idlib 0.0.1.dev19__tar.gz → 0.0.1.dev21__tar.gz

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: idlib
-Version: 0.0.1.dev19
+Version: 0.0.1.dev21
 Summary: A library for working with identifiers of all kinds.
 Home-page: https://github.com/tgbugs/idlib
 Author: Tom Gillespie
@@ -22,11 +22,21 @@ Classifier: Operating System :: MacOS :: MacOS X
 Classifier: Operating System :: Microsoft :: Windows
 Requires-Python: >=3.6
 Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: orthauth[sxpr,yaml]>=0.0.13
+Requires-Dist: requests
 Provides-Extra: test
+Requires-Dist: pytest; extra == "test"
+Requires-Dist: joblib>=1.1.0; extra == "test"
+Requires-Dist: beautifulsoup4[html5lib]; extra == "test"
+Requires-Dist: pyontutils>=0.1.28; extra == "test"
+Requires-Dist: google-auth-oauthlib; extra == "test"
 Provides-Extra: org
+Requires-Dist: beautifulsoup4[html5lib]; extra == "org"
 Provides-Extra: rdf
+Requires-Dist: pyontutils>=0.1.28; extra == "rdf"
 Provides-Extra: oauth
-License-File: LICENSE
+Requires-Dist: google-auth-oauthlib; extra == "oauth"
 
 # idlib
 [![PyPI version](https://badge.fury.io/py/idlib.svg)](https://pypi.org/project/idlib/)

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/idlib/__init__.py

@@ -55,4 +55,4 @@ def get_right_id(uri):
     return pi
 
 
-__version__ = '0.0.1.dev19'
+__version__ = '0.0.1.dev21'

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/idlib/from_oq.py

@@ -149,7 +149,9 @@ class PioId(oq.OntId, idlib.Identifier, idlib.Stream):
 
     @property
     def uri_human(self):
-        return self.__class__(prefix='pio.view', suffix=self.slug)
+        pid = self.__class__(prefix='pio.view', suffix=self.slug)
+        pid._id_converted_from = self
+        return pid
 
     @property
     def uri_api(self):
@@ -157,15 +159,21 @@ class PioId(oq.OntId, idlib.Identifier, idlib.Stream):
             #prefix = 'pio.folders.api'  # XXX prefix won't work here, requires a suffix /ids SIGH
         #else:
             #prefix = 'pio.api'
-        return self.__class__(prefix='pio.api', suffix=self.slug)
+        pid = self.__class__(prefix='pio.api', suffix=self.slug)
+        pid._id_converted_from = self
+        return pid
 
     @property
     def uri_api1(self):
-        return self.__class__(prefix='pio.api1', suffix=self.slug)
+        pid = self.__class__(prefix='pio.api1', suffix=self.slug)
+        pid._id_converted_from = self
+        return pid
 
     @property
     def uri_api3(self):
-        return self.__class__(prefix='pio.api3', suffix=self.slug)
+        pid = self.__class__(prefix='pio.api3', suffix=self.slug)
+        pid._id_converted_from = self
+        return pid
 
     def normalize(self):
         return self
@@ -584,8 +592,11 @@ class Pio(formats.Rdf, idlib.Stream):
         if noh:
             # XXX noh assumes that the originating form was a uri_human use at own risk
             return self._get_direct(self.identifier.uri_api)
-        else:
+        elif self.uri_human:
             return self.uri_human._get_direct(self.identifier.uri_api)
+        else:
+            msg = 'no beans'
+            raise NotImplementedError(msg)
 
     def data1(self, fail_ok=False, noh=False):
         # FIXME this depends on data4 for uri_human and cannot be used alone at the moment
@@ -595,11 +606,38 @@ class Pio(formats.Rdf, idlib.Stream):
         # because there is no oauth for it
         # get data from the api/v1 endpoint which is is needed for unmangled steps
         resp = self._data_direct(noh=noh)
+        # amazingly we get json even on fail ... 500 will not though
+        # I'm sure so some day this will break
         j = resp.json()
         if 'protocol' not in j:  # FIXME SIGH
-            if 'status_code' in j and j['status_code'] in (250, 205):
-                message = j['text'] + ' ' + self.identifier.asStr()
-                raise exc.NotAuthorizedError(message)
+            if 'status_code' in j:
+                # FIXME this is not the right way to do it
+                # we could be handed an id that was an api
+                # id that is clearly using the len 32 private
+                # id and we should probably just detect that
+                # instead of trying to chase conversion prov
+                # which can get broken for any number of reasons
+                id_was_private = False
+                _id = self.identifier
+                while True:
+                    if _id.is_private():
+                        id_was_private = True
+                        break
+                    elif not hasattr(_id, '_id_converted_from'):
+                        break
+                    else:
+                        _id = _id._id_converted_from
+
+                if j['status_code'] in (250, 205):
+                    message = j['status_text'] + ' ' + self.identifier.asStr()
+                    raise exc.NotAuthorizedError(message)
+                elif j['status_code'] == 1 and id_was_private:
+                    msg = (
+                        f'{j["status_text"]} private id might have existed before but '
+                        f'does not exist any more {self.identifier.asStr()}')
+                    raise exc.IdDoesNotExistError(msg)
+                else:
+                    raise NotImplementedError((resp.url, j))
 
             return None
 
@@ -767,10 +805,17 @@ class Pio(formats.Rdf, idlib.Stream):
             elif self.identifier.is_private():
                 try:
                     return self.data4(fail_ok=fail_ok)
-                except:
-                    return self.data1(fail_ok=fail_ok)
+                except exc.IdDoesNotExistError as ene:
+                    raise ene
+                except Exception as e0:
+                    try:
+                        return self.data1(fail_ok=fail_ok)
+                    except Exception as e1:
+                        raise exc.TransportError() from e1
             else:
                 return self.data4(fail_ok=fail_ok)
+        except (AttributeError, NotImplementedError) as e:
+            raise e
         except Exception as e:
             wait_until = time() + self._wait_time
             self._data_in_error_sigh = wait_until, e
@@ -2174,7 +2219,9 @@ class PioUser(idlib.HelperNoData, idlib.Stream):
 
     @property
     def orcid(self):
-        orcid = self.metadata()['orcid']
+        md = self.metadata()
+        k = 'orcid'
+        orcid =  md[k]if k in md and md[k] else None
         if orcid is not None:
             return idlib.Orcid.fromIdInit(prefix='orcid', suffix=orcid)
 
@@ -2186,7 +2233,9 @@ class PioUser(idlib.HelperNoData, idlib.Stream):
 
     @property
     def bio(self):
-        return self.metadata()['bio']
+        md = self.metadata()
+        k = 'bio'
+        return md[k]if k in md and md[k] else None
 
     description = bio
 

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/idlib/streams.py

@@ -451,8 +451,10 @@ class HelpTestStreams:
         # a separate loop?
         cypher = self._hashlib.blake2b
         bads = []
+        ds = []
         for i in self.ids:
             d = self.stream(i)
+            ds.append(d)
             d.identifier
             if d.identifier and hasattr(d.identifier, 'prefix'):
                 print(d.identifier)
@@ -547,10 +549,12 @@ class HelpTestStreams:
         assert i1.identifier is not i2
 
     def test_asDict(self):
+        ds = []
         for id in self.ids:
             s = self.stream(id)
             try:
                 d = s.asDict()
+                ds.append(d)
             except self._requests.exceptions.ConnectionError as e:
                 self._pytest.skip('Internet done goofed')
 
@@ -572,7 +576,6 @@ class StreamUri(Stream):
 
     _id_class = None  # NOTE asssigned to idlib.Uri in idlib.__init__
 
-
     @property
     def identifier(self):
         return self._identifier

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/idlib/systems/orcid.py

@@ -29,10 +29,9 @@ OrcidPrefixes({'orcid': 'https://orcid.org/',
 class OrcidId(oq.OntId, idlib.Identifier):
     _namespaces = OrcidPrefixes
     _local_conventions = _namespaces
-    local_regex = ('^0000-000(1-[5-9]|2-[0-9]|3-[0-4])'
-                   '[0-9][0-9][0-9]-[0-9][0-9][0-9]([0-9]|X)$')
-    canonical_regex = ('^https://orcid.org/0000-000(1-[5-9]|2-[0-9]|3-'
-                       '[0-4])[0-9][0-9][0-9]-[0-9][0-9][0-9]([0-9]|X)$')
+    _common = '000[09]-00[01][0-9]-[0-9]{4}-[0-9]{3}([0-9]|X)'
+    local_regex = (f'^{_common}$')
+    canonical_regex = (f'^https://orcid.org/{_common}$')
 
     class OrcidMalformedError(exc.IdlibError):
         """ WHAT HAVE YOU DONE!? """

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/idlib/systems/rrid.py

@@ -57,7 +57,9 @@ class Rrid(formats.Rdf, idlib.HelperNoData, idlib.Stream):
     def id_bound_metadata(self):  # FIXME bound_id_metadata bound_id_data
         metadata = self.metadata()
         # wouldn't it be nice if all the metadata schemas had a common field called 'identifier' ?
-        id = metadata['rrid']['curie']
+        ids = set(m['rrid']['curie'] for m in metadata)
+        assert len(ids) == 1, ids
+        id = next(iter(ids))
         return self._id_class(id)
 
     identifier_bound_metadata = id_bound_metadata
@@ -78,9 +80,12 @@ class Rrid(formats.Rdf, idlib.HelperNoData, idlib.Stream):
         if metadata is not None:
             self._path_metadata = path
             self._progenitor_metadata_blob = metadata
-            source = metadata['hits']['hits'][0]['_source']
+            self._total_hits = metadata['hits']['total']
+            source = [h['_source'] for h in metadata['hits']['hits']]
             return source
 
+        return []
+
     def _cooldown(self):
         self._COOLDOWN = True
         metadata, path = self._metadata(self.identifier)
@@ -109,12 +114,11 @@ class Rrid(formats.Rdf, idlib.HelperNoData, idlib.Stream):
         # the identity of the RRID record ...
         m = cypher()
         metadata = self.metadata()
-        proper_citation = metadata['rrid']['properCitation']
+        proper_citations = sorted([m['rrid']['properCitation'] for m in metadata])
         m.update(self.identifier.checksum(cypher))
         m.update(self.id_bound_metadata.checksum(cypher))
-        m.update(proper_citation.encode())
-        for vuri in self.vendorUris:
-            m.update(vuri.encode())
+        for proper_citation in proper_citations:
+            m.update(proper_citation.encode())
 
         return m.digest()
 
@@ -122,33 +126,54 @@ class Rrid(formats.Rdf, idlib.HelperNoData, idlib.Stream):
     def vendorUris(self):
         metadata = self.metadata()
         if 'vendors' in metadata:  # FIXME SCR continues to be a bad citizen >_<
-            return [v['uri'] for v in metadata['vendors']]
+            return [v['uri'] if 'uri' in v else None for v in metadata['vendors']]
         else:
             return []
 
     @property
     def name(self):
-        m = self.metadata()
-        if m is not None:
-            return m['item']['name']
+        ms = self.metadata()
+        if ms:
+            names = set(m['item']['name'] for m in ms)
+            return sorted(names, key=len, reverse=True)[0]
 
     label = name
 
     @property
     def synonyms(self):
-        m = self.metadata()['item']
-        fs = 'label', 'synonyms', 'abbreviations'
-        out = []
-        for f in fs:
-            if f in m:
-                for v in m[f]:
-                    out.append(v)
-
-        return out
+        ms = self.metadata()
+        out = set()
+        for m in ms:
+            it = m['item']
+            fs = 'label', 'name', 'nomenclature', 'synonyms', 'abbreviations'
+            for f in fs:
+                if f in it:
+                    thing = it[f]
+                    if isinstance(thing, list):
+                        for v in thing:
+                            if type(v) is dict:
+                                if 'name' in v:
+                                    for sigh in v['name'].split('. '):
+                                        syn = sigh.strip()
+                                        if syn:
+                                            out.add(syn)
+                                else:
+                                    breakpoint()
+                                    raise NotImplementedError(f'TODO {v}')
+                            else:
+                                out.add(v)
+                    else:
+                        out.add(thing)
+
+        return sorted(out)
 
     @property
     def description(self):
-        return self.metadata()['item']['description']
+        ms = self.metadata()
+        if ms:
+            descs = set(m['item']['description'] for m in ms if 'description' in m['item'])
+            if descs:
+                return sorted(descs, key=len, reverse=True)[0]
 
     # alternate representations
 

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/idlib.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: idlib
-Version: 0.0.1.dev19
+Version: 0.0.1.dev21
 Summary: A library for working with identifiers of all kinds.
 Home-page: https://github.com/tgbugs/idlib
 Author: Tom Gillespie
@@ -22,11 +22,21 @@ Classifier: Operating System :: MacOS :: MacOS X
 Classifier: Operating System :: Microsoft :: Windows
 Requires-Python: >=3.6
 Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: orthauth[sxpr,yaml]>=0.0.13
+Requires-Dist: requests
 Provides-Extra: test
+Requires-Dist: pytest; extra == "test"
+Requires-Dist: joblib>=1.1.0; extra == "test"
+Requires-Dist: beautifulsoup4[html5lib]; extra == "test"
+Requires-Dist: pyontutils>=0.1.28; extra == "test"
+Requires-Dist: google-auth-oauthlib; extra == "test"
 Provides-Extra: org
+Requires-Dist: beautifulsoup4[html5lib]; extra == "org"
 Provides-Extra: rdf
+Requires-Dist: pyontutils>=0.1.28; extra == "rdf"
 Provides-Extra: oauth
-License-File: LICENSE
+Requires-Dist: google-auth-oauthlib; extra == "oauth"
 
 # idlib
 [![PyPI version](https://badge.fury.io/py/idlib.svg)](https://pypi.org/project/idlib/)

{idlib-0.0.1.dev19 → idlib-0.0.1.dev21}/test/test_identifiers.py

@@ -58,6 +58,7 @@ class TestOrcid(HelpTestStreams, unittest.TestCase):
         'https://orcid.org/0000-0002-1825-0097',
         'https://orcid.org/0000-0001-5109-3700',
         'https://orcid.org/0000-0002-1694-233X',
+        'https://orcid.org/0009-0005-9822-0865',
     ]
     ids_bad = []
     for __i in ids:  # LOL PYTHON class scope list comprehensions
@@ -175,6 +176,7 @@ class TestPio(HelpTestStreams, unittest.TestCase):
         'lol not an identifier',
         'https://www.protocols.io/view/18980',  # deleted
         'https://www.protocols.io/api/v3/protocols/18980',  # deleted
+        'https://www.protocols.io/private/D327A68038D5B4ECB32964C22D904270',  # deleted
     ]
 
     @skipif_no_net