1 """Interface with encoded software for ENCODE3 data submission & warehouse
3 This allows retrieving blocks
5 from __future__ import print_function
15 from six.moves.urllib.parse import urljoin, urlparse, urlunparse
17 LOGGER = logging.getLogger(__name__)
20 # The None context will get added to the root of the tree and will
21 # provide common defaults.
23 # terms in multiple encoded objects
24 'award': {'@type': '@id'},
25 'dataset': {'@type': '@id'},
26 'description': 'rdf:description',
27 'documents': {'@type': '@id'},
28 'experiment': {'@type': '@id'},
29 'href': {'@type': '@id'},
30 'lab': {'@type': '@id'},
31 'library': {'@type': '@id'},
32 'pi': {'@type': '@id'},
33 'platform': {'@type': '@id'},
34 'replicates': {'@type': '@id'},
35 'submitted_by': {'@type': '@id'},
36 'url': {'@type': '@id'},
38 # Identify and markup contained classes.
39 # e.g. in the tree there was a sub-dictionary named 'biosample'
40 # That dictionary had a term 'biosample_term_id, which is the
41 # term that should be used as the @id.
43 'biosample_term_id': {'@type': '@id'},
46 "assay_term_id": {"@type": "@id"},
47 "files": {"@type": "@id"},
48 "original_files": {"@type": "@id"},
50 # I tried to use the JSON-LD mapping capabilities to convert the lab
51 # contact information into a vcard record, but the encoded model
52 # didn't lend itself well to the vcard schema
54 # "address1": "vcard:street-address",
55 # "address2": "vcard:street-address",
56 # "city": "vcard:locality",
57 # "state": "vcard:region",
58 # "country": "vcard:country"
61 'nucleic_acid_term_id': {'@type': '@id'}
65 #FIXME: this needs to be initialized from rdfns
66 ENCODED_NAMESPACES = {
67 # JSON-LD lets you define namespaces so you can used the shorted url syntax.
68 # (instead of http://www.w3.org/2000/01/rdf-schema#label you can do
70 "rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
71 "rdfs": "http://www.w3.org/2000/01/rdf-schema#",
72 "owl": "http://www.w3.org/2002/07/owl#",
73 "dc": "htp://purl.org/dc/elements/1.1/",
74 "xsd": "http://www.w3.org/2001/XMLSchema#",
75 "vcard": "http://www.w3.org/2006/vcard/ns#",
77 # for some namespaces I made a best guess for the ontology root.
78 "EFO": "http://www.ebi.ac.uk/efo/", # EFO ontology
79 "OBO": "http://purl.obolibrary.org/obo/", # OBO ontology
80 "OBI": "http://purl.obolibrary.org/obo/OBI_", # Ontology for Biomedical Investigations
81 # OBI: available from http://svn.code.sf.net/p/obi/code/releases/2012-07-01/merged/merged-obi-comments.owl
82 "SO": "http://purl.obolibrary.org/obo/SO_", # Sequence ontology
83 # SO: available from http://www.berkeleybop.org/ontologies/so.owl
84 # NTR: New Term Request space for DCC to implement new ontology terms
88 ENCODED_SCHEMA_ROOT = '/profiles/'
92 '''Programatic access encoded, the software powering ENCODE3's submit site.
94 def __init__(self, server, contexts=None, namespaces=None):
99 self.contexts = contexts if contexts else ENCODED_CONTEXT
100 self.namespaces = namespaces if namespaces else ENCODED_NAMESPACES
101 self.json_headers = {'content-type': 'application/json', 'accept': 'application/json'}
105 return (self.username, self.password)
106 auth = property(get_auth)
108 def load_netrc(self):
110 session = netrc.netrc()
111 authenticators = session.authenticators(self.server)
113 self.username = authenticators[0]
114 self.password = authenticators[2]
116 def add_jsonld_context(self, tree, default_base):
117 """Add contexts to various objects in the tree.
119 tree is a json tree returned from the DCC's encoded database.
120 contexts is a dictionary of dictionaries containing contexts
121 for the various possible encoded classes.
122 base, if supplied allows setting the base url that relative
123 urls will be resolved against.
125 self.add_jsonld_child_context(tree, default_base)
126 self.add_jsonld_namespaces(tree['@context'])
128 def add_jsonld_child_context(self, obj, default_base):
129 '''Add JSON-LD context to the encoded JSON.
131 This is recursive because some of the IDs were relative URLs
132 and I needed a way to properly compute a the correct base URL.
134 # pretend strings aren't iterable
135 if type(obj) in types.StringTypes:
138 # recurse on container types
139 if isinstance(obj, collections.Sequence):
140 # how should I update lists?
142 self.add_jsonld_child_context(v, default_base)
145 if isinstance(obj, collections.Mapping):
146 for v in obj.values():
147 self.add_jsonld_child_context(v, default_base)
149 # we have an object. attach a context to it.
150 if self._is_encoded_object(obj):
151 context = self.create_jsonld_context(obj, default_base)
153 obj.setdefault('@context', {}).update(context)
155 def add_jsonld_namespaces(self, context):
156 '''Add shortcut namespaces to a context
158 Only needs to be run on the top-most context
160 context.update(self.namespaces)
162 def create_jsonld_context(self, obj, default_base):
163 '''Synthesize the context for a encoded type
165 self.contexts[None] = default context attributes added to any type
166 self.contexts[type] = context attributes for this type.
168 obj_type = self.get_object_type(obj)
169 context = {'@base': urljoin(default_base, obj['@id']),
170 '@vocab': self.get_schema_url(obj_type)}
172 context.update(self.contexts[None])
173 for t in obj['@type']:
174 if t in self.contexts:
175 context.update(self.contexts[t])
178 def get_json(self, obj_id, **kwargs):
179 '''GET an ENCODE object as JSON and return as dict
181 Uses prepare_url to allow url short-cuts
182 if no keyword arguments are specified it will default to adding limit=all
183 Alternative keyword arguments can be passed in and will be sent to the host.
186 limit - (integer or 'all') how many records to return, all for all of them
187 embed - (bool) if true expands linking ids into their associated object.
188 format - text/html or application/json
191 kwargs['limit'] = 'all'
193 url = self.prepare_url(obj_id)
194 LOGGER.info('requesting url: {}'.format(url))
198 LOGGER.debug('username: %s, password: %s', self.username, self.password)
200 if self.username and self.password:
201 arguments['auth'] = self.auth
202 response = requests.get(url, headers=self.json_headers,
205 if not response.status_code == requests.codes.ok:
206 LOGGER.error("Error http status: {}".format(response.status_code))
207 response.raise_for_status()
208 return response.json()
210 def get_jsonld(self, obj_id, **kwargs):
211 '''Get ENCODE object as JSONLD annotated with classses contexts
213 see get_json for documentation about what keywords can be passed.
215 url = self.prepare_url(obj_id)
216 json = self.get_json(obj_id, **kwargs)
217 self.add_jsonld_context(json, url)
220 def get_object_type(self, obj):
221 """Return type for a encoded object
223 obj_type = obj.get('@type')
225 raise ValueError('None type')
226 if type(obj_type) in types.StringTypes:
227 raise ValueError('@type should be a list, not a string')
228 if not isinstance(obj_type, collections.Sequence):
229 raise ValueError('@type is not a sequence')
232 def get_schema_url(self, object_type):
233 return self.prepare_url(ENCODED_SCHEMA_ROOT + object_type + '.json') + '#'
235 def _is_encoded_object(self, obj):
236 '''Test to see if an object is a JSON-LD object
238 Some of the nested dictionaries lack the @id or @type
239 information necessary to convert them.
241 if not isinstance(obj, collections.Iterable):
244 if '@id' in obj and '@type' in obj:
248 def patch_json(self, obj_id, changes):
249 """Given a dictionary of changes push them as a HTTP patch request
251 url = self.prepare_url(obj_id)
252 LOGGER.info('PATCHing to %s', url)
253 payload = json.dumps(changes)
254 response = requests.patch(url, auth=self.auth, headers=self.json_headers, data=payload)
255 if response.status_code != requests.codes.ok:
256 LOGGER.error("Error http status: {}".format(response.status_code))
257 LOGGER.error("Response: %s", response.text)
258 response.raise_for_status()
259 return response.json()
261 def put_json(self, obj_id, new_object):
262 url = self.prepare_url(obj_id)
263 LOGGER.info('PUTing to %s', url)
264 payload = json.dumps(new_object)
265 response = requests.put(url, auth=self.auth, headers=self.json_headers, data=payload)
266 if response.status_code != requests.codes.created:
267 LOGGER.error("Error http status: {}".format(response.status_code))
268 response.raise_for_status()
269 return response.json()
271 def post_json(self, collection_id, new_object):
272 url = self.prepare_url(collection_id)
273 LOGGER.info('POSTing to %s', url)
274 payload = json.dumps(new_object)
276 response = requests.post(url, auth=self.auth, headers=self.json_headers, data=payload)
277 if response.status_code != requests.codes.created:
278 LOGGER.error("Error http status: {}".format(response.status_code))
279 response.raise_for_status()
280 return response.json()
282 def prepare_url(self, request_url):
283 '''This attempts to provide some convienence for accessing a URL
285 Given a url fragment it will default to :
287 * requests to self.server
289 This allows fairly flexible urls. e.g.
291 prepare_url('/experiments/ENCSR000AEG')
292 prepare_url('submit.encodedcc.org/experiments/ENCSR000AEG')
293 prepare_url('http://submit.encodedcc.org/experiments/ENCSR000AEG?limit=all')
295 should all return the same url
297 # clean up potentially messy urls
298 url = urlparse(request_url)._asdict()
299 if not url['scheme']:
300 url['scheme'] = self.scheme
301 if not url['netloc']:
302 url['netloc'] = self.server
303 url = urlunparse(url.values())
306 def search_jsonld(self, **kwargs):
307 '''Send search request to ENCODED
309 to do a general search do
312 url = self.prepare_url('/search/')
313 result = self.get_json(url, **kwargs)
314 self.convert_search_to_jsonld(result)
317 def convert_search_to_jsonld(self, result):
318 '''Add the context to search result
320 Also remove hard to handle nested attributes
321 e.g. remove object.term when we have no id
323 graph = result['@graph']
324 for i, obj in enumerate(graph):
325 # suppress nested attributes
326 graph[i] = {k: v for k, v in obj.items() if '.' not in k}
328 self.add_jsonld_context(result, self.prepare_url(result['@id']))
331 def validate(self, obj, object_type=None):
332 object_type = object_type if object_type else self.get_object_type(obj)
333 schema_url = self.get_schema_url(object_type)
335 raise ValueError("Unable to construct schema url")
337 schema = self.schemas.setdefault(object_type, self.get_json(schema_url))
341 if '@type' in hidden:
343 jsonschema.validate(hidden, schema)
345 class TypedColumnParser(object):
347 def parse_sheet_array_type(value):
348 """Helper function to parse :array columns in sheet
350 return value.split(', ')
353 def parse_sheet_integer_type(value):
354 """Helper function to parse :integer columns in sheet
359 def parse_sheet_boolean_type(value):
360 """Helper function to parse :boolean columns in sheet
365 def parse_sheet_timestamp_type(value):
366 """Helper function to parse :date columns in sheet
368 return value.strftime('%Y-%m-%d')
371 def parse_sheet_string_type(value):
372 """Helper function to parse :string columns in sheet (the default)
374 return unicode(value)
376 def __getitem__(self, name):
378 'array': self.parse_sheet_array_type,
379 'boolean': self.parse_sheet_boolean_type,
380 'integer': self.parse_sheet_integer_type,
381 'date': self.parse_sheet_timestamp_type,
382 'string': self.parse_sheet_string_type
387 raise RuntimeError("unrecognized column type")
389 def __call__(self, header, value):
390 header = header.split(':')
391 column_type = 'string'
393 if header[1] == 'skip':
396 column_type = header[1]
397 return header[0], self[column_type](value)
399 typed_column_parser = TypedColumnParser()
401 class Document(object):
402 """Helper class for registering documents
405 lysis_uuid = 'f0cc5a7f-96a5-4970-9f46-317cc8e2d6a4'
406 lysis = Document(url_to_pdf, 'extraction protocol', 'Lysis Protocol')
407 lysis.create_if_needed(server, lysis_uuid)
409 award = 'U54HG006998'
410 lab = '/labs/barbara-wold'
412 def __init__(self, url, document_type, description, aliases=None):
414 self.filename = os.path.basename(url)
415 self.document_type = document_type
416 self.description = description
419 self.aliases = aliases if aliases is not None else []
420 self.content_type = None
428 def get_document(self):
429 if os.path.exists(self.url):
430 with open(self.url, 'r') as instream:
431 assert self.url.endswith('pdf')
432 self.content_type = 'application/pdf'
433 self.document = instream.read()
434 self.md5sum = hashlib.md5(self.document)
436 req = requests.get(self.url)
437 if req.status_code == 200:
438 self.content_type = req.headers['content-type']
439 self.document = req.content
440 self.md5sum = hashlib.md5(self.document)
441 self.urls = [self.url]
443 def create_payload(self):
446 'download': self.filename,
447 'type': self.content_type,
448 'href': 'data:'+self.content_type+';base64,' + base64.b64encode(self.document),
449 'md5sum': self.md5sum.hexdigest()
451 'document_type': self.document_type,
452 'description': self.description,
457 document_payload['aliases'] = self.aliases
459 document_payload['references'] = self.references
461 document_payload['urls'] = self.urls
463 return document_payload
465 def post(self, server):
466 document_payload = self.create_payload()
467 return server.post_json('/documents/', document_payload)
469 def save(self, filename):
470 payload = self.create_payload()
471 with open(filename, 'w') as outstream:
472 outstream.write(pformat(payload))
474 def create_if_needed(self, server, uuid):
477 return self.post(server)
479 return server.get_json(uuid, embed=False)
481 if __name__ == '__main__':
483 from htsworkflow.util.rdfhelp import get_model, dump_model
484 from htsworkflow.util.rdfjsonld import load_into_model
485 from pprint import pprint
487 logging.basicConfig(level=logging.DEBUG)
488 encoded = ENCODED('test.encodedcc.org')
490 body = encoded.get_jsonld('/experiments/ENCSR000AEC/')
492 load_into_model(model, body)