The body message is not properly captured from the tag message.

Change:

# Use tag subject and message body for release description
body: |
  ${{ env.TAG_SUBJECT }}
  ${{ env.TAG_BODY }}

To:

# No body: defaults to tag message and body

Also, need to strip the "Get tag info" step as that is no longer needed

Describe the bug (REQUIRED)
The quota.ual_ldap_quota_query ldap3 query strings will return faculty, grad and ugrad members. However, there is overlaps between these groups (e.g., a grad member can also be staff). To fix this issue, it's best to mutually exclusive with the highest rank determining the quota level.

Version information (please complete the following, default in brackets): (REQUIRED)

• ReQUIAM version: 0.12.2`
• ReQUIAM branch info: feature/limited_set

Reproduction Steps (REQUIRED)

from requiam import quota                                                                                                                
from requiam import ldap_query                                                                                                           

# Get query string
grad = quota.ual_ldap_quota_query('grad')
faculty = quota.ual_ldap_quota_query('faculty') 
ugrad = quota.ual_ldap_quota_query('ugrad')

ldc = ldap_query.LDAPConnection(ldap_host='eds.iam.arizona.edu',
                                ldap_base_dn='dc=eds,dc=arizona,dc=edu',
                                ldap_user='**override**', ldap_password='**override**')

# Query for members
faculty_uaid = ldap_query.ldap_search(ldc, faculty)                                                                                      
grad_uaid = ldap_query.ldap_search(ldc, grad)
ugrad_uaid = ldap_query.ldap_search(ldc, ugrad)

# Info
In [10]: len(faculty_uaid)                                                                                                                       
Out[10]: 23558

In [11]: len(grad_uaid)                                                                                                                          
Out[11]: 15450

In [19]: len(ugrad_uaid)                                                                                                                         
Out[19]: 46777

In [13]: len(faculty_uaid & grad_uaid)                                                                                                                            
Out[13]: 860

In [17]: len(faculty_uaid & ugrad_uaid)                                                                                                              
Out[17]: 435

In [18]: len(grad_uaid & ugrad_uaid)                                                                                                             
Out[18]: 4217

Expected behavior (Optional)
Exclusive list for grad and ugrad

From DataRepository_patrons to ReQUIAM (ReDATA EDS Query and Update for Identity and Access Management)

We need some statistics on the number of individuals in an organization code with Library privileges. We should retrieve:

1. Total number of users part of ual-pgrp Grouper groups
   • Total numbers in organization
   • Total number for faculty
   • Total number for staff
   • Total number for grad students

• Need to update figshare.ini [15c19cf]
• Create a module called org_code_numbers [0908038, 497b57f, 3f9fe8f, 5428bf4, a21cea5]
• Changes to LDAP queries for greater flexibility [a3ed448, f94e8c9, 4ff531a, 9013e10]

The manual_override script enable the ability to add users in Grouper portal and quota groups. However, it does not have the capability to add users to the active Grouper group. This ensure IAM access for the ReDATA platform. Note that this is a temporary solution for large group of users before the service is entirely live. This provides an alternative solution from APPS-IAM or the Grouper interface.

Some feature to include:

• Script has the ability to receive a single netid and add and remove the user [26e12e4, 3436120]
• Script has the ability to receive a list via a CSV file and add multiple users [Migrated to #52]
• Script has the ability to check if the user is present and skips the addition process [26e12e4]

The following error was received.

11:26:36 -  WARNING: portal not found on Grouper!
Traceback (most recent call last):
  File "scripts/user_update", line 167, in <module>
    raise SystemError
SystemError

This is simply due to check_grouper_exists() which does not find a portal group name root.

Solution:

• if vargs['portal'] != 'root' exception needed here

The user_update script automatically add portal and quota settings to the ManualOverride dataframe regardless. We use this for individual handling; however, we should have an option to disable this with the default of adding it.

This is a quick fix and can be implemented in feature/limited_set

By default, running the user_update script with --portal root will strip current portal association from the CSV. However, we now came across the case when a user is by default assigned to a given sub-portal but belongs at the top university level (i.e., 'root'). In this case, we need a simple fix that prompts whether to add 'root' to the portal file.

This will ensure that the override is in place and that the user is not returned to the sub-portal based on organization classification

This is the main issue thread for Travis CI integration.

Steps include:

• Implementing the integration with ualibraries/ReQUIAM
• Creating a .travis.yml file
• Creating a pytest.ini file
• Building pytests - additional tests needed, but that will be a separate issue
• Creating a .coveragerc file to coverage report
• Update README.md to include build state

This is implemented in the feature/travis_ci branch

Deploy actions/create-release action to create a new release on any tagged version.

Since this touches on many codes/scripts, it's best to have this in release/v0.12.0 rather than a feature as that will include feature/grouper_api_admin and feature/multi_user

The following actions are needed:

• Separating stdout from file logging in terms of verbosity. See LD-Cool-P for examples [d9d49fa]
• Adding information about the requiam version, branches, and commit hash. The first two are already included. [9264182]
• Add information about system (e.g., IP, username, hostname, etc) [a3fcd9a]
• Enable log optional input for the following objects and functions:
  • Delta
  • GrouperAPI
  • grouper_query.grouper_delta_user
  • ldap_query.LDAPConnection
  • ManualOverride, read_manual_file, update_entries, get_current_groups
  • GrouperQuery
• The following scripts will need to be updated
  • add_grouper_groups
  • script_run
  • user_update

Currently, user_update directly add members to figshare:active.

Moving forward, figshare:active should only had indirect membership. The direct membership makes it complicated. With new groups being added, it's perhaps better to create a new group that has those members directly and to indirectly add that group in figshare:active. This is best as an interactive prompt option that adds for the names, etc.

• Re-purpose add_groups function to create_groups and have it outside of add_grouper_groups script [3def1a5, e3ca412]

• Create a function called create_active_group to enable creating such a group in figtest:group_active stem [d501e26]

• Add option in user_update to use create_active_group and add members [bd937dc]

• Note that this is related to #65 as the limited use should also consider adding active members by at least creating the group and adding them there.

For script_run, only the pandas DataFrame is available via stdout. It would be easy to save this similar to what we did in LD-Cool-P

Note that this requires updating pandas to >= 1.0.2 for .to_markdown(). I will use 1.1.0 since this will have index=False option. Also, tabulate is needed. I suggest using tabulate=0.8.3 since it worked for LD-Cool-P

This issue has split off from #43.

Primary work will be in user_update script. At the moment, this script can only handle a single patron.
Minimum: a comma-separated list of netid
Desired: Minimum + ability to use simple txt file with netid's (one per row).

• Update user_update:
  - [x] Split out netid by commas. Best to have as a set to avoid any duplicate entries
  - [x] Handling of incorrect ID (via user_uaid)
• Update `ldap_query.uid_query() to provide appropriate search criteria. This will be tough because of handling the users that have access and the users that do not have access (though set operations should avoid conflicts). Larger issue is handling incorrect netid's.
• Calls to get_current_groups needs to handle multi-user. Best to use dict within dict
• For full compatibility, the quota case needs a root option which implies no association with a quota
• Improvements needed in ManualOverride to handle multi-user add to tables, etc.

There are two parts of this feature request:

First, this most likely will be used with quota increases, but can also be used with figshare groups / "portals".

When a quota request is approved, there needs to be a query to retrieve the specific user and perform the synchronization based on the change in a file that provides manual changes.

The simple solution is:

1. LDAP query that retrieves a given user EDS uid and old figshare Grouper attributes based on the EDSuaid (NetID)
2. Remove the user from appropriate figshare Grouper group using delta
3. Add the user to the new figshare Grouper group using delta

The removal/addition can be done by creating a set from the LDAP query and stripping/adding the uid and passing to delta with the Grouper query. delta will identify the drops and additions.

There are a number of steps:

• Create a user_update script that accepts inputs, such as the NetID, and the specified changes to quota or/and portal [5361b6f, 14de970, cccc483, 1492ea1, 33603a7, 3594a5e, d3bce32]
• The script should have a number of handling such as incorrect netid [d3bce32], changing to the same portal [6699787], checking for whether the portal/quota exists first [65ea345, 225e580].
• Update the manual CSV file to include the changes. Note that we want to keep the commented heading [ce16056, 8142215]
• Add root option such that it simply removes current portal settings and defaults to the main/root level. This would require that we also remove it from the CSV files. Unsure if I should do something similar for quota [fd446c7 e4a2be0]
• Retrieve current user Grouper figshare ismemberof attributes [0e560ac, 7d63c2b, 078385c, 7dd4195]
• Remove from user from Grouper figshare ismemberof attributes [078385c, 14de970]
• Factoring out common codes between manual_override and grouper_query to avoid circular imports [17fae2c, 70ca119, f899bd0]
• Refactor delta codes to handle addition and deletion of a user via manual override [4228eae, 1492ea1]

Second, is the automation of handling changes to quota and/or portal based on an input database that is a CSV file. This ensures that when someone has a portal/quota change, it stays when the automated script is run. This requires:

• Updating config/figshare.ini [531ca48]
• Creating manual override templates [55dcdd1]
• Excluding private CSV files in gitignore [d116f0f]
• A manual_override module with a ManualOverride class [1102aa1, 7af93a4]
• Update to script_run [b031f3e]
  - This will check whether the manual CSV files are present. If not, no manual override handling will happen. Note that additional testing is needed at a later point. A separate ticket will be created if the manual override handling fails (#47).

Note that this has been well-developed and available here.

Code is available in feature/delta

Received this error:

  File "/Users/cly/codes/UALibraries/DataRepository_patrons/tests/ldap_connection.py", line 89
    return ldap_query
         ^
SyntaxError: invalid syntax

The use of .format() did not have a complete set of parentheses.

Describe the bug (REQUIRED)
No production option is set when calling GrouperAPI, thus default is False

Version information (please complete the following, default in brackets): (REQUIRED)

• ReQUIAM version: 0.12.1
• ReQUIAM branch info: master
• pandas version: 1.1.0
• requests version: 2.22.0
• numpy version: 1.18.0

Reproduction Steps (REQUIRED)
Steps to reproduce the behavior:

1. This is fairly clear in user_update script

Expected behavior (Optional)

• Defaults to production unless an option in the script grouper_figtest indicate otherwise [8f0b7b7]
• Note that manual_override.get_current_groups needs a production option [937aa85]
• Calls to manual_override.grouper_delta_user will need to specify production option. Default is True [8f0b7b7]

Screenshots (Optional)
If applicable, add screenshots to help explain your problem.

Additional context (Optional)
Add any other context about the problem here.

Did a dry run which updated portals and quotas. Portals ran without errors but quotas failed. These are the last few lines of the log

21:58:36 - INFO: Working on grad quota : 536870912 bytes
22:00:38 - INFO: Removing : ['figshare01-test', 'kalcock']
22:00:38 - INFO: EDS size 16082
Traceback (most recent call last):
File "scripts/script_run", line 238, in
grouper_group=grouper_quota)
File "/home/mozart/codes/ReQUIAM/requiam/grouper_query.py", line 42, in init
if 'wsSubjects' in rsp.json()['WsGetMembersLiteResult']:
File "/opt/anaconda3/envs/figshare_patrons/lib/python3.7/site-packages/requests-2.22.0-py3.7.egg/requests/models.py", line 897, in json
return complexjson.loads(self.text, **kwargs)
File "/opt/anaconda3/envs/figshare_patrons/lib/python3.7/json/init.py", line 348, in loads
return _default_decoder.decode(s)
File "/opt/anaconda3/envs/figshare_patrons/lib/python3.7/json/decoder.py", line 337, in decode
obj, end = self.raw_decode(s, idx=_w(s, 0).end())
File "/opt/anaconda3/envs/figshare_patrons/lib/python3.7/json/decoder.py", line 355, in raw_decode
raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

I also have a question about why 'figshare01-test', 'kalcock' and sometimes 'chunly' were going to be removed from every group. Presumably they should only be removed from groups they're actually in? It could be a symptom of me not fully understanding the log output. I can make a separate issue for this if needed

Currently script_run queries Grouper for membership in various Grouper groups. However, there's no failsafe check to make sure those group exists. We have within GrouperAPI the check_grouper_exists method that can be easily used.

This is straightforward and should be a hotfix

Rather than run the full script, an option to limit it to certain organization codes, sub-portal names.

This should first be working with:

• main themes and sub-portals (see --groups option) [3b1b713]
• organization code (see --org_codes option) [3b1b713]
• Able to provide a list of codes or main themes or sub-portals (comma separated) [3b1b713]
• Able to only run the quota for those members [d9598b8, a312eda]
  • One of the trick is only to add those members and not delete existing members. This is also the case for portals. One way to enable this is having an add_only option in Delta. It might be easier to simply update the ldap_members to include existing ones to avoid deletion [2649160]
• Create Grouper group with users for indirect adds to figshare:active. Related to #72 [d055ecb, ef30dd3, dcb0a94]
  • This is possible using the create_active_group function
• For readability, change grouper_query module to import [05d5fd0]
• Fix bug with quota running [e6355d2]

Add a simple test that uses a test NetID account and a Grouper group to synchronize and then undo. This will confirm that the code is functioning.

This was listed in #16, but will be a separate issue.

Check list:

1. Add --test flag and implementation with if args.test
2. Add option to reverse (--test_reverse)
3. Update README with examples

Describe the bug (REQUIRED)
When running script_run the list of remove and adds are noted. For the adds, this is generally consistent but the removal often includes accounts that don't apply due to the conditional requirement. This can be improved by stripping down to only the list of users that is in the group where it applies.

Version information (please complete the following, default in brackets): (REQUIRED)

• ReQUIAM version: 0.12.2
• ReQUIAM branch info: feature/limited_set

Reproduction Steps (REQUIRED)
Steps to reproduce the behavior:

1. Execute script_run

Expected behavior (Optional)
A less verbose message containing applicableA clear and concise description of what you expected to happen.

Screenshots (Optional)
If applicable, add screenshots to help explain your problem.

Note: This will be implemented as part of the feature/limited_set branch

By default, script_run executes on the production stems. An option should be added to disable this and perform updates on the stage/test.

We had renamed DataRepository_research_themes to ReQUIAM_csv. As such, the current URL will return a 404 error.

Currently we have two grouper modules, grouper_query and grouper_admin. It was done in a modular manner to avoid any conflicts while implementing admin tools. Now that both modules are stable and well-tested, it's time to merge the two so there's only one grouper module.

Action Items:

• Rename grouper_admin module to grouper
• Rename GrouperAPI class to Grouper
• Move grouper_query.GrouperQuery into a query method for grouper_admin.GrouperAPI
• Move figshare_group function to be part of commons module
• Move grouper_delta_user function into grouper module. Note that this is heavily used in user_update script

Implemented in feature/87_grouper_refactor (PR #117)

In the user_update script, multiple calls to grouper_delta_user is made and a set of sync/nosync commands are included.

Now that grouper_delta_user can receive a LogClass object, let's refactor into grouper_delta_user. This will make the code more easy to read.

Quota tiers are:

1. faculty, staff, and dcc
2. graduate students
3. undergraduate students

These queries will retrieve the full list and later be modified to exclude individuals with different tiers.

Describe the bug (REQUIRED)

The problem seems to stem from set vs list. The former is not in the same order as the latter. This creates some discrepancies.

I have performed tests of LDAP queries, Grouper queries, and delta synchronization in the tests folder.

Next step is to package it together into a working module(s).

This is in the feature/ldap_grouper_delta branch

1. Modules will be within a DataRepository_patrons folder
2. setup.py is needed [7abba06, dd202e4]
3. requirements.txt is needed [152e664, dd202e4]
4. A primary script (script_run) to run the query [777ba4f, 306b1c8, 3303169, 4a6a103 8882d20, 1385487, aa296b8, bdb6cb3, ec10d9e, 01a5d18, 8f5dfe2, 3b636da, c6317b0, 6223407, 452aeed, d9f60df, ba1e8cb, fa48fc0, 6413065, 89af40a, 3469d3a, a64c668, fe1f397, bfe373c, e2099ad, 8d9788a, 3e1a5b0, 47e45e3, 7aa387f]
5. A config file is needed containing the CSV info [c3413f5, 2a853a7, c6b2b72, 8ef9944]
6. Add a logging module (logger.py) and use it in script_run [66821d8, 0093af9, 6413065, 433364b]
7. Update README to document the package [851780, 82643c9, 3fe29ae, 2640fe6, f78e085, 812f0ed, 262d8f7, c50fe34, 06e8bff, 210b0fd, e701bed, 69035f7, b57ba63, 8ef9944, 14c85f5]
8. A timing class to indicate duration of execution [b3a6cd5, 5f0353b, 47e45e3, 7aa387f]
9. Add a simple test that uses a test NetID account and a Grouper group to synchronize and undo. This will confirm that the code is functioning. [This is a separate issue, #18]

• All scripts currently live in the requiam folder. Let's have a scripts folder in the parent path
• Include a banner message in scripts
• Include ReQUIAM version info and current active git branch (requires a get_branch_name module)

Retrieves Grouper membership information

General issue for best practices. This ticket will remain open indefinitely.

We need to add a column that allows for additional comments. This could include when the addition was done or changes that was made. The ManualOverride.update_dataframe function would mark changes done through user_update

Describe the bug (REQUIRED)
The follow error message was returned:

22:17:50 - INFO: Removing ___ from current quota: 2147483648
Traceback (most recent call last):
File "scripts/user_update", line 397, in
production=grouper_production)
File "/home/mozart/codes/ReQUIAM/requiam/grouper_query.py", line 135, in grouper_delta_user
member_set = update_entries(member_set, netid, uaid, action, log=log)
File "/home/mozart/codes/ReQUIAM/requiam/manual_override.py", line 210, in update_entries
remove_netid = [netid[i] for i in range(len(netid)) if
File "/home/mozart/codes/ReQUIAM/requiam/manual_override.py", line 211, in
uaid[i] in remove_uaid]
TypeError: 'set' object is not subscriptable

Currently Grouper use is limited to addition and deletion of users in a Grouper group. We should extend the API use to include:

• Creating groups within a specified Grouper stem. This will require providing the metadata for the description. I created add_grouper_groups script [793fd16]
• Setting permissions/ownership for groups
• Add testing Grouper stem definitions to enable easy switch between Grouper stage and production [89241df, 92be48e]
• Make stage/production as part of the grouper_dict
• Check if a group exists (This is being implemented in feature/manual_override as a MVP)

The Grouper API can be found here.

A Good demonstration of the Grouper API is available in this github repository

passwd -> password

The config file should include:

1. The URL to retrieve the CSV-formatted spreadsheet

In addition, __init__ function is needed to run the script via a single command line

This is in feature/service_architecture.

In a number of scripts (e.g., add_grouper_groups), we need the list of quotas to execute all possible cases.

For this to work, we need:

• Spreadsheet that has all combinations
• Updates to config file for the settings
• Capability to read in as pandas DataFrame
• Use in add_grouper_groups

The above list can be implemented within feature/grouper_api_admin

read_manual_file is a separate function and can easily be refactored into ManualOverride as it is not used elsewhere. This would simplify input calls with a simple portal or quota to refer to which CSV files to read in.

The following scripts should be made executable (chmod +x):

1. add_grouper_groups
2. get_org_code_numbers
3. grouper_permission_checks (not part of v0.13.4)
4. script_run
5. user_update

This can be implemented as hotfix/v0.13.5

Also, updates to README.md for simpler execution.

Although outside this issue, for deployment aliases on patron server needs to be modified. This is not crucial but helps to keep things clean.

When performing portal and quota updates, please add the option for additional verbosity which not only includes summary information about what will change, but include the actual changes as well. This increases our capacity to troubleshoot issues should they arise.

For example, current log info when performing a portal update on the tech_eng portal (similarly for quota updates)

11:33:07 -     INFO: Working on Technology & Engineering (tech_eng) portal
<snip>
11:33:11 -     INFO:  EDS size 1250
11:33:11 -     INFO:  Grouper size 1
11:33:11 -     INFO: ldap and grouper have 0 members in common
11:33:11 -     INFO: synchronization will drop 1 entries from grouper group
11:33:11 -     INFO: synchronization will add 1250 entries to grouper group
11:33:11 -     INFO: dry run, not performing synchronization

Desired logging info

11:33:07 -     INFO: Working on Technology & Engineering (tech_eng) portal
<snip>
11:33:11 -     INFO:  EDS size 1250
11:33:11 -     INFO:  Grouper size 1
11:33:11 -     INFO: ldap and grouper have 0 members in common
11:33:11 -     INFO: synchronization will drop 1 entries from grouper group
11:33:11 -     DEBUG: Dropping (name : employeeprimarydept )
                     Jane R. Smith : 0121
11:33:11 -     INFO: synchronization will add 1250 entries to grouper group
11:33:11 -     DEBUG: Adding (name : employeeprimarydept )
                     Mary A. Researcher : 2308
                     Mary B. Researcher : 2308
                     Mary C. Researcher : 2308
<snip>
11:33:11 -     INFO: dry run, not performing synchronization

Logs are currently stored locally. As such, this poses a problem for ephemeral design.

Persistent storage was provisioned. To store logs, we need:

• Add to configuration files the parent directory path for logs
• Update scripts to handle path
• Also change the permission on the log to avoid any possible write issue (had an unusual encounter)

This can be implemented as hotfix/v0.11.3

Objective:

A Grouper portal group can have association from multiple organizations. Thus a direct comparison between LDAP and Grouper to identify additions and deletions requires that multiple LDAP queries be conducted and merging those membership.

The following error message was encountered when both --portal and --quota were not specified:

Traceback (most recent call last):
  File "scripts/user_update", line 164, in <module>
    if not portal_check or not quota_check:
NameError: name 'portal_check' is not defined

Simple solution is to pre-defined portal_check and quota_check. This avoids the SystemError when either of them do not exists on Grouper

A separate warning should be made to indicate the neither of them are set.

The script could simply be executed after the main script script_run to show summary data.

This would help to identify any sync issue.

This would be beneficial to also have it when user_update script is executed. There may be a few minute delayed needed.

Undergrads will by default have a quota of zero. Only those with privileges are manually added using the user_update script. Thus an option should be added to disable running it the ugrad quota.

Options are:

1. Change the figshare.ini
2. Add an option to warn of not running in the logs and execute other groups as is. For the latter, this means that the number of members satisfying the EDS query is known.

I will implement this in feature/limited_set

Not necessarily a bug, but the path for the grouper_base_path is specific for the tool to add members: grouper-ws/servicesRest/json/v2_2_001/groups

As such, it will not work with some components of the Grouper API, specifically permission. There's a simple hack to strip the trailing group, but a more ideal situation is to keep it generic and set the proper endpoint as needed.

This should be implemented in release/v0.12.0

Notes:

• It is implemented in feature/grouper_api_admin for GrouperAPI and GrouperQuery
• Since feature/multi_user uses GrouperQuery, it should be fine when we everything is merged into release/v0.12.0

This was seen in release/v0.11.0.

             netid                 uaid       portal
1  figshare01-test  T440413500927388672  arts_design
Traceback (most recent call last):
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/ops/__init__.py", line 968, in na_op
    result = expressions.evaluate(op, str_rep, x, y, **eval_kwargs)
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/computation/expressions.py", line 221, in evaluate
    return _evaluate(op, op_str, a, b, **eval_kwargs)
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/computation/expressions.py", line 70, in _evaluate_standard
    return op(a, b)
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/ops/roperator.py", line 13, in rsub
    return right - left
TypeError: unsupported operand type(s) for -: 'set' and 'str'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "requiam/script_run", line 164, in <module>
    ldap_members = mo.identify_changes(ldap_members, portal, 'portal')
  File "/Users/cly/codes/UALibraries/ReQUIAM/requiam/manual_override.py", line 81, in identify_changes
    outside_df['uaid'], 'remove', self.log)
  File "/Users/cly/codes/UALibraries/ReQUIAM/requiam/manual_override.py", line 199, in update_entries
    new_ldap_set = ldap_set - uaid
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/ops/__init__.py", line 1048, in wrapper
    result = na_op(lvalues, rvalues)
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/ops/__init__.py", line 970, in na_op
    result = masked_arith_op(x, y, op)
  File "/Applications/anaconda3/envs/ual-patron-groups/lib/python3.7/site-packages/pandas/core/ops/__init__.py", line 450, in masked_arith_op
    assert is_scalar(y), type(y)
AssertionError: <class 'set'>