def test_arbitrary_mapping(self, mapping):
builder = CatalogBuilder(mapping)
catalog = builder.create()
shared_items = set(mapping.items()) & set(catalog.items())
self.assertEqual(len(shared_items), len(mapping))
def catalog_traces(fh, bps, trace_header_format=TraceHeaderRev1, endian='>', progress=None):
"""Build catalogs to facilitate random access to trace_samples data.
Note:
This function can take significant time to run, proportional
to the number of traces in the SEG Y file.
Four catalogs will be build:
1. A catalog mapping trace_samples index (0-based) to the position of that
trace_samples header in the file.
2. A catalog mapping trace_samples index (0-based) to the number of
samples in that trace_samples.
3. A catalog mapping CDP number to the trace_samples index.
4. A catalog mapping an (inline, crossline) number 2-tuple to
trace_samples index.
Args:
fh: A file-like-object open in binary mode, positioned at the
start of the first trace_samples header.
bps: The number of bytes per sample, such as obtained by a call
to bytes_per_sample()
trace_header_format: The class defining the trace header format.
Defaults to TraceHeaderRev1.
endian: '>' for big-endian data (the standard and default), '<'
for little-endian (non-standard)
progress: A unary callable which will be passed a number
between zero and one indicating the progress made. If
provided, this callback will be invoked at least once with
an argument equal to 1
Returns:
A 4-tuple of the form (trace_samples-offset-catalog,
trace_samples-length-catalog,
cdp-catalog,
line-catalog)` where
each catalog is an instance of ``collections.Mapping`` or None
if no catalog could be built.
"""
progress_callback = progress if progress is not None else lambda p: None
if not callable(progress_callback):
raise TypeError("catalog_traces(): progress callback must be callable")
class CatalogSubFormat(metaclass=SubFormatMeta,
parent_format=trace_header_format,
parent_field_names=(
'file_sequence_num',
'ensemble_num',
'num_samples',
'inline_number',
'crossline_number',
)):
pass
trace_header_packer = make_header_packer(CatalogSubFormat, endian)
length = file_length(fh)
pos_begin = fh.tell()
trace_offset_catalog_builder = CatalogBuilder()
trace_length_catalog_builder = CatalogBuilder()
line_catalog_builder = CatalogBuilder()
alt_line_catalog_builder = CatalogBuilder()
cdp_catalog_builder = CatalogBuilder()
for trace_number in count():
progress_callback(_READ_PROPORTION * pos_begin / length)
fh.seek(pos_begin)
data = fh.read(TRACE_HEADER_NUM_BYTES)
if len(data) < TRACE_HEADER_NUM_BYTES:
break
trace_header = trace_header_packer.unpack(data)
num_samples = trace_header.num_samples
trace_length_catalog_builder.add(trace_number, num_samples)
samples_bytes = num_samples * bps
trace_offset_catalog_builder.add(trace_number, pos_begin)
# Should we check the data actually exists?
line_catalog_builder.add((trace_header.inline_number,
trace_header.crossline_number),
trace_number)
alt_line_catalog_builder.add((trace_header.file_sequence_num,
trace_header.ensemble_num),
trace_number)
cdp_catalog_builder.add(trace_header.ensemble_num, trace_number)
pos_end = pos_begin + TRACE_HEADER_NUM_BYTES + samples_bytes
pos_begin = pos_end
progress_callback(_READ_PROPORTION)
trace_offset_catalog = trace_offset_catalog_builder.create()
progress_callback(_READ_PROPORTION + (_READ_PROPORTION / 4))
trace_length_catalog = trace_length_catalog_builder.create()
progress_callback(_READ_PROPORTION + (_READ_PROPORTION / 2))
cdp_catalog = cdp_catalog_builder.create()
progress_callback(_READ_PROPORTION + (_READ_PROPORTION * 3 / 4))
line_catalog = line_catalog_builder.create()
if line_catalog is None:
# Some 3D files put Inline and Crossline numbers in (TraceSequenceFile, cdp) pair
line_catalog = alt_line_catalog_builder.create()
progress_callback(1)
return (trace_offset_catalog,
trace_length_catalog,
cdp_catalog,
line_catalog)
def catalog_fixed_length_traces(fh, binary_reel_header, trace_header_format=TraceHeaderRev1, endian='>', progress=None):
"""Build catalogs to for a fixed length SEG Y file. This is much faster
than the full catalog, but has limitations. No CDP, or inline, xline
catalogs, and it only works for segy files with fixed legth SEG Y files.
Note:
This function is faster than the full catalog, but has limitations.
No CDP, or inline, xline catalogs, and it only works for SEG Y files
with a fixed number of samples in each trace.
Two catalogs will be built:
1. A catalog mapping trace_samples index (0-based) to the position of that
trace_samples header in the file.
2. A catalog mapping trace_samples index (0-based) to the number of
samples in that trace_samples.
Args:
fh: A file-like-object open in binary mode, positioned at the
start of the first trace_samples header.
bps: The number of bytes per sample, such as obtained by a call
to bytes_per_sample()
trace_header_format: The class defining the trace header format.
Defaults to TraceHeaderRev1.
endian: '>' for big-endian data (the standard and default), '<'
for little-endian (non-standard)
progress: A unary callable which will be passed a number
between zero and one indicating the progress made. If
provided, this callback will be invoked at least once with
an argument equal to 1
Returns:
A 4-tuple of the form (trace_samples-offset-catalog,
trace_samples-length-catalog,
None,
None)` where
each catalog is an instance of ``collections.Mapping`` or None
if no catalog could be built.
"""
revision = extract_revision(binary_reel_header)
bps = bytes_per_sample(binary_reel_header, revision)
progress_callback = progress if progress is not None else lambda p: None
if not callable(progress_callback):
raise TypeError("catalog_traces(): progress callback must be callable")
class CatalogSubFormat(metaclass=SubFormatMeta,
parent_format=trace_header_format,
parent_field_names=(
'file_sequence_num',
'ensemble_num',
'num_samples',
'inline_number',
'crossline_number',
)):
pass
num_file_bytes = file_length(fh)
num_samples=binary_reel_header.num_samples
num_traces_float = (num_file_bytes-REEL_HEADER_NUM_BYTES)/(TRACE_HEADER_NUM_BYTES+num_samples*bps)
num_traces = int(num_traces_float)
if num_traces != num_traces_float:
raise ValueError(
"SEG Y file {!r} of {} bytes is not consistent with a fixed trace length".format(
filename_from_handle(fh),
num_file_bytes))
trace_offset_catalog_builder = CatalogBuilder()
trace_length_catalog_builder = CatalogBuilder()
for trace_index in range(num_traces):
pos_begin=REEL_HEADER_NUM_BYTES+(num_samples * bps+TRACE_HEADER_NUM_BYTES) * trace_index
trace_length_catalog_builder.add(trace_index, num_samples)
trace_offset_catalog_builder.add(trace_index, pos_begin)
trace_offset_catalog = trace_offset_catalog_builder.create()
trace_length_catalog = trace_length_catalog_builder.create()
progress_callback(1)
return (trace_offset_catalog,
trace_length_catalog,
None,
None)
