dryad2dataverse 0.7.1__tar.gz → 0.7.5__tar.gz

{dryad2dataverse-0.7.1/src/dryad2dataverse.egg-info → dryad2dataverse-0.7.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dryad2dataverse
-Version: 0.7.1
+Version: 0.7.5
 Summary: Utility for copying and syncing data from a Dryad data repository to a Dataverse repository
 Author-email: Paul Lesack <paul.lesack@ubc.ca>
 Project-URL: Homepage, https://ubc-library-rc.github.io/dryad2dataverse
@@ -20,7 +20,7 @@ Description-Content-Type: text/markdown
 Requires-Dist: certifi>=2022.12.7
 Requires-Dist: charset-normalizer>=2.0.4
 Requires-Dist: chardet>=3.0.4
-Requires-Dist: idna>=2.10
+Requires-Dist: idna>=2.10.0
 Requires-Dist: pycryptodome>=3.20.0
 Requires-Dist: requests>=2.26.0
 Requires-Dist: requests-toolbelt>=0.9.1

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/requirements.txt

@@ -1,7 +1,7 @@
 certifi>=2022.12.7
 charset-normalizer>=2.0.4
 chardet>=3.0.4
-idna>=2.10
+idna>=2.10.0
 pycryptodome>=3.20.0
 requests>=2.26.0
 requests-toolbelt>=0.9.1

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/src/dryad2dataverse/__init__.py

@@ -22,7 +22,9 @@ Modules included:
 
     dryad2dataverse.exceptions : Custom exceptions.
 '''
+import sys
 
-VERSION = (0, 7, 1)
-
+VERSION = (0, 7, 5)
 __version__ = '.'.join([str(x) for x in VERSION])
+USERAGENT = (f'dryad2dataverse/v{__version__} ({sys.platform.capitalize()}); '
+             f'Python {sys.version[:sys.version.find("(")-1]}')

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/src/dryad2dataverse/monitor.py

@@ -325,7 +325,10 @@ class Monitor():
                 # because of code duplication below.
                 for f in oldFiles:
                     #Download links are not persistent. Be warned
-                    downLink = f['_links']['stash:file-download']['href']
+                    try:
+                        downLink = f['_links']['stash:file-download']['href']
+                    except KeyError:
+                        downLink = f['_links']['stash:download']['href']
                     downLink = f'{constants.DRYURL}{downLink}'
                     name = f['path']
                     mimeType = f['mimeType']

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/src/dryad2dataverse/scripts/dryadd.py

@@ -14,6 +14,7 @@ import glob
 import logging
 import logging.handlers
 import os
+import pathlib
 import shutil
 import smtplib
 import sys
@@ -27,7 +28,7 @@ import dryad2dataverse.serializer
 import dryad2dataverse.transfer
 from dryad2dataverse.handlers import SSLSMTPHandler
 
-VERSION = (0, 5, 3)
+VERSION = (0, 6, 2)
 __version__ = '.'.join([str(x) for x in VERSION])
 
 DRY = 'https://datadryad.org/api/v2'
@@ -372,6 +373,21 @@ def argp():
                         type=int,
                         dest='warn',
                         default=15)
+    parser.add_argument('--testmode-on',
+                        help=('Turn on test mode. '
+                              'Number of transfers will be limited '
+                              'to the value in --testmode-limit '
+                              'or 5 if you don\'t set --testmode-limit '),
+                        action='store_true',
+                        dest='testmode')
+    parser.add_argument('--testmode-limit',
+                        help=('Test mode - only transfer first [n] '
+                              'of the total number of (new) records. Old ones will '
+                              'still be updated, though. '
+                              'Default: 5'),
+                        type=int,
+                        default=5,
+                        dest='testlimit')
     parser.add_argument('--version', action='version',
                         version='%(prog)s '+__version__
                         +'; dryad2dataverse '+
@@ -416,7 +432,8 @@ def email_log(mailhost, fromaddr, toaddrs, credentials, port=465, secure=(),
     '''
     #pylint: disable=too-many-arguments
     #Because consistency is for suckers and yahoo requires full hostname
-    subject = 'Dryad to Dataverse transfer error'
+    #subject = 'Dryad to Dataverse transfer error'
+    subject = 'Dryad to Dataverse logger message'
     elog = logging.getLogger('email_log')
     mailer = SSLSMTPHandler(mailhost=(mailhost, port),
                             fromaddr=fromaddr,
@@ -491,11 +508,11 @@ def checkwarn(val:int, **kwargs) -> None:
         mess = ('Large number of updates detected. '
                 f'{val} new studies exceeds threshold of {kwargs.get("warn", 0)}. '
                 'Program execution halted.')
-        subject = ('Dryad to Dataverse large update warning')
+        subject = 'Dryad to Dataverse large update warning'
         for logme in kwargs.get('loggers'):
             logme.warning(mess)
         notify(msgtxt=(subject, mess),
-               **vars(kwargs))
+               **kwargs)
         sys.exit()
 
 def verbo(verbosity:bool, **kwargs)->None:
@@ -536,9 +553,18 @@ def main(log='/var/log/dryadd.log', level=logging.WARNING):
     monitor = dryad2dataverse.monitor.Monitor(args.dbase)
     #copy the database to make a backup, because paranoia is your friend
     if os.path.exists(dryad2dataverse.constants.DBASE):
-        shutil.copyfile(dryad2dataverse.constants.DBASE,
-                        dryad2dataverse.constants.DBASE+'.'+
-                        datetime.datetime.now().strftime('%Y-%m-%d-%H%M'))
+        bu_db = pathlib.Path(dryad2dataverse.constants.DBASE)
+        try:
+            shutil.copyfile( bu_db,
+                            pathlib.Path(bu_db.parent,
+                                         bu_db.stem + '_' +
+                                         datetime.datetime.now().strftime('%Y-%m-%d-%H%M') +
+                                         bu_db.suffix)
+                            )
+        except FileNotFoundError:
+            print(dryad2dataverse.constants.DBASE)
+            print(bu_db)
+            sys.exit()
     #list comprehension includes untimestamped dbase name, hence 2+
     fnames = glob.glob(os.path.abspath(dryad2dataverse.constants.DBASE)
                        +'*')
@@ -554,15 +580,24 @@ def main(log='/var/log/dryadd.log', level=logging.WARNING):
     logger.info('Total new files: %s', len(updates))
     elog.info('Total new files: %s', len(updates))
 
-    checkwarn(val=len(updates),
+    checkwarn(val=len(updates) if not args.testmode else
+              min(args.testlimit, len(updates)),
               loggers=[logger],
               **vars(args))
+    if args.testmode:
+        logger.warning('Test mode is ON - number of updates limited to %s', args.testlimit)
+        elog.warning('Test mode is ON - number of updates limited to %s', args.testlimit)
 
     #update all the new files
     verbo(args.verbosity, **{'Total to process': len(updates)})
+
     try:
         count = 0
+        testcount = 0
         for doi in updates:
+            if args.testmode and (testcount >= args.testlimit):
+                logger.info('Test limit of %s reached', args.testlimit)
+                break
             count += 1
             logger.info('Start processing %s of %s', count, len(updates))
             logger.info('DOI: %s, Dryad URL: https://datadryad.org/stash/dataset/%s',
@@ -608,6 +643,7 @@ def main(log='/var/log/dryadd.log', level=logging.WARNING):
                 transfer.set_correct_date()
                 notify(new_content(study),
                        **vars(args))
+                testcount+=1
 
             elif update_type == 'updated':
                 logger.info('Updated metadata: %s', doi[0])
@@ -673,3 +709,7 @@ def main(log='/var/log/dryadd.log', level=logging.WARNING):
 
 if __name__ == '__main__':
     main()
+    _parser = argp()
+    _args = _parser.parse_args()
+    print('This is what you would have done had you actually run this')
+    print(_args)

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/src/dryad2dataverse/serializer.py

@@ -11,11 +11,13 @@ import requests
 from requests.adapters import HTTPAdapter
 
 from  dryad2dataverse import constants
+from dryad2dataverse import USERAGENT
 
 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}
 
 class Serializer():
     '''
@@ -72,6 +74,7 @@ class Serializer():
         try:
             headers = {'accept':'application/json',
                        'Content-Type':'application/json'}
+            headers.update(USER_AGENT)
             doiClean = urllib.parse.quote(self.doi, safe='')
             resp = self.session.get(f'{url}/api/v2/datasets/{doiClean}',
                                     headers=headers, timeout=timeout)
@@ -164,6 +167,7 @@ class Serializer():
                 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=headers,
                                             timeout=timeout)

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/src/dryad2dataverse/transfer.py

@@ -2,6 +2,7 @@
 This module handles data downloads and uploads from a Dryad instance to a Dataverse instance
 '''
 
+#TODO harmonize headers instead of hideous copypasta
 import hashlib
 import io
 import json
@@ -18,7 +19,9 @@ from requests_toolbelt.multipart.encoder import MultipartEncoder
 
 from dryad2dataverse import constants
 from dryad2dataverse import exceptions
+from dryad2dataverse import USERAGENT
 
+USER_AGENT = {'User-agent': USERAGENT}
 LOGGER = logging.getLogger(__name__)
 URL_LOGGER = logging.getLogger('urllib3')
 
@@ -85,6 +88,7 @@ class Transfer():
         if not url:
             url = constants.DVURL
         headers = {'X-Dataverse-key': apikey if apikey else constants.APIKEY}
+        headers.update(USER_AGENT)
         bad_test = self.session.get(f'{url}/api/datasets/:persistentId',
                                 headers=headers,
                                 params=params)
@@ -232,6 +236,7 @@ class Transfer():
             else:
                 headers = {'X-Dataverse-key' : constants.APIKEY}
 
+            headers.update(USER_AGENT)
             params = {'persistentId': hdl}
             set_date = self.session.put(f'{url}/api/datasets/:persistentId/citationdate',
                                         headers=headers,
@@ -284,7 +289,7 @@ class Transfer():
         if not apikey:
             apikey = constants.APIKEY
         headers = {'X-Dataverse-key' : apikey}
-
+        headers.update(USER_AGENT)
         targetDv = kwargs.get('targetDv')
         dvpid = kwargs.get('dvpid')
         #dryFid = kwargs.get('dryFid') #Why did I put this here?
@@ -355,7 +360,7 @@ class Transfer():
     @staticmethod
     def _check_md5(infile, dig_type):
         '''
-        Returns the md5 checksum of a file.
+        Returns the hex digest of a file (formerly just md5sum).
 
         ----------------------------------------
         Parameters:
@@ -442,6 +447,7 @@ class Transfer():
 
         if size:
             if size > constants.MAX_UPLOAD:
+                #TOO BIG
                 LOGGER.warning('%s: File %s exceeds '
                                'Dataverse MAX_UPLOAD size. Skipping download.',
                                self.doi, filename)
@@ -485,6 +491,7 @@ class Transfer():
                 if url == i[0]:
                     i[-1] = md5
             LOGGER.debug('Complete download sequence')
+            #This doesn't actually return an md5, just the hash value
             return md5
         except (requests.exceptions.HTTPError,
                 requests.exceptions.ConnectionError) as err:
@@ -560,6 +567,8 @@ class Transfer():
             headers = {'X-Dataverse-key': apikey}
         else:
             headers = self.auth
+
+        headers.update(USER_AGENT)
         params = {'persistentId': study}
         try:
             lock_status = self.session.get(f'{dv_url}/api/datasets/:persistentId/locks',
@@ -608,6 +617,8 @@ class Transfer():
             headers = {'X-Dataverse-key': apikey}
         else:
             headers = self.auth
+
+        headers.update(USER_AGENT)
         params = {'persistentId': study}
         lock_status = self.session.get(f'{dv_url}/api/datasets/:persistentId/locks',
                                        headers=headers,
@@ -641,7 +652,9 @@ class Transfer():
 
     def upload_file(self, dryadUrl=None, filename=None,
                     mimetype=None, size=None, descr=None,
-                    md5=None, studyId=None, dest=None,
+                    hashtype=None,
+                    #md5=None, studyId=None, dest=None,
+                    digest=None, studyId=None, dest=None,
                     fprefix=None, force_unlock=False, timeout=300):
         '''
         Uploads file to Dataverse study. Returns a tuple of the
@@ -668,8 +681,11 @@ class Transfer():
         dest : str
             — Destination dataverse installation url.
               Defaults to constants.DVURL.
+        hashtype: str
+            original Dryad hash type
 
-        md5 : str
+        #md5 : str
+        digest
             — md5 checksum for file.
 
         fprefix : str
@@ -692,6 +708,8 @@ class Transfer():
 
         ----------------------------------------
         '''
+        #return locals()
+        #TODONE remove above
         if not studyId:
             studyId = self.dvpid
         if not dest:
@@ -730,6 +748,7 @@ class Transfer():
         ctype = {'Content-type' : multi.content_type}
         tmphead = self.auth.copy()
         tmphead.update(ctype)
+        tmphead.update(USER_AGENT)
         url = dest + '/api/datasets/:persistentId/add'
         try:
             upload = self.session.post(url, params=params,
@@ -739,9 +758,22 @@ class Transfer():
             upload.raise_for_status()
             self.fileUpRecord.append((fid, upload.json()))
             upmd5 = upload.json()['data']['files'][0]['dataFile']['checksum']['value']
-            if md5 and upmd5 != md5:
+            #Dataverse hash type
+            _type = upload.json()['data']['files'][0]['dataFile']['checksum']['type']
+            if _type.lower() != hashtype.lower():
+                comparator = self._check_md5(upfile, _type.lower())
+            else:
+                comparator = digest
+            #if hashtype.lower () != 'md5':
+            #    #get an md5 because dataverse uses md5s. Or most of them do anyway.
+            #    #One day this will be rewritten properly.
+            #    md5 = self._check_md5(filename, 'md5')
+            #else:
+            #    md5 = digest
+            #if md5 and (upmd5 != md5):
+            if upmd5 != comparator:
                 try:
-                    raise exceptions.HashError(f'md5sum mismatch:\nlocal: {md5}\nuploaded: {upmd5}')
+                    raise exceptions.HashError(f'{_type} mismatch:\nlocal: {comparator}\nuploaded: {upmd5}')
                 except exceptions.HashError as e:
                     LOGGER.exception(e)
                     raise

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5/src/dryad2dataverse.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dryad2dataverse
-Version: 0.7.1
+Version: 0.7.5
 Summary: Utility for copying and syncing data from a Dryad data repository to a Dataverse repository
 Author-email: Paul Lesack <paul.lesack@ubc.ca>
 Project-URL: Homepage, https://ubc-library-rc.github.io/dryad2dataverse
@@ -20,7 +20,7 @@ Description-Content-Type: text/markdown
 Requires-Dist: certifi>=2022.12.7
 Requires-Dist: charset-normalizer>=2.0.4
 Requires-Dist: chardet>=3.0.4
-Requires-Dist: idna>=2.10
+Requires-Dist: idna>=2.10.0
 Requires-Dist: pycryptodome>=3.20.0
 Requires-Dist: requests>=2.26.0
 Requires-Dist: requests-toolbelt>=0.9.1

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/src/dryad2dataverse.egg-info/requires.txt

@@ -1,7 +1,7 @@
 certifi>=2022.12.7
 charset-normalizer>=2.0.4
 chardet>=3.0.4
-idna>=2.10
+idna>=2.10.0
 pycryptodome>=3.20.0
 requests>=2.26.0
 requests-toolbelt>=0.9.1

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/tests/tests_monitor.py

@@ -70,7 +70,9 @@ class TestMonitor(unittest.TestCase):
     def test_07_added_files(self):
         self.assertEqual.__self__.maxDiff = None
         newdata=copy.copy(self.testCase.fileJson[0]['_embedded']['stash:files'][-1])
-        newdata['_links']['stash:file-download']['href'] = '/api/v2/files/999999/download' 
+        #newdata['_links']['stash:file-download']['href'] = '/api/v2/files/999999/download' 
+        #let's change API output for no reason!
+        newdata['_links']['stash:download']['href'] = '/api/v2/files/999999/download' 
         newdata['path'] = 'ubc_rand1.csv'
         newdata['description'] = 'UBC random data 1'
         newdata['digestType'] = 'md5'
@@ -100,7 +102,8 @@ class TestMonitor(unittest.TestCase):
                'application/x-zip-compressed',
                23787587,
                '',
-               '', '')]}
+               #'', '')]}
+               'sha-256', 'cb1c4f28de8aaec8c8c4b1eca498eace1c2c8c847dd443bd5314d42a70d6e4ae')]}
         self.assertEqual.__self__.maxDiff = None 
         self.assertEqual(expect, diff)
         #and restore

{dryad2dataverse-0.7.1 → dryad2dataverse-0.7.5}/tests/tests_serializer.py

@@ -155,7 +155,9 @@ class TestSerialize(unittest.TestCase):
         notes = {'typeName': 'notesText', 
                  'typeClass':'primitive',
                  'multiple':False,
-                 'value': '<p><b>Dryad version number:</b> 4</p>\n<p><b>Version status:</b> submitted</p>\n<p><b>Dryad curation status:</b> Published</p>\n<p><b>Sharing link:</b> https://datadryad.org/stash/share/anFoRwjUzvjvpH8RA2T0mNipNsVst0s0N5mFzcTTcJE</p>\n<p><b>Storage size:</b> 23874866</p>\n<p><b>Visibility:</b> public</p>\n'}
+                 #'value': '<p><b>Dryad version number:</b> 4</p>\n<p><b>Version status:</b> submitted</p>\n<p><b>Dryad curation status:</b> Published</p>\n<p><b>Sharing link:</b> https://datadryad.org/stash/share/anFoRwjUzvjvpH8RA2T0mNipNsVst0s0N5mFzcTTcJE</p>\n<p><b>Storage size:</b> 23874866</p>\n<p><b>Visibility:</b> public</p>\n'}
+                 #Why keep something the same over time? That's for suckers!
+                 'value': '<p><b>Dryad version number:</b> 4</p>\n<p><b>Version status:</b> submitted</p>\n<p><b>Dryad curation status:</b> Published</p>\n<p><b>Sharing link:</b> http://datadryad.org/stash/dataset/doi:10.5061/dryad.2rbnzs7jp</p>\n<p><b>Storage size:</b> 23874866</p>\n<p><b>Visibility:</b> public</p>\n'}
         self.assertEqual(self.testCase._convert_notes(self.testCase.dryadJson), notes )
 
     def test_pub_date(self):