Cleanup and better DOI handling

Author: tmorrell

README.md

@@ -1,7 +1,5 @@
 # caltechdata_api
 
-[![DOI](https://data.caltech.edu/badge/87960443.svg)](https://data.caltech.edu/badge/latestdoi/87960443)
-
 Python library for using the CaltechDATA API
 
 - caltechdata_write write files and a DataCite 4 standard json record to CaltechDATA repository
@@ -18,49 +16,47 @@ Create a record:
 
 ```shell
 python write.py example.json -fnames logo.gif
-Successfully created record https://cd-sandbox.tind.io/records/352.  
+pbkn6-m9y63
 ```
+The response will be the unique identifier for the record. You can put this at
+the end of a url to visit the record (e.g.
+https://data.caltechlibrary.dev/records/pbkn6-m9y63)
 
 Edit a record (make changes to the example.json file to see a change)
 ```
-python edit.py example.json -ids 352 -fnames logo.gif
-Successfully modified record https://cd-sandbox.tind.io/records/352
+python edit.py example.json -id pbkn6-m9y63
+10.33569/pbkn6-m9y63
 ```
+The response is the DOI for the record, which includes the unique identifier
+for the record in the default configuration.
 
-Get geographic metadata from CaltechDATA with WKT representations in a csv file. 
-You can import this to a GIS program like QGIS
-using a delimited text import and projection epsg:4326. You'll have to do one
-import for Geometry type Point and another for Geometry type Polygon. 
+Some groups have worked with the library to create custom DOIs. These can be
+passed in the metadata like:
 
-```
-python get_geo.py caltechdata_geo.csv
+```shell
+python write.py example_custom.json -fnames logo.gif
+m6zxz-p4j22
 ```
 
-You can filter by keyword
-
+And then you can edit with
 ```
-python get_geo.py caltechdata_geo.csv -keywords TCCON
+python edit.py example_custom.json -id m6zxz-p4j22
+10.5281/inveniordm.1234
 ```
 
+This returns the custom DOI of the record if it is successful.
 
 ## Setup 
 
 Install by typing 'pip install caltechdata_api'
 
 ## Usage
 
-Write API access is controlled by repository staff.  Email us at [email protected] 
-with your request if you want to use the write API.
-
 You need to acquire a personal access token from your CaltechDATA account
 (find it at the top right of your screen under "Applications").
-Make sure you include the "deposit_api:write" and "file_manager:upload"
-scopes.  Then copy the token to token.bash.  Type source token.bash in 
+Then copy the token to token.bash.  Type `source token.bash` in 
 the command line to load the token.  
 
-Only test your application on the test repository.  Testing the API on the public 
+Only test your application on the test repository (data.caltechlibrary.dev).  Testing the API on the public 
 repository will generate junk records that are annoying to delete.
 
-## TODO
-
-Handle incorrect token on file upload

caltechdata_api/caltechdata_edit.py

@@ -82,12 +82,13 @@ def caltechdata_edit(
         repo_prefix = "10.33569"
     pids = {}
     oai = False
+    doi = False
     if "identifiers" in metadata:
         for identifier in metadata["identifiers"]:
             if identifier["identifierType"] == "DOI":
+                doi = True
                 doi = identifier["identifier"]
                 prefix = doi.split("/")[0]
-
                 if prefix == repo_prefix:
                     pids["doi"] = {
                         "identifier": doi,
@@ -105,11 +106,20 @@ def caltechdata_edit(
                     "provider": "oai",
                 }
                 oai = True
+    #Records are not happy without the auto-assigned oai identifier
     if oai == False:
         pids["oai"] = {
             "identifier": f"oai:data.caltech.edu:{idv}",
             "provider": "oai",
         }
+    #We do not want to lose the auto-assigned DOI
+    #Users with custom DOIs must pass them in the metadata
+    if doi == False:
+        pids["doi"] = {
+                        "identifier": f'{repo_prefix}/{idv}',
+                        "provider": "datacite",
+                        "client": "datacite",
+                    }
     metadata["pids"] = pids
 
     data = customize_schema.customize_schema(copy.deepcopy(metadata), schema=schema)

caltechdata_api/caltechdata_write.py

@@ -129,8 +129,8 @@ def caltechdata_write(
         repo_prefix = "10.22002"
     else:
         repo_prefix = "10.33569"
+    pids = {}
     if "identifiers" in metadata:
-        pids = {}
         for identifier in metadata["identifiers"]:
             if identifier["identifierType"] == "DOI":
                 doi = identifier["identifier"]

edit.py

@@ -11,7 +11,7 @@
     default=None,
     help="file name for json DataCite metadata file",
 )
-parser.add_argument("-ids", nargs="*", help="CaltechDATA IDs")
+parser.add_argument("-id", help="CaltechDATA IDs")
 parser.add_argument("-fnames", nargs="*", help="New Files")
 parser.add_argument("-flinks", nargs="*", help="New File Links")
 parser.add_argument("-schema", default="43", help="Metadata Schema")
@@ -26,11 +26,11 @@
 else:
     metadata = {}
 
-production = True
+production = False
 publish = True
 
 response = caltechdata_edit(
-    args.ids,
+    args.id,
     metadata,
     token,
     args.fnames,

example.json

@@ -1,37 +1,48 @@
 {
-    "alternateIdentifiers": [
-        {
-            "alternateIdentifier": "Alt_Identifier",
-            "alternateIdentifierType": "Alt_Identifier_Type"
-        }
+    "identifiers": [
+            {"identifier": "1924MNRAS..84..308E", "identifierType": "bibcode"}
     ],
     "contributors": [
         {
-            "affiliations": [ "Affiliation"  ],
-            "contributorName": "Contributor Name",
+          "nameType": "Personal",
+          "affiliation": [
+              {
+                  "name": "DataCitea",
+                  "affiliationIdentifier": "https://ror.org/04wxnsj81",
+                  "affiliationIdentifierScheme": "ROR"
+                }
+              ],
+            "name": "Contributor Name",
+            "familyName": "Family Name",
+            "givenName": "Given Name",
             "contributorType": "ContactPerson",
             "nameIdentifiers": [
                 {
-                    "nameIdentifier": "Contributor Identifier",
-                    "nameIdentifierScheme": "GRID",
-                    "schemeURI": "https://www.grid.ac/institutes/"
+                    "nameIdentifier": "0000-0002-1825-0097",
+                    "nameIdentifierScheme": "ORCID",
+                    "schemeUri": "https://orcid.org/"
                 }
             ]
         }
     ],
     "creators": [
         {
-            "affiliations": [
-                "Affiliation"
+          "nameType": "Personal",
+          "affiliation": [
+              {
+                "name": "DataCite",
+                "affiliationIdentifier": "https://ror.org/04wxnsj81",
+                "affiliationIdentifierScheme": "ROR"
+              }
             ],
-            "creatorName": "Name",
+            "name": "Name",
             "familyName": "Family Name",
             "givenName": "Given Name",
             "nameIdentifiers": [
                 {
-                    "nameIdentifier": "Name Identifier",
+                    "nameIdentifier": "0000-0002-1825-0097",
                     "nameIdentifierScheme": "ORCID",
-                    "schemeURI": "http://orcid.org/"
+                    "schemeUri": "https://orcid.org/"
                 }
             ]
         }
@@ -41,10 +52,6 @@
             "date": "2014-10-01",
             "dateType": "Created"
         },
-        {
-            "date": "2017-03-01",
-            "dateType": "Updated"
-        },
         {
             "date": "2012-05-22/2016-12-21",
             "dateType": "Collected"
@@ -60,24 +67,24 @@
         "format"
     ],
     "fundingReferences": [
-        {
-            "funderIdentifier": {
-                "funderIdentifier": "Funding Identifier",
-                "funderIdentifierType": "GRID"
-            },
-            "funderName": "Funder Name"
-        }
+          {
+            "awardTitle": "Measurement of Column-Averaged CO2",
+            "funderName": "National Aeronautics and Space Administration",
+            "funderIdentifierType": "GRID",
+            "funderIdentifier": "grid.238252.c",
+            "awardNumber": "NAG5-12247"
+          }
     ],
     "geoLocations": [
         {
             "geoLocationPlace": "Place Name",
             "geoLocationPoint": {
-                "pointLatitude": 0.000,
-                "pointLongitude": 0.000
+                "pointLatitude": "34.138",
+                "pointLongitude": "-118.1258"
             }
         }
     ],
-    "language": "en",
+    "language": "eng",
     "publicationYear": "2017",
     "publisher": "Publisher",
     "relatedIdentifiers": [
@@ -92,8 +99,9 @@
             "relationType": "IsDocumentedBy"
         }
     ],
-    "resourceType": {
-        "resourceTypeGeneral": "Dataset"
+    "types": {
+        "resourceTypeGeneral": "Dataset",
+        "resourceType": "Dataset"
     },
     "rightsList": [
         {
@@ -118,5 +126,6 @@
             "titleType": "AlternativeTitle"
         }
     ],
-    "version": "0"
+    "version": "0",
+    "schemaVersion": "http://datacite.org/schema/kernel-4"
 }

example43.json renamed to example_custom.json

@@ -1,32 +1,27 @@
 {
-  "identifiers": [
-    {
-    "identifier": "10.14291/1",
-    "identifierType": "DOI"
-    }],
-    "alternateIdentifiers": [
-        {
-            "alternateIdentifier": "Alt_Identifier",
-            "alternateIdentifierType": "Alt_Identifier_Type"
-        }
+    "identifiers": [
+            {"identifier": "10.5281/inveniordm.1234", "identifierType": "DOI"},
+            {"identifier": "1924MNRAS..84..308E", "identifierType": "bibcode"}
     ],
     "contributors": [
         {
           "nameType": "Personal",
           "affiliation": [
               {
-                  "name": "DataCite",
+                  "name": "DataCitea",
                   "affiliationIdentifier": "https://ror.org/04wxnsj81",
                   "affiliationIdentifierScheme": "ROR"
                 }
               ],
             "name": "Contributor Name",
+            "familyName": "Family Name",
+            "givenName": "Given Name",
             "contributorType": "ContactPerson",
             "nameIdentifiers": [
                 {
-                    "nameIdentifier": "Contributor Identifier",
-                    "nameIdentifierScheme": "GRID",
-                    "schemeUri": "https://www.grid.ac/institutes/"
+                    "nameIdentifier": "0000-0002-1825-0097",
+                    "nameIdentifierScheme": "ORCID",
+                    "schemeUri": "https://orcid.org/"
                 }
             ]
         }
@@ -46,9 +41,9 @@
             "givenName": "Given Name",
             "nameIdentifiers": [
                 {
-                    "nameIdentifier": "Name Identifier",
+                    "nameIdentifier": "0000-0002-1825-0097",
                     "nameIdentifierScheme": "ORCID",
-                    "schemeUri": "http://orcid.org/"
+                    "schemeUri": "https://orcid.org/"
                 }
             ]
         }
@@ -58,10 +53,6 @@
             "date": "2014-10-01",
             "dateType": "Created"
         },
-        {
-            "date": "2017-03-01",
-            "dateType": "Updated"
-        },
         {
             "date": "2012-05-22/2016-12-21",
             "dateType": "Collected"
@@ -89,8 +80,8 @@
         {
             "geoLocationPlace": "Place Name",
             "geoLocationPoint": {
-                "pointLatitude": "0.000",
-                "pointLongitude": "0.000"
+                "pointLatitude": "34.138",
+                "pointLongitude": "-118.1258"
             }
         }
     ],

outdated/get_geo.py

@@ -1,6 +1,5 @@
 import os, json, csv, argparse
 import requests
-from caltechdata_api import decustomize_schema
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(

write.py

@@ -20,7 +20,7 @@
 metaf = open(args.json_file[0], "r")
 metadata = json.load(metaf)
 
-production = True
+production = False
 publish = False
 
 response = caltechdata_write(