ODS harvesting via simple URL not working

[tkohr]

Describe the bug
Harvesting the following ODS catalog via the simple url harvester (which works on version 4.2.2) does not seem to work anymore. I have the feeling, this is related to the change that the recordIdPath input now expects a path /datasets/datasetid (from the document root?). Or is it just me indicating the wrong path? In version 4.2.2 only the property key datasetid is indicated here.

To Reproduce
Steps to reproduce the behavior:

1. Go in the admin UI to the harvester settings
2. Add a new harvester of type simple URL with the following params

• URL : https://opendata.lillemetropole.fr/api/datasets/1.0/search?refine.publisher=M%C3%A9tropole+Europ%C3%A9enne+de+Lille&start=0&rows=20
• loopElement: /datasets
• numberOfRecordPath: /nhits
• recordIdPath: /datasets/datasetid
• pageFromParam: start
• pageSizeParam: rows
• toISOConversion : schema:iso19155-3.2018:convert/fromJsonOpenDataSoft

4. Save and harvest

Expected behavior
Harvest ~208 records from the catalog.

Log file
harvester_simpleUrl_MEL_ODS_GN_main_202303301528.log

Desktop (please complete the following information):

• Browser: Chromium, Version 111.0.5563.110 (Build officiel) snap (64 bits)
• GeoNetwork Version: main
• Server Application: Jetty with Java 8

[jahow]

Related to break change introduced in #6677

[fxprunayre]

Definitely not an ODS expert but it looks like depending on if you request API version 1 and version 2 the datased id path is different.

The PR you're pointing at ODS API v2 support was added a3db440 and should have preserved compability with version 1 API.

So before 4.2.3, harvesting an ODS API v2 was not working.

Running on main the following harvester config

{"@id":"228","@type":"simpleurl","owner":["1"],"ownerGroup":["2"],"ownerUser":["undefined"],"site":{"name":"6962","uuid":"d3e54543-097d-4bb8-bfe6-0fa9c04bb73d","account":{"use":false,"username":[],"password":[]},"url":"https://opendata.lillemetropole.fr/api/datasets/1.0/search?refine.publisher=M%C3%A9tropole+Europ%C3%A9enne+de+Lille&start=0&rows=20","icon":"blank.png","loopElement":"/datasets","numberOfRecordPath":"/nhits","recordIdPath":"/datasetid","pageSizeParam":"rows","pageFromParam":"start","toISOConversion":"schema:iso19115-3.2018:convert/fromJsonOpenDataSoft"},"content":{"validate":"NOVALIDATION","importxslt":"none","batchEdits":"[]"},"options":{"every":"0 0 0 ? * *","oneRunOnly":false,"overrideUuid":"SKIP","status":"active"},"privileges":[{"@id":"1","operation":[{"@name":"view"},{"@name":"dynamic"},{"@name":"download"}]}],"ifRecordExistAppendPrivileges":false,"info":{"lastRun":"2023-05-05T05:25:19.923Z","running":false,"result":{"added":"224","atomicDatasetRecords":"0","badFormat":"0","collectionDatasetRecords":"0","datasetUuidExist":"0","privilegesAppendedOnExistingRecord":"0","doesNotValidate":"0","xpathFilterExcluded":"0","duplicatedResource":"0","fragmentsMatched":"0","fragmentsReturned":"0","fragmentsUnknownSchema":"0","incompatible":"0","recordsBuilt":"0","recordsUpdated":"0","removed":"0","serviceRecords":"0","subtemplatesAdded":"0","subtemplatesRemoved":"0","subtemplatesUpdated":"0","total":"224","unchanged":"0","unknownSchema":"0","unretrievable":"0","updated":"0","thumbnails":"0","thumbnailsFailed":"0"}}}

for v1 API collects 224 records.

and playing

{"@id":"373","@type":"simpleurl","owner":["1"],"ownerGroup":["2"],"ownerUser":["undefined"],"site":{"name":"6962 v2","uuid":"cc6c2ae1-34a8-4ac6-bd19-8df33098f61b","account":{"use":false,"username":[],"password":[]},"url":"https://opendata.lillemetropole.fr/api/explore/v2.0/catalog/datasets?rows=100","icon":"blank.png","loopElement":"/datasets","numberOfRecordPath":"/nhits","recordIdPath":"/dataset/dataset_id","pageSizeParam":"rows","pageFromParam":"start","toISOConversion":"schema:iso19115-3.2018:convert/fromJsonOpenDataSoft"},"content":{"validate":"NOVALIDATION","importxslt":"none","batchEdits":"[]"},"options":{"every":"0 0 0 ? * *","oneRunOnly":false,"overrideUuid":"SKIP","status":"active"},"privileges":[{"@id":"1","operation":[{"@name":"view"},{"@name":"dynamic"},{"@name":"download"}]}],"ifRecordExistAppendPrivileges":false,"info":{"lastRun":"2023-05-05T05:46:25.882Z","running":false,"result":{"added":"10","atomicDatasetRecords":"0","badFormat":"0","collectionDatasetRecords":"0","datasetUuidExist":"0","privilegesAppendedOnExistingRecord":"0","doesNotValidate":"0","xpathFilterExcluded":"0","duplicatedResource":"0","fragmentsMatched":"0","fragmentsReturned":"0","fragmentsUnknownSchema":"0","incompatible":"0","recordsBuilt":"0","recordsUpdated":"0","removed":"1","serviceRecords":"0","subtemplatesAdded":"0","subtemplatesRemoved":"0","subtemplatesUpdated":"0","total":"10","unchanged":"0","unknownSchema":"0","unretrievable":"0","updated":"0","thumbnails":"0","thumbnailsFailed":"0"}}}

collect 100 records

So this seems fine to me, no?

[fxprunayre]

So your issue was related to

String uuid = this.extractUuidFromIdentifier(record.get(params.recordIdPath).asText());

which only works if the property you need is a property of the loopElement node which is not the case in all JSON harvester and not in ODS API v2. So it was indeed changed to

String uuid = this.extractUuidFromIdentifier(record.at(params.recordIdPath).asText());

This explains why your config in 4.2.2 did not work in 4.2.3. By the way, a quite clear error is reported in the harvester log

2023-05-05T13:42:40,976 ERROR [geonetwork.harvester] -
 Failed to collect record UUID at path datasetid. 
 Error is: Invalid input:
 JSON Pointer expression must start with '/': "datasetid"

[tkohr]

Thanks for looking into this @fxprunayre. Indeed, in the end, it's just the missing / that breaks the ODS config from GN 4.2.2 to > 4.2.2.

I didn't pay attention that the mentioned PR was using ODS v2 having a different hierarchy and keys datasetid, dataset_id from V1, which obscured the problem a little, despite the rather clear error message.

[jahow]

Just to clarify, this had nothing to do with ODS API v2 (which we don't use). It was an error on our side, indeed it works with the new format for the record id pointer.

Thanks @fxprunayre

[tkohr]

FYI, I opened geonetwork/doc#240 regarding this.