omero.columns module

Concrete implementations of the omero.grid.Column type hierarchy which know how to convert themselves to PyTables types.

class columns.AbstractArrayColumn

Bases: AbstractColumn

Additional base logic for array columns

arrays()

Arrays of size 1 have to be converted to scalars, otherwise the column-to-row conversion in HdfStorage.append() will fail. This is messy, but I can’t think of a better way.

dtypes()

Overriding to correct for size.

settable(tbl)

Called by tables.py when first initializing columns. Can be used to complete further initialization.

class columns.AbstractColumn

Bases: object

Base logic for all columns

append(tbl)

Called by tables.py to give columns. By default, does nothing.

arrays()

Any method which does not use the “values” field will need to override this method.

dtypes()

Override this method if descriptor() doesn’t return the correct data type/size at initialisation- this is mostly a problem for array types

fromrows(rows, field_only=False)

Any method which does not use the “values” field will need to override this method.

getsize()

Any method which does not use the “values” field will need to override this method.

read(tbl, start, stop)

readCoordinates(tbl, rowNumbers)

setsize(size)

Any method which does not use the “values” field will need to override this method.

settable(tbl)

Called by tables.py when first initializing columns. Can be used to complete further initialization.

class columns.BoolColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, BoolColumn

descriptor(pos)

class columns.DatasetColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, DatasetColumn

descriptor(pos)

class columns.DoubleArrayColumnI(*args: Any, **kwargs: Any)

Bases: AbstractArrayColumn, DoubleArrayColumn

descriptor(pos)

class columns.DoubleColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, DoubleColumn

descriptor(pos)

class columns.FileColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, FileColumn

descriptor(pos)

class columns.FloatArrayColumnI(*args: Any, **kwargs: Any)

Bases: AbstractArrayColumn, FloatArrayColumn

descriptor(pos)

class columns.ImageColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, ImageColumn

descriptor(pos)

class columns.LongArrayColumnI(*args: Any, **kwargs: Any)

Bases: AbstractArrayColumn, LongArrayColumn

descriptor(pos)

class columns.LongColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, LongColumn

descriptor(pos)

class columns.MaskColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, MaskColumn

append(tbl)

Called by tables.py to give columns. By default, does nothing.

arrays()

Any method which does not use the “values” field will need to override this method.

descriptor(pos)

fromrows(rows, field_only=False)

Any method which does not use the “values” field will need to override this method.

getbytes(masks, rowNumbers)

getsize()

Any method which does not use the “values” field will need to override this method.

read(tbl, start, stop)

readCoordinates(tbl, rowNumbers)

setsize(size)

Any method which does not use the “values” field will need to override this method.

class columns.ObjectFactory(*args: Any, **kwargs: Any)

Bases: ObjectFactory

create(string)

destroy()

register(ic)

class columns.PlateColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, PlateColumn

descriptor(pos)

class columns.RoiColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, RoiColumn

descriptor(pos)

class columns.StringColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, StringColumn

StringColumns are actually numpy dtype ‘S’: “zero-terminated bytes (not recommended)” https://github.com/ome/omero-py/blob/v5.6.dev8/src/omero/columns.py#L269 https://docs.scipy.org/doc/numpy-1.15.1/reference/arrays.dtypes.html#specifying-and-constructing-data-types In any case HDF5 doesn’t seem to properly support unicode, and numexpr doesn’t even pretend to support it: - https://github.com/PyTables/PyTables/issues/499 - https://github.com/pydata/numexpr/issues/142 - https://github.com/pydata/numexpr/issues/150 - https://github.com/pydata/numexpr/issues/263 - https://github.com/pydata/numexpr/blob/v2.7.0/numexpr/necompiler.py#L340-L341

> import numexpr > a = “£” > numexpr.evaluate(‘a==”£”’) ValueError: unknown type str32 > b = “£”.encode() > numexpr.evaluate(‘b==”£”’) UnicodeEncodeError: ‘ascii’ codec can’t encode character ‘£’ in position 0: ordinal not in range(128)

You should be able to store/load unicode data but you can’t use unicode in a where condition

arrays()

Check for strings longer than the initialised column width This will always return bytes

descriptor(pos)

dtypes()

Overriding to correct for size. (Testing suggests this may not be necessary, the size appears to be correctly set at initialisation)

fromrows(rows, field_only=False)

Any method which does not use the “values” field will need to override this method.

settable(tbl)

Called by tables.py when first initializing columns. Can be used to complete further initialization.

class columns.WellColumnI(*args: Any, **kwargs: Any)

Bases: AbstractColumn, WellColumn

descriptor(pos)

columns.columns2definition(cols)

Takes a list of columns and converts them into a map from names to tables.* column descriptors