def depth(self, depth):
"data of a depth slice"
depth_idx = int((depth - self.startDepth) // self.stepDepth)
with segyio.open(self.segy_file, 'r') as segyfile:
segyfile.mmap()
data = segyfile.depth_slice[depth_idx]
return data
def cdp(self, cdp):
"data of a cdp"
with segyio.open(self.segy_file, 'r') as segyfile:
segyfile.mmap()
data = segyfile.gather[cdp]
data = data.reshape((data.shape[-1],))
return data
def _parse_segy(self):
with segyio.open(self.segy_file, 'r') as segyfile:
segyfile.mmap()
self.startInline = segyfile.ilines[0]
self.endInline = segyfile.ilines[-1]
self.nEast = len(segyfile.ilines)
self.stepInline = (self.endInline - self.startInline) // \
(self.nEast - 1)
self.startCrline = segyfile.xlines[0]
self.endCrline = segyfile.xlines[-1]
self.nNorth = len(segyfile.xlines)
self.stepCrline = (self.endCrline - self.startCrline) // \
(self.nNorth - 1)
self.startDepth = segyfile.samples[0]
self.endDepth = segyfile.samples[-1]
self.nDepth = len(segyfile.samples)
self.stepDepth = (self.endDepth - self.startDepth) // \
(self.nDepth - 1)
inline_A = self.startInline
crline_A = self.startCrline
index_A = 0
x_A = segyfile.header[index_A][segyio.su.cdpx]
y_A = segyfile.header[index_A][segyio.su.cdpy]
inline_B = inline_A
crline_B = self.startCrline + 2 * self.stepCrline
index_B = 2
x_B = segyfile.header[index_B][segyio.su.cdpx]
y_B = segyfile.header[index_B][segyio.su.cdpy]
inline_C = self.startInline + 2 * self.stepInline
crline_C = crline_B
index_C = 2 * self.nNorth + 2
x_C = segyfile.header[index_C][segyio.su.cdpx]
y_C = segyfile.header[index_C][segyio.su.cdpy]
setting_dict = {
"inline_range": [
self.startInline, self.endInline, self.stepInline],
"crline_range": [
self.startCrline, self.endCrline, self.stepCrline],
"z_range": [
self.startDepth, self.endDepth, self.stepDepth, "unknown"],
"point_A": [inline_A, crline_A, x_A, y_A],
"point_B": [inline_B, crline_B, x_B, y_B],
"point_C": [inline_C, crline_C, x_C, y_C]
}
self.survey_setting = SurveySetting(ThreePoints(setting_dict))
def test_header_dict_methods(self):
with segyio.open(self.filename, "r") as f:
self.assertEqual(89, len(list(f.header[0].keys())))
self.assertEqual(89, len(list(f.header[1].values())))
self.assertEqual(89, len(list(f.header[2].items())))
self.assertEqual(89, len(list(f.header[3])))
self.assertTrue(0 not in f.header[0])
self.assertTrue(1 in f.header[0])
self.assertTrue(segyio.su.cdpx in f.header[0])
iter(f.header[0])
self.assertEqual(30, len(f.bin.keys()))
self.assertEqual(30, len(list(f.bin.values())))
self.assertEqual(30, len(list(f.bin.items())))
self.assertEqual(30, len(f.bin))
iter(f.bin)
def set_source_filename(self, filename, **kwargs):
if filename:
try:
source = segyio.open(filename, "r", **kwargs)
except:
raise
else:
self._close_current_file()
self._source = source
self._source.mmap()
samples = self._source.samples
else:
self._close_current_file()
self._source = EmptyDataSource()
self.slice_data_source_changed.emit()
def test_header_dict_methods():
with segyio.open("test-data/small.sgy") as f:
assert 89 == len(list(f.header[0].keys()))
assert 89 == len(list(f.header[1].values()))
assert 89 == len(list(f.header[2].items()))
assert 89 == len(list(f.header[3]))
assert 0 not in f.header[0]
assert 1 in f.header[0]
assert segyio.su.cdpx in f.header[0]
iter(f.header[0])
assert 30 == len(f.bin.keys())
assert 30 == len(list(f.bin.values()))
assert 30 == len(list(f.bin.items()))
assert 30 == len(f.bin)
iter(f.bin)
def test_traces_slicing():
with segyio.open("test-data/small.sgy") as f:
traces = list(map(np.copy, f.trace[0:6:2]))
assert len(traces) == 3
assert traces[0][49] == f.trace[0][49]
assert traces[1][49] == f.trace[2][49]
assert traces[2][49] == f.trace[4][49]
rev_traces = list(map(np.copy, f.trace[4::-2]))
assert rev_traces[0][49] == f.trace[4][49]
assert rev_traces[1][49] == f.trace[2][49]
assert rev_traces[2][49] == f.trace[0][49]
# make sure buffers can be reused
for i, trace in enumerate(f.trace[0:6:2]):
assert np.array_equal(trace, traces[i])
def resample(segy, sample_rate=4, outpath=pathlib.Path.cwd()):
'''Create a resample segy file (increase sampling rate)
Creates a resampled segy file from an original segy. Note that the new
sampling rate has to be a strict multiple of the original one.
Parameters
----------
segy : vmlib.seis.segy.io.Segy()
instance of the source segyfile as loaded by vmlib.seis.segy.io
sample_rate : float, optional
destination sampling rate in milliseconds (defaults to 4 [ms])
outpath : pathlib.Path, optional
path to destination folder (defaults is pathlib.Path.cwd())
Returns
-------
pathlib.Path
path to the created, resampled, segy file
Raises
------
ValueError
if the desired sample rate is not a strict multiple of the original one
'''
# Resolve destination file
filename, extension = segy.info['filename'].split('.')
destination = outpath.joinpath(f'{filename}_resample.{extension}')
# Check new sampling rate
if (sample_rate % segy.info['sample_rate']) == 0:
ratio = int(sample_rate / segy.info['sample_rate'])
# Process file
with segyio.open(segy.info['file'], ignore_geometry=True) as src:
spec = segyio.tools.metadata(src)
spec.samples = spec.samples[:int(src.samples.size / ratio) - 1]
with segyio.create(destination, spec) as dst:
dst.text[0] = src.text[0]
dst.bin = src.bin
for ix, trace in enumerate(src.trace):
dst.trace[ix] = sps.resample(trace,
int(len(trace) / ratio))
dst.bin.update(hdt=sample_rate * 1000)
dst.bin.update(hns=len(spec.samples))
dst.header = src.header
return destination
else:
raise ValueError('Check sample rate to be multiple of the original')
def put_segy_texthead(segyfile,
ebcidc,
ext_headers=False,
line_counter=True,
**segyio_kwargs):
header = ""
if isinstance(ebcidc, dict):
for key in ebcidc:
if not isinstance(key, int):
warn(
"ebcidc dict contains not integer keys that will be ignored",
UserWarning,
)
for line in range(1, 41):
if line_counter:
lc = f"C{line:02d} "
n = 75
content = "{:<76}"
else:
lc = ""
n = 79
content = "{:<80}"
try:
test = ebcidc[line]
if len(test) > 75:
warn(f"EBCIDC line {line} is too long - truncating",
UserWarning)
header = header + lc + content.format(ebcidc[line][:n])
except KeyError:
# line not specified in dictionary
header = header + lc + " " * n
header = bytes(header, "utf8")
elif isinstance(ebcidc, bytes):
if len(ebcidc) > 3200:
warn("Byte EBCIDC is too large - truncating", UserWarning)
header = ebcidc[:3200]
elif isinstance(ebcidc, str):
if len(ebcidc) > 3200:
warn("String EBCIDC is too large - truncating", UserWarning)
header = bytes(ebcidc[:3200], "utf8")
else:
raise ValueError("Unknown ebcidc type")
segyio_kwargs["ignore_geometry"] = True
with segyio.open(segyfile, "r+", **segyio_kwargs) as segyf:
segyf.text[0] = header
def get_segy_texthead(segyfile,
ext_headers=False,
no_richstr=False,
**segyio_kwargs):
"""Return the ebcidc
Args:
segyfile (str): Segy File Path
ext_headers (bool): Return EBCIDC and extended headers in list.
Defaults to False
no_richstr (bool, optional): Defaults to False. If true the returned string
will not be updated for pretty HTML printing.
segyio_kwargs: Key word arguments to pass to segyio.open
Returns:
str: Returns the EBCIDC text as a formatted paragraph.
"""
with open(segyfile, mode="rb") as f:
f.seek(0, 0) # Locate our position to first byte of file
data = f.read(3200) # Read the first 3200 byte from our position
if _isascii(data) and ext_headers == False:
text = data.decode("ascii") # EBCDIC encoding
text = _text_fixes(text)
text = segyio.tools.wrap(text)
elif ext_headers == False:
text = data.decode("cp500") # text is ebcidc
text = _text_fixes(text)
text = segyio.tools.wrap(text)
else:
segyio_kwargs["ignore_geometry"] = True
try: # pray that the encoding is ebcidc
with segyio.open(segyfile, "r", **segyio_kwargs) as segyf:
text = segyf.text[0].decode("ascii", "replace")
text = _text_fixes(text)
text = segyio.tools.wrap(text)
if segyf.ext_headers and ext_headers:
text2 = segyf.text[1].decode("ascii", "replace")
text = [text, text2]
except UnicodeDecodeError as err:
print(err)
print("The segy text header could not be decoded.")
if no_richstr:
return text
else:
return _upgrade_txt_richstr(text)
def metadata(f):
"""Get survey structural properties and metadata
Create a description object that, when passed to ``segyio.create()``, would
create a new file with the same structure, dimensions, and metadata as
``f``.
Takes an open segy file (created with segyio.open) or a file name.
Parameters
----------
f : str or segyio.SegyFile
Returns
-------
spec : segyio.spec
Notes
-----
.. versionadded:: 1.4
"""
if not isinstance(f, segyio.SegyFile):
with segyio.open(f) as fl:
return metadata(fl)
spec = segyio.spec()
spec.iline = f._il
spec.xline = f._xl
spec.samples = f.samples
spec.format = f.format
spec.ilines = f.ilines
spec.xlines = f.xlines
spec.offsets = f.offsets
spec.sorting = f.sorting
spec.tracecount = f.tracecount
spec.ext_headers = f.ext_headers
spec.endian = f.endian
return spec
def test_create_from_naught_prestack(self):
with TestContext("create_from_naught_prestack"):
fname = "mk-ps.sgy"
spec = segyio.spec()
spec.format = 5
spec.sorting = 2
spec.samples = range(7)
spec.ilines = range(1, 4)
spec.xlines = range(1, 3)
spec.offsets = range(1, 6)
with segyio.create(fname, spec) as dst:
arr = np.arange(start=0.000,
stop=0.007,
step=0.001,
dtype=np.single)
arr = np.concatenate([[arr + 0.01], [arr + 0.02]], axis=0)
lines = [arr + i for i in spec.ilines]
cube = [(off * 100) + line for line in lines
for off in spec.offsets]
dst.iline[:, :] = cube
for of in spec.offsets:
for il in spec.ilines:
dst.header.iline[il, of] = {
TraceField.INLINE_3D: il,
TraceField.offset: of
}
for xl in spec.xlines:
dst.header.xline[xl, of] = {
TraceField.CROSSLINE_3D: xl
}
with segyio.open(fname, "r") as f:
self.assertAlmostEqual(101.010, f.trace[0][0], places=4)
self.assertAlmostEqual(101.011, f.trace[0][1], places=4)
self.assertAlmostEqual(101.016, f.trace[0][-1], places=4)
self.assertAlmostEqual(503.025, f.trace[-1][5], places=4)
self.assertNotEqual(f.header[0][TraceField.offset],
f.header[1][TraceField.offset])
self.assertEqual(1, f.header[0][TraceField.offset])
self.assertEqual(2, f.header[1][TraceField.offset])
for x, y in zip(f.iline[:, :], cube):
self.assertListEqual(list(x.flatten()), list(y.flatten()))
def test_attributes():
with segyio.open("test-data/small.sgy") as f:
il = TraceField.INLINE_3D
xl = TraceField.CROSSLINE_3D
assert 1 == f.attributes(il)[0]
assert 20 == f.attributes(xl)[0]
assert f.tracecount == len(f.attributes(il))
assert iter(f.attributes(il))
ils = [(i // 5) + 1 for i in range(25)]
attrils = list(map(int, f.attributes(il)[:]))
assert ils == attrils
xls = [(i % 5) + 20 for i in range(25)]
attrxls = list(map(int, f.attributes(xl)[:]))
assert xls == attrxls
ils = [(i // 5) + 1 for i in range(25)][::-1]
attrils = list(map(int, f.attributes(il)[::-1]))
assert ils == attrils
xls = [(i % 5) + 20 for i in range(25)][::-1]
attrxls = list(map(int, f.attributes(xl)[::-1]))
assert xls == attrxls
assert f.header[0][il] == f.attributes(il)[0]
f.mmap()
assert f.header[0][il] == f.attributes(il)[0]
ils = [(i // 5) + 1 for i in range(25)][1:21:3]
attrils = list(map(int, f.attributes(il)[1:21:3]))
assert ils == attrils
xls = [(i % 5) + 20 for i in range(25)][2:17:5]
attrxls = list(map(int, f.attributes(xl)[2:17:5]))
assert xls == attrxls
ils = [1, 2, 3, 4, 5]
attrils = list(map(int, f.attributes(il)[[0, 5, 11, 17, 23]]))
assert ils == attrils
ils = [1, 2, 3, 4, 5]
indices = np.asarray([0, 5, 11, 17, 23])
attrils = list(map(int, f.attributes(il)[indices]))
assert ils == attrils
def _segy2d_xr(
segyfile,
ds,
segyio_kwargs,
n0,
ns,
head_df,
head_loc,
vert_domain="TWT",
silent=False,
):
"""Helper function to load 2d data into an xarray with the seisnc form."""
if vert_domain == "TWT":
dims = DimensionKeyField.twod_twt
elif vert_domain == "DEPTH":
dims = DimensionKeyField.twod_depth
else:
raise ValueError(f"Unknown vert_domain: {vert_domain}")
with segyio.open(segyfile, "r", **segyio_kwargs) as segyf:
segyf.mmap()
pb = tqdm(total=segyf.tracecount,
desc="Converting SEGY",
disable=silent,
**TQDM_ARGS)
shape = [ds.dims[d] for d in dims]
volume = np.zeros(shape, dtype=np.float32)
percentiles = np.zeros_like(PERCENTILES)
# this can probably be done as a block - leaving for now just incase sorting becomes an issue
for trc, val in head_df.iterrows():
volume[int(val.cdp_index), :] = segyf.trace[trc][n0:ns + 1]
pb.update()
percentiles = (np.percentile(segyf.trace[trc][n0:ns + 1],
PERCENTILES) + percentiles) / 2.0
pb.close()
ds[VariableKeyField.data] = (
dims,
volume[:, n0:ns + 1],
)
ds.attrs[AttrKeyField.percentiles] = list(percentiles)
return ds
def test_fopen_error():
# non-existent file
with pytest.raises(IOError):
segyio.open("no_dir/no_file")
# non-existant mode
with pytest.raises(ValueError):
segyio.open("test-data/small.sgy", "foo")
with pytest.raises(ValueError):
segyio.open("test-data/small.sgy", "r+b+toolong")
def test_fopen_error(self):
# non-existent file
with self.assertRaises(IOError):
segyio.open("no_dir/no_file", "r")
# non-existant mode
with self.assertRaises(ValueError):
segyio.open(self.filename, "foo")
with self.assertRaises(ValueError):
segyio.open(self.filename, "r+b+toolong")
def learn5():
path2 = 'E:/Research/data/F3_entire.segy'
path3 = 'E:/Research/data/test1.segy'
with segyio.open(path2) as src:
spec = segyio.spec()
spec.sorting = src.sorting
spec.format = src.format
spec.samples = src.samples
spec.ilines = src.ilines
spec.xlines = src.xlines
spec.tracecount = src.tracecount
print(len(src.samples))
with segyio.create(path3, spec) as dst:
dst.text[0] = src.text[0]
dst.bin = src.bin
dst.header = src.header
dst.trace = src.trace
def test_dt_fallback(self):
with TestContext("dt_fallback") as context:
context.copy_file(self.filename)
with segyio.open("small.sgy", "r+") as f:
# Both zero
f.bin[BinField.Interval] = 0
f.header[0][TraceField.TRACE_SAMPLE_INTERVAL] = 0
f.flush()
fallback_dt = 4
np.testing.assert_almost_equal(segyio.dt(f, fallback_dt), fallback_dt)
# dt in bin header different from first trace
f.bin[BinField.Interval] = 6000
f.header[0][TraceField.TRACE_SAMPLE_INTERVAL] = 1000
f.flush()
fallback_dt = 4
np.testing.assert_almost_equal(segyio.dt(f, fallback_dt), fallback_dt)
def load_segy(fn_sgy, *args, **kwargs):
"""Load segy data.
This is very generic for now, need to do work if there are particular
types of segy files that need to be read. We cannot yet support a
generic SEGY file, even if complies to some standard set.
"""
if not SEGY:
raise ImportError('Need segyio to load SGY files')
segy_data = RadarData(None)
segy_data.fn = fn_sgy
f = segyio.open(fn_sgy, ignore_geometry=True)
segy_data.data = segyio.tools.collect(f.trace).transpose()
segy_data.snum = f.bin[segyio.BinField.Samples]
segy_data.tnum = segy_data.data.shape[1]
segy_data.dt = f.bin[segyio.BinField.Interval] * 1.0e-12
segy_data.travel_time = np.arange(segy_data.snum) * segy_data.dt * 1.0e6
segy_data.trace_num = np.arange(segy_data.data.shape[1]) + 1
segy_data.flags = RadarFlags()
# segy_data.travel_time = np.atleast_2d(np.arange(0,
# segy_data.dt * segy_data.snum, segy_data.dt)).transpose()
# + segy_data.dt
# TODO these next ones are filler
# right now, they are selected so that they read in the delores data
segy_data.trace_int = 1
segy_data.chan = 1
segy_data.trig = np.zeros((segy_data.tnum, ))
segy_data.decday = np.zeros((segy_data.tnum, ))
segy_data.x_coord = f.attributes(segyio.TraceField.CDP_X)[:] / 10.0
segy_data.y_coord = f.attributes(segyio.TraceField.CDP_Y)[:] / 10.0
segy_data.dist = np.hstack(
([0],
np.cumsum(
np.sqrt(
np.diff(segy_data.x_coord)**2.0 +
np.diff(segy_data.y_coord)**2.0)))) / 1000.
segy_data.long = np.zeros((segy_data.tnum, ))
segy_data.lat = np.zeros((segy_data.tnum, ))
segy_data.elev = np.zeros((segy_data.tnum, ))
segy_data.trig_level = np.zeros((segy_data.tnum, ))
segy_data.pressure = np.zeros((segy_data.tnum, ))
segy_data.check_attrs()
return segy_data
def getFileInfo(dirName):
"""
This function will read all of the *.segy or & *.sgy files in a given
directory. It returns a list with the file name and the shot location.
This information will be passed to the GUI to display the file names. At
a latter time it might be worth extracting other things from the headers
and storing them in this list.
DEPENDENCIES:
GLOB - this is used to get the file names in the directory
segyio - this is used to read the segy files and extract header info
INPUTS:
dirName (str) = this is a string to the directory that contains all of
the segy files from the survey.
OUTPUTS:
fileInfo is a list that is total Files by 2.
Column 1 (str) = file name
Column 2 (float) = shot location (units assumed to be m)
NOTES:
At this stage I use two if statemetns to check for segy files. If there
are no segy files fileInfo will be an empty list and the user will get
an error. Though I am not sure where error goes in a GUI?
- It depends, but we will be able to use try-except blocks for them
It might be worth adding columns to this list if we need more info from
the files later on
"""
files = glb.glob(os.path.join(dirName, "*.sgy"))
if files == []:
files = glb.glob(os.path.join(dirName, "*.segy"))
if files == []:
print("No files with *.sgy or *.segy exist in this directory")
# Column 1: File Name (str)
# Column 2: SX (float)
fileInfo = []
for file in files:
filename = os.path.basename(file)
# print(filename)
with segyio.open(file, strict=False) as f:
shotLoc = f.header[0][segyio.TraceField.SourceX]
# print(shotLoc)
fileInfo.append([filename, shotLoc])
return fileInfo
def test_depth_slice_writing(tmpdir):
buf = np.empty(shape=(5, 5), dtype=np.single)
def value(x, y):
return x + (1.0 // 5) * y
for x, y in itertools.product(range(5), range(5)):
buf[x][y] = value(x, y)
with segyio.open(tmpdir / "small.sgy", "r+") as f:
f.depth_slice[7] = buf * 3.14 # assign to depth 7
assert np.allclose(f.depth_slice[7], buf * 3.14)
f.depth_slice = [buf * i for i in range(len(f.depth_slice))] # assign to all depths
for index, depth_slice in enumerate(f.depth_slice):
assert np.allclose(depth_slice, buf * index)
def btnArquivoPressed(self):
options = QFileDialog.Options()
#options |= QFileDialog.FileType("*.segy")
fileName, _ = QFileDialog.getOpenFileName(
self,
"QFileDialog.getOpenFileName()",
"",
"Seg-Y (*.segy);; Seg-Y (*.sgy)",
options=options)
if fileName:
self.txtArquivo.setText(fileName)
sgy = segyio.open(self.txtArquivo.text(), ignore_geometry=True)
Dx = np.array(sgy.trace)
self.dado = Dx
[self.nt, self.nr] = self.dado.shape
self.txtNt.setText(str(self.nt))
self.txtNr.setText(str(self.nr))
def location(self):
if self._data_type[0][-3: ] == 'bin':
self._location['inline_start'] = self._data_type[2][0]
self._location['xline_start'] = self._data_type[2][1]
self._location['time_start'] = self._data_type[2][2]
self._location['sample_interval'] = self._data_type[2][3]
self.location_out = self._location
pass
elif self._data_type[0][-3: ] == 'sgy':
with segyio.open(self._data_type[0]) as f:
self._location['inline_start'] = f.ilines[0]
self._location['xline_start'] = f.xlines[0]
self._location['time_start'] = int(f.samples[0])
self._location['sample_interval'] = int(f.samples[1] - f.samples[0])
self.location_out = self._location
pass
return
def openSegy3D(filename):
"""
Open 3D seismic volume in SEGY or SGY format
"""
import segyio
try:
with segyio.open(filename) as f:
data = segyio.tools.cube(f)
inlines = f.ilines
crosslines = f.xlines
twt = f.samples
sample_rate = segyio.tools.dt(f) / 1000
print('Successfully read \n')
print('Inline range from', inlines[0], 'to', inlines[-1])
print('Crossline range from', crosslines[0], 'to', crosslines[-1])
print('TWT from', twt[0], 'to', twt[-1])
print('Sample rate:', sample_rate, 'ms')
try:
rot, cdpx, cdpy = segyio.tools.rotation(f, line="fast")
print('Survey rotation: {:.2f} deg'.format(rot))
except:
print("Survey rotation not recognized")
cube = dict({"data": data,
"inlines": inlines,
"crosslines": crosslines,
"twt": twt,
"sample_rate": sample_rate})
# See Stackoverflow: https://stackoverflow.com/questions/4984647/accessing-dict-keys-like-an-attribute
class AttrDict(dict):
def __init__(self, *args, **kwargs):
super(AttrDict, self).__init__(*args, **kwargs)
self.__dict__ = self
cube = AttrDict(cube)
except:
print("openSegy cannot read your data")
return cube
def make_segy_index(filename, extra_headers=None, limits=None):
"""Index traces in a single SEGY file.
Parameters
----------
filename : str
Path to SEGY file.
extra_headers : array-like or str
Additional headers to put unto DataFrme. If 'all', all headers are included.
limits : slice or int, default to None
If int, index only first ```limits``` traces. If slice, index only traces
within given range. If None, index all traces.
Returns
-------
df : pandas.DataFrame
DataFrame with trace indexing.
"""
if not isinstance(limits, slice):
limits = slice(limits)
with segyio.open(filename, strict=False) as segyfile:
segyfile.mmap()
if extra_headers == 'all':
headers = [h.__str__() for h in segyio.TraceField.enums()]
elif extra_headers is None:
headers = DEFAULT_SEGY_HEADERS + SUPPORT_SEGY_HEADERS
else:
extra_headers = [extra_headers] if isinstance(
extra_headers, str) else list(extra_headers)
headers = set(DEFAULT_SEGY_HEADERS + extra_headers +
SUPPORT_SEGY_HEADERS)
meta = dict()
for k in headers:
meta[k] = segyfile.attributes(getattr(segyio.TraceField,
k))[limits]
meta['file_id'] = np.repeat(filename, segyfile.tracecount)[limits]
meta['RecieverID'] = np.array(
[hash(pair) for pair in zip(meta['GroupX'], meta['GroupY'])])
df = pd.DataFrame(meta)
return df
def convert_segy_stream(self, bits_per_voxel, blockshape, reduce_iops=False):
"""Memory-efficient method of compressing SEG-Y file larger than machine memory.
Requires at least n_crosslines x n_samples x blockshape[2] x 4 bytes of available memory"""
t0 = time.time()
if not os.path.exists(self.in_filename):
msg = "With searching comes loss, and the presence of absence: 'My Segy' not found."
raise FileNotFoundError(msg)
with segyio.open(self.in_filename, mode='r', strict=False) as segyfile:
if self.geom is None:
if segyfile.unstructured:
print("SEG-Y file is unstructured and no geometry provided. Determining this may take some time...")
traces_ref = {(h[189], h[193]): i for i, h in enumerate(segyfile.header)}
self.geom = InferredGeometry(traces_ref)
print("... inferred geometry is:", self.geom)
else:
self.geom = Geometry(0, len(segyfile.ilines), 0, len(segyfile.xlines))
n_traces = len(self.geom.ilines) * len(self.geom.xlines)
inline_set_bytes = blockshape[0] * (len(self.geom.xlines) * len(segyfile.samples)) * 4
headers_to_store = get_unique_headerwords(segyfile)
numpy_headers_arrays = [np.zeros(n_traces, dtype=np.int32) for _ in range(len(headers_to_store))]
if inline_set_bytes > virtual_memory().total // 2:
print("One inline set is {} bytes, machine memory is {} bytes".format(inline_set_bytes, virtual_memory().total))
print("Try using fewer inlines in the blockshape, or compressing a subcube")
raise RuntimeError("ABORTED effort: Close all that you have. You ask way too much.")
max_queue_length = min(16, (virtual_memory().total // 2) // inline_set_bytes)
print("VirtualMemory={}MB, InlineSet={}MB : Using queue of length {}".format(virtual_memory().total/(1024*1024),
inline_set_bytes/(1024*1024),
max_queue_length))
run_conversion_loop(self.in_filename, self.out_filename, bits_per_voxel, blockshape,
headers_to_store, numpy_headers_arrays, self.geom,
queuesize=max_queue_length, reduce_iops=reduce_iops)
with open(self.out_filename, 'ab') as f:
for header_array in numpy_headers_arrays:
f.write(header_array.tobytes())
t3 = time.time()
print("Total conversion time: {} ".format(t3-t0))
def _read_velocity(self, vcube, cell_corners):
"""
Read velocity from segy file. Upscale.
:param vcube:
:param cell_corners:
:return:
"""
with segyio.open(vcube) as f:
x = np.empty(f.tracecount, dtype=np.float64)
y = np.empty(f.tracecount, dtype=np.float64)
dt = segyio.dt(f) / 1e6
nt = len(f.samples)
for i, h in enumerate(f.header):
scalar = h[TraceField.SourceGroupScalar]
if scalar < 0:
scalar = -1.0 / scalar
x[i] = h[TraceField.CDP_X] * scalar
y[i] = h[TraceField.CDP_Y] * scalar
ny = len(f.xlines)
nx = len(f.ilines)
# memory issue might happen if volume becomes too large
traces = f.trace.raw[:]
if f.sorting == segyio.TraceSortingFormat.INLINE_SORTING:
x = x.reshape(ny, nx)
y = y.reshape(ny, nx)
traces = traces.reshape(ny, nx, nt)
x = np.transpose(x)
y = np.transpose(y)
traces = np.transpose(traces, (1, 0, 2))
elif f.sorting == segyio.TraceSortingFormat.CROSSLINE_SORTING:
x = x.reshape(nx, ny)
y = y.reshape(nx, ny)
traces = traces.reshape(nx, ny, nt)
else:
raise RuntimeError
x, y, traces, nt, dt = self._upscale_velocity(cell_corners, x, y,
traces, nt, dt)
return x, y, traces, nt, dt
def read_2Dmodel(self, path):
"""
Build a 2D velocity model from a SEG-Y format.
It uses the 'segyio' from https://github.com/equinor/segyio
"""
with segyio.open(path, ignore_geometry=True) as f:
n_samples = len(f.samples)
n_traces = len(f.trace)
data = np.zeros(shape=(n_samples, n_traces), dtype=np.float32)
index = 0
for trace in f.trace:
data[:, index] = trace
index += 1
# data = 2d data
# n_samples = height
# n_traces = length
self.model = data
def test_open_fails_unstructured(self):
with segyio.open(self.filename, "r", 37, strict=False) as f:
with self.assertRaises(ValueError):
f.iline[10]
with self.assertRaises(ValueError):
f.iline[:, :]
with self.assertRaises(ValueError):
f.xline[:, :]
with self.assertRaises(ValueError):
f.depth_slice[2]
# operations that don't rely on geometry still works
self.assertEqual(f.header[2][189], 1)
self.assertListEqual(list(f.attributes(189)[:]),
[(i // 5) + 1 for i in range(len(f.trace))])
def test_traces_slicing(self):
with segyio.open(self.filename, "r") as f:
traces = list(map(np.copy, f.trace[0:6:2]))
self.assertEqual(len(traces), 3)
self.assertEqual(traces[0][49], f.trace[0][49])
self.assertEqual(traces[1][49], f.trace[2][49])
self.assertEqual(traces[2][49], f.trace[4][49])
rev_traces = list(map(np.copy, f.trace[4::-2]))
self.assertEqual(rev_traces[0][49], f.trace[4][49])
self.assertEqual(rev_traces[1][49], f.trace[2][49])
self.assertEqual(rev_traces[2][49], f.trace[0][49])
# make sure buffers can be reused
buf = None
for i, trace in enumerate(f.trace[0:6:2, buf]):
self.assertTrue(np.array_equal(trace, traces[i]))
def compress_and_compare_zgy(zgy_file, sgy_file, tmp_path, bits_per_voxel,
rtol):
out_sgz = os.path.join(
str(tmp_path), 'test_{}_{}_.sgz'.format(
os.path.splitext(os.path.basename(zgy_file))[0], bits_per_voxel))
with ZgyConverter(zgy_file) as converter:
converter.run(out_sgz, bits_per_voxel=bits_per_voxel)
with SgzReader(out_sgz) as reader:
sgz_data = reader.read_volume()
sgz_ilines = reader.ilines
with segyio.open(sgy_file) as f:
ref_ilines = f.ilines
assert np.allclose(sgz_data, segyio.tools.cube(sgy_file), rtol=rtol)
assert all([a == b for a, b in zip(sgz_ilines, ref_ilines)])
def main():
if len(sys.argv) < 2:
sys.exit("Usage: trace_headers_to_dataframe.py [file]")
filename = sys.argv[1]
# Open file
with segyio.open(filename, ignore_geometry=True) as f:
# Get all header keys:
header_keys = segyio.tracefield.keys
# Initialize df with trace id as index and headers as columns
trace_headers = pd.DataFrame(index=range(1, f.tracecount + 1),
columns=header_keys.keys())
# Fill dataframe with all trace headers values
for k, v in header_keys.items():
trace_headers[k] = f.attributes(v)[:]
print(trace_headers.head())
def test_open_fails_unstructured():
with segyio.open("test-data/small.sgy", "r", 37, strict=False) as f:
with pytest.raises(ValueError):
_ = f.iline[10]
with pytest.raises(ValueError):
_ = f.iline[:, :]
with pytest.raises(ValueError):
_ = f.xline[:, :]
with pytest.raises(ValueError):
_ = f.depth_slice[2]
# operations that don't rely on geometry still works
assert f.header[2][189] == 1
assert (list(f.attributes(189)[:]) ==
[(i // 5) + 1 for i in range(len(f.trace))])
def update(self, index, data):
"""
Update data with ndarray
Parameters
----------
index : InlineIndex
data : 2-d ndarray
data for updating Inline
"""
try:
if not isinstance(index, InlineIndex):
raise TypeError("has to be InlineIndex")
if data.shape != (self.nNorth, self.nDepth):
raise ValueError
with segyio.open(self.segy_file, 'r+') as segyfile:
segyfile.mmap()
segyfile.iline[index.value] = data
except Exception as er:
print(er.message)
def inline(self, inline):
"data of a inline section"
with segyio.open(self.segy_file, 'r') as segyfile:
segyfile.mmap()
data = segyfile.iline[inline]
return data
def crline(self, crline):
"data of a crossline section"
with segyio.open(self.segy_file, 'r') as segyfile:
segyfile.mmap()
data = segyfile.xline[crline]
return data