def getShuffle(app, dset_id, dset_json):
shuffle_size = 0
if isShuffle(dset_json):
type_json = dset_json["type"]
item_size = getItemSize(type_json)
if item_size == 'H5T_VARIABLE':
log.warn(
f"shuffle filter can't be used on variable datatype for datasset: {dset_id}"
)
else:
shuffle_size = item_size
log.debug(f"got shuffle_size: {shuffle_size}")
else:
log.debug("isShuffle is false")
if shuffle_size > 1:
shuffle_map = app['shuffle_map']
if dset_id not in shuffle_map:
# save the shuffle size so the lazy chunk writer can access it
shuffle_map[dset_id] = shuffle_size
log.debug(f"update shuffle_map {dset_id}: {shuffle_size}")
return shuffle_size
async def PUT_AttributeValue(request):
"""HTTP method to update an attributes data"""
log.request(request)
log.info("PUT_AttributeValue")
app = request.app
collection = getRequestCollectionName(request) # returns datasets|groups|datatypes
obj_id = request.match_info.get('id')
if not obj_id:
msg = "Missing object id"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if not isValidUuid(obj_id, obj_class=collection):
msg = f"Invalid object id: {obj_id}"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
attr_name = request.match_info.get('name')
log.debug(f"Attribute name: [{attr_name}]")
validateAttributeName(attr_name)
log.info(f"PUT Attribute Value id: {obj_id} name: {attr_name}")
username, pswd = getUserPasswordFromRequest(request)
# write actions need auth
await validateUserPassword(app, username, pswd)
if not request.has_body:
msg = "PUT AttributeValue with no body"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
domain = getDomainFromRequest(request)
if not isValidDomain(domain):
msg = f"Invalid domain: {domain}"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
bucket = getBucketForDomain(domain)
# get domain JSON
domain_json = await getDomainJson(app, domain)
if "root" not in domain_json:
log.error(f"Expected root key for domain: {domain}")
raise HTTPInternalServerError()
# TBD - verify that the obj_id belongs to the given domain
await validateAction(app, domain, obj_id, username, "update")
req = getDataNodeUrl(app, obj_id)
req += '/' + collection + '/' + obj_id + "/attributes/" + attr_name
log.debug("get Attribute: " + req)
params = {}
if bucket:
params["bucket"] = bucket
dn_json = await http_get(app, req, params=params)
log.debug("got attributes json from dn for obj_id: " + str(obj_id))
log.debug(f"got dn_json: {dn_json}")
attr_shape = dn_json["shape"]
if attr_shape["class"] == 'H5S_NULL':
msg = "Null space attributes can not be updated"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
np_shape = getShapeDims(attr_shape)
type_json = dn_json["type"]
np_dtype = createDataType(type_json) # np datatype
request_type = "json"
if "Content-Type" in request.headers:
# client should use "application/octet-stream" for binary transfer
content_type = request.headers["Content-Type"]
if content_type not in ("application/json", "application/octet-stream"):
msg = f"Unknown content_type: {content_type}"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if content_type == "application/octet-stream":
log.debug("PUT AttributeValue - request_type is binary")
request_type = "binary"
else:
log.debug("PUT AttribueValue - request type is json")
binary_data = None
if request_type == "binary":
item_size = getItemSize(type_json)
if item_size == 'H5T_VARIABLE':
msg = "Only JSON is supported for variable length data types"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
# read binary data
binary_data = await request.read()
if len(binary_data) != request.content_length:
msg = f"Read {len(binary_data)} bytes, expecting: {request.content_length}"
log.error(msg)
raise HTTPInternalServerError()
arr = None # np array to hold request data
if binary_data:
npoints = getNumElements(np_shape)
if npoints*item_size != len(binary_data):
msg = "Expected: " + str(npoints*item_size) + " bytes, but got: " + str(len(binary_data))
log.warn(msg)
raise HTTPBadRequest(reason=msg)
arr = np.fromstring(binary_data, dtype=np_dtype)
arr = arr.reshape(np_shape) # conform to selection shape
# convert to JSON for transmission to DN
data = arr.tolist()
value = bytesArrayToList(data)
else:
body = await request.json()
if "value" not in body:
msg = "PUT attribute value with no value in body"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
value = body["value"]
# validate that the value agrees with type/shape
try:
arr = jsonToArray(np_shape, np_dtype, value)
except ValueError:
msg = "Bad Request: input data doesn't match selection"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
log.info(f"Got: {arr.size} array elements")
# ready to add attribute now
attr_json = {}
attr_json["type"] = type_json
attr_json["shape"] = attr_shape
attr_json["value"] = value
req = getDataNodeUrl(app, obj_id)
req += '/' + collection + '/' + obj_id + "/attributes/" + attr_name
log.info(f"PUT Attribute Value: {req}")
dn_json["value"] = value
params = {}
params = {"replace": 1} # let the DN know we can overwrite the attribute
if bucket:
params["bucket"] = bucket
put_rsp = await http_put(app, req, params=params, data=attr_json)
log.info(f"PUT Attribute Value resp: {put_rsp}")
hrefs = [] # TBD
req_rsp = { "hrefs": hrefs }
# attribute creation successful
resp = await jsonResponse(request, req_rsp)
log.response(request, resp=resp)
return resp
async def POST_Dataset(request):
"""HTTP method to create a new dataset object"""
log.request(request)
app = request.app
username, pswd = getUserPasswordFromRequest(request)
# write actions need auth
await validateUserPassword(app, username, pswd)
if not request.has_body:
msg = "POST Datasets with no body"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
body = await request.json()
# get domain, check authorization
domain = getDomainFromRequest(request)
if not isValidDomain(domain):
msg = "Invalid host value: {}".format(domain)
log.warn(msg)
raise HTTPBadRequest(reason=msg)
domain_json = await getDomainJson(app, domain, reload=True)
root_id = domain_json["root"]
aclCheck(domain_json, "create",
username) # throws exception if not allowed
if "root" not in domain_json:
msg = "Expected root key for domain: {}".format(domain)
log.warn(msg)
raise HTTPBadRequest(reason=msg)
#
# validate type input
#
if "type" not in body:
msg = "POST Dataset has no type key in body"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
datatype = body["type"]
if isinstance(datatype, str) and datatype.startswith("t-"):
# Committed type - fetch type json from DN
ctype_id = datatype
log.debug("got ctypeid: {}".format(ctype_id))
ctype_json = await getObjectJson(app, ctype_id)
log.debug("ctype: {}".format(ctype_json))
if ctype_json["root"] != root_id:
msg = "Referenced committed datatype must belong in same domain"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
datatype = ctype_json["type"]
# add the ctype_id to type type
datatype["id"] = ctype_id
elif isinstance(datatype, str):
try:
# convert predefined type string (e.g. "H5T_STD_I32LE") to
# corresponding json representation
datatype = getBaseTypeJson(datatype)
log.debug("got datatype: {}".format(datatype))
except TypeError:
msg = "POST Dataset with invalid predefined type"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
validateTypeItem(datatype)
item_size = getItemSize(datatype)
#
# Validate shape input
#
dims = None
shape_json = {}
if "shape" not in body:
shape_json["class"] = "H5S_SCALAR"
else:
shape = body["shape"]
if isinstance(shape, int):
shape_json["class"] = "H5S_SIMPLE"
dims = [
shape,
]
shape_json["dims"] = dims
elif isinstance(shape, str):
# only valid string value is H5S_NULL
if shape != "H5S_NULL":
msg = "POST Datset with invalid shape value"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
shape_json["class"] = "H5S_NULL"
elif isinstance(shape, list):
if len(shape) == 0:
shape_json["class"] = "H5S_SCALAR"
else:
shape_json["class"] = "H5S_SIMPLE"
shape_json["dims"] = shape
dims = shape
else:
msg = "Bad Request: shape is invalid"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if dims is not None:
for i in range(len(dims)):
extent = dims[i]
if not isinstance(extent, int):
msg = "Invalid shape type"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if extent < 0:
msg = "shape dimension is negative"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
maxdims = None
if "maxdims" in body:
if dims is None:
msg = "Maxdims cannot be supplied if space is NULL"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
maxdims = body["maxdims"]
if isinstance(maxdims, int):
dim1 = maxdims
maxdims = [dim1]
elif isinstance(maxdims, list):
pass # can use as is
else:
msg = "Bad Request: maxdims is invalid"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if len(dims) != len(maxdims):
msg = "Maxdims rank doesn't match Shape"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if maxdims is not None:
for extent in maxdims:
if not isinstance(extent, int):
msg = "Invalid maxdims type"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if extent < 0:
msg = "maxdims dimension is negative"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if len(maxdims) != len(dims):
msg = "Bad Request: maxdims array length must equal shape array length"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
shape_json["maxdims"] = []
for i in range(len(dims)):
maxextent = maxdims[i]
if not isinstance(maxextent, int):
msg = "Bad Request: maxdims must be integer type"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
elif maxextent == 0:
# unlimited dimension
shape_json["maxdims"].append(0)
elif maxextent < dims[i]:
msg = "Bad Request: maxdims extent can't be smaller than shape extent"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
else:
shape_json["maxdims"].append(maxextent)
layout = None
min_chunk_size = int(config.get("min_chunk_size"))
max_chunk_size = int(config.get("max_chunk_size"))
if 'creationProperties' in body:
creationProperties = body["creationProperties"]
if 'layout' in creationProperties:
layout = creationProperties["layout"]
validateChunkLayout(shape_json, item_size, layout)
if layout is None and shape_json["class"] != "H5S_NULL":
# default to chunked layout
layout = {"class": "H5D_CHUNKED"}
if layout and layout["class"] == 'H5D_CONTIGUOUS_REF':
chunk_dims = getContiguousLayout(shape_json,
item_size,
chunk_min=min_chunk_size,
chunk_max=max_chunk_size)
layout["dims"] = chunk_dims
log.debug(f"autoContiguous layout: {layout}")
if layout and layout["class"] == 'H5D_CHUNKED' and "dims" not in layout:
# do autochunking
chunk_dims = guessChunk(shape_json, item_size)
layout["dims"] = chunk_dims
log.debug(f"initial autochunk layout: {layout}")
if layout and layout["class"] == 'H5D_CHUNKED':
chunk_dims = layout["dims"]
chunk_size = getChunkSize(chunk_dims, item_size)
log.debug("chunk_size: {}, min: {}, max: {}".format(
chunk_size, min_chunk_size, max_chunk_size))
# adjust the chunk shape if chunk size is too small or too big
adjusted_chunk_dims = None
if chunk_size < min_chunk_size:
log.debug(
"chunk size: {} less than min size: {}, expanding".format(
chunk_size, min_chunk_size))
adjusted_chunk_dims = expandChunk(chunk_dims,
item_size,
shape_json,
chunk_min=min_chunk_size,
layout_class=layout["class"])
elif chunk_size > max_chunk_size:
log.debug(
"chunk size: {} greater than max size: {}, expanding".format(
chunk_size, max_chunk_size, layout_class=layout["class"]))
adjusted_chunk_dims = shrinkChunk(chunk_dims,
item_size,
chunk_max=max_chunk_size)
if adjusted_chunk_dims:
log.debug(
f"requested chunk_dimensions: {chunk_dims} modified dimensions: {adjusted_chunk_dims}"
)
layout["dims"] = adjusted_chunk_dims
if layout and layout["class"] in ('H5D_CHUNKED_REF',
'H5D_CHUNKED_REF_INDIRECT'):
chunk_dims = layout["dims"]
chunk_size = getChunkSize(chunk_dims, item_size)
log.debug("chunk_size: {}, min: {}, max: {}".format(
chunk_size, min_chunk_size, max_chunk_size))
# adjust the chunk shape if chunk size is too small or too big
if chunk_size < min_chunk_size:
log.warn(
"chunk size: {} less than min size: {} for H5D_CHUNKED_REF dataset"
.format(chunk_size, min_chunk_size))
elif chunk_size > max_chunk_size:
log.warn(
"chunk size: {} greater than max size: {}, for H5D_CHUNKED_REF dataset"
.format(chunk_size,
max_chunk_size,
layout_class=layout["class"]))
link_id = None
link_title = None
if "link" in body:
link_body = body["link"]
if "id" in link_body:
link_id = link_body["id"]
if "name" in link_body:
link_title = link_body["name"]
if link_id and link_title:
log.info("link id: {}".format(link_id))
# verify that the referenced id exists and is in this domain
# and that the requestor has permissions to create a link
await validateAction(app, domain, link_id, username, "create")
dset_id = createObjId("datasets", rootid=root_id)
log.info("new dataset id: {}".format(dset_id))
dataset_json = {
"id": dset_id,
"root": root_id,
"type": datatype,
"shape": shape_json
}
if "creationProperties" in body:
# TBD - validate all creationProperties
creationProperties = body["creationProperties"]
if "fillValue" in creationProperties:
# validate fill value compatible with type
dt = createDataType(datatype)
fill_value = creationProperties["fillValue"]
if isinstance(fill_value, list):
fill_value = tuple(fill_value)
try:
np.asarray(fill_value, dtype=dt)
except (TypeError, ValueError):
msg = "Fill value {} not compatible with dataset type: {}".format(
fill_value, datatype)
log.warn(msg)
raise HTTPBadRequest(reason=msg)
dataset_json["creationProperties"] = creationProperties
if layout is not None:
dataset_json["layout"] = layout
log.debug("create dataset: " + json.dumps(dataset_json))
req = getDataNodeUrl(app, dset_id) + "/datasets"
post_json = await http_post(app, req, data=dataset_json)
# create link if requested
if link_id and link_title:
link_json = {}
link_json["id"] = dset_id
link_json["class"] = "H5L_TYPE_HARD"
link_req = getDataNodeUrl(app, link_id)
link_req += "/groups/" + link_id + "/links/" + link_title
log.info("PUT link - : " + link_req)
put_rsp = await http_put(app, link_req, data=link_json)
log.debug("PUT Link resp: {}".format(put_rsp))
# dataset creation successful
resp = await jsonResponse(request, post_json, status=201)
log.response(request, resp=resp)
return resp
async def GET_AttributeValue(request):
"""HTTP method to return an attribute value"""
log.request(request)
app = request.app
log.info("GET_AttributeValue")
collection = getRequestCollectionName(request) # returns datasets|groups|datatypes
obj_id = request.match_info.get('id')
if not obj_id:
msg = "Missing object id"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
if not isValidUuid(obj_id, obj_class=collection):
msg = f"Invalid object id: {obj_id}"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
attr_name = request.match_info.get('name')
validateAttributeName(attr_name)
username, pswd = getUserPasswordFromRequest(request)
if username is None and app['allow_noauth']:
username = "******"
else:
await validateUserPassword(app, username, pswd)
domain = getDomainFromRequest(request)
if not isValidDomain(domain):
msg = f"Invalid domain value: {domain}"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
bucket = getBucketForDomain(domain)
# get domain JSON
domain_json = await getDomainJson(app, domain)
if "root" not in domain_json:
log.error(f"Expected root key for domain: {domain}")
raise HTTPBadRequest(reason="Unexpected Error")
# TBD - verify that the obj_id belongs to the given domain
await validateAction(app, domain, obj_id, username, "read")
req = getDataNodeUrl(app, obj_id)
req += '/' + collection + '/' + obj_id + "/attributes/" + attr_name
log.debug("get Attribute: " + req)
params = {}
if bucket:
params["bucket"] = bucket
dn_json = await http_get(app, req, params=params)
log.debug("got attributes json from dn for obj_id: " + str(dn_json))
attr_shape = dn_json["shape"]
log.debug(f"attribute shape: {attr_shape}")
if attr_shape["class"] == 'H5S_NULL':
msg = "Null space attributes can not be read"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
accept_type = getAcceptType(request)
response_type = accept_type # will adjust later if binary not possible
type_json = dn_json["type"]
shape_json = dn_json["shape"]
item_size = getItemSize(type_json)
if item_size == 'H5T_VARIABLE' and accept_type != "json":
msg = "Client requested binary, but only JSON is supported for variable length data types"
log.info(msg)
response_type = "json"
if response_type == "binary":
arr_dtype = createDataType(type_json) # np datatype
np_shape = getShapeDims(shape_json)
try:
arr = jsonToArray(np_shape, arr_dtype, dn_json["value"])
except ValueError:
msg = "Bad Request: input data doesn't match selection"
log.warn(msg)
raise HTTPBadRequest(reason=msg)
output_data = arr.tobytes()
log.debug(f"GET AttributeValue - returning {len(output_data)} bytes binary data")
# write response
try:
resp = StreamResponse()
resp.content_type = "application/octet-stream"
resp.content_length = len(output_data)
# allow CORS
resp.headers['Access-Control-Allow-Origin'] = '*'
resp.headers['Access-Control-Allow-Methods'] = "GET, POST, DELETE, PUT, OPTIONS"
resp.headers['Access-Control-Allow-Headers'] = "Content-Type, api_key, Authorization"
await resp.prepare(request)
await resp.write(output_data)
except Exception as e:
log.error(f"Got exception: {e}")
raise HTTPInternalServerError()
finally:
await resp.write_eof()
else:
resp_json = {}
if "value" in dn_json:
resp_json["value"] = dn_json["value"]
hrefs = []
obj_uri = '/' + collection + '/' + obj_id
attr_uri = obj_uri + '/attributes/' + attr_name
hrefs.append({'rel': 'self', 'href': getHref(request, attr_uri)})
hrefs.append({'rel': 'home', 'href': getHref(request, '/')})
hrefs.append({'rel': 'owner', 'href': getHref(request, obj_uri)})
resp_json["hrefs"] = hrefs
resp = await jsonResponse(request, resp_json)
log.response(request, resp=resp)
return resp