dryad2dataverse 0.7.11a0__py3-none-any.whl → 0.8.1__py3-none-any.whl

dryad2dataverse/serializer.py

@@ -2,22 +2,22 @@
 Serializes Dryad study JSON to Dataverse JSON, as well as
 producing associated file information.
 '''
-
-
 import logging
 import urllib.parse
 
 import requests
 from requests.adapters import HTTPAdapter
 
-from  dryad2dataverse import constants
-from dryad2dataverse import USERAGENT
+from dryad2dataverse import config
+import dryad2dataverse.auth
 
 LOGGER = logging.getLogger(__name__)
 #Connection monitoring as per
 #https://stackoverflow.com/questions/16337511/log-all-requests-from-the-python-requests-module
 URL_LOGGER = logging.getLogger('urllib3')
-USER_AGENT = {'User-agent': USERAGENT}
+
+#pylint: disable=invalid-name, line-too-long
+#Note: Metadata downloads do not (as of 2026-01) require authentication
 
 class Serializer():
     '''
@@ -27,20 +27,44 @@ class Serializer():
     <img src="https://licensebuttons.net/p/zero/1.0/88x31.png" title="Creative Commons CC0 1.0 Universal Public Domain Dedication. " style="display:none" onload="this.style.display='inline'" />
     <a href="http://creativecommons.org/publicdomain/zero/1.0" title="Creative Commons CC0 1.0 Universal Public Domain Dedication. " target="_blank">CC0 1.0</a>
     </p>'''
-    
-    def __init__(self, doi):
+
+    def __init__(self, doi:str, **kwargs):
         '''
         Creates Dryad study metadata instance.
 
-        ----------------------------------------
-        Parameters:
+        Parameters
+        ----------
+        doi : str
+            DOI of Dryad study. Required for downloading.
+            eg: 'doi:10.5061/dryad.2rbnzs7jp'
+
+        kwargs : dict
+            Other keyword parameters
+
+        Other parameters
+        ----------------
+        token : dryad2dataverse.auth.Token
+            If present, will use authenticated API
+
+        Notes
+        -----
+        Unpacking a dryad2dataverse.config.Config instance holding
+        global setup should give all of the
+        required kwargs. ie, Serializer(doi, **config_instance)
 
-    doi : str
-        — DOI of Dryad study. Required for downloading.
-        eg: 'doi:10.5061/dryad.2rbnzs7jp'
-    ----------------------------------------
         '''
         self.doi = doi
+        self.kwargs = kwargs
+        self.kwargs['dry_url'] = kwargs.get('dry_url', 'https://datadryad.org')
+        self.kwargs['api_path'] = kwargs.get('api_path', '/api/v2')
+        self.kwargs['max_upload'] = kwargs.get('max_upload', 3221225472)
+        self.kwargs['dv_contact_name'] = kwargs.get('dv_contact_name')
+        self.kwargs['dv_contact_email'] = kwargs.get('dv_contact_email')
+        if self.kwargs.get('token'):
+            if not isinstance(self.kwargs['token'],dryad2dataverse.auth.Token):
+                raise ValueError('Token must be a dryad2dataverse.auth.Token instance')
+        #Don't need timeout if have RETRY_STRATEGY
+        self.kwargs['timeout'] = kwargs.get('timeout', 100)
         self._dryadJson = None
         self._fileJson = None
         self._dvJson = None
@@ -49,35 +73,28 @@ class Serializer():
         self.dvpid = None
         self.session = requests.Session()
         self.session.mount('https://',
-                           HTTPAdapter(max_retries=constants.RETRY_STRATEGY))
+                           HTTPAdapter(max_retries=config.RETRY_STRATEGY))
         LOGGER.debug('Creating Serializer instance object')
 
-    def fetch_record(self, url=None, timeout=45):
+    def fetch_record(self, url=None) :
         '''
         Fetches Dryad study record JSON from Dryad V2 API at
         https://datadryad.org/api/v2/datasets/.
         Saves to self._dryadJson. Querying Serializer.dryadJson
         will call this function automatically.
 
-        ----------------------------------------
-        Parameters:
-
+        Parameters
+        ----------
         url : str
-            — Dryad instance base URL (eg: 'https://datadryad.org').
-
-        timeout : int
-            — Timeout in seconds. Default 45.
-        ----------------------------------------
+            Dryad instance base URL (eg: 'https://datadryad.org').
         '''
         if not url:
-            url = constants.DRYURL
+            url = self.kwargs['dry_url']
         try:
-            headers = {'accept':'application/json',
-                       'Content-Type':'application/json'}
-            headers.update(USER_AGENT)
+            headers = config.Config.update_headers(**self.kwargs)
             doiClean = urllib.parse.quote(self.doi, safe='')
-            resp = self.session.get(f'{url}/api/v2/datasets/{doiClean}',
-                                    headers=headers, timeout=timeout)
+            resp = self.session.get(f'{url}{self.kwargs["api_path"]}/datasets/{doiClean}',
+                                    headers=headers, timeout=self.kwargs['timeout'])
             resp.raise_for_status()
             self._dryadJson = resp.json()
         except (requests.exceptions.HTTPError,
@@ -118,12 +135,10 @@ class Serializer():
         If supplying it, make sure it's correct or you will run into trouble
         with processing later.
 
-        ----------------------------------------
-        Parameters:
-
+        Parameters
+        ----------
         value : dict
-            — Dryad JSON.
-
+            Dryad JSON.
         '''
         if value:
             self._dryadJson = value
@@ -131,9 +146,10 @@ class Serializer():
             self.fetch_record()
 
     @property
-    def embargo(self):
+    def embargo(self)->bool:
         '''
         Check embargo status. Returns boolean True if embargoed.
+
         '''
         if self.dryadJson.get('curationStatus') == 'Embargoed':
             return True
@@ -148,39 +164,31 @@ class Serializer():
         return self._dvJson
 
     @property
-    def fileJson(self, timeout=45):
+    def fileJson(self):
         '''
         Returns a list of file JSONs from call to Dryad API /files/{id},
         where the ID is parsed from the Dryad JSON. Dryad file listings
         are paginated, so the return consists of a list of dicts, one
         per page.
-
-        ----------------------------------------
-        Parameters:
-
-        timeout : int
-            — Request timeout in seconds.
-        ----------------------------------------
         '''
         if not self._fileJson:
             try:
                 self._fileJson = []
-                headers = {'accept':'application/json',
-                           'Content-Type':'application/json'}
-                headers.update(USER_AGENT)
-                fileList = self.session.get(f'{constants.DRYURL}/api/v2/versions/{self.id}/files',
+                headers = config.Config.update_headers(**self.kwargs)
+                fileList = self.session.get(f'{self.kwargs["dry_url"]}'
+                                            f'{self.kwargs["api_path"]}/versions/{self.id}/files',
                                             headers=headers,
-                                            timeout=timeout)
+                                            timeout=self.kwargs['timeout'])
                 fileList.raise_for_status()
                 #total = fileList.json()['total'] #Not needed
                 lastPage = fileList.json()['_links']['last']['href']
                 pages = int(lastPage[lastPage.rfind('=')+1:])
                 self._fileJson.append(fileList.json())
                 for i in range(2, pages+1):
-                    fileCont = self.session.get(f'{constants.DRYURL}/api/v2'
+                    fileCont = self.session.get(f'{self.kwargs["dry_url"]}{self.kwargs["api_path"]}'
                                                 f'/versions/{self.id}/files?page={i}',
                                                 headers=headers,
-                                                timeout=timeout)
+                                                timeout=self.kwargs['timeout'])
                     fileCont.raise_for_status()
                     self._fileJson.append(fileCont.json())
             except Exception as e:
@@ -189,7 +197,7 @@ class Serializer():
         return self._fileJson
 
     @property
-    def files(self):
+    def files(self)->list:
         '''
         Returns a list of tuples with:
 
@@ -211,7 +219,7 @@ class Serializer():
 
                     #downLink = f['_links']['stash:file-download']['href']
                     downLink = f['_links']['stash:download']['href']
-                    downLink = f'{constants.DRYURL}{downLink}'
+                    downLink = f'{self.kwargs["dry_url"]}{downLink}'
                     name = f['path']
                     mimeType = f['mimeType']
                     size = f['size']
@@ -235,22 +243,13 @@ class Serializer():
         return out
 
     @property
-    def oversize(self, maxsize=None):
+    def oversize(self):
         '''
         Returns a list of Dryad files whose size value
         exceeds maxsize. Maximum size defaults to
-        dryad2dataverse.constants.MAX_UPLOAD
-
-        ----------------------------------------
-        Parameters:
-
-        maxsize : int
-            — Size in bytes in which to flag as oversize.
-              Defaults to constants.MAX_UPLOAD.
-        ----------------------------------------
+        dryad2dataverse.config.MAX_UPLOAD
         '''
-        if not maxsize:
-            maxsize = constants.MAX_UPLOAD
+        maxsize = self.kwargs['max_upload']
         toobig = []
         for f in self.files:
             if f[3] >= maxsize:
@@ -264,19 +263,17 @@ class Serializer():
         Creates wrapper around single or multiple Dataverse JSON objects.
         Returns a dict *without* the  Dataverse 'value' key'.
 
-        ----------------------------------------
-        Parameters:
-
+        Parameters
+        ----------
         typeName : str
-            — Dataverse typeName (eg: 'author').
+            Dataverse typeName (eg: 'author').
 
         multiple : boolean
-            — "Multiple" value in Dataverse JSON.
+            "Multiple" value in Dataverse JSON.
 
         typeClass : str
-            — Dataverse typeClass. Usually one of 'compound', 'primitive,
+            Dataverse typeClass. Usually one of 'compound', 'primitive,
               'controlledVocabulary').
-        ----------------------------------------
         '''
         return {'typeName':typeName, 'multiple':multiple,
                 'typeClass':typeClass}
@@ -293,29 +290,32 @@ class Serializer():
         Suitable for generalized conversions. Only provides fields with
         multiple: False and typeclass:Primitive
 
-        ----------------------------------------
-        Parameters:
+        Parameters
+        ----------
+        kwargs : dict
+            Dict from Dataverse JSON segment
 
+        Other parameters
+        ----------------
         dvField : str
-            — Dataverse output field
+            Dataverse output field
 
         dryField : str
-            — Dryad JSON field to convert
+            Dryad JSON field to convert
 
         inJson : dict
-            — Dryad JSON **segment** to convert
+            Dryad JSON **segment** to convert
 
         addJSON : dict (optional)
-            — any other JSON required to complete (cf ISNI)
+        	any other JSON required to complete (cf ISNI)
 
         rType : str
-            — 'dict' (default) or 'list'.
+        	'dict' (default) or 'list'.
             Returns 'value' field as dict value or list.
 
         pNotes : str
             Notes to be prepended to list type values.
             No trailing space required.
-        ----------------------------------------
         '''
 
         dvField = kwargs.get('dvField')
@@ -359,12 +359,11 @@ class Serializer():
         '''
         Produces required author json fields.
         This is a special case, requiring concatenation of several fields.
-        ----------------------------------------
-        Parameters:
 
+        Parameters
+        ----------
         author : dict
-            — dryad['author'] JSON segment.
-        ----------------------------------------
+        	dryad['author'] JSON segment.
         '''
         first = author.get('firstName')
         last = author.get('lastName')
@@ -381,13 +380,12 @@ class Serializer():
         Produces the insane keyword structure Dataverse JSON segment
         from a list of words.
 
-        ----------------------------------------
-        Parameters:
-
-        args : list with str elements
-            — Generally input is Dryad JSON 'keywords', ie *Dryad['keywords'].
-              Don't forget to expand the list using *.
-        ----------------------------------------
+        Parameters
+        ----------
+        args : list
+            List with elements as strings.
+        	Generally input is Dryad JSON 'keywords', ie *Dryad['keywords'].
+            Don't forget to expand the list using *.
         '''
         outlist = []
         for arg in args:
@@ -402,12 +400,10 @@ class Serializer():
         Returns formatted notes field with Dryad JSON values that
         don't really fit anywhere into the Dataverse JSON.
 
-        ----------------------------------------
-        Parameters:
-
+        Parameters
+        ----------
         dryJson : dict
-            — Dryad JSON as dict.
-        ----------------------------------------
+        	Dryad JSON as dict.
         '''
         notes = ''
         #these fields should be concatenated into notes
@@ -423,34 +419,24 @@ class Serializer():
                    'storageSize',
                    'visibility',
                    'skipEmails']
+        lookup = {'versionNumber': '<b>Dryad version number:</b>',
+                  'versionStatus': '<b>Version status:</b>',
+                  'manuscriptNumber': '<b>Manuscript number:</b>',
+                  'curationStatus': '<b>Dryad curation status:</b>',
+                  'preserveCurationStatus': '<b>Dryad preserve curation status:</b>',
+                  'invoiceId': '<b>Invoice ID:</b>',
+                  'sharingLink': '<b>Sharing link:</b>',
+                  'loosenValidation': '<b>Loosen validation:</b>',
+                  'skipDataciteUpdate': '<b>Skip Datacite update:</b>',
+                  'storageSize': '<b>Storage size:</b>',
+                  'visibility': '<b>Visibility:</b>',
+                  'skipEmails': '<b>Skip emails:</b>'}
         for note in notable:
             text = dryJson.get(note)
             if text:
                 text = str(text).strip()
-                if note == 'versionNumber':
-                    text = f'<b>Dryad version number:</b> {text}'
-                if note == 'versionStatus':
-                    text = f'<b>Version status:</b> {text}'
-                if note == 'manuscriptNumber':
-                    text = f'<b>Manuscript number:</b> {text}'
-                if note == 'curationStatus':
-                    text = f'<b>Dryad curation status:</b> {text}'
-                if note == 'preserveCurationStatus':
-                    text = f'<b>Dryad preserve curation status:</b> {text}'
-                if note == 'invoiceId':
-                    text = f'<b>Invoice ID:</b> {text}'
-                if note == 'sharingLink':
-                    text = f'<b>Sharing link:</b> {text}'
-                if note == 'loosenValidation':
-                    text = f'<b>Loosen validation:</b> {text}'
-                if note == 'skipDataciteUpdate':
-                    text = f'<b>Skip Datacite update:</b> {text}'
-                if note == 'storageSize':
-                    text = f'<b>Storage size:</b> {text}'
-                if note == 'visibility':
-                    text = f'<b>Visibility:</b> {text}'
-                if note == 'skipEmails':
-                    text = f'<b>Skip emails:</b> {text}'
+                if note in lookup:
+                    text = f'{lookup.get(note)} {text}'
 
                 notes += f'<p>{text}</p>\n'
         concat = {'typeName':'notesText',
@@ -465,12 +451,16 @@ class Serializer():
         Makes a Dataverse bounding box from appropriate coordinates.
         Returns Dataverse JSON segment as dict.
 
-        ----------------------------------------
-        Parameters:
+        Parameters
+        ----------
+        north : float
+        south : float
+        east : float
+        west : float
 
-        north, south, east, west : float
-            — Coordinates in decimal degrees.
-        ----------------------------------------
+        Notes
+        -----
+            Coordinates in decimal degrees.
         '''
         names = ['north', 'south', 'east', 'west']
         points = [str(x) for x in [north, south, east, west]]
@@ -482,7 +472,8 @@ class Serializer():
             out.append(Serializer._convert_generic(inJson=coord[1],
                                                    dvField=coord[0],
                                                    #dryField='north'))
-                                                   dryField=[k for k in coord[1].keys()][0]))
+                                                   #dryField=[k for k in coord[1].keys()][0]))
+                                                   dryField=list(coord[1].keys())[0]))
         return out
 
     @staticmethod
@@ -490,12 +481,10 @@ class Serializer():
         '''
         Outputs Dataverse geospatial metadata block.
 
-        ----------------------------------------
-        Parameters:
-
+        Parameters
+        ----------
         dryJson : dict
-            — Dryad json as dict.
-        ----------------------------------------
+        	Dryad json as dict.
         '''
         if dryJson.get('locations'):
             #out = {}
@@ -552,30 +541,30 @@ class Serializer():
 
     def _assemble_json(self, dryJson=None, dvContact=None,
                        dvEmail=None, defContact=True):
+        #pylint: disable = too-many-statements, too-many-locals, too-many-branches
         '''
+
         Assembles Dataverse json from Dryad JSON components.
         Dataverse JSON is a nightmare, so this function is too.
 
-        ----------------------------------------
-        Parameters:
-
+        Parameters
+        ----------
         dryJson : dict
-            — Dryad json as dict.
+        	Dryad json as dict.
 
         dvContact : str
-            — Default Dataverse contact name.
+        	Default Dataverse contact name.
 
         dvEmail : str
-            — Default Dataverse 4 contact email address.
+        	Default Dataverse 4 contact email address.
 
         defContact : boolean
-            — Flag to include default contact information with record.
-        ----------------------------------------
+        	Flag to include default contact information with record.
         '''
         if not dvContact:
-            dvContact = constants.DV_CONTACT_NAME
+            dvContact = self.kwargs['dv_contact_name']
         if not dvEmail:
-            dvEmail = constants.DV_CONTACT_EMAIL
+            dvEmail = self.kwargs['dv_contact_email']
         if not dryJson:
             dryJson = self.dryadJson
         LOGGER.debug(dryJson)