# -*- coding: utf-8 -*-
"""
Mini-SEED specific utilities.
"""
from __future__ import (absolute_import, division, print_function,
unicode_literals)
from future.builtins import * # NOQA
from future.utils import native_str
import collections
import ctypes as C
import math
import os
import sys
import warnings
from datetime import datetime
from struct import pack, unpack
import numpy as np
from obspy import UTCDateTime
from obspy.core.util import score_at_percentile
from obspy.core.util.decorator import deprecated
from .headers import (ENCODINGS, ENDIAN, FIXED_HEADER_ACTIVITY_FLAGS,
FIXED_HEADER_DATA_QUAL_FLAGS,
FIXED_HEADER_IO_CLOCK_FLAGS, FRAME, HPTMODULUS,
SAMPLESIZES, UNSUPPORTED_ENCODINGS, clibmseed)
@deprecated("'getStartAndEndTime' has been renamed to "
"'get_start_and_end_time'. Use that instead.") # noqa
[docs]def getStartAndEndTime(*args, **kwargs):
return get_start_and_end_time(*args, **kwargs)
[docs]def get_start_and_end_time(file_or_file_object):
"""
Returns the start and end time of a Mini-SEED file or file-like object.
:type file_or_file_object: str or file
:param file_or_file_object: Mini-SEED file name or open file-like object
containing a Mini-SEED record.
:return: tuple (start time of first record, end time of last record)
This method will return the start time of the first record and the end time
of the last record. Keep in mind that it will not return the correct result
if the records in the Mini-SEED file do not have a chronological ordering.
The returned end time is the time of the last data sample and not the
time that the last sample covers.
.. rubric:: Example
>>> from obspy.core.util import get_example_file
>>> filename = get_example_file(
... "BW.BGLD.__.EHE.D.2008.001.first_10_records")
>>> get_start_and_end_time(filename) # doctest: +NORMALIZE_WHITESPACE
(UTCDateTime(2007, 12, 31, 23, 59, 59, 915000),
UTCDateTime(2008, 1, 1, 0, 0, 20, 510000))
It also works with an open file pointer. The file pointer itself will not
be changed.
>>> f = open(filename, 'rb')
>>> get_start_and_end_time(f) # doctest: +NORMALIZE_WHITESPACE
(UTCDateTime(2007, 12, 31, 23, 59, 59, 915000),
UTCDateTime(2008, 1, 1, 0, 0, 20, 510000))
And also with a Mini-SEED file stored in a BytesIO
>>> import io
>>> file_object = io.BytesIO(f.read())
>>> get_start_and_end_time(file_object) # doctest: +NORMALIZE_WHITESPACE
(UTCDateTime(2007, 12, 31, 23, 59, 59, 915000),
UTCDateTime(2008, 1, 1, 0, 0, 20, 510000))
>>> file_object.close()
If the file pointer does not point to the first record, the start time will
refer to the record it points to.
>>> _ = f.seek(512)
>>> get_start_and_end_time(f) # doctest: +NORMALIZE_WHITESPACE
(UTCDateTime(2008, 1, 1, 0, 0, 1, 975000),
UTCDateTime(2008, 1, 1, 0, 0, 20, 510000))
The same is valid for a file-like object.
>>> file_object = io.BytesIO(f.read())
>>> get_start_and_end_time(file_object) # doctest: +NORMALIZE_WHITESPACE
(UTCDateTime(2008, 1, 1, 0, 0, 1, 975000),
UTCDateTime(2008, 1, 1, 0, 0, 20, 510000))
>>> f.close()
"""
# Get the starttime of the first record.
info = get_record_information(file_or_file_object)
starttime = info['starttime']
# Get the end time of the last record.
info = get_record_information(
file_or_file_object,
(info['number_of_records'] - 1) * info['record_length'])
endtime = info['endtime']
return starttime, endtime
@deprecated("'getTimingAndDataQuality' has been renamed to "
"'get_timing_and_data_quality'. Use that instead.") # noqa
[docs]def getTimingAndDataQuality(*args, **kwargs):
return get_timing_and_data_quality(*args, **kwargs)
[docs]def get_timing_and_data_quality(file_or_file_object):
"""
Counts all data quality flags of the given Mini-SEED file and returns
statistics about the timing quality if applicable.
:type file_or_file_object: str or file
:param file_or_file_object: Mini-SEED file name or open file-like object
containing a Mini-SEED record.
:return: Dictionary with information about the timing quality and the data
quality flags.
.. rubric:: Data quality
This method will count all set data quality flag bits in the fixed section
of the data header in a Mini-SEED file and returns the total count for each
flag type.
======== =================================================
Bit Description
======== =================================================
[Bit 0] Amplifier saturation detected (station dependent)
[Bit 1] Digitizer clipping detected
[Bit 2] Spikes detected
[Bit 3] Glitches detected
[Bit 4] Missing/padded data present
[Bit 5] Telemetry synchronization error
[Bit 6] A digital filter may be charging
[Bit 7] Time tag is questionable
======== =================================================
.. rubric:: Timing quality
If the file has a Blockette 1001 statistics about the timing quality will
also be returned. See the doctests for more information.
This method will read the timing quality in Blockette 1001 for each
record in the file if available and return the following statistics:
Minima, maxima, average, median and upper and lower quantile.
Quantiles are calculated using a integer round outwards policy: lower
quantiles are rounded down (probability < 0.5), and upper quantiles
(probability > 0.5) are rounded up.
This gives no more than the requested probability in the tails, and at
least the requested probability in the central area.
The median is calculating by either taking the middle value or, with an
even numbers of values, the average between the two middle values.
.. rubric:: Example
>>> from obspy.core.util import get_example_file
>>> filename = get_example_file("qualityflags.mseed")
>>> tq = get_timing_and_data_quality(filename)
>>> for k, v in tq.items():
... print(k, v)
data_quality_flags [9, 8, 7, 6, 5, 4, 3, 2]
Also works with file pointers and BytesIOs.
>>> f = open(filename, 'rb')
>>> tq = get_timing_and_data_quality(f)
>>> for k, v in tq.items():
... print(k, v)
data_quality_flags [9, 8, 7, 6, 5, 4, 3, 2]
>>> import io
>>> file_object = io.BytesIO(f.read())
>>> f.close()
>>> tq = get_timing_and_data_quality(file_object)
>>> for k, v in tq.items():
... print(k, v)
data_quality_flags [9, 8, 7, 6, 5, 4, 3, 2]
If the file pointer or BytesIO position does not correspond to the first
record the omitted records will be skipped.
>>> _ = file_object.seek(1024, 1)
>>> tq = get_timing_and_data_quality(file_object)
>>> for k, v in tq.items():
... print(k, v)
data_quality_flags [8, 8, 7, 6, 5, 4, 3, 2]
>>> file_object.close()
Reading a file with Blockette 1001 will return timing quality statistics.
The data quality flags will always exists because they are part of the
fixed Mini-SEED header and therefore need to be in every Mini-SEED file.
>>> filename = get_example_file("timingquality.mseed")
>>> tq = get_timing_and_data_quality(filename)
>>> for k, v in sorted(tq.items()):
... print(k, v)
data_quality_flags [0, 0, 0, 0, 0, 0, 0, 0]
timing_quality_average 50.0
timing_quality_lower_quantile 25.0
timing_quality_max 100.0
timing_quality_median 50.0
timing_quality_min 0.0
timing_quality_upper_quantile 75.0
Also works with file pointers and BytesIOs.
>>> f = open(filename, 'rb')
>>> tq = get_timing_and_data_quality(f)
>>> for k, v in sorted(tq.items()):
... print(k, v)
data_quality_flags [0, 0, 0, 0, 0, 0, 0, 0]
timing_quality_average 50.0
timing_quality_lower_quantile 25.0
timing_quality_max 100.0
timing_quality_median 50.0
timing_quality_min 0.0
timing_quality_upper_quantile 75.0
>>> file_object = io.BytesIO(f.read())
>>> f.close()
>>> tq = get_timing_and_data_quality(file_object)
>>> for k, v in sorted(tq.items()):
... print(k, v)
data_quality_flags [0, 0, 0, 0, 0, 0, 0, 0]
timing_quality_average 50.0
timing_quality_lower_quantile 25.0
timing_quality_max 100.0
timing_quality_median 50.0
timing_quality_min 0.0
timing_quality_upper_quantile 75.0
>>> file_object.close()
"""
# Read the first record to get a starting point and.
info = get_record_information(file_or_file_object)
# Keep track of the extracted information.
quality_count = [0, 0, 0, 0, 0, 0, 0, 0]
timing_quality = []
offset = 0
# Loop over each record. A valid record needs to have a record length of at
# least 256 bytes.
while offset <= (info['filesize'] - 256):
this_info = get_record_information(file_or_file_object, offset)
# Add the timing quality.
if 'timing_quality' in this_info:
timing_quality.append(float(this_info['timing_quality']))
# Add the value of each bit to the quality_count.
for _i in range(8):
if (this_info['data_quality_flags'] & (1 << _i)) != 0:
quality_count[_i] += 1
offset += this_info['record_length']
# Collect the results in a dictionary.
result = {'data_quality_flags': quality_count}
# Parse of the timing quality list.
count = len(timing_quality)
timing_quality = sorted(timing_quality)
# If no timing_quality was collected just return an empty dictionary.
if count == 0:
return result
# Otherwise calculate some statistical values from the timing quality.
result['timing_quality_min'] = min(timing_quality)
result['timing_quality_max'] = max(timing_quality)
result['timing_quality_average'] = sum(timing_quality) / count
result['timing_quality_median'] = \
score_at_percentile(timing_quality, 50, issorted=False)
result['timing_quality_lower_quantile'] = \
score_at_percentile(timing_quality, 25, issorted=False)
result['timing_quality_upper_quantile'] = \
score_at_percentile(timing_quality, 75, issorted=False)
return result
@deprecated("'getRecordInformation' has been renamed to "
"'get_record_information'. Use that instead.") # noqa
[docs]def _ctypes_array_2_numpy_array(buffer_, buffer_elements, sampletype):
"""
Takes a Ctypes array and its length and type and returns it as a
NumPy array.
:param buffer_: Ctypes c_void_p pointer to buffer.
:param buffer_elements: length of the whole buffer
:param sampletype: type of sample, on of "a", "i", "f", "d"
"""
# Allocate NumPy array to move memory to
numpy_array = np.empty(buffer_elements, dtype=sampletype)
datptr = numpy_array.ctypes.get_data()
# Manually copy the contents of the C allocated memory area to
# the address of the previously created NumPy array
C.memmove(datptr, buffer_, buffer_elements * SAMPLESIZES[sampletype])
return numpy_array
[docs]def _convert_msr_to_dict(m):
"""
Internal method used for setting header attributes.
"""
h = {}
attributes = ('network', 'station', 'location', 'channel',
'dataquality', 'starttime', 'samprate',
'samplecnt', 'numsamples', 'sampletype')
# loop over attributes
for _i in attributes:
h[_i] = getattr(m, _i)
return h
[docs]def _convert_datetime_to_mstime(dt):
"""
Takes a obspy.util.UTCDateTime object and returns an epoch time in ms.
:param dt: obspy.util.UTCDateTime object.
"""
_fsec, _sec = math.modf(dt.timestamp)
return int(round(_fsec * HPTMODULUS)) + int(_sec * HPTMODULUS)
[docs]def _convert_mstime_to_datetime(timestring):
"""
Takes a Mini-SEED timestamp and returns a obspy.util.UTCDateTime object.
:param timestamp: Mini-SEED timestring (Epoch time string in ms).
"""
return UTCDateTime(timestring / HPTMODULUS)
[docs]def _unpack_steim_1(data_string, npts, swapflag=0, verbose=0):
"""
Unpack steim1 compressed data given as string.
:param data_string: data as string
:param npts: number of data points
:param swapflag: Swap bytes, defaults to 0
:return: Return data as numpy.ndarray of dtype int32
"""
dbuf = data_string
datasize = len(dbuf)
samplecnt = npts
datasamples = np.empty(npts, dtype=np.int32)
diffbuff = np.empty(npts, dtype=np.int32)
x0 = C.c_int32()
xn = C.c_int32()
nsamples = clibmseed.msr_unpack_steim1(
C.cast(dbuf, C.POINTER(FRAME)), datasize,
samplecnt, samplecnt, datasamples, diffbuff,
C.byref(x0), C.byref(xn), swapflag, verbose)
if nsamples != npts:
raise Exception("Error in unpack_steim1")
return datasamples
[docs]def _unpack_steim_2(data_string, npts, swapflag=0, verbose=0):
"""
Unpack steim2 compressed data given as string.
:param data_string: data as string
:param npts: number of data points
:param swapflag: Swap bytes, defaults to 0
:return: Return data as numpy.ndarray of dtype int32
"""
dbuf = data_string
datasize = len(dbuf)
samplecnt = npts
datasamples = np.empty(npts, dtype=np.int32)
diffbuff = np.empty(npts, dtype=np.int32)
x0 = C.c_int32()
xn = C.c_int32()
nsamples = clibmseed.msr_unpack_steim2(
C.cast(dbuf, C.POINTER(FRAME)), datasize,
samplecnt, samplecnt, datasamples, diffbuff,
C.byref(x0), C.byref(xn), swapflag, verbose)
if nsamples != npts:
raise Exception("Error in unpack_steim2")
return datasamples
[docs]def _check_flag_value(flag_value):
"""
Search for a given flag in a given blockette for the current record.
This is a utility function for set_flags_in_fixed_headers and is not
designed to be called by someone else.
This function checks for valid entries for a flag. A flag can be either
* ``bool`` value to be always True or False for all the records
* ``datetime`` or ``UTCDateTime`` value to add a single 'INSTANT' datation
(see below)
* ``dict`` to allow complex flag datation
** The dict keys may be the keyword INSTANT to mark arbitrarly short
duration flags, or the keyword DURATION to mark events that span across
time.
** The dict values are:
*** for the INSTANT value, a single UTCDateTime or datetime object, or a
list of these datation objects
*** for the DURATION value, either a list of
[start1, end1, start2, end2, ...] or a list of tuples
[(start1, end1), (start2, end2), ...]
This function then returns all datation events as a list of tuples
[(start1, end1), ...] to ease the work of _convert_flags_to_raw_byte. Bool
values are unchanged, instant events become a tuple
(event_date, event_date).
If the flag value is incorrect, a ValueError is raised with a (hopefully)
explicit enough message.
:type flag_value: bool or dict
:param flag_value: the flag value to check.
:return: corrected value of the flag.
:raises: If the flag is not the one expected, a ``ValueError`` is raised
"""
if isinstance(flag_value, bool):
# bool allowed
corrected_flag = flag_value
elif isinstance(flag_value, datetime) or \
isinstance(flag_value, UTCDateTime):
# A single instant value is allowed
utc_val = UTCDateTime(flag_value)
corrected_flag = [(utc_val, utc_val)]
elif isinstance(flag_value, collections.Mapping):
# dict allowed if it has the right format
corrected_flag = []
for flag_key in flag_value:
if flag_key == "INSTANT":
# Expected: list of UTCDateTime
inst_values = flag_value[flag_key]
if isinstance(inst_values, datetime) or \
isinstance(inst_values, UTCDateTime):
# Single value : ensure it's UTCDateTime and store it
utc_val = UTCDateTime(inst_values)
corrected_flag.append((utc_val, utc_val))
elif isinstance(inst_values, collections.Sequence):
# Several instant values : check their types
# and add each of them
for value in inst_values:
if isinstance(value, datetime) or \
isinstance(value, UTCDateTime):
utc_val = UTCDateTime(value)
corrected_flag.append((utc_val, utc_val))
else:
msg = "Unexpected type for flag duration " +\
"'INSTANT' %s"
raise ValueError(msg % str(type(inst_values)))
else:
msg = "Unexpected type for flag duration 'INSTANT' %s"
raise ValueError(msg % str(type(inst_values)))
elif flag_key == "DURATION":
# Expecting either a list of tuples (start, end) or
# a list of (start1, end1, start1, end1)
dur_values = flag_value[flag_key]
if isinstance(dur_values, collections.Sequence):
if len(dur_values) != 0:
# Check first item
if isinstance(dur_values[0], datetime) or \
isinstance(dur_values[0], UTCDateTime):
# List of [start1, end1, start2, end2, etc]
# Check len
if len(dur_values) % 2 != 0:
msg = "Expected even length of duration " +\
"values, got %s"
raise ValueError(msg % len(dur_values))
# Add values
duration_iter = iter(dur_values)
for value in duration_iter:
start = value
end = dur_values[dur_values.index(value) + 1]
# Check start type
if not isinstance(start, datetime) and \
not isinstance(start, UTCDateTime):
msg = "Incorrect type for duration " +\
"start %s"
raise ValueError(msg % str(type(start)))
# Check end type
if not isinstance(end, datetime) and \
not isinstance(end, UTCDateTime):
msg = "Incorrect type for duration " +\
"end %s"
raise ValueError(msg % str(type(end)))
# Check duration validity
start = UTCDateTime(start)
end = UTCDateTime(end)
if start <= end:
corrected_flag.append((start, end))
else:
msg = "Flag datation: expected end of " +\
"duration after its start"
raise ValueError(msg)
next(duration_iter)
elif isinstance(dur_values[0], collections.Sequence):
# List of tuples (start, end)
for value in dur_values:
if not isinstance(value, collections.Sequence):
msg = "Incorrect type %s for flag duration"
raise ValueError(msg % str(type(value)))
elif len(value) != 2:
msg = "Incorrect len %s for flag duration"
raise ValueError(msg % len(value))
else:
start = value[0]
end = value[1]
# Check start type
if not isinstance(start, datetime) and \
not isinstance(start, UTCDateTime):
msg = "Incorrect type for duration " +\
"start %s"
raise ValueError(msg %
str(type(start)))
# Check end type
if not isinstance(end, datetime) and \
not isinstance(end, UTCDateTime):
msg = "Incorrect type for duration " +\
"end %s"
raise ValueError(msg % str(type(end)))
if start <= end:
corrected_flag.append((start, end))
else:
msg = "Flag datation: expected end " +\
"of duration after its start"
raise ValueError(msg)
# Else: len(dur_values) == 0, empty duration list:
# do nothing
else:
msg = "Incorrect DURATION value: expected a list of " +\
"tuples (start, end), got %s"
raise ValueError(msg % str(type(dur_values)))
else:
msg = "Invalid key %s for flag value. One of " +\
"'INSTANT', 'DURATION' is expected."
raise ValueError(msg % flag_key)
else:
msg = "Invalid type %s for flag value. Allowed values " +\
"are bool or dict"
raise ValueError(msg % str(type(flag_value)))
return corrected_flag
[docs]def _search_flag_in_blockette(mseed_file_desc, first_blockette_offset,
blockette_number, field_offset, field_length):
"""
Search for a given flag in a given blockette for the current record.
This is a utility function for set_flags_in_fixed_headers and is not
designed to be called by someone else.
This function uses the file descriptor``mseed_file_desc``, seeks
``first_blockette_offset`` to go to the first blockette, reads through all
the blockettes until it finds the one with number ``blockette_number``,
then skips ``field_offset`` bytes to read ``field_length`` bytes and
returns them. If the blockette does not exist, it returns None
Please note that this function does not decommute the binary value into an
exploitable data (int, float, string, ...)
:type mseed_file_desc: File object
:param mseed_file_desc: a File descriptor to the current miniseed file.
The value of mseed_file_desc.tell() is set back by this funcion before
returning, use in multithread applications at your own risk.
:type first_blockette_offset: int
:param first_blockette_offset: tells the function where the first blockette
of the record is compared to the mseed_file_desc current position in the
file. A positive value means the blockette is after the current position.
:type blockette_number: int
:param blockette_number: the blockette number to search for
:type field_offset: int
:param field_offset: how many bytes we have to skip before attaining the
wanted field. Please note that it also counts blockette number and next
blockette index's field.
:type field_length: int
:param field_length: length of the wanted field, in bytes
:return: bytes containing the field's value in this record
"""
previous_position = mseed_file_desc.tell()
try:
# Go to first blockette
mseed_file_desc.seek(first_blockette_offset, os.SEEK_CUR)
mseed_record_start = mseed_file_desc.tell() - 48
read_data = mseed_file_desc.read(4)
# Read info in the first blockette
[cur_blkt_number, next_blkt_offset] = unpack(native_str(">HH"),
read_data)
while cur_blkt_number != blockette_number \
and next_blkt_offset != 0:
# Nothing here, read next blockette
mseed_file_desc.seek(mseed_record_start + next_blkt_offset,
os.SEEK_SET)
read_data = mseed_file_desc.read(4)
[cur_blkt_number, next_blkt_offset] = unpack(native_str(">HH"),
read_data)
if cur_blkt_number == blockette_number:
# Blockette found: we want to skip ``field_offset`` bytes but we
# have already read 4 of the offset to get informations about the
# current blockette, so we remove them from skipped data
mseed_file_desc.seek(field_offset - 4, os.SEEK_CUR)
returned_bytes = mseed_file_desc.read(field_length)
else:
returned_bytes = None
finally:
mseed_file_desc.seek(previous_position, os.SEEK_SET)
return returned_bytes
[docs]def _convert_flags_to_raw_byte(expected_flags, user_flags, recstart, recend):
"""
Converts a flag dictionary to a byte, ready to be encoded in a MiniSEED
header.
This is a utility function for set_flags_in_fixed_headers and is not
designed to be called by someone else.
expected_signals describes all the possible bit names for the user flags
and their place in the result byte. Expected: dict { exponent: bit_name }.
The fixed header flags are available in obspy.io.mseed.headers as
FIXED_HEADER_ACTIVITY_FLAGS, FIXED_HEADER_DATA_QUAL_FLAGS and
FIXED_HEADER_IO_CLOCK_FLAGS.
This expects a user_flags as a dictionary { bit_name : value }. bit_name is
compared to the expected_signals, and its value is converted to bool.
Missing values are considered false.
:type expected_flags: dict {int: str}
:param expected_flags: every possible flag in this field, with its offset
:type user_flags: dict {str: bool}
:param user_flags: user defined flags and its value
:type recstart: UTCDateTime
:param recstart: date of the first sample of the current record
:type recstart: UTCDateTime
:param recend: date of the last sample of the current record
:return: raw int value for the flag group
"""
flag_byte = 0
for (bit, key) in expected_flags.items():
use_in_this_record = False
if key in user_flags:
if isinstance(user_flags[key], bool) and user_flags[key]:
# Boolean value, we accept it for all records
use_in_this_record = True
elif isinstance(user_flags[key], collections.Sequence):
# List of tuples (start, end)
use_in_this_record = False
for tuple_value in user_flags[key]:
# Check wether this record is concerned
event_start = tuple_value[0]
event_end = tuple_value[1]
if(event_start < recend) and (recstart <= event_end):
use_in_this_record = True
break
if use_in_this_record:
flag_byte += 2**bit
return flag_byte
@deprecated("'shiftTimeOfFile' has been renamed to "
"'shift_time_of_file'. Use that instead.") # noqa
[docs]def shiftTimeOfFile(*args, **kwargs):
return shift_time_of_file(*args, **kwargs)
[docs]def shift_time_of_file(input_file, output_file, timeshift):
"""
Takes a MiniSEED file and shifts the time of every record by the given
amount.
The same could be achieved by reading the MiniSEED file with ObsPy,
modifying the starttime and writing it again. The problem with this
approach is that some record specific flags and special blockettes might
not be conserved. This function directly operates on the file and simply
changes some header fields, not touching the rest, thus preserving it.
Will only work correctly if all records have the same record length which
usually should be the case.
All times are in 0.0001 seconds, that is in 1/10000 seconds. NOT ms but one
order of magnitude smaller! This is due to the way time corrections are
stored in the MiniSEED format.
:type input_file: str
:param input_file: The input filename.
:type output_file: str
:param output_file: The output filename.
:type timeshift: int
:param timeshift: The time-shift to be applied in 0.0001, e.g. 1E-4
seconds. Use an integer number.
Please do NOT use identical input and output files because if something
goes wrong, your data WILL be corrupted/destroyed. Also always check the
resulting output file.
.. rubric:: Technical details
The function will loop over every record and change the "Time correction"
field in the fixed section of the MiniSEED data header by the specified
amount. Unfortunately a further flag (bit 1 in the "Activity flags" field)
determines whether or not the time correction has already been applied to
the record start time. If it has not, all is fine and changing the "Time
correction" field is enough. Otherwise the actual time also needs to be
changed.
One further detail: If bit 1 in the "Activity flags" field is 1 (True) and
the "Time correction" field is 0, then the bit will be set to 0 and only
the time correction will be changed thus avoiding the need to change the
record start time which is preferable.
"""
timeshift = int(timeshift)
# A timeshift of zero makes no sense.
if timeshift == 0:
msg = "The timeshift must to be not equal to 0."
raise ValueError(msg)
# Get the necessary information from the file.
info = get_record_information(input_file)
record_length = info["record_length"]
byteorder = info["byteorder"]
sys_byteorder = "<" if (sys.byteorder == "little") else ">"
do_swap = False if (byteorder == sys_byteorder) else True
# This is in this scenario somewhat easier to use than BytesIO because one
# can directly modify the data array.
data = np.fromfile(input_file, dtype=np.uint8)
array_length = len(data)
record_offset = 0
# Loop over every record.
while True:
remaining_bytes = array_length - record_offset
if remaining_bytes < 48:
if remaining_bytes > 0:
msg = "%i excessive byte(s) in the file. " % remaining_bytes
msg += "They will be appended to the output file."
warnings.warn(msg)
break
# Use a slice for the current record.
current_record = data[record_offset: record_offset + record_length]
record_offset += record_length
activity_flags = current_record[36]
is_time_correction_applied = bool(activity_flags & 2)
current_time_shift = current_record[40:44]
current_time_shift.dtype = np.int32
if do_swap:
current_time_shift = current_time_shift.byteswap(False)
current_time_shift = current_time_shift[0]
# If the time correction has been applied, but there is no actual
# time correction, then simply set the time correction applied
# field to false and process normally.
# This should rarely be the case.
if current_time_shift == 0 and is_time_correction_applied:
# This sets bit 2 of the activity flags to 0.
current_record[36] = current_record[36] & (~2)
is_time_correction_applied = False
# This is the case if the time correction has been applied. This
# requires some more work by changing both, the actual time and the
# time correction field.
elif is_time_correction_applied:
msg = "The timeshift can only be applied by actually changing the "
msg += "time. This is experimental. Please make sure the output "
msg += "file is correct."
warnings.warn(msg)
# The whole process is not particularly fast or optimized but
# instead intends to avoid errors.
# Get the time variables.
time = current_record[20:30]
year = time[0:2]
julday = time[2:4]
hour = time[4:5]
minute = time[5:6]
second = time[6:7]
msecs = time[8:10]
# Change dtype of multibyte values.
year.dtype = np.uint16
julday.dtype = np.uint16
msecs.dtype = np.uint16
if do_swap:
year = year.byteswap(False)
julday = julday.byteswap(False)
msecs = msecs.byteswap(False)
dtime = UTCDateTime(year=year[0], julday=julday[0], hour=hour[0],
minute=minute[0], second=second[0],
microsecond=msecs[0] * 100)
dtime += (float(timeshift) / 10000)
year[0] = dtime.year
julday[0] = dtime.julday
hour[0] = dtime.hour
minute[0] = dtime.minute
second[0] = dtime.second
msecs[0] = dtime.microsecond / 100
# Swap again.
if do_swap:
year = year.byteswap(False)
julday = julday.byteswap(False)
msecs = msecs.byteswap(False)
# Change dtypes back.
year.dtype = np.uint8
julday.dtype = np.uint8
msecs.dtype = np.uint8
# Write to current record.
time[0:2] = year[:]
time[2:4] = julday[:]
time[4] = hour[:]
time[5] = minute[:]
time[6] = second[:]
time[8:10] = msecs[:]
current_record[20:30] = time[:]
# Now modify the time correction flag.
current_time_shift += timeshift
current_time_shift = np.array([current_time_shift], np.int32)
if do_swap:
current_time_shift = current_time_shift.byteswap(False)
current_time_shift.dtype = np.uint8
current_record[40:44] = current_time_shift[:]
# Write to the output file.
data.tofile(output_file)
[docs]def _convert_and_check_encoding_for_writing(encoding):
"""
Helper function to handle and test encodings.
If encoding is a string, it will be converted to the appropriate
integer. It will furthermore be checked if the specified encoding can be
written using libmseed. Appropriate errors will be raised if necessary.
"""
# Check if encoding kwarg is set and catch invalid encodings.
encoding_strings = {v[0]: k for k, v in ENCODINGS.items()}
try:
encoding = int(encoding)
except:
pass
if isinstance(encoding, int):
if (encoding in ENCODINGS and ENCODINGS[encoding][3] is False) or \
encoding in UNSUPPORTED_ENCODINGS:
msg = ("Encoding %i cannot be written with ObsPy. Please "
"use another encoding.") % encoding
raise ValueError(msg)
elif encoding not in ENCODINGS:
raise ValueError("Unknown encoding: %i." % encoding)
else:
if encoding not in encoding_strings:
raise ValueError("Unknown encoding: '%s'." % str(encoding))
elif ENCODINGS[encoding_strings[encoding]][3] is False:
msg = ("Encoding '%s' cannot be written with ObsPy. Please "
"use another encoding.") % encoding
raise ValueError(msg)
encoding = encoding_strings[encoding]
return encoding
if __name__ == '__main__':
import doctest
doctest.testmod(exclude_empty=True)