dlisio

DLIS/LIS reading and controlled LAS export

Use dlisio for binary RP66 DLIS and LIS79 input. It returns structured NumPy arrays and metadata objects; it does not write DLIS. Use lasio for LAS.

Select a logical file and frame

Inspect the physical file before selecting data. Logical files and frames may have different depths, sampling rates and channel identities. dlis.load() returns a PhysicalFile context manager, not a generator. Read array data inside the context, then retain copied arrays or DataFrames after closing it.

from dlisio import dlis

def inventory(path):
    items = []
    with dlis.load(str(path)) as files:
        for logical_index, logical in enumerate(files):
            for frame_index, frame in enumerate(logical.frames):
                items.append({
                    'logical_file': logical_index, 'frame': frame_index,
                    'fingerprint': frame.fingerprint,
                    'index_type': frame.index_type, 'index': frame.index,
                    'channels': [(ch.fingerprint, ch.units, ch.dimension)
                                 for ch in frame.channels],
                })
    return items

A channel's identity includes type, mnemonic, origin and copy number. Names alone need not be unique. Use channel.fingerprint to access its array field; frame.curves() uses fingerprints as dtype titles even when names are disambiguated. The FRAMENO field records frame sequence numbers, not measured depth.

Read scalar and array channels

from dlisio import dlis
import pandas as pd

def read_frame(path, logical_file_index=0, frame_index=0):
    with dlis.load(str(path)) as files:
        if not 0 <= logical_file_index < len(files):
            raise ValueError('Logical file index out of range')
        logical = files[logical_file_index]
        if not 0 <= frame_index < len(logical.frames):
            raise ValueError('Frame index out of range')
        frame = logical.frames[frame_index]
        records = frame.curves()
        scalar = {name: records[name].copy() for name in records.dtype.names
                  if records[name].ndim == 1}
        arrays = {ch.fingerprint: records[ch.fingerprint].copy()
                  for ch in frame.channels if records[ch.fingerprint].ndim > 1}
        metadata = {'frame': frame.fingerprint, 'index_type': frame.index_type,
                    'index': frame.index,
                    'channels': {ch.fingerprint: {'name': ch.name, 'units': ch.units,
                                 'dimension': ch.dimension} for ch in frame.channels}}
        return pd.DataFrame(scalar), arrays, metadata