1 """Interface with encoded software for ENCODE3 data submission & warehouse
3 This allows retrieving blocks
7 from __future__ import print_function
14 from urlparse import urljoin, urlparse, urlunparse
16 LOGGER = logging.getLogger(__name__)
19 # The None context will get added to the root of the tree and will
20 # provide common defaults.
22 # terms in multiple encoded objects
23 'award': {'@type': '@id'},
24 'dataset': {'@type': '@id'},
25 'description': 'rdf:description',
26 'documents': {'@type': '@id'},
27 'experiment': {'@type': '@id'},
28 'href': {'@type': '@id'},
29 'lab': {'@type': '@id'},
30 'library': {'@type': '@id'},
31 'pi': {'@type': '@id'},
32 'platform': {'@type': '@id'},
33 'replicates': {'@type': '@id'},
34 'submitted_by': {'@type': '@id'},
35 'url': {'@type': '@id'},
37 # Identify and markup contained classes.
38 # e.g. in the tree there was a sub-dictionary named 'biosample'
39 # That dictionary had a term 'biosample_term_id, which is the
40 # term that should be used as the @id.
42 'biosample_term_id': {'@type': '@id'},
45 "assay_term_id": {"@type": "@id"},
46 "files": {"@type": "@id"},
47 "original_files": {"@type": "@id"},
49 # I tried to use the JSON-LD mapping capabilities to convert the lab
50 # contact information into a vcard record, but the encoded model
51 # didn't lend itself well to the vcard schema
53 # "address1": "vcard:street-address",
54 # "address2": "vcard:street-address",
55 # "city": "vcard:locality",
56 # "state": "vcard:region",
57 # "country": "vcard:country"
60 'nucleic_acid_term_id': {'@type': '@id'}
64 #FIXME: this needs to be initialized from rdfns
65 ENCODED_NAMESPACES = {
66 # JSON-LD lets you define namespaces so you can used the shorted url syntax.
67 # (instead of http://www.w3.org/2000/01/rdf-schema#label you can do
69 "rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
70 "rdfs": "http://www.w3.org/2000/01/rdf-schema#",
71 "owl": "http://www.w3.org/2002/07/owl#",
72 "dc": "htp://purl.org/dc/elements/1.1/",
73 "xsd": "http://www.w3.org/2001/XMLSchema#",
74 "vcard": "http://www.w3.org/2006/vcard/ns#",
76 # for some namespaces I made a best guess for the ontology root.
77 "EFO": "http://www.ebi.ac.uk/efo/", # EFO ontology
78 "OBO": "http://purl.obolibrary.org/obo/", # OBO ontology
79 "OBI": "http://purl.obolibrary.org/obo/OBI_", # Ontology for Biomedical Investigations
80 # OBI: available from http://svn.code.sf.net/p/obi/code/releases/2012-07-01/merged/merged-obi-comments.owl
81 "SO": "http://purl.obolibrary.org/obo/SO_", # Sequence ontology
82 # SO: available from http://www.berkeleybop.org/ontologies/so.owl
83 # NTR: New Term Request space for DCC to implement new ontology terms
87 ENCODED_SCHEMA_ROOT = '/profiles/'
91 '''Programatic access encoded, the software powering ENCODE3's submit site.
93 def __init__(self, server, contexts=None, namespaces=None):
98 self.contexts = contexts if contexts else ENCODED_CONTEXT
99 self.namespaces = namespaces if namespaces else ENCODED_NAMESPACES
100 self.json_headers = {'content-type': 'application/json', 'accept': 'application/json'}
104 return (self.username, self.password)
105 auth = property(get_auth)
107 def load_netrc(self):
109 session = netrc.netrc()
110 authenticators = session.authenticators(self.server)
112 self.username = authenticators[0]
113 self.password = authenticators[2]
115 def add_jsonld_context(self, tree, default_base):
116 """Add contexts to various objects in the tree.
118 tree is a json tree returned from the DCC's encoded database.
119 contexts is a dictionary of dictionaries containing contexts
120 for the various possible encoded classes.
121 base, if supplied allows setting the base url that relative
122 urls will be resolved against.
124 self.add_jsonld_child_context(tree, default_base)
125 self.add_jsonld_namespaces(tree['@context'])
127 def add_jsonld_child_context(self, obj, default_base):
128 '''Add JSON-LD context to the encoded JSON.
130 This is recursive because some of the IDs were relative URLs
131 and I needed a way to properly compute a the correct base URL.
133 # pretend strings aren't iterable
134 if type(obj) in types.StringTypes:
137 # recurse on container types
138 if isinstance(obj, collections.Sequence):
139 # how should I update lists?
141 self.add_jsonld_child_context(v, default_base)
144 if isinstance(obj, collections.Mapping):
145 for v in obj.values():
146 self.add_jsonld_child_context(v, default_base)
148 # we have an object. attach a context to it.
149 if self._is_encoded_object(obj):
150 context = self.create_jsonld_context(obj, default_base)
152 obj.setdefault('@context', {}).update(context)
154 def add_jsonld_namespaces(self, context):
155 '''Add shortcut namespaces to a context
157 Only needs to be run on the top-most context
159 context.update(self.namespaces)
161 def create_jsonld_context(self, obj, default_base):
162 '''Synthesize the context for a encoded type
164 self.contexts[None] = default context attributes added to any type
165 self.contexts[type] = context attributes for this type.
167 context = {'@base': urljoin(default_base, obj['@id']),
168 '@vocab': self.get_schema_url(obj)}
170 context.update(self.contexts[None])
171 for t in obj['@type']:
172 if t in self.contexts:
173 context.update(self.contexts[t])
176 def get_json(self, obj_id, **kwargs):
177 '''GET an ENCODE object as JSON and return as dict
179 Uses prepare_url to allow url short-cuts
180 if no keyword arguments are specified it will default to adding limit=all
181 Alternative keyword arguments can be passed in and will be sent to the host.
184 limit - (integer or 'all') how many records to return, all for all of them
185 embed - (bool) if true expands linking ids into their associated object.
186 format - text/html or application/json
189 kwargs['limit'] = 'all'
191 url = self.prepare_url(obj_id)
192 LOGGER.info('requesting url: {}'.format(url))
196 LOGGER.debug('username: %s, password: %s', self.username, self.password)
197 response = requests.get(url, auth=self.auth, headers=self.json_headers, params=kwargs)
198 if not response.status_code == requests.codes.ok:
199 LOGGER.error("Error http status: {}".format(response.status_code))
200 response.raise_for_status()
201 return response.json()
203 def get_jsonld(self, obj_id, **kwargs):
204 '''Get ENCODE object as JSONLD annotated with classses contexts
206 see get_json for documentation about what keywords can be passed.
208 url = self.prepare_url(obj_id)
209 json = self.get_json(obj_id, **kwargs)
210 self.add_jsonld_context(json, url)
213 def get_object_type(self, obj):
214 """Return type for a encoded object
216 obj_type = obj.get('@type')
218 raise ValueError('None type')
219 if type(obj_type) in types.StringTypes:
220 raise ValueError('@type should be a list, not a string')
221 if not isinstance(obj_type, collections.Sequence):
222 raise ValueError('@type is not a sequence')
225 def get_schema_url(self, obj):
226 obj_type = self.get_object_type(obj)
228 return self.prepare_url(ENCODED_SCHEMA_ROOT + obj_type + '.json') + '#'
230 def _is_encoded_object(self, obj):
231 '''Test to see if an object is a JSON-LD object
233 Some of the nested dictionaries lack the @id or @type
234 information necessary to convert them.
236 if not isinstance(obj, collections.Iterable):
239 if '@id' in obj and '@type' in obj:
243 def patch_json(self, obj_id, changes):
244 """Given a dictionary of changes push them as a HTTP patch request
246 url = self.prepare_url(obj_id)
247 LOGGER.info('PATCHing to %s', url)
248 payload = json.dumps(changes)
249 response = requests.patch(url, auth=self.auth, headers=self.json_headers, data=payload)
250 if response.status_code != requests.codes.ok:
251 LOGGER.error("Error http status: {}".format(response.status_code))
252 LOGGER.error("Response: %s", response.text)
253 response.raise_for_status()
254 return response.json()
256 def put_json(self, obj_id, new_object):
257 url = self.prepare_url(obj_id)
258 LOGGER.info('PUTing to %s', url)
259 payload = json.dumps(new_object)
260 response = requests.put(url, auth=self.auth, headers=self.json_headers, data=payload)
261 if response.status_code != requests.codes.created:
262 LOGGER.error("Error http status: {}".format(response.status_code))
263 response.raise_for_status()
264 return response.json()
266 def post_json(self, collection_id, new_object):
267 url = self.prepare_url(collection_id)
268 LOGGER.info('POSTing to %s', url)
269 payload = json.dumps(new_object)
271 response = requests.post(url, auth=self.auth, headers=self.json_headers, data=payload)
272 if response.status_code != requests.codes.created:
273 LOGGER.error("Error http status: {}".format(response.status_code))
274 response.raise_for_status()
275 return response.json()
277 def prepare_url(self, request_url):
278 '''This attempts to provide some convienence for accessing a URL
280 Given a url fragment it will default to :
282 * requests to self.server
284 This allows fairly flexible urls. e.g.
286 prepare_url('/experiments/ENCSR000AEG')
287 prepare_url('submit.encodedcc.org/experiments/ENCSR000AEG')
288 prepare_url('http://submit.encodedcc.org/experiments/ENCSR000AEG?limit=all')
290 should all return the same url
292 # clean up potentially messy urls
293 url = urlparse(request_url)._asdict()
294 if not url['scheme']:
295 url['scheme'] = self.scheme
296 if not url['netloc']:
297 url['netloc'] = self.server
298 url = urlunparse(url.values())
301 def search_jsonld(self, term, **kwargs):
302 '''Send search request to ENCODED
304 url = self.prepare_url('/search/')
305 result = self.get_json(url, searchTerm=term, **kwargs)
306 self.convert_search_to_jsonld(result)
309 def convert_search_to_jsonld(self, result):
310 '''Add the context to search result
312 Also remove hard to handle nested attributes
313 e.g. remove object.term when we have no id
315 graph = result['@graph']
316 for i, obj in enumerate(graph):
317 # suppress nested attributes
318 graph[i] = {k: v for k, v in obj.items() if '.' not in k}
320 self.add_jsonld_context(result, self.prepare_url(result['@id']))
323 def validate(self, obj):
324 obj_type = self.get_object_type(obj)
325 schema_url = self.get_schema_url(obj)
327 raise ValueError("Unable to construct schema url")
329 schema = self.schemas.setdefault(obj_type, self.get_json(schema_url))
331 if '@id' in hidden: del hidden['@id']
332 if '@type' in hidden: del hidden['@type']
333 jsonschema.validate(hidden, schema)
336 class Document(object):
337 """Helper class for registering documents
340 lysis_uuid = 'f0cc5a7f-96a5-4970-9f46-317cc8e2d6a4'
341 lysis = Document(url_to_pdf, 'extraction protocol', 'Lysis Protocol')
342 lysis.create_if_needed(server, lysis_uuid)
344 award = 'U54HG006998'
345 lab = '/labs/barbara-wold'
347 def __init__(self, url, document_type, description, aliases=None):
349 self.filename = os.path.basename(url)
350 self.document_type = document_type
351 self.description = description
354 self.aliases = aliases if aliases is not None else []
355 self.content_type = None
363 def get_document(self):
364 if os.path.exists(self.url):
365 with open(self.url, 'r') as instream:
366 assert self.url.endswith('pdf')
367 self.content_type = 'application/pdf'
368 self.document = instream.read()
369 self.md5sum = hashlib.md5(self.document)
371 req = requests.get(self.url)
372 if req.status_code == 200:
373 self.content_type = req.headers['content-type']
374 self.document = req.content
375 self.md5sum = hashlib.md5(self.document)
376 self.urls = [self.url]
378 def create_payload(self):
381 'download': self.filename,
382 'type': self.content_type,
383 'href': 'data:'+self.content_type+';base64,' + base64.b64encode(self.document),
384 'md5sum': self.md5sum.hexdigest()
386 'document_type': self.document_type,
387 'description': self.description,
392 document_payload['aliases'] = self.aliases
394 document_payload['references'] = self.references
396 document_payload['urls'] = self.urls
398 return document_payload
400 def post(self, server):
401 document_payload = self.create_payload()
402 return server.post_json('/documents/', document_payload)
404 def save(self, filename):
405 payload = self.create_payload()
406 with open(filename, 'w') as outstream:
407 outstream.write(pformat(payload))
409 def create_if_needed(self, server, uuid):
412 return self.post(server)
414 return server.get_json(uuid, embed=False)
416 if __name__ == '__main__':
418 from htsworkflow.util.rdfhelp import get_model, dump_model
419 from htsworkflow.util.rdfjsonld import load_into_model
420 from pprint import pprint
422 logging.basicConfig(level=logging.DEBUG)
423 encoded = ENCODED('test.encodedcc.org')
425 body = encoded.get_jsonld('/experiments/ENCSR000AEC/')
427 load_into_model(model, body)