Newer
Older
"""returns the list of files including their directories as an array of strings. Just folders are not
listed.
"""
files = []
for file in self.get_file_list(recursive=True):
if file['isDirectory']:
pass
else:
files.append(file['pathInDataSet'])
return files
def get_files(self, start_folder='/'):
"""Returns a DataFrame of all files in this dataset
"""
def createRelativePath(pathInDataSet):
if self.shareId is None:
return ''
else:
return os.path.join(self.shareId, self.location, pathInDataSet)
def signed_to_unsigned(sig_int):
"""openBIS delivers crc32 checksums as signed integers.
If the number is negative, we just have to add 2**32
We display the hex number to match with the classic UI
"""
if sig_int < 0:
sig_int += 2**32
return "%x"%(sig_int & 0xFFFFFFFF)
files = self.get_file_list(start_folder=start_folder)
df = DataFrame(files)
df['relativePath'] = df['pathInDataSet'].map(createRelativePath)
df['crc32Checksum'] = df['crc32Checksum'].fillna(0.0).astype(int).map(signed_to_unsigned)
return df[['isDirectory', 'pathInDataSet', 'fileSize', 'crc32Checksum']]
def get_file_list(self, recursive=True, start_folder="/"):
"""Lists all files of a given dataset. You can specifiy a start_folder other than "/".
By default, all directories and their containing files are listed recursively. You can
turn off this option by setting recursive=False.
"""
request = {
"method" : "listFilesForDataSet",
"params" : [
self.openbis.token,
"id":"1"
self.data["dataStore"]["downloadUrl"] + '/datastore_server/rmi-dss-api-v1.json',
json.dumps(request),
verify=self.openbis.verify_certificates
)
data = resp.json()
if 'error' in data:
Swen Vermeul
committed
raise ValueError('Error from openBIS: ' + data['error'] )
elif 'result' in data:
return data['result']
Swen Vermeul
committed
raise ValueError('request to openBIS did not return either result nor error')
Swen Vermeul
committed
raise ValueError('internal error while performing post request')
class AttrHolder():
""" General class for both samples and experiments that hold all common attributes, such as:
- space
Swen Vermeul
committed
- experiment (sample)
- samples (experiment)
Swen Vermeul
committed
- parents (sample, dataset)
- children (sample, dataset)
- tags
"""
def __init__(self, openbis_obj, entity, type=None):
self.__dict__['_openbis'] = openbis_obj
self.__dict__['_entity'] = entity
if type is not None:
self.__dict__['_allowed_attrs'] = _definitions(entity)['attrs']
self.__dict__['_identifier'] = None
self.__dict__['_is_new'] = True
Swen Vermeul
committed
def __call__(self, data):
self.__dict__['_is_new'] = False
Swen Vermeul
committed
for attr in self._allowed_attrs:
if attr in ["code","permId","identifier",
"type", "container","components"]:
Swen Vermeul
committed
self.__dict__['_'+attr] = data.get(attr, None)
Swen Vermeul
committed
d = data.get(attr, None)
if d is not None:
d = d['permId']
self.__dict__['_'+attr] = d
elif attr in ["sample", "experiment", "project"]:
Swen Vermeul
committed
d = data.get(attr, None)
if d is not None:
d = d['identifier']
self.__dict__['_'+attr] = d
elif attr in ["parents","children","samples"]:
Swen Vermeul
committed
self.__dict__['_'+attr] = []
for item in data[attr]:
if 'identifier' in item:
self.__dict__['_'+attr].append(item['identifier'])
elif 'permId' in item:
self.__dict__['_'+attr].append(item['permId'])
Swen Vermeul
committed
Swen Vermeul
committed
for item in data[attr]:
Swen Vermeul
committed
"code": item['code'],
"@type": "as.dto.tag.id.TagCode"
})
self.__dict__['_tags'] = tags
import copy
self.__dict__['_prev_tags'] = copy.deepcopy(tags)
else:
self.__dict__['_'+attr] = data.get(attr, None)
def _new_attrs(self):
defs = _definitions(self.entity)
attr2ids = _definitions('attr2ids')
new_obj = {
"@type" : "as.dto.{}.create.{}Creation".format(self.entity.lower(), self.entity)
}
for attr in defs['attrs_new']:
items = None
if attr == 'type':
new_obj['typeId'] = self._type['permId']
continue
elif attr == 'attachments':
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
attachments = getattr(self,'_new_attachments')
if attachments is None:
continue
atts_data = [ attachment.get_data() for attachment in attachments ]
items = atts_data
elif attr in defs['multi']:
items = getattr(self, '_'+attr)
if items is None:
items = []
else:
items = getattr(self, '_'+attr)
key = None
if attr in attr2ids:
key = attr2ids[attr]
else:
key = attr
new_obj[key] = items
# create a new entity
request = {
"method": "create{}s".format(self.entity),
"params": [
self.openbis.token,
[ new_obj ]
]
}
return request
def _up_attrs(self):
attr2ids = _definitions('attr2ids')
up_obj = {
"@type" : "as.dto.{}.update.{}Update".format(self.entity.lower(), self.entity),
defs["identifier"]: self._permId
}
# look at all attributes available for that entity
# that can be updated
for attr in defs['attrs_up']:
items = None
if attr == 'attachments':
# v3 API currently only supports adding attachments
attachments = self.__dict__.get('_new_attachments', None)
if attachments is None:
continue
atts_data = [ attachment.get_data() for attachment in attachments ]
if self._is_new:
up_obj['attachments'] = atts_data
"actions": [ {
"items": atts_data,
"@type": "as.dto.common.update.ListUpdateActionAdd"
} ],
"@type": "as.dto.attachment.update.AttachmentListUpdateValue"
}
elif attr == 'tags':
# look which tags have been added or removed and update them
if getattr(self,'_prev_tags') is None:
self.__dict__['_prev_tags'] = []
actions = []
for tagId in self._prev_tags:
if tagId not in self._tags:
actions.append({
"items": [ tagId ],
"@type": "as.dto.common.update.ListUpdateActionRemove"
})
for tagId in self._tags:
if tagId not in self._prev_tags:
actions.append({
"items": [ tagId ],
"@type": "as.dto.common.update.ListUpdateActionAdd"
})
"@type": "as.dto.common.update.IdListUpdateValue",
"actions": actions
}
# handle multivalue attributes (parents, children, tags etc.)
# we only cover the Set mechanism, which means we always update
# all items in a list
if attr in defs['multi']:
items = self.__dict__.get('_'+attr, [])
if items == None:
items = []
up_obj[attr2ids[attr]] = {
"actions": [
{
"items": items,
"@type": "as.dto.common.update.ListUpdateActionSet",
}
],
"@type": "as.dto.common.update.IdListUpdateValue"
}
else:
# handle single attributes (space, experiment, project, container, etc.)
value = self.__dict__.get('_'+attr, {})
if value is None:
pass
isModified=False
if 'isModified' in value:
isModified=True
del value['isModified']
up_obj[attr2ids[attr]] = {
"@type": "as.dto.common.update.FieldUpdateValue",
"isModified": isModified,
"value": value,
}
# update a new entity
request = {
"method": "update{}s".format(self.entity),
"params": [
self.openbis.token,
[ up_obj ]
]
}
return request
def __getattr__(self, name):
""" handles all attribute requests dynamically. Values are returned in a sensible way,
for example the identifiers of parents, children and components are returned
as an array of values.
"""
int_name = '_'+name
if int_name in self.__dict__:
if int_name in ['_attachments']:
return [
{
"fileName": x['fileName'],
"title": x['title'],
"description": x['description']
} for x in self._attachments
]
return self.__dict__[int_name].get('userId', None)
elif int_name in ['_registrationDate', '_modificationDate']:
return format_timestamp(self.__dict__[int_name])
# if the attribute contains a list,
# return a list of either identifiers, codes or
# permIds (whatever is available first)
elif isinstance(self.__dict__[int_name], list):
values = []
for item in self.__dict__[int_name]:
if "identifier" in item:
values.append(item['identifier'])
elif "code" in item:
values.append(item['code'])
elif "permId" in item:
values.append(item['permId'])
else:
pass
return values
# attribute contains a dictionary: same procedure as above.
elif isinstance(self.__dict__[int_name], dict):
if "identifier" in self.__dict__[int_name]:
return self.__dict__[int_name]['identifier']
elif "code" in self.__dict__[int_name]:
return self.__dict__[int_name]['code']
elif "permId" in self.__dict__[int_name]:
return self.__dict__[int_name]['permId']
else:
return self.__dict__[int_name]
else:
return None
def __setattr__(self, name, value):
if name in ["parents", "children", "components"]:
if not isinstance(value, list):
value = [value]
objs = []
for val in value:
# fetch objects in openBIS, make sure they actually exists
obj = getattr(self._openbis, 'get_'+self._entity.lower())(val)
objs.append(obj)
self.__dict__['_'+name] = {
"@type": "as.dto.common.update.IdListUpdateValue",
"actions": [{
"@type": "as.dto.common.update.ListUpdateActionSet",
"items": [item._permId for item in objs]
}]
}
elif name in ["tags"]:
self.set_tags(value)
elif name in ["attachments"]:
if isinstance(value, list):
for item in value:
if isinstance(item, dict):
self.add_attachment(**item)
else:
self.add_attachment(item)
else:
self.add_attachment(value)
elif name in ["sample", "experiment", "space", "project"]:
obj = None
# fetch object in openBIS, make sure it actually exists
obj = getattr(self._openbis, "get_"+name)(value)
self.__dict__['_'+name] = obj._identifier
# mark attribute as modified, if it's an existing entity
pass
else:
self.__dict__['_'+name]['isModified'] = True
elif name in ["identifier"]:
raise KeyError("you can not modify the {}".format(name))
elif name == "code":
try:
if self._type.data['autoGeneratedCode']:
raise KeyError("for this {}Type you can not set a code".format(self.entity))
except AttributeError:
pass
self.__dict__['_code'] = value
elif name == "description":
self.__dict__['_description'] = value
else:
raise KeyError("no such attribute: {}".format(name))
def get_type(self):
def get_parents(self):
Swen Vermeul
committed
# e.g. self._openbis.get_samples(withChildren=self.identifier)
return getattr(self._openbis, 'get_'+self._entity.lower()+'s')(withChildren=self.identifier)
def get_children(self):
Swen Vermeul
committed
# e.g. self._openbis.get_samples(withParents=self.identifier)
return getattr(self._openbis, 'get_'+self._entity.lower()+'s')(withParents=self.identifier)
@property
def tags(self):
if getattr(self, '_tags') is not None:
return [ x['code'] for x in self._tags ]
def set_tags(self, tags):
if getattr(self, '_tags') is None:
self.__dict__['_tags'] = []
tagIds = _create_tagIds(tags)
# remove tags that are not in the new tags list
for tagId in self.__dict__['_tags']:
if tagId not in tagIds:
self.__dict__['_tags'].remove(tagId)
# add all new tags that are not in the list yet
for tagId in tagIds:
if tagId not in self.__dict__['_tags']:
self.__dict__['_tags'].append(tagId)
def add_tags(self, tags):
if getattr(self, '_tags') is None:
self.__dict__['_tags'] = []
# add the new tags to the _tags and _new_tags list,
# if not listed yet
tagIds = _create_tagIds(tags)
for tagId in tagIds:
if not tagId in self.__dict__['_tags']:
self.__dict__['_tags'].append(tagId)
def del_tags(self, tags):
if getattr(self, '_tags') is None:
self.__dict__['_tags'] = []
# remove the tags from the _tags and _del_tags list,
# if listed there
tagIds = _create_tagIds(tags)
for tagId in tagIds:
if tagId in self.__dict__['_tags']:
self.__dict__['_tags'].remove(tagId)
def get_attachments(self):
if getattr(self, '_attachments') is None:
return None
else:
return DataFrame(self._attachments)[['fileName','title','description','version']]
def add_attachment(self, fileName, title=None, description=None):
att = Attachment(filename=fileName, title=title, description=description)
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
if getattr(self, '_attachments') is None:
self.__dict__['_attachments'] = []
self._attachments.append(att.get_data_short())
if getattr(self, '_new_attachments') is None:
self.__dict__['_new_attachments'] = []
self._new_attachments.append(att)
def download_attachments(self):
method = 'get'+self.entity+'s'
entity = self.entity.lower()
request = {
"method": method,
"params": [ self._openbis.token,
[ self._permId ],
{
"attachments": fetch_option['attachmentsWithContent'],
**fetch_option[entity]
}
]
}
resp = self._openbis._post_request(self._openbis.as_v3, request)
attachments = resp[self.permId]['attachments']
file_list = []
for attachment in attachments:
filename = os.path.join(
self._openbis.hostname,
self.permId,
attachment['fileName']
)
os.makedirs(os.path.dirname(filename), exist_ok=True)
with open(filename, 'wb') as att:
content = base64.b64decode(attachment['content'])
att.write(content)
file_list.append(filename)
return file_list
def _repr_html_(self):
Swen Vermeul
committed
def nvl(val, string=''):
if val is None:
return string
return val
html = """
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th>attribute</th>
<th>value</th>
</tr>
</thead>
<tbody>
"""
Swen Vermeul
committed
for attr in self._allowed_attrs:
if attr == 'attachments':
continue
html += "<tr> <td>{}</td> <td>{}</td> </tr>".format(
attr, nvl(getattr(self, attr, ''),'')
)
if getattr(self, '_attachments') is not None:
Swen Vermeul
committed
html += "<tr><td>attachments</td><td>"
html += "<br/>".join(att['fileName'] for att in self._attachments)
html += "</td></tr>"
html += """
</tbody>
</table>
"""
return html
class Sample():
""" A Sample is one of the most commonly used objects in openBIS.
"""
def __init__(self, openbis_obj, type, data=None, **kwargs):
self.__dict__['openbis'] = openbis_obj
self.__dict__['type'] = type
self.__dict__['p'] = PropertyHolder(openbis_obj, type)
self.__dict__['a'] = AttrHolder(openbis_obj, 'Sample', type)
if data is not None:
self._set_data(data)
if kwargs is not None:
for key in kwargs:
setattr(self, key, kwargs[key])
def _set_data(self, data):
# assign the attribute data to self.a by calling it
# (invoking the AttrHolder.__call__ function)
self.a(data)
self.__dict__['data'] = data
# put the properties in the self.p namespace (without checking them)
for key, value in data['properties'].items():
self.p.__dict__[key.lower()] = value
Swen Vermeul
committed
def __dir__(self):
return [
'props', 'get_parents()', 'get_children()',
'get_datasets()', 'get_experiment()',
'space', 'project', 'experiment', 'project', 'tags',
'set_tags()', 'add_tags()', 'del_tags()',
'add_attachment()', 'get_attachments()', 'download_attachments()'
]
Swen Vermeul
committed
@property
def props(self):
return self.__dict__['p']
Swen Vermeul
committed
@property
def type(self):
@type.setter
def type(self, type_name):
sample_type = self.openbis.get_sample_type(type_name)
self.p.__dict__['_type'] = sample_type
self.a.__dict__['_type'] = sample_type
def __getattr__(self, name):
return getattr(self.__dict__['a'], name)
def __setattr__(self, name, value):
if name in ['set_properties', 'set_tags', 'add_tags']:
raise ValueError("These are methods which should not be overwritten")
setattr(self.__dict__['a'], name, value)
def _repr_html_(self):
html = self.a._repr_html_()
return html
def set_properties(self, properties):
self.openbis.update_sample(self.permId, properties=properties)
def save(self):
props = self.p._all_props()
attrs["properties"] = props
if self.identifier is None:
# create a new sample
attrs["@type"] = "as.dto.sample.create.SampleCreation"
attrs["typeId"] = self.__dict__['type'].data['permId']
request = {
"method": "createSamples",
"params": [ self.openbis.token,
[ attrs ]
]
}
resp = self.openbis._post_request(self.openbis.as_v3, request)
new_sample_data = self.openbis.get_sample(resp[0]['permId'], only_data=True)
self._set_data(new_sample_data)
return self
else:
attrs["@type"] = "as.dto.sample.update.SampleUpdate"
attrs["sampleId"] = {
"permId": self.permId,
"@type": "as.dto.sample.id.SamplePermId"
}
request = {
"method": "updateSamples",
"params": [ self.openbis.token,
[ attrs ]
]
}
resp = self.openbis._post_request(self.openbis.as_v3, request)
print('Sample successfully updated')
self.openbis.delete_entity('sample', self.permId, reason)
def get_datasets(self):
return self.openbis.get_datasets(sample=self.permId)
def get_projects(self):
return self.openbis.get_project(withSamples=[self.permId])
Swen Vermeul
committed
def get_experiment(self):
try:
return self.openbis.get_experiment(self._experiment['identifier'])
except Exception:
pass
@property
def experiment(self):
try:
return self.openbis.get_experiment(self._experiment['identifier'])
except Exception:
pass
Swen Vermeul
committed
""" managing openBIS spaces
"""
def __init__(self, openbis_obj, type=None, data=None, **kwargs):
self.__dict__['openbis'] = openbis_obj
self.__dict__['a'] = AttrHolder(openbis_obj, 'Space', type)
if data is not None:
self.a(data)
self.__dict__['data'] = data
if kwargs is not None:
for key in kwargs:
setattr(self, key, kwargs[key])
def __dir__(self):
"""all the available methods and attributes that should be displayed
when using the autocompletion feature (TAB) in Jupyter
"""
return['code','permId', 'description', 'registrator', 'registrationDate',
'modificationDate', 'get_projects()', 'new_project()', 'get_samples()', 'delete()']
def __str__(self):
return self.data.get('code', None)
def get_samples(self):
return self.openbis.get_samples(space=self.code)
def get_projects(self):
return self.openbis.get_projects(space=self.code)
def new_project(self, code, description=None, **kwargs):
return self.openbis.new_project(self.code, code, description, **kwargs)
def delete(self, reason):
self.openbis.delete_entity('Space', self.permId, reason)
class Things():
"""An object that contains a DataFrame object about an entity available in openBIS.
"""
def __init__(self, openbis_obj, entity, df, identifier_name='code'):
self.df = df
self.identifier_name = identifier_name
def _repr_html_(self):
return self.df._repr_html_()
def __getitem__(self, key):
if self.df is not None and len(self.df) > 0:
row = None
if isinstance(key, int):
# get thing by rowid
row = self.df.loc[[key]]
elif isinstance(key, list):
# treat it as a normal dataframe
return self.df[key]
else:
# get thing by code
row = self.df[self.df[self.identifier_name]==key.upper()]
if row is not None:
# invoke the openbis.get_entity() method
return getattr(self.openbis, 'get_'+self.entity)(row[self.identifier_name].values[0])
Swen Vermeul
committed
"""
Swen Vermeul
committed
def __init__(self, openbis_obj, type, data=None, **kwargs):
self.__dict__['openbis'] = openbis_obj
self.__dict__['type'] = type
self.__dict__['p'] = PropertyHolder(openbis_obj, type)
self.__dict__['a'] = AttrHolder(openbis_obj, 'Experiment', type)
Swen Vermeul
committed
if data is not None:
self._set_data(data)
if kwargs is not None:
for key in kwargs:
setattr(self, key, kwargs[key])
def _set_data(self, data):
# assign the attribute data to self.a by calling it
# (invoking the AttrHolder.__call__ function)
self.a(data)
self.__dict__['data'] = data
# put the properties in the self.p namespace (without checking them)
for key, value in data['properties'].items():
self.p.__dict__[key.lower()] = value
def __str__(self):
return self.data['code']
Swen Vermeul
committed
def __dir__(self):
# the list of possible methods/attributes displayed
# when invoking TAB-completition
return [
Swen Vermeul
committed
'project','tags', 'attachments', 'data',
'set_tags()', 'add_tags()', 'del_tags()',
'add_attachment()', 'get_attachments()', 'download_attachments()'
Swen Vermeul
committed
]
@property
def props(self):
return self.__dict__['p']
@property
def type(self):
Swen Vermeul
committed
@type.setter
def type(self, type_name):
experiment_type = self.openbis.get_experiment_type(type_name)
self.p.__dict__['_type'] = experiment_type
self.a.__dict__['_type'] = experiment_type
def __getattr__(self, name):
return getattr(self.__dict__['a'], name)
def __setattr__(self, name, value):
if name in ['set_properties', 'add_tags()', 'del_tags()', 'set_tags()']:
Swen Vermeul
committed
raise ValueError("These are methods which should not be overwritten")
setattr(self.__dict__['a'], name, value)
def _repr_html_(self):
html = self.a._repr_html_()
return html
def set_properties(self, properties):
self.openbis.update_experiment(self.permId, properties=properties)
Swen Vermeul
committed
def save(self):
if self.is_new:
request = self._new_attrs()
props = self.p._all_props()
request["properties"] = props
self.openbis._post_request(self.openbis.as_v3, request)
self.a.__dict__['_is_new'] = False
print("Experiment successfully created.")
Swen Vermeul
committed
else:
request = self._up_attrs()
props = self.p._all_props()
request["properties"] = props
self.openbis._post_request(self.openbis.as_v3, request)
print("Experiment successfully updated.")
def delete(self, reason):
self.openbis.delete_entity('experiment', self.permId, reason)
Swen Vermeul
committed
def get_datasets(self):
return self.openbis.get_datasets(experiment=self.permId)
Swen Vermeul
committed
def get_projects(self):
return self.openbis.get_project(experiment=self.permId)
Swen Vermeul
committed
def get_samples(self):
return self.openbis.get_samples(experiment=self.permId)
Swen Vermeul
committed
def __init__(self, filename, title=None, description=None):
if not os.path.exists(filename):
raise ValueError("File not found: {}".format(filename))
self.fileName = filename
self.title = title
self.description = description
Swen Vermeul
committed
def get_data_short(self):
return {
"fileName" : self.fileName,
"title" : self.title,
"description" : self.description,
}
Swen Vermeul
committed
def get_data(self):
with open(self.fileName, 'rb') as att:
content = att.read()
contentb64 = base64.b64encode(content).decode()
return {
"fileName" : self.fileName,
"title" : self.title,
"description" : self.description,
"content" : contentb64,
"@type" : "as.dto.attachment.create.AttachmentCreation",
}
Swen Vermeul
committed
def __init__(self, openbis_obj, data=None, **kwargs):
self.__dict__['openbis'] = openbis_obj
self.__dict__['a'] = AttrHolder(openbis_obj, 'Project')
if data is not None:
self.a(data)
self.__dict__['data'] = data
if kwargs is not None:
for key in kwargs:
setattr(self, key, kwargs[key])
def _modifiable_attrs(self):
return
def __dir__(self):
"""all the available methods and attributes that should be displayed
when using the autocompletion feature (TAB) in Jupyter
"""
return['code','permId', 'identifier', 'description', 'space', 'registrator',
'registrationDate', 'modifier', 'modificationDate', 'add_attachment()',
'get_attachments()', 'download_attachments()',
'get_experiments()', 'get_samples()', 'get_datasets()',
'delete()'
]
def get_samples(self):
return self.openbis.get_samples(project=self.permId)
def get_experiments(self):
return self.openbis.get_experiments(project=self.permId)
def get_datasets(self):
return self.openbis.get_datasets(project=self.permId)
def delete(self, reason):
self.openbis.delete_entity('project', self.permId, reason)
if self.is_new:
request = self._new_attrs()
self.openbis._post_request(self.openbis.as_v3, request)
self.a.__dict__['_is_new'] = False
print("Project successfully created.")
request = self._up_attrs()
self.openbis._post_request(self.openbis.as_v3, request)
print("Project successfully updated.")