep.append(None)
fields.append(timestamps)
types.append((name, timestamps.dtype))
field_names.add(name)
offset += t_size
ch_cntr += 1
gp_sig_types.append(0)
for signal in df:
sig = df[signal]
name = signal
sig_type = v23c.SIGNAL_TYPE_SCALAR
gp_sig_types.append(sig_type)
min_val, max_val = get_min_max(sig)
new_source = ce_block
# compute additional byte offset for large records size
if offset > v23c.MAX_UINT16:
additional_byte_offset = ceil(
(offset - v23c.MAX_UINT16) / 8)
start_bit_offset = offset - additional_byte_offset * 8
else:
start_bit_offset = offset
additional_byte_offset = 0
s_type, s_size = fmt_to_datatype_v3(
sig.dtype,
sig.shape,
)
kargs = {
'channel_type': v23c.CHANNEL_TYPE_VALUE,
'data_type': s_type,
'min_raw_value': min_val if min_val <= max_val else 0,
'max_raw_value': max_val if min_val <= max_val else 0,
'start_offset': start_bit_offset,
'bit_count': s_size,
'aditional_byte_offset': additional_byte_offset,
'block_len': channel_size,
'version': version,
}
if s_size < 8:
s_size = 8
channel = Channel(**kargs)
channel.name = name
channel.source = new_source
unit = units.get(name, b'')
if unit:
if hasattr(unit, 'encode'):
unit = unit.encode('latin-1')
# conversion for time channel
kargs = {
'conversion_type': v23c.CONVERSION_TYPE_NONE,
'unit': unit,
'min_phy_value': min_val if min_val <= max_val else 0,
'max_phy_value': max_val if min_val <= max_val else 0,
}
conversion = ChannelConversion(**kargs)
conversion.unit = unit
if memory != 'minimum':
gp_channels.append(channel)
else:
channel.to_stream(file, defined_texts, cc_map, si_map)
gp_channels.append(channel.address)
offset += s_size
self.channels_db.add(name, dg_cntr, ch_cntr)
# update the parents as well
field_name = get_unique_name(field_names, name)
parents[ch_cntr] = field_name, 0
if sig.dtype.kind == 'S':
gp['string_dtypes'].append(sig.dtype)
fields.append(sig)
types.append((field_name, sig.dtype))
field_names.add(field_name)
ch_cntr += 1
# simple channels don't have channel dependencies
gp_dep.append(None)
# channel group
kargs = {
'cycles_nr': cycles_nr,
'samples_byte_nr': offset >> 3,
'ch_nr': ch_cntr,
}
if self.version >= '3.30':
kargs['block_len'] = v23c.CG_POST_330_BLOCK_SIZE
else:
kargs['block_len'] = v23c.CG_PRE_330_BLOCK_SIZE
gp['channel_group'] = ChannelGroup(**kargs)
gp['channel_group'].comment = source_info
gp['size'] = cycles_nr * (offset >> 3)
# data group
if self.version >= '3.20':
block_len = v23c.DG_POST_320_BLOCK_SIZE
else:
block_len = v23c.DG_PRE_320_BLOCK_SIZE
gp['data_group'] = DataGroup(block_len=block_len)
# data block
if PYVERSION == 2:
types = fix_dtype_fields(types, 'latin-1')
types = dtype(types)
gp['types'] = types
gp['parents'] = parents
gp['sorted'] = True
if df.shape[0]:
samples = fromarrays(fields, dtype=types)
else:
samples = array([])
block = samples.tostring()
if memory == 'full':
gp['data_location'] = v23c.LOCATION_MEMORY
kargs = {'data': block}
gp['data_block'] = DataBlock(**kargs)
else:
gp['data_location'] = v23c.LOCATION_TEMPORARY_FILE
if cycles_nr:
data_address = tell()
gp['data_group']['data_block_addr'] = data_address
gp['data_block_addr'] = [data_address, ]
gp['data_block_size'] = [len(block), ]
self._tempfile.write(block)
else:
gp['data_group']['data_block_addr'] = 0
gp['data_block_addr'] = [0, ]
gp['data_block_size'] = [0, ]
# data group trigger
gp['trigger'] = None
[docs] def close(self):
""" if the MDF was created with memory='minimum' and new
channels have been appended, then this must be called just before the
object is not used anymore to clean-up the temporary file
"""
if self._tempfile is not None:
self._tempfile.close()
if self._file is not None and not self._from_filelike:
self._file.close()
[docs] def extend(self, index, signals):
"""
Extend a group with new samples. The first signal is the master channel's samples, and the
next signals must respect the same order in which they were appended. The samples must have raw
or physical values according to the *Signals* used for the initial append.
Parameters
----------
index : int
group index
signals : list
list on numpy.ndarray objects
Examples
--------
>>> # case 1 conversion type None
>>> s1 = np.array([1, 2, 3, 4, 5])
>>> s2 = np.array([-1, -2, -3, -4, -5])
>>> s3 = np.array([0.1, 0.04, 0.09, 0.16, 0.25])
>>> t = np.array([0.001, 0.002, 0.003, 0.004, 0.005])
>>> names = ['Positive', 'Negative', 'Float']
>>> units = ['+', '-', '.f']
>>> s1 = Signal(samples=s1, timstamps=t, unit='+', name='Positive')
>>> s2 = Signal(samples=s2, timstamps=t, unit='-', name='Negative')
>>> s3 = Signal(samples=s3, timstamps=t, unit='flts', name='Floats')
>>> mdf = MDF3('new.mdf')
>>> mdf.append([s1, s2, s3], 'created by asammdf v1.1.0')
>>> t = np.array([0.006, 0.007, 0.008, 0.009, 0.010])
>>> mdf2.extend(0, [t, s1, s2, s3])
"""
memory = self.memory
new_group_offset = 0
gp = self.groups[index]
if not signals:
message = '"append" requires a non-empty list of Signal objects'
raise MdfException(message)
if gp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
canopen_time_fields = (
'ms',
'days',
)
canopen_date_fields = (
'ms',
'min',
'hour',
'day',
'month',
'year',
'summer_time',
'day_of_week',
)
fields = []
types = []
cycles_nr = len(signals[0][0])
string_counter = 0
for signal, _ in signals:
sig = signal
names = sig.dtype.names
if len(sig.shape) <= 1:
if names is None:
sig_type = v23c.SIGNAL_TYPE_SCALAR
else:
if names in (canopen_time_fields, canopen_date_fields):
sig_type = v23c.SIGNAL_TYPE_CANOPEN
elif names[0] != sig.name:
sig_type = v23c.SIGNAL_TYPE_STRUCTURE_COMPOSITION
else:
sig_type = v23c.SIGNAL_TYPE_ARRAY
else:
sig_type = v23c.SIGNAL_TYPE_ARRAY
if sig_type == v23c.SIGNAL_TYPE_SCALAR:
if signal.dtype.kind == 'S':
str_dtype = gp['string_dtypes'][string_counter]
signal = signal.astype(str_dtype)
string_counter += 1
fields.append(signal)
if signal.shape[1:]:
types.append(('', signal.dtype, signal.shape[1:]))
else:
types.append(('', signal.dtype))
# second, add the composed signals
elif sig_type in (
v23c.SIGNAL_TYPE_CANOPEN,
v23c.SIGNAL_TYPE_STRUCTURE_COMPOSITION):
new_group_offset += 1
new_gp = self.groups[index + new_group_offset]
new_fields = []
new_types = []
names = signal.dtype.names
for name in names:
new_fields.append(signal[name])
new_types.append(('', signal.dtype))
# data block
if PYVERSION == 2:
new_types = fix_dtype_fields(new_types)
new_types = dtype(new_types)
samples = fromarrays(new_fields, dtype=new_types)
samples = samples.tostring()
record_size = new_gp['channel_group']['samples_byte_nr']
extended_size = cycles_nr * record_size
new_gp['size'] += extended_size
if memory == 'full':
if samples:
data = new_gp['data_block']['data'] + samples
new_gp['data_block'] = DataBlock(data=data)
else:
if samples:
stream.seek(0, 2)
data_address = stream.tell()
new_gp['data_block_addr'].append(data_address)
new_gp['data_block_size'].append(extended_size)
stream.write(samples)
else:
names = signal.dtype.names
component_samples = []
if names:
samples = signal[names[0]]
else:
samples = signal
shape = samples.shape[1:]
dims = [list(range(size)) for size in shape]
for indexes in product(*dims):
subarray = samples
for idx in indexes:
subarray = subarray[:, idx]
component_samples.append(subarray)
if names:
new_samples = [signal[fld] for fld in names[1:]]
component_samples.extend(new_samples)
for samples in component_samples:
shape = samples.shape[1:]
fields.append(samples)
types.append(('', samples.dtype, shape))
record_size = gp['channel_group']['samples_byte_nr']
extended_size = cycles_nr * record_size
gp['size'] += extended_size
# data block
if PYVERSION == 2:
types = fix_dtype_fields(types, 'latin-1')
types = dtype(types)
samples = fromarrays(fields, dtype=types)
samples = samples.tostring()
if memory == 'full':
if samples:
data = gp['data_block']['data'] + samples
gp['data_block'] = DataBlock(data=data)
gp['channel_group']['cycles_nr'] += cycles_nr
else:
if cycles_nr:
stream.seek(0, 2)
data_address = stream.tell()
gp['data_block_addr'].append(data_address)
gp['data_block_size'].append(extended_size)
stream.write(samples)
gp['channel_group']['cycles_nr'] += cycles_nr
[docs] def get_channel_name(self, group, index):
"""Gets channel name.
Parameters
----------
group : int
0-based group index
index : int
0-based channel index
Returns
-------
name : str
found channel name
"""
gp_nr, ch_nr = self._validate_channel_selection(
None,
group,
index,
)
grp = self.groups[gp_nr]
if grp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
if self.memory == 'minimum':
channel = Channel(
address=grp['channels'][ch_nr],
stream=stream,
load_metadata=False,
)
else:
channel = grp['channels'][ch_nr]
return channel.name
def get_channel_metadata(
self,
name=None,
group=None,
index=None):
gp_nr, ch_nr = self._validate_channel_selection(
name,
group,
index,
)
grp = self.groups[gp_nr]
if grp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
channel = grp['channels'][ch_nr]
if self.memory == 'minimum':
channel = Channel(
address=channel,
stream=stream,
)
channel = deepcopy(channel)
return channel
[docs] def get_channel_unit(self, name=None, group=None, index=None):
"""Gets channel unit.
Channel can be specified in two ways:
* using the first positional argument *name*
* if there are multiple occurances for this channel then the
*group* and *index* arguments can be used to select a specific
group.
* if there are multiple occurances for this channel and either the
*group* or *index* arguments is None then a warning is issued
* using the group number (keyword argument *group*) and the channel
number (keyword argument *index*). Use *info* method for group and
channel numbers
If the *raster* keyword argument is not *None* the output is
interpolated accordingly.
Parameters
----------
name : string
name of channel
group : int
0-based group index
index : int
0-based channel index
Returns
-------
unit : str
found channel unit
"""
gp_nr, ch_nr = self._validate_channel_selection(
name,
group,
index,
)
grp = self.groups[gp_nr]
if grp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
if self.memory == 'minimum':
channel = Channel(
address=grp['channels'][ch_nr],
stream=stream,
)
else:
channel = grp['channels'][ch_nr]
if channel.conversion:
unit = channel.conversion.unit
else:
unit = ''
return unit
[docs] def get(self,
name=None,
group=None,
index=None,
raster=None,
samples_only=False,
data=None,
raw=False,
ignore_invalidation_bits=False):
"""Gets channel samples.
Channel can be specified in two ways:
* using the first positional argument *name*
* if there are multiple occurances for this channel then the
*group* and *index* arguments can be used to select a specific
group.
* if there are multiple occurances for this channel and either the
*group* or *index* arguments is None then a warning is issued
* using the group number (keyword argument *group*) and the channel
number (keyword argument *index*). Use *info* method for group and
channel numbers
If the *raster* keyword argument is not *None* the output is
interpolated accordingly.
Parameters
----------
name : string
name of channel
group : int
0-based group index
index : int
0-based channel index
raster : float
time raster in seconds
samples_only : bool
if *True* return only the channel samples as numpy array; if
*False* return a *Signal* object
data : bytes
prevent redundant data read by providing the raw data group samples
raw : bool
return channel samples without appling the conversion rule; default
`False`
ignore_invalidation_bits : bool
only defined to have the same API with the MDF v4
Returns
-------
res : (numpy.array, None) | Signal
returns *Signal* if *samples_only*=*False* (default option),
otherwise returns a (numpy.array, None) tuple (for compatibility
with MDF v4 class.
The *Signal* samples are
* numpy recarray for channels that have CDBLOCK or BYTEARRAY
type channels
* numpy array for all the rest
Raises
------
MdfException :
* if the channel name is not found
* if the group index is out of range
* if the channel index is out of range
Examples
--------
>>> from asammdf import MDF, Signal
>>> import numpy as np
>>> t = np.arange(5)
>>> s = np.ones(5)
>>> mdf = MDF(version='3.30')
>>> for i in range(4):
... sigs = [Signal(s*(i*10+j), t, name='Sig') for j in range(1, 4)]
... mdf.append(sigs)
...
>>> # first group and channel index of the specified channel name
...
>>> mdf.get('Sig')
UserWarning: Multiple occurances for channel "Sig". Using first occurance from data group 4. Provide both "group" and "index" arguments to select another data group
<Signal Sig:
samples=[ 1. 1. 1. 1. 1.]
timestamps=[0 1 2 3 4]
unit=""
info=None
comment="">
>>> # first channel index in the specified group
...
>>> mdf.get('Sig', 1)
<Signal Sig:
samples=[ 11. 11. 11. 11. 11.]
timestamps=[0 1 2 3 4]
unit=""
info=None
comment="">
>>> # channel named Sig from group 1 channel index 2
...
>>> mdf.get('Sig', 1, 2)
<Signal Sig:
samples=[ 12. 12. 12. 12. 12.]
timestamps=[0 1 2 3 4]
unit=""
info=None
comment="">
>>> # channel index 1 or group 2
...
>>> mdf.get(None, 2, 1)
<Signal Sig:
samples=[ 21. 21. 21. 21. 21.]
timestamps=[0 1 2 3 4]
unit=""
info=None
comment="">
>>> mdf.get(group=2, index=1)
<Signal Sig:
samples=[ 21. 21. 21. 21. 21.]
timestamps=[0 1 2 3 4]
unit=""
info=None
comment="">
"""
gp_nr, ch_nr = self._validate_channel_selection(
name,
group,
index,
)
original_data = data
memory = self.memory
grp = self.groups[gp_nr]
if grp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
channel = grp['channels'][ch_nr]
if memory == 'minimum':
channel = Channel(
address=grp['channels'][ch_nr],
stream=stream,
)
conversion = channel.conversion
name = channel.name
display_name = channel.display_name
bit_count = channel['bit_count'] or 64
dep = grp['channel_dependencies'][ch_nr]
cycles_nr = grp['channel_group']['cycles_nr']
# get data group record
if data is None:
data = self._load_group_data(grp)
else:
data = (data, )
# check if this is a channel array
if dep:
if dep['dependency_type'] == v23c.DEPENDENCY_TYPE_VECTOR:
shape = [dep['sd_nr'], ]
elif dep['dependency_type'] >= v23c.DEPENDENCY_TYPE_NDIM:
shape = []
i = 0
while True:
try:
dim = dep['dim_{}'.format(i)]
shape.append(dim)
i += 1
except KeyError:
break
shape = shape[::-1]
record_shape = tuple(shape)
arrays = [
self.get(group=dg_nr, index=ch_nr, samples_only=True, raw=raw, data=original_data)[0]
for ch_nr, dg_nr in dep.referenced_channels
]
if cycles_nr:
shape.insert(0, cycles_nr)
vals = column_stack(arrays).flatten().reshape(tuple(shape))
arrays = [vals, ]
types = [(channel.name, vals.dtype, record_shape), ]
if PYVERSION == 2:
types = fix_dtype_fields(types, 'latin-1')
types = dtype(types)
vals = fromarrays(arrays, dtype=types)
if not samples_only or raster:
timestamps = self.get_master(gp_nr, original_data)
if raster and len(timestamps):
t = arange(
timestamps[0],
timestamps[-1],
raster,
)
vals = Signal(
vals,
timestamps,
name='_',
).interp(t).samples
timestamps = t
else:
# get channel values
channel_values = []
timestamps = []
count = 0
for fragment in data:
data_bytes, _ = fragment
try:
parents, dtypes = grp['parents'], grp['types']
except KeyError:
grp['parents'], grp['types'] = self._prepare_record(grp)
parents, dtypes = grp['parents'], grp['types']
try:
parent, bit_offset = parents[ch_nr]
except KeyError:
parent, bit_offset = None, None
bits = channel['bit_count']
if parent is not None:
if 'record' not in grp:
if dtypes.itemsize:
record = fromstring(data_bytes, dtype=dtypes)
else:
record = None
if memory == 'full':
grp['record'] = record
else:
record = grp['record']
record.setflags(write=False)
vals = record[parent]
data_type = channel['data_type']
size = vals.dtype.itemsize
if data_type == v23c.DATA_TYPE_BYTEARRAY:
size *= vals.shape[1]
vals_dtype = vals.dtype.kind
if vals_dtype not in 'ui' and (bit_offset or not bits == size * 8):
vals = self._get_not_byte_aligned_data(data_bytes, grp, ch_nr)
else:
if bit_offset:
dtype_ = vals.dtype
if dtype_.kind == 'i':
vals = vals.astype(dtype('<u{}'.format(size)))
vals >>= bit_offset
else:
vals = vals >> bit_offset
if not bits == size * 8:
if data_type in v23c.SIGNED_INT:
vals = as_non_byte_sized_signed_int(vals, bits)
else:
mask = (1 << bits) - 1
if vals.flags.writeable:
vals &= mask
else:
vals = vals & mask
else:
vals = self._get_not_byte_aligned_data(data_bytes, grp, ch_nr)
if not samples_only or raster:
timestamps.append(self.get_master(gp_nr, fragment))
if bits == 1 and self._single_bit_uint_as_bool:
vals = array(vals, dtype=bool)
else:
data_type = channel['data_type']
channel_dtype = array([], dtype=get_fmt_v3(data_type, bits))
if vals.dtype != channel_dtype.dtype:
vals = vals.astype(channel_dtype.dtype)
channel_values.append(vals.copy())
count += 1
if count > 1:
vals = concatenate(channel_values)
elif count == 1:
vals = channel_values[0]
else:
vals = []
if not samples_only or raster:
if count > 1:
timestamps = concatenate(timestamps)
else:
timestamps = timestamps[0]
if raster and len(timestamps):
t = arange(
timestamps[0],
timestamps[-1],
raster,
)
vals = Signal(
vals,
timestamps,
name='_',
).interp(t).samples
timestamps = t
if conversion is None:
conversion_type = v23c.CONVERSION_TYPE_NONE
else:
conversion_type = conversion['conversion_type']
if conversion_type == v23c.CONVERSION_TYPE_NONE:
pass
elif conversion_type in (
v23c.CONVERSION_TYPE_LINEAR,
v23c.CONVERSION_TYPE_TABI,
v23c.CONVERSION_TYPE_TAB,
v23c.CONVERSION_TYPE_EXPO,
v23c.CONVERSION_TYPE_LOGH,
v23c.CONVERSION_TYPE_RAT,
v23c.CONVERSION_TYPE_POLY,
v23c.CONVERSION_TYPE_FORMULA):
if not raw:
try:
vals = conversion.convert(vals)
except:
print(channel, conversion)
raise
elif conversion_type in (
v23c.CONVERSION_TYPE_TABX,
v23c.CONVERSION_TYPE_RTABX):
raw = True
if samples_only:
res = vals, None
else:
if conversion:
unit = conversion.unit
else:
unit = ''
comment = channel.comment
description = (
channel['description']
.decode('latin-1')
.strip(' \t\n\0')
)
if comment:
comment = '{}\n{}'.format(comment, description)
else:
comment = description
source = channel.source
if source:
if source['type'] == v23c.SOURCE_ECU:
source = SignalSource(
source.name,
source.path,
source.comment,
0, # source type other
0, # bus type none
)
else:
source = SignalSource(
source.name,
source.path,
source.comment,
2, # source type bus
2, # bus type CAN
)
master_metadata = self._master_channel_metadata.get(gp_nr, None)
res = Signal(
samples=vals,
timestamps=timestamps,
unit=unit,
name=channel.name,
comment=comment,
conversion=conversion,
raw=raw,
master_metadata=master_metadata,
display_name=display_name,
source=source,
bit_count=bit_count,
)
return res
[docs] def get_master(self, index, data=None, raster=None):
""" returns master channel samples for given group
Parameters
----------
index : int
group index
data : (bytes, int)
(data block raw bytes, fragment offset); default None
raster : float
raster to be used for interpolation; default None
Returns
-------
t : numpy.array
master channel samples
"""
original_data = data
fragment = data
if fragment:
data_bytes, offset = fragment
try:
timestamps = self._master_channel_cache[(index, offset)]
if raster and timestamps:
timestamps = arange(
timestamps[0],
timestamps[-1],
raster,
)
return timestamps
except KeyError:
pass
else:
try:
timestamps = self._master_channel_cache[index]
if raster and timestamps:
timestamps = arange(
timestamps[0],
timestamps[-1],
raster,
)
return timestamps
except KeyError:
pass
group = self.groups[index]
if group['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
memory = self.memory
time_ch_nr = self.masters_db.get(index, None)
cycles_nr = group['channel_group']['cycles_nr']
if time_ch_nr is None:
t = arange(cycles_nr, dtype=float64)
metadata = (
'time',
1,
)
else:
time_ch = group['channels'][time_ch_nr]
if memory == 'minimum':
time_ch = Channel(
address=group['channels'][time_ch_nr],
stream=stream,
)
metadata = (
time_ch.name,
1,
)
if time_ch['bit_count'] == 0:
if time_ch['sampling_rate']:
sampling_rate = time_ch['sampling_rate']
else:
sampling_rate = 1
t = arange(cycles_nr, dtype=float64) * sampling_rate
else:
# get data group parents and dtypes
try:
parents, dtypes = group['parents'], group['types']
except KeyError:
parents, dtypes = self._prepare_record(group)
group['parents'], group['types'] = parents, dtypes
# get data group record
if data is None:
data = self._load_group_data(group)
else:
data = (data, )
time_values = []
count = 0
for fragment in data:
data_bytes, offset = fragment
parent, _ = parents.get(time_ch_nr, (None, None))
if parent is not None:
not_found = object()
record = group.get('record', not_found)
if record is not_found:
if dtypes.itemsize:
record = fromstring(data_bytes, dtype=dtypes)
else:
record = None
if memory == 'full':
group['record'] = record
record.setflags(write=False)
t = record[parent]
else:
t = self._get_not_byte_aligned_data(
data_bytes,
group,
time_ch_nr,
)
time_values.append(t.copy())
count += 1
if count > 1:
t = concatenate(time_values)
elif count == 1:
t = time_values[0]
else:
t = array([], dtype=float64)
# get timestamps
conversion = time_ch.conversion
if conversion is None:
time_conv_type = v23c.CONVERSION_TYPE_NONE
else:
time_conv_type = conversion['conversion_type']
if time_conv_type == v23c.CONVERSION_TYPE_LINEAR:
time_a = conversion['a']
time_b = conversion['b']
t = t * time_a
if time_b:
t += time_b
if not t.dtype == float64:
t = t.astype(float64)
self._master_channel_metadata[index] = metadata
if original_data is None:
self._master_channel_cache[index] = t
else:
data_bytes, offset = original_data
self._master_channel_cache[(index, offset)] = t
if raster and t.size:
timestamps = arange(
t[0],
t[-1],
raster,
)
else:
timestamps = t
return timestamps
[docs] def iter_get_triggers(self):
""" generator that yields triggers
Returns
-------
trigger_info : dict
trigger information with the following keys:
* comment : trigger comment
* time : trigger time
* pre_time : trigger pre time
* post_time : trigger post time
* index : trigger index
* group : data group index of trigger
"""
for i, gp in enumerate(self.groups):
trigger = gp['trigger']
if trigger:
for j in range(trigger['trigger_events_nr']):
trigger_info = {
'comment': trigger.comment,
'index': j,
'group': i,
'time': trigger['trigger_{}_time'.format(j)],
'pre_time': trigger['trigger_{}_pretime'.format(j)],
'post_time': trigger['trigger_{}_posttime'.format(j)],
}
yield trigger_info
[docs] def info(self):
"""get MDF information as a dict
Examples
--------
>>> mdf = MDF3('test.mdf')
>>> mdf.info()
"""
info = {}
for key in ('author',
'department',
'project',
'subject'):
value = self.header[key].decode('latin-1').strip(' \n\t\0')
info[key] = value
info['version'] = self.version
info['groups'] = len(self.groups)
for i, gp in enumerate(self.groups):
if gp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
elif gp['data_location'] == v23c.LOCATION_TEMPORARY_FILE:
stream = self._tempfile
inf = {}
info['group {}'.format(i)] = inf
inf['cycles'] = gp['channel_group']['cycles_nr']
inf['channels count'] = len(gp['channels'])
for j, channel in enumerate(gp['channels']):
if self.memory == 'minimum':
channel = Channel(
address=channel,
stream=stream,
)
name = channel.name
if channel['channel_type'] == v23c.CHANNEL_TYPE_MASTER:
ch_type = 'master'
else:
ch_type = 'value'
inf['channel {}'.format(j)] = 'name="{}" type={}'.format(
name,
ch_type,
)
return info
[docs] def save(self, dst='', overwrite=False, compression=0):
"""Save MDF to *dst*. If *dst* is not provided the the destination file
name is the MDF name. If overwrite is *True* then the destination file
is overwritten, otherwise the file name is appended with '_<cntr>',
were '<cntr>' is the first counter that produces a new file name (that
does not already exist in the filesystem).
Parameters
----------
dst : str
destination file name, Default ''
overwrite : bool
overwrite flag, default *False*
compression : int
does nothing for mdf version3; introduced here to share the same
API as mdf version 4 files
Returns
-------
output_file : str
output file name
"""
if self.name is None and dst == '':
message = ('Must specify a destination file name '
'for MDF created from scratch')
raise MdfException(message)
destination_dir = os.path.dirname(dst)
if destination_dir and not os.path.exists(destination_dir):
os.makedirs(destination_dir)
if self.memory == 'minimum':
output_file = self._save_without_metadata(
dst,
overwrite,
compression,
)
else:
output_file = self._save_with_metadata(
dst,
overwrite,
compression,
)
if self._callback:
self._callback(100, 100)
return output_file
def _save_with_metadata(self, dst, overwrite, compression):
"""Save MDF to *dst*. If *dst* is not provided the the destination file
name is the MDF name. If overwrite is *True* then the destination file
is overwritten, otherwise the file name is appended with '_<cntr>',
were '<cntr>' is the first counter that produces a new file name (that
does not already exist in the filesystem).
Parameters
----------
dst : str
destination file name, Default ''
overwrite : bool
overwrite flag, default *False*
compression : int
does nothing for mdf version3; introduced here to share the same
API as mdf version 4 files
"""
# pylint: disable=unused-argument
if not self.header.comment:
self.header.comment = '''<FHcomment>
<TX>created</TX>
<tool_id>asammdf</tool_id>
<tool_vendor> </tool_vendor>
<tool_version>{}</tool_version>
</FHcomment>'''.format(__version__)
else:
text = '{}\n{}: updated by asammdf {}'
old_history = self.header.comment
timestamp = time.asctime().encode('latin-1')
text = text.format(
old_history,
timestamp,
__version__,
)
self.header.comment = text
if self.name is None and dst == '':
message = (
'Must specify a destination file name '
'for MDF created from scratch'
)
raise MdfException(message)
defined_texts, cc_map, si_map = {}, {}, {}
dst = dst if dst else self.name
if not dst.endswith(('mdf', 'MDF')):
dst = dst + '.mdf'
if overwrite is False:
if os.path.isfile(dst):
cntr = 0
while True:
name = os.path.splitext(dst)[0] + '_{}.mdf'.format(cntr)
if not os.path.isfile(name):
break
else:
cntr += 1
message = (
'Destination file "{}" already exists '
'and "overwrite" is False. Saving MDF file as "{}"'
)
message = message.format(dst, name)
logger.warning(message)
dst = name
# all MDF blocks are appended to the blocks list in the order in which
# they will be written to disk. While creating this list, all the
# relevant block links are updated so that once all blocks have been
# added to the list they can be written using the bytes protocol.
# DataGroup blocks are written first after the identification and
# header blocks. When memory='low' we need to restore the
# original data block addresses within the data group block. This is
# needed to allow further work with the object after the save method
# call (eq. new calls to get method). Since the data group blocks are
# written first, it is safe to restor the original links when the data
# blocks are written. For memory=False the blocks list will
# contain a tuple instead of a DataBlock instance; the tuple will have
# the reference to the data group object and the original link to the
# data block in the soource MDF file.
if self.memory == 'low' and dst == self.name:
destination = dst + '.temp'
else:
destination = dst
with open(destination, 'wb+') as dst_:
groups_nr = len(self.groups)
write = dst_.write
seek = dst_.seek
# list of all blocks
blocks = []
address = 0
write(bytes(self.identification))
address += v23c.ID_BLOCK_SIZE
write(bytes(self.header))
address += self.header['block_len']
if self.header.program:
write(bytes(self.header.program))
self.header['program_addr'] = address
address += self.header.program['block_len']
else:
self.header['program_addr'] = 0
comment = TextBlock(text=self.header.comment)
write(bytes(comment))
self.header['comment_addr'] = address
address += comment['block_len']
# DataGroup
# put them first in the block list so they will be written first to
# disk this way, in case of memory=False, we can safely
# restore he original data block address
gp_rec_ids = []
original_data_block_addrs = [
group['data_group']['data_block_addr']
for group in self.groups
]
for idx, gp in enumerate(self.groups):
dg = gp['data_group']
gp_rec_ids.append(dg['record_id_len'])
dg['record_id_len'] = 0
# DataBlock
for (data_bytes, _) in self._load_group_data(gp):
if self.memory == 'full':
data = memoryview(data_bytes)
read_size = 4 * 2**20
count = int(ceil(len(data_bytes) / read_size))
for j in range(count):
write(data[j*read_size: (j+1)*read_size])
else:
write(data_bytes)
if gp['size']:
gp['data_group']['data_block_addr'] = address
else:
gp['data_group']['data_block_addr'] = 0
address += gp['size'] - gp_rec_ids[idx] * gp['channel_group']['cycles_nr']
if self._callback:
self._callback(int(33 * (idx+1) / groups_nr), 100)
if self._terminate:
dst_.close()
self.close()
return
for gp in self.groups:
dg = gp['data_group']
blocks.append(dg)
dg.address = address
address += dg['block_len']
if self.groups:
for i, dg in enumerate(self.groups[:-1]):
addr = self.groups[i + 1]['data_group'].address
dg['data_group']['next_dg_addr'] = addr
self.groups[-1]['data_group']['next_dg_addr'] = 0
for idx, gp in enumerate(self.groups):
# Channel Dependency
cd = gp['channel_dependencies']
for dep in cd:
if dep:
dep.address = address
blocks.append(dep)
address += dep['block_len']
for channel, dep in zip(gp['channels'], gp['channel_dependencies']):
if dep:
channel['ch_depend_addr'] = dep.address = address
blocks.append(dep)
address += dep['block_len']
else:
channel['ch_depend_addr'] = 0
address = channel.to_blocks(address, blocks, defined_texts, cc_map, si_map)
count = len(gp['channels'])
if count:
for i in range(count-1):
gp['channels'][i]['next_ch_addr'] = gp['channels'][i+1].address
gp['channels'][-1]['next_ch_addr'] = 0
# ChannelGroup
cg = gp['channel_group']
if gp['channels']:
cg['first_ch_addr'] = gp['channels'][0].address
else:
cg['first_ch_addr'] = 0
cg['next_cg_addr'] = 0
address = cg.to_blocks(address, blocks, defined_texts, si_map)
# TriggerBLock
trigger = gp['trigger']
if trigger:
address = trigger.to_blocks(address, blocks)
if self._callback:
self._callback(int(33 * (idx+1) / groups_nr) + 33, 100)
if self._terminate:
dst_.close()
self.close()
return
# update referenced channels addresses in the channel dependecies
for gp in self.groups:
for dep in gp['channel_dependencies']:
if not dep:
continue
for i, pair_ in enumerate(dep.referenced_channels):
ch_nr, dg_nr = pair_
grp = self.groups[dg_nr]
ch = grp['channels'][ch_nr]
dep['ch_{}'.format(i)] = ch.address
dep['cg_{}'.format(i)] = grp['channel_group'].address
dep['dg_{}'.format(i)] = grp['data_group'].address
# DataGroup
for gp in self.groups:
gp['data_group']['first_cg_addr'] = gp['channel_group'].address
if gp['trigger']:
gp['data_group']['trigger_addr'] = gp['trigger'].address
else:
gp['data_group']['trigger_addr'] = 0
if self.groups:
address = self.groups[0]['data_group'].address
self.header['first_dg_addr'] = address
self.header['dg_nr'] = len(self.groups)
if self._terminate:
dst_.close()
self.close()
return
if self._callback:
blocks_nr = len(blocks)
threshold = blocks_nr / 33
count = 1
for i, block in enumerate(blocks):
write(bytes(block))
if i >= threshold:
self._callback(66 + count, 100)
count += 1
threshold += blocks_nr / 33
else:
for block in blocks:
write(bytes(block))
for gp, rec_id, original_address in zip(
self.groups,
gp_rec_ids,
original_data_block_addrs):
gp['data_group']['record_id_len'] = rec_id
gp['data_group']['data_block_addr'] = original_address
seek(0)
write(bytes(self.identification))
write(bytes(self.header))
if self.memory == 'low' and dst == self.name:
self.close()
os.remove(self.name)
os.rename(destination, self.name)
self.groups = []
self.header = None
self.identification = None
self.channels_db = {}
self.masters_db = {}
self._master_channel_cache = {}
self._tempfile = TemporaryFile()
self._file = open(self.name, 'rb')
self._read()
return dst
def _save_without_metadata(self, dst, overwrite, compression):
"""Save MDF to *dst*. If *dst* is not provided the the destination file
name is the MDF name. If overwrite is *True* then the destination file
is overwritten, otherwise the file name is appended with '_<cntr>',
were '<cntr>' is the first counter that produces a new file name (that
does not already exist in the filesystem).
Parameters
----------
dst : str
destination file name, Default ''
overwrite : bool
overwrite flag, default *False*
compression : int
does nothing for mdf version3; introduced here to share the same
API as mdf version 4 files
"""
# pylint: disable=unused-argument
if not self.header.comment:
self.header.comment = '''<FHcomment>
<TX>created</TX>
<tool_id>asammdf</tool_id>
<tool_vendor> </tool_vendor>
<tool_version>{}</tool_version>
</FHcomment>'''.format(__version__)
else:
text = '{}\n{}: updated by asammdf {}'
old_history = self.header.comment
timestamp = time.asctime().encode('latin-1')
text = text.format(
old_history,
timestamp,
__version__,
)
self.header.comment = text
if self.name is None and dst == '':
message = (
'Must specify a destination file name '
'for MDF created from scratch'
)
raise MdfException(message)
defined_texts, cc_map, si_map = {}, {}, {}
dst = dst if dst else self.name
if not dst.endswith(('mdf', 'MDF')):
dst = dst + '.mdf'
if overwrite is False:
if os.path.isfile(dst):
cntr = 0
while True:
name = os.path.splitext(dst)[0] + '_{}.mdf'.format(cntr)
if not os.path.isfile(name):
break
else:
cntr += 1
message = (
'Destination file "{}" already exists '
'and "overwrite" is False. Saving MDF file as "{}"'
)
message = message.format(dst, name)
logger.warning(message)
dst = name
if dst == self.name:
destination = dst + '.temp'
else:
destination = dst
with open(destination, 'wb+') as dst_:
groups_nr = len(self.groups)
write = dst_.write
tell = dst_.tell
seek = dst_.seek
# list of all blocks
blocks = []
address = 0
write(bytes(self.identification))
self.header.to_stream(dst_, defined_texts, si_map)
# DataGroup
# put them first in the block list so they will be written first to
# disk this way, in case of memory=False, we can safely
# restore he original data block address
data_address = []
for idx, gp in enumerate(self.groups):
gp['temp_channels'] = ch_addrs = []
gp['temp_channel_dependencies'] = cd_addrs = []
if gp['data_location'] == v23c.LOCATION_ORIGINAL_FILE:
stream = self._file
else:
stream = self._tempfile
# Channel Dependency
for dep in gp['channel_dependencies']:
if dep:
address = tell()
cd_addrs.append(address)
write(bytes(dep))
else:
cd_addrs.append(0)
next_ch_addr = 0
size = len(gp['channels'])
for i in range(size-1, -1, -1):
channel = gp['channels'][i]
channel = Channel(
address=channel,
stream=stream,
)
channel['next_ch_addr'] = next_ch_addr
channel['ch_depend_addr'] = cd_addrs[i]
address = channel.to_stream(dst_, defined_texts, cc_map, si_map)
ch_addrs.append(channel.address)
next_ch_addr = channel.address
address = tell()
ch_addrs.reverse()
# ChannelGroup
cg = gp['channel_group']
cg.address = address
cg['next_cg_addr'] = 0
if ch_addrs:
cg['first_ch_addr'] = ch_addrs[0]
else:
cg['first_ch_addr'] = 0
address = cg.to_stream(dst_, defined_texts, si_map)
# TriggerBLock
trigger = gp['trigger']
if trigger:
address = trigger.to_stream(dst_)
# DataBlock
data = self._load_group_data(gp)
dat_addr = tell()
for (data_bytes, _) in data:
write(data_bytes)
if tell() - dat_addr:
data_address.append(dat_addr)
else:
data_address.append(0)
if self._callback:
self._callback(int(100 * (idx+1) / groups_nr), 100)
if self._terminate:
dst_.close()
self.close()
return
orig_addr = [
gp['data_group']['data_block_addr']
for gp in self.groups
]
address = tell()
gp_rec_ids = []
for i, gp in enumerate(self.groups):
dg = gp['data_group']
gp_rec_ids.append(dg['record_id_len'])
dg['record_id_len'] = 0
dg['data_block_addr'] = data_address[i]
dg.address = address
address += dg['block_len']
gp['data_group']['first_cg_addr'] = gp['channel_group'].address
if gp['trigger']:
gp['data_group']['trigger_addr'] = gp['trigger'].address
else:
gp['data_group']['trigger_addr'] = 0
if self.groups:
for i, gp in enumerate(self.groups[:-1]):
addr = self.groups[i + 1]['data_group'].address
gp['data_group']['next_dg_addr'] = addr
self.groups[-1]['data_group']['next_dg_addr'] = 0
for i, gp in enumerate(self.groups):
write(bytes(gp['data_group']))
gp['data_block_addr'] = orig_addr[i]
for gp, rec_id in zip(self.groups, gp_rec_ids):
gp['data_group']['record_id_len'] = rec_id
if self.groups:
address = self.groups[0]['data_group'].address
self.header['first_dg_addr'] = address
self.header['dg_nr'] = len(self.groups)
# update referenced channels addresses in the channel dependecies
for gp in self.groups:
for dep in gp['channel_dependencies']:
if not dep:
continue
for i, pair_ in enumerate(dep.referenced_channels):
_, dg_nr = pair_
grp = self.groups[dg_nr]
dep['ch_{}'.format(i)] = grp['temp_channels'][i]
dep['cg_{}'.format(i)] = grp['channel_group'].address
dep['dg_{}'.format(i)] = grp['data_group'].address
seek(dep.address)
write(bytes(dep))
seek(v23c.ID_BLOCK_SIZE)
self.header.to_stream(dst_, defined_texts, si_map)
for gp in self.groups:
del gp['temp_channels']
if dst == self.name:
self.close()
os.remove(self.name)
os.rename(destination, self.name)
self.groups = []
self.header = None
self.identification = None
self.channels_db = {}
self.masters_db = {}
self._master_channel_cache = {}
self._tempfile = TemporaryFile()
self._file = open(self.name, 'rb')
self._read()
return dst
if __name__ == '__main__':
pass