Add doi date script and pass through for emails

Author: tmorrell

add_doi_minting_date.py

@@ -0,0 +1,41 @@
+import os,requests
+from progressbar import progressbar
+from caltechdata_api import get_metadata, caltechdata_edit
+
+def get_datacite_dates(prefix):
+    '''Get sumbitted date for DataCite DOIs with specific prefix'''
+    doi_dates = {}
+    doi_urls = {}
+    url = 'https://api.datacite.org/dois?query=prefix:'+prefix+'&page[cursor]=1&page[size]=500'
+    next_link = url
+    meta = requests.get(next_link).json()['meta']
+    for j in progressbar(range(meta['totalPages'])):
+        r = requests.get(next_link)
+        data = r.json()
+        for doi in data['data']:
+            date = doi['attributes']['registered'].split('T')[0]
+            doi_dates[doi['id']] = date
+            doi_urls[doi['id']] =  doi['attributes']['url']
+        if 'next' in data['links']:
+            next_link = data['links']['next']
+        else:
+            next_link = None
+    return doi_dates,doi_urls
+
+token = os.environ['TINDTOK']
+
+doi_dates,doi_urls = get_datacite_dates('10.14291')
+for doi in doi_urls:
+    if 'data.caltech.edu' in doi_urls[doi]:
+        caltech_id = doi_urls[doi].split('/')[-1]
+        if caltech_id not in ['252','253','254','255']:
+            metadata = get_metadata(caltech_id,emails=True)
+            print(caltech_id)
+            #print(metadata['dates'])
+            for date in metadata['dates']:
+                if date['dateType'] == 'Issued':
+                    print(date['date'],doi_dates[doi])
+                    date['date'] = doi_dates[doi]
+            response = caltechdata_edit(token, caltech_id,metadata,production=True)
+            print(response)
+

caltechdata_api/get_metadata.py

@@ -3,8 +3,8 @@
 from datacite import DataCiteMDSClient, schema40
 from caltechdata_api import decustomize_schema
 
-def get_metadata(idv,production=True,auth=None):
-    # Returns just DataCite metadata
+def get_metadata(idv,production=True,auth=None,emails=False):
+    # Returns just DataCite metadata or DataCite metadata with emails
 
     if production==True:
         api_url = "https://data.caltech.edu/api/record/"
@@ -20,15 +20,18 @@ def get_metadata(idv,production=True,auth=None):
         raise AssertionError('expected as metadata property in response, got '+r_data)
     metadata = r_data['metadata']
 
-    metadata = decustomize_schema(metadata)
-    try: 
-        assert schema40.validate(metadata)
-    except AssertionError:
-        v = schema40.validator.validate(metadata)
-        errors = sorted(v.iter_errors(instance), key=lambda e:e.path)
-        for error in errors:
-            print(error.message)
-        exit()
+    if emails == True:
+        metadata = decustomize_schema(metadata,pass_emails=True)
+    else:
+        metadata = decustomize_schema(metadata)
+        try: 
+            assert schema40.validate(metadata)
+        except AssertionError:
+            v = schema40.validator.validate(metadata)
+            errors = sorted(v.iter_errors(instance), key=lambda e:e.path)
+            for error in errors:
+                print(error.message)
+            exit()
 
     return metadata
 
@@ -38,6 +41,7 @@ def get_metadata(idv,production=True,auth=None):
     and returns DataCite-compatable metadata")
     parser.add_argument('ids', metavar='ID', type=int, nargs='+',\
     help='The CaltechDATA ID for each record of interest')
+    parser.add_argument('-emails',dest='emails', action='store_true')
     parser.add_argument('-test',dest='production', action='store_false')
     parser.add_argument('-xml',dest='save_xml', action='store_true')
     parser.add_argument('-auth_user',help='Username for basic authentication')
@@ -46,12 +50,13 @@ def get_metadata(idv,production=True,auth=None):
     args = parser.parse_args()
 
     production = args.production
+    emails = args.emails
     auth = None
     if args.auth_user != None:
         auth = (args.auth_user,args.auth_pass)
 
     for idv in args.ids:
-        metadata = get_metadata(idv,production,auth)
+        metadata = get_metadata(idv,emails,production,auth)
         outfile = open(str(idv)+'.json','w')
         outfile.write(json.dumps(metadata))
         outfile.close()