The QueryLayer¶

You can import from ./ or with bfly installed.

from bfly import QueryLayer

Or, in scripts in some directories:

# if './bfly/' in sys.path
from CoreLayer.AccessLayer import QueryLayer
# if './bfly/CoreLayer/' in sys.path
from AccessLayer import QueryLayer
# if './bfly/CoreLayer/AccessLayer' in sys.path
import QueryLayer

You can from QueryLayer import both ImageLayer and UtilityLayer.

# if './bfly/CoreLayer/AccessLayer' in sys.path
from QueryLayer import ImageLayer, UtilityLayer

QueryLayer¶

You can import QueryLayer from bfly, CoreLayer, and AccessLayer.

QueryLayer classes¶

class QueryLayer.DataQuery(*args, **keywords)[source]¶

Describe INPUT.METHODS.IMAGE_LIST requests

Parameters:

args (list) – unused
keywords (dict) –
- OUTPUT.INFO.PATH (str) –
  
  the path to the given Datasource
- INPUT.POSITION.Z (int) –
  
  the scaled resolution z start of the image request
- INPUT.POSITION.Y (int) –
  
  the scaled resolution y start of the image request
- INPUT.POSITION.X (int) –
  
  the scaled resolution x start of the image request
- INPUT.POSITION.DEPTH (int) –
  
  the requested scaled depth of the image request
- INPUT.POSITION.HEIGHT (int) –
  
  the requested scaled height of the image request
- INPUT.POSITION.WIDTH (int) –
  
  the requested scaled width of the image request
- INPUT.RESOLUTION.XY (int) –
  
  the number of halvings along the X and Y axes
- INPUT.RESOLUTION.FORMAT (str) –
  
  the requested output image file format
- INPUT.RESOLUTION.VIEW (str) –
  
  the requested coloring method for images

_basic_mime = 'image/{}'¶: default mime type for HTTP response

all_in_some(t_index)[source]¶

The tile region inside the scaled target volume

Parameters:	t_index (numpy.ndarray) – The 3x1 tile offset of a tile from the origin
Returns:	The 2x3 offset of `target_bounds()` from the origin of `tile_origin` repetitions of `blocksize()` such that either row gives an upper or lower corner of the total target volume within the requested `t_index` tile.
Return type:	numpy.ndarray

some_in_all(t_origin, t_shape)[source]¶

The target subvolume needed by a given tile

Parameters:	t_origin (numpy.ndarray) – The 3x1 origin of the tile in the scaled target image t_shape (numpy.ndarray) – The 3x1 shape of the tile in the scaled target image
Returns:	The 2x3 joining of `t_origin` and `t_shape` offset from `target_origin()` and clipped by `target_shape()` such that either row gives the upper or lower corner of the tile in the full scaled target volume.
Return type:	numpy.ndarray

blocksize¶

get the size of a Datasource tile

Returns:	The 3x1 block value for this resolution
Return type:	numpy.ndarray

blocksizes¶

get the sizes of each Datasource tile

Returns:	The Nx3 block values from `RUNTIME.IMAGE`
Return type:	numpy.ndarray

dtype¶

return actual numpy dtype object

Returns:	from the type value from `OUTPUT.INFO`
Return type:	numpy.dtype

key¶

return the key for the database

Returns:	the path value from `OUTPUT.INFO`
Return type:	str

scales¶

convert a number of halvings to a scale array

Returns:	A 3x1 array of the target/source voxel ratios
Return type:	numpy.ndarray

source_off¶

Get the data offset at full resolution

Returns:	[z_off, y_off, x_off]
Return type:	numpy.ndarray

source_origin¶

Full source offset of the requested origin

Returns:	The 3x1 `target_origin()` multiplied by the voxel scales from `scales()`.
Return type:	numpy.ndarray

source_shape¶

Full source size of the requested volume

Returns:	The 3x1 `target_shape()` multiplied by the voxel scales from `scales()`.
Return type:	numpy.ndarray

target_bounds¶

Scaled bounding box of the requested volume

Returns:	The 2x3 joining of `target_origin()` and `target_shape()` transposed such that each row gives one of two bounding corners.
Return type:	numpy.ndarray

target_off¶

Get the data offset at target resolution

Returns:	[z_off, y_off, x_off]
Return type:	numpy.ndarray

target_origin¶

Scaled target offset of the requested origin

Returns:	The 3x1 block of Z, Y, and X in INPUT.POSITION
Return type:	numpy.ndarray

target_shape¶

Scaled target size of the requested volume

Returns:	The 3x1 block of DEPTH, HEIGHT, WIDTH values from INPUT.POSITION
Return type:	numpy.ndarray

tile_bounds¶

Bounding box given by counting whole tiles

Returns:	The 2x3 joining of the lowermost tile and the uppermost tile transposed such that a row gives the upper or lower limit of tiles.
Return type:	numpy.ndarray

tile_shape¶

The ranges of tiles needed for the request

Returns:	The 3x1 difference of `tile_bounds()` to give three numbers that count the number of tiles needed in all three directions.
Return type:	numpy.ndarray

class QueryLayer.InfoQuery(*args, **keywords)[source]¶

Describe INPUT.METHODS.INFO_LIST requests

Parameters:	args (list) – unused keywords (dict) – OUTPUT.INFO.NAMES (list) – all answers to group and feature queries OUTPUT.INFO.PATH (str) – the path to the given `Datasource` OUTPUT.INFO.SIZE (numpy.ndarray) – 3x1 for full volume shape OUTPUT.INFO.CHANNEL (str) – The name of the most specific image group OUTPUT.INFO.FORMAT (str) – The requested format for HTTP output

_basic_mime = 'text/plain'¶

_form = [{'indent': 4, 'cls': <class 'QueryLayer.InfoQuery.NumpyEncoder'>}, {'default_flow_style': False}]¶: Options for all _write

_write¶: list – All write functions for output formats

dump_dataset(all_channels)[source]¶

format result() with write

Parameters:	all_channels (list) – Contains keywords for each channel
Returns:	The `result()` formatted as a string
Return type:	str

set_channel(channel)[source]¶

Change the query to a given channel

channel: dict: self.OUTPUT.INFO.CHANNEL: str self.OUTPUT.INFO.PATH: str

websocket_edit(msg={})[source]¶

Change internal state based on websocket message

Parameters:	msg (dict) – The commands controlling the message
Returns:	The values changed in the internal state
Return type:	dict

websocket_result(action)[source]¶

Results to a websocket query

Parameters:	action (str) – The type of result to send to client
Returns:	The result to send to client
Return type:	dict

channels¶

return the list of channels

Returns:	All channel dictionaries
Return type:	list

dump¶

format result() with write

Returns:	The `result()` formatted as a string
Return type:	str

get_format¶

get the index of the output format

Returns:	index for `_form` or `_write`
Return type:	int

result¶

get the answer to the info query

Returns:

dict – For the method INPUT.METHODS.META.NAME:
- OUTPUT.INFO.PATH.NAME –
  
  (str) OUTPUT.INFO.PATH.VALUE
- OUTPUT.INFO.TYPE.NAME –
  
  (str) OUTPUT.INFO.TYPE.VALUE
- OUTPUT.INFO.SIZE.NAME –
  
  (dict) OUTPUT.INFO.SIZE.VALUE
- OUTPUT.INFO.CHANNEL.NAME –
  
  (str) OUTPUT.INFO.CHANNEL.VALUE
list – For all methods INPUT.METHODS.GROUP_LIST: For the method INPUT.METHODS.FEATURE.NAME:
- all (str) in OUTPUT.INFO.NAMES.VALUE

class QueryLayer.TileQuery(*args, **keywords)[source]¶

Describe requsts for tiles from a Datasource

Parameters:	args (list) – (`DataQuery`) gives the full volume request (numpy.ndarray) 3x1 tile offset of the given tile (numpy.ndarray) 2x3 subregion within given tile keywords (dict) – unused

_basic_mime = 'text/plain'¶

source_list¶: list – from RUNTIME .IMAGE.SOURCE.LIST

SOURCES¶

dict –

source_list[0]: HDF5
source_list[1]: BossGrid

get_source(name)[source]¶

Gets a Datasource from SOURCES

Parameters:	name (str) – The key to a `Datasource` subclass as listed in `SOURCES`
Returns:	A `Datasource` subclass given by `name`
Return type:	`Datasource`

all_scales¶

A scale array from RUNTIME.TILE

Returns:	A 3x1 array of the target/source voxel ratios
Return type:	numpy.ndarray

index_zyx¶

The tile offsets needed to get the tile

Returns:	3x1 tile offset of the given tile
Return type:	numpy.ndarray

key¶

return the key for the database

Returns:	the path value from `path()` joined with the tile_values from `index_zyx()` and `all_scales()`
Return type:	str

my_source¶

return the source that loads this tile

Returns:	the source value from `RUNTIME.IMAGE` as an actual `Datasource` that can load a `TileQuery`.
Return type:	`Datasource`

pixels_kji¶

The pixels bounds needed from the tile

Returns:	The 2x3 subregion within the given tile
Return type:	numpy.ndarray

preload_source¶

Get a valid_source() with a measured size

Returns:	Empty if no `valid_source()` `RUNTIME` `.CACHE.META.NAME` (int) – The byte size of this dictionary `RUNTIME` `.IMAGE.SOURCE.NAME` (str) – The `Datasource` subclass `RUNTIME` `.IMAGE.BLOCK.NAME` (numpy.ndarray) – 3x1 for any give tile shape `OUTPUT` `.INFO.TYPE.NAME` (str) – numpy dtype of any given tile `OUTPUT` `.INFO.SIZE.NAME` (numpy.ndarray) – 3x1 for full volume shape
Return type:	dict

source_bounds¶

The bounds of query in full coordinates

Returns:	2x3 full source pixel bounds in tile
Return type:	numpy.ndarray

source_tile_bounds¶

The full image bounds of the tile

Returns:	2x3 full image pixel bounds
Return type:	numpy.ndarray

target_bounds¶

The bounds of query in target coordinates

Returns:	numpy.ndarray 2x3 target pixel boundaries

target_origin¶

The origin of query in target coordinates

Returns:	3x1 target pixel lower bound
Return type:	numpy.ndarray

target_tile_bounds¶

The scaled target bounds of the tile

Returns:	2x3 scaled target pixel bounds
Return type:	numpy.ndarray

tile¶

Load the requested tile from the source

Returns:	The full image loaded by `my_source()`
Return type:	numpy.ndarray

valid_source¶

Call preload_source for all source_list

Returns:	Empty if no `Datasource.preload_source()` can find this filname to give a valid volume. `RUNTIME` `.IMAGE.SOURCE.NAME` (str) – The `Datasource` subclass `RUNTIME` `.IMAGE.BLOCK.NAME` (numpy.ndarray) – 3x1 for any give tile shape `OUTPUT` `.INFO.TYPE.NAME` (str) – numpy dtype of any given tile `OUTPUT` `.INFO.SIZE.NAME` (numpy.ndarray) – 3x1 for full volume shape
Return type:	dict

Query base class¶

class QueryLayer.Query(*args, **keywords)[source]¶

Describe content of AccessLayer requests

Parameters:	args (list) – See `TileQuery` for use of positional arguments keywords (dict) – Each derived class may use any keyword arguments given keys each match a NAME of a NamedStruct`in :class:`INPUT, `RUNTIME`, or `OUTPUT`.

_basic_mime = 'text/plain'¶: default mime type for HTTP response

INPUT¶: INPUT

RUNTIME¶: RUNTIME

OUTPUT¶: OUTPUT

keywords¶: dict – Taken from keywords argument

check_list(whitelist, value, term)[source]¶

Checks that a value is in a given list

Parameters:	whitelist (list-like) – The list of desired values value (anything) – The value should be in the `whitelist` term (str) – The name of the attribute to test

check_vector(value, term='list', dims=3)[source]¶

Checks that a vector has three dimensions

Parameters:	term (str) – The name of the attribute to test value (list-like) – The value should be a list or array dims (int) – The desired number of dimensions

set_key(struct, key, empty='')[source]¶

Copy the key from keywords to the structure

Parameters:	struct (`NamelessStruct`) – Where to store a value from `keywords` key (str) – The key to transfer from `keywords` empty (anything) – The value if `keywords` has no key``and the ``key has no default in the struct

update_keys(keywords)[source]¶

Update stored keywords

Parameters:	keywords (dict) – Updates stored keywords

update_source(keywords)[source]¶

Set all attribute values to match keywords

Parameters:

keywords (dict) –

RUNTIME .IMAGE.SOURCE.NAME

(str) – The subclass of Datasource
RUNTIME .IMAGE.BLOCK.NAME

(numpy.ndarray) – 3x1 for any given tile shape
OUTPUT .INFO.TYPE.NAME

(str) – numpy dtype of any given tile
OUTPUT .INFO.SIZE.NAME

(numpy.ndarray) – 3x1 for full volume shape

Keyword arguments only for HDF5

RUNTIME .IMAGE.SOURCE.HDF5.OUTER.NAME

(str) – The direct filename to an hdf5 file
RUNTIME .IMAGE.SOURCE.HDF5.INNER.NAME

(str) – The dataset in the file with image data

Keyword arguments if editable * RUNTIME .IMAGE.MERGE.NAME

lil_matrix – The matrix of merged ids

RUNTIME .IMAGE.MERGE.NAME

lil_matrix – The matrix of split regions

edit_path¶

return the path to edit the volume

Returns:	the path under the EDIT_PATH root
Return type:	str

is_data¶

Checks whether the method requests images

Returns:
Return type:	bool

is_dataset¶

Checks if requesting dataset info

Returns:
Return type:	bool

is_group¶

Checks whether the method requests groups

Returns:
Return type:	bool

is_websocket¶

Checks whether websocket method

Returns:
Return type:	bool

key¶

return the key for the database

Returns:	the path value from `OUTPUT.INFO`
Return type:	str

mime_type¶

Gets the mime type for the file_type

The mime type is from INPUT.IMAGE if is_data(), or INPUT.INFO otherwise.

Returns:	The mime type from `mimetypes.types_map`
Return type:	str

path¶

return the path to the whole volume

Returns:	the path value from `OUTPUT.INFO`
Return type:	str

ImageLayer¶

You can import ImageLayer from bfly, CoreLayer, AccessLayer, and QueryLayer.

ImageLayer classes¶

class ImageLayer.HDF5[source]¶

Loads images from hdf5 files

_meta_files = ['.json']¶: All extensions of files pointing to h5

_read¶: list – All load functions for _meta_files

static get_details(h5_info, file_dict)[source]¶

Get all needed h5 file info from a pointer file

Parameters:	file_dict (dict) – Contains keys for INNER, OUTER, and OFF values
Returns:	All INNER, OUTER, OFF values in a flat list
Return type:	list

static load_file(h5_file)[source]¶

Load the needed volume from a single h5 File

Parameters:	t_query (`TileQuery`) – With file path and image position

static load_info(t_query)[source]¶

Gets the h5 volume filename and datapath

If the t_query path has an extension in the _meta_files and the file contains RUNTIME.IMAGE.SOURCE.HDF5.OUTER.NAME and RUNTIME.IMAGE.SOURCE.HDF5.INNER.NAME keys, then the values of those keys are returned. If any of those statements is not true, then the original t_query path is returned along with the default dataset given by RUNTIME.IMAGE.SOURCE.HDF5.INNER.VALUE.

Parameters:	t_query (`TileQuery`) – Only the file path is needed
Returns:	The direct filename to an hdf5 file The datset in the file with image data
Return type:	list

static load_tile(t_query)[source]¶

load a single tile (image)

Gets the image path from the TileQuery.RUNTIME. IMAGE attribute.

Gets the position of the image with the whole volume from TileQuery.all_scales(), TileQuery.tile_origin(), and TileQuery.blocksize().

Parameters:	t_query (`TileQuery`) – With file path and image position
Returns:	An image array that may be as large as an entire full resolution slice of the whole hdf5 volume. Based on the value of `TileQuery.all_scales()`, this array will likely be downsampled by to a small fraction of the full tile resolution.
Return type:	np.ndarray

static preload_source(t_query)[source]¶

load info from example tile (image)

Calls valid_path() to get filename and inner dataset path for the full h5 image volume.

Then gets three needed values from the given path from the TileQuery t_query

Parameters:	t_query (`TileQuery`) – Only the file path is needed
Returns:	Will be empty if `valid_path()` findsthis filname to not give a valid h5 volume. `RUNTIME` `.IMAGE.BLOCK.NAME` (numpy.ndarray) – 3x1 for any give tile shape `OUTPUT` `.INFO.TYPE.NAME` (str) – numpy dtype of any given tile `OUTPUT` `.INFO.SIZE.NAME` (numpy.ndarray) – 3x1 for full volume shape
Return type:	dict

static valid_path(t_query)[source]¶

Check if filename can access h5 data

The filename can be a path to a json file that lists an h5 file and dataset path, or the filename can be a direct path to an h5 file. In either case the ‘outer’ file path directly to the h5 file and the ‘inner’ dataset path will be returned.

Parameters:	t_query (`TileQuery`) – Only the file path is needed
Returns:	Empty if not a valid h5 volume `RUNTIME` `.IMAGE.SOURCE.HDF5.OUTER.NAME` (str) – The direct filename to an hdf5 file `RUNTIME` `.IMAGE.SOURCE.HDF5.INNER.NAME` (str) – The datset in the file with image data
Return type:	dict

Datasource base class¶

class ImageLayer.Datasource[source]¶

Loads images from files on the server

static load_tile(t_query)[source]¶

load a single tile (image)

Parameters:	t_query (`TileQuery`) – With file path and image position

static preload_source(t_query)[source]¶

load info from example tile (image)

Parameters:	t_query (`TileQuery`) – Only the file path is needed
Returns:	`RUNTIME` `.IMAGE.BLOCK.NAME` (numpy.ndarray) – 3x1 for any given tile shape `OUTPUT` `.INFO.TYPE.NAME` (str) – numpy dtype of any given tile `OUTPUT` `.INFO.SIZE.NAME` (numpy.ndarray) – 3x1 for full volume shape `OUTPUT` `.IMAGE.MERGE.NAME` (lil_matrix) – A matrix of merged ids `OUTPUT` `.IMAGE.SPLIT.NAME` (lil_matrix) – A matrix of plit regions `OUTPUT` `.IMAGE.ERROR.NAME` dict – All error messages
Return type:	dict