Merge pull request #1 from datopian/feat/user-provide-doi

User can provide custom doi identifier

ckanext/doi/lib/api.py

@@ -75,29 +75,23 @@ def get_prefix(cls):
     def generate_doi(self, identifier=None):
         """
         Generate a new DOI which isn't currently in use.
-
+    
         The database is checked for previous
         usage, as is Datacite itself. Use whatever value is retuned from this function quickly to
         avoid double use as this function uses no locking.
         :return: the full, unique DOI
         """
-        # the list of valid characters is larger than just lowercase and the digits but we don't
-        # need that many options and URLs with just alphanumeric characters in them are nicer. We
-        # just use lowercase characters to avoid any issues with case being ignored
         valid_characters = string.ascii_lowercase + string.digits
-
         attempts = 5
-
+    
         while attempts > 0:
-
-            year = dt.now().year
-            # form the doi using the prefix
             if identifier:
                 doi = identifier
             else:
-                # generate a random 8 character identifier
                 random_identifier = ''.join(random.choice(valid_characters) for _ in range(8))
+                year = dt.now().year
                 doi = f'{self.prefix}/{year}.{random_identifier}'
+
             if DOIQuery.read_doi(doi) is None:
                 try:
                     self.client.metadata_get(doi)
@@ -109,7 +103,7 @@ def generate_doi(self, identifier=None):
                         f'error: {e}'
                     )
             attempts -= 1
-        raise Exception('Failed to generate a DOI')
+        raise toolkit.ValidationError('Could not generate DOI')
 
     def mint_doi(self, doi, package_id):
         """

ckanext/doi/model/crud.py

@@ -7,6 +7,7 @@
 from ckan.model import Session
 
 from ckanext.doi.model.doi import DOI, doi_table
+from ckan.plugins import toolkit as tk
 
 
 class DOIQuery:
@@ -43,7 +44,12 @@ def read_doi(cls, identifier):
 
     @classmethod
     def read_package(
-        cls, package_id, version=None, is_version=None, create_if_none=False
+        cls,
+        package_id,
+        version=None,
+        is_version=None,
+        requested_identifier=None,
+        create_if_none=False,
     ):
         """
         Retrieve a record associated with a given package.
@@ -57,7 +63,7 @@ def read_package(
 
         doi_exist = Session.query(DOI).filter(DOI.package_id == package_id).first()
         identifier = None
-        
+
         if is_version and version:
             # Find the DOI for the version
             version_doi = (
@@ -68,14 +74,19 @@ def read_package(
                 # Remove existing version suffix and append the new version
                 identifier = re.sub(r'\.v\d+$', '', version_doi.identifier)
                 identifier = f"{identifier}.v{version}"
-
-		# If no DOI exists and creation is allowed
-        if doi_exist is None and create_if_none:
-            client = DataciteClient()
-            new_doi = client.generate_doi(identifier=identifier)
-            new_doi = cls.create(new_doi, package_id)
-            return new_doi
 
+        # If no DOI exists and creation is allowed
+        if doi_exist is None and create_if_none:
+            try:
+                if requested_identifier:
+                    new_doi = cls.create(requested_identifier, package_id)
+                else:
+                    client = DataciteClient()
+                    new_doi = client.generate_doi(identifier=identifier)
+                    new_doi = cls.create(new_doi, package_id)
+                return new_doi
+            except Exception as e:
+                raise tk.ValidationError(f"Error creating DOI: {e}")
         return doi_exist
 
     @classmethod

ckanext/doi/plugin.py

@@ -3,7 +3,7 @@
 #
 # This file is part of ckanext-doi
 # Created by the Natural History Museum in London, UK
-
+import re
 from datetime import datetime
 from logging import getLogger
 
@@ -25,6 +25,15 @@
 log = getLogger(__name__)
 
 
+def doi_validator(key, data, errors, context):
+    value = data.get(key)
+    PATTERN = re.compile("^10\.[0-9]{4,9}/[-._;()/:a-zA-Z0-9]+$")
+
+    if value:
+        if not PATTERN.match(value):
+            errors[key].append(toolkit._("Invalid DOI format"))
+            return
+
 class DOIPlugin(SingletonPlugin, toolkit.DefaultDatasetForm):
     """
     CKAN DOI Extension.
@@ -35,6 +44,7 @@ class DOIPlugin(SingletonPlugin, toolkit.DefaultDatasetForm):
     implements(interfaces.ITemplateHelpers, inherit=True)
     implements(interfaces.IBlueprint)
     implements(interfaces.IClick)
+    implements(interfaces.IValidators)
 
     ## IClick
     def get_commands(self):
@@ -55,11 +65,11 @@ def after_dataset_create(self, context, pkg_dict):
         NB: This is called after creation of a dataset, before resources have been
         added, so state = draft.
         """
-
         DOIQuery.read_package(
             pkg_dict['id'],
             version=pkg_dict.get('version'),
             is_version=pkg_dict.get('is_version_of'),
+            requested_identifier= pkg_dict.get('doi', None),
             create_if_none=True,
         )
 
@@ -77,7 +87,6 @@ def after_dataset_update(self, context, pkg_dict):
         )
         if is_active:
             package_id = pkg_dict['id']
-
             # remove user-defined update schemas first (if needed)
             context.pop('schema', None)
             client = DataciteClient()
@@ -123,6 +132,12 @@ def get_blueprint(self):
         Return the blueprint to be registered.
         """
         return registred_views()
+
+    # IValidators
+    def get_validators(self):
+        return {
+            'doi_validator': doi_validator,
+        }
 
     # ITemplateHelpers
     def get_helpers(self):