Source code for SDF.file_io.lsm

import logging
import os

import numpy as np
from PIL import Image

from SDF.config import CONFIG
from SDF.data_model import Workspace, SourceParameters, Instrument, ArrayDataset1D, Parameter, ImageDataset, \
    ImageData, ParameterSet, ArrayDataset2D, ArrayData2D

if CONFIG.use_system_tifffile:
    import tifffile
else:
    from SDF.extern import tifffile

logger = logging.getLogger(__name__)


# The parameters read from the lsm file lack units.
# The following dictionary assigns a few units to
# parameters manually. We are not 100% sure they
# are always right, but those were the fixed units
# shown by Zeiss' ZEN software for our examples
# lsm files.
MANUALLY_INSERTED_UNITS = {"plane_spacing": "um",
                           "line_spacing": "um",
                           "sample_spacing": "um",
                           "pinhole_diameter": "um"}


def load_from_lsm(filename: str) -> Workspace:
    logger.info(f'Reading {filename}')
    with tifffile.TiffFile(filename) as lsm:

        if not lsm.is_lsm:
            raise ValueError("File is no lsm file")

        logger.info(f'Found {len(lsm.pages)} pages (images)')

        ws = Workspace(name=os.path.split(filename)[1])
        ws.parameters.add(SourceParameters(converter_name="lsm2sdf", source_file_type="lsm-file", source_file=filename))

        # the first page contains global information not present in the other pages
        first_page = lsm.pages[0]

        instrument = Instrument(name="Carl Zeiss LSM")
        instrument.add_from_structure(first_page.cz_lsm_scan_info, parse_names_from_dicts=True)
        ws.instruments.add(instrument)

        ws.datasets.add(ArrayDataset1D(name="time-stamps", data=first_page.cz_lsm_time_stamps))

        # all attributes of the second page are present in all other pages
        attr_keys = {key for key in dir(lsm.pages[1]) if not key.startswith("_")}
        attr_keys.remove("is_lsm")  # only true for first page, irrelevant for the others
        attr_keys.remove("asarray")  # image content, handled separately
        # TODO: These are just copied from old converter
        #       I (Niklas) have no clue, so I copied the old comments for future reference
        attr_keys.remove("tags")  # dictionary of tags, needs some special treatment
        attr_keys.remove("imagej_tags")  # there is some problem
        attr_keys.remove("uic_tags")  # there is some problem

        # handle series data
        for series_index, series in enumerate(lsm.series):
            series_ws = Workspace(name=f"series_{series_index}")
            ws.workspaces.add(series_ws)

            # find attributes that vary per layer in the series
            per_layer_attr_keys = set()
            for key in attr_keys:
                for layer in series[1:]:
                    # if key has different values in different layers: handle later
                    if getattr(layer, key) != getattr(series[0], key):
                        per_layer_attr_keys.add(key)
                        break
                else:  # if key has the same values for all layers: take value from first page
                    series_ws.parameters.add(Parameter(key, getattr(first_page, key)))

            # handle layers in series
            n_layers = series[0].asarray().shape[0]
            for layer_index in range(n_layers):
                layer_ws = Workspace(name=f"layer_{layer_index}")
                series_ws.workspaces.add(layer_ws)

                # handle images in layer
                for image_index in range(len(series)):
                    image = Image.fromarray(series[image_index].asarray()[layer_index].astype(np.uint8))
                    dataset = ImageDataset(name=f"img_{image_index}", data=ImageData(image))
                    layer_ws.datasets.add(dataset)

                    for key in per_layer_attr_keys:
                        dataset.parameters.add(Parameter(key, getattr(series[image_index], key)))

        # global lsm attributes
        lsm_keys = {key for key in dir(lsm) if not key.startswith("_")}
        lsm_keys.remove("byteorder")  # irrelevant since tifffile handled this for us already
        lsm_keys.remove("pages")
        lsm_keys.remove("filehandle")
        lsm_keys.remove("asarray")
        lsm_keys.remove("fstat")
        lsm_keys.remove("close")
        lsm_keys.remove("series")
        for key in lsm_keys:
            value = getattr(lsm, key)
            ws.parameters.add(Parameter(key, value))

        # other attributes unique to first page:
        # 1. cz_lsm_info
        cz_lsm_info_params = ParameterSet(name="cz_lsm_info")
        ws.parameters.add(cz_lsm_info_params)
        for index, value in enumerate(first_page.cz_lsm_info):
            cz_lsm_info_params.add(Parameter(name=str(index), value=value))

        # 2. cz_lsm_channel_colors
        ws.datasets.add(ArrayDataset2D(name="channel colors", data=ArrayData2D(first_page.cz_lsm_channel_colors)))

        # 3. cz_lsm_event_list: empty in example, skipped for now  # TODO

        return ws