diff options
| -rw-r--r-- | python/fatcat_tools/importers/datacite.py | 12 | ||||
| -rw-r--r-- | python/fatcat_tools/workers/elasticsearch.py | 15 | ||||
| -rw-r--r-- | python/tests/files/datacite/datacite_doc_31.json | 53 | ||||
| -rw-r--r-- | python/tests/files/datacite/datacite_doc_32.json | 53 | ||||
| -rw-r--r-- | python/tests/files/datacite/datacite_result_31.json | 24 | ||||
| -rw-r--r-- | python/tests/files/datacite/datacite_result_32.json | 23 | ||||
| -rw-r--r-- | python/tests/import_datacite.py | 2 |
7 files changed, 177 insertions, 5 deletions
diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py index 81f00876..d998f266 100644 --- a/python/fatcat_tools/importers/datacite.py +++ b/python/fatcat_tools/importers/datacite.py @@ -496,10 +496,12 @@ class DataciteImporter(EntityImporter): if not desc.get('descriptionType') == 'Abstract': continue - # Description maybe a string or list. + # Description maybe a string, int or list. text = desc.get('description', '') if not text: continue + if isinstance(text, int): + text = '{}'.format(text) if isinstance(text, list): try: text = "\n".join(text) @@ -758,6 +760,14 @@ class DataciteImporter(EntityImporter): given_name = clean(given_name) if surname: surname = clean(surname) + + # Perform a final assertion that name does not reduce to zero + # (e.g. whitespace only name). + if name: + name = name.strip() + if not name: + continue + if raw_affiliation == '': continue diff --git a/python/fatcat_tools/workers/elasticsearch.py b/python/fatcat_tools/workers/elasticsearch.py index 525f372b..e58b3da1 100644 --- a/python/fatcat_tools/workers/elasticsearch.py +++ b/python/fatcat_tools/workers/elasticsearch.py @@ -19,7 +19,7 @@ class ElasticsearchReleaseWorker(FatcatWorker): def __init__(self, kafka_hosts, consume_topic, poll_interval=10.0, offset=None, elasticsearch_backend="http://localhost:9200", elasticsearch_index="fatcat", - batch_size=200): + batch_size=200, api_host="https://api.fatcat.wiki/v0"): super().__init__(kafka_hosts=kafka_hosts, consume_topic=consume_topic) self.consumer_group = "elasticsearch-updates3" @@ -30,9 +30,11 @@ class ElasticsearchReleaseWorker(FatcatWorker): self.entity_type = ReleaseEntity self.elasticsearch_document_name = "release" self.transform_func = release_to_elasticsearch + self.api_host = api_host def run(self): ac = ApiClient() + api = public_api(self.api_host) def fail_fast(err, partitions): if err is not None: @@ -103,13 +105,20 @@ class ElasticsearchReleaseWorker(FatcatWorker): if entity_dict.get('name') and not entity_dict.get('title'): continue entity = entity_from_json(json_str, self.entity_type, api_client=ac) + if self.elasticsearch_document_name == "changelog": + key = entity.index + # might need to fetch from API + if not (entity.editgroup and entity.editgroup.editor): + entity = api.get_changelog_entry(entity.index) + else: + key = entity.ident # TODO: handle deletions from index bulk_actions.append(json.dumps({ - "index": { "_id": entity.ident, }, + "index": { "_id": key, }, })) bulk_actions.append(json.dumps( self.transform_func(entity))) - print("Upserting, eg, {} (of {} releases in elasticsearch)".format(entity.ident, len(batch))) + print("Upserting, eg, {} (of {} {} in elasticsearch)".format(key, len(batch), self.elasticsearch_document_name)) elasticsearch_endpoint = "{}/{}/{}/_bulk".format( self.elasticsearch_backend, self.elasticsearch_index, diff --git a/python/tests/files/datacite/datacite_doc_31.json b/python/tests/files/datacite/datacite_doc_31.json new file mode 100644 index 00000000..83af3e4d --- /dev/null +++ b/python/tests/files/datacite/datacite_doc_31.json @@ -0,0 +1,53 @@ +{ + "id": "10.17912/micropub.biology.000143", + "type": "dois", + "attributes": { + "doi": "10.17912/micropub.biology.000143", + "identifiers": null, + "creators": [ + { + "raw_name": " ", + "givenName": "", + "familyName": "", + "affiliation": [], + "role": "author" + } + ], + "titles": [ + { + "title": "Sample" + } + ], + "publisher": "microPublication Biology", + "publicationYear": 2019, + "types": { + "resourceTypeGeneral": "DataPaper" + }, + "relatedIdentifiers": [], + "sizes": [], + "formats": [], + "version": null, + "rightsList": [], + "descriptions": [ + { + "description": "Biological liquid-liquid phase separation", + "descriptionType": "Abstract" + } + ], + "geoLocations": [], + "fundingReferences": [], + "url": "https://www.micropublication.org/journals/biology/micropub.biology.000143", + "created": "2019-08-19T14:43:08.000Z", + "registered": "2019-08-19T14:43:09.000Z", + "published": "2019", + "updated": "2019-11-09T12:32:02.000Z" + }, + "relationships": { + "client": { + "data": { + "id": "caltech.micropub", + "type": "clients" + } + } + } +} diff --git a/python/tests/files/datacite/datacite_doc_32.json b/python/tests/files/datacite/datacite_doc_32.json new file mode 100644 index 00000000..7ea7e873 --- /dev/null +++ b/python/tests/files/datacite/datacite_doc_32.json @@ -0,0 +1,53 @@ +{ + "id": "10.17912/micropub.biology.000143", + "type": "dois", + "attributes": { + "doi": "10.17912/micropub.biology.000143", + "identifiers": null, + "creators": [ + { + "raw_name": " ", + "givenName": "", + "familyName": "", + "affiliation": [], + "role": "author" + } + ], + "titles": [ + { + "title": "Sample" + } + ], + "publisher": "microPublication Biology", + "publicationYear": 2019, + "types": { + "resourceTypeGeneral": "DataPaper" + }, + "relatedIdentifiers": [], + "sizes": [], + "formats": [], + "version": null, + "rightsList": [], + "descriptions": [ + { + "description": 1234567890, + "descriptionType": "Abstract" + } + ], + "geoLocations": [], + "fundingReferences": [], + "url": "https://www.micropublication.org/journals/biology/micropub.biology.000143", + "created": "2019-08-19T14:43:08.000Z", + "registered": "2019-08-19T14:43:09.000Z", + "published": "2019", + "updated": "2019-11-09T12:32:02.000Z" + }, + "relationships": { + "client": { + "data": { + "id": "caltech.micropub", + "type": "clients" + } + } + } +} diff --git a/python/tests/files/datacite/datacite_result_31.json b/python/tests/files/datacite/datacite_result_31.json new file mode 100644 index 00000000..193104b0 --- /dev/null +++ b/python/tests/files/datacite/datacite_result_31.json @@ -0,0 +1,24 @@ +{ + "abstracts": [ + { + "content": "Biological liquid-liquid phase separation", + "lang": "fr", + "mimetype": "text/plain" + } + ], + "contribs": [], + "ext_ids": { + "doi": "10.17912/micropub.biology.000143" + }, + "extra": { + "datacite": { + "resourceTypeGeneral": "DataPaper" + }, + "container_name": "microPublication Biology" + }, + "refs": [], + "release_stage": "published", + "release_year": 2019, + "publisher": "microPublication Biology", + "title": "Sample" +} diff --git a/python/tests/files/datacite/datacite_result_32.json b/python/tests/files/datacite/datacite_result_32.json new file mode 100644 index 00000000..1a84a043 --- /dev/null +++ b/python/tests/files/datacite/datacite_result_32.json @@ -0,0 +1,23 @@ +{ + "abstracts": [ + { + "content": "1234567890", + "mimetype": "text/plain" + } + ], + "contribs": [], + "ext_ids": { + "doi": "10.17912/micropub.biology.000143" + }, + "extra": { + "datacite": { + "resourceTypeGeneral": "DataPaper" + }, + "container_name": "microPublication Biology" + }, + "refs": [], + "release_stage": "published", + "release_year": 2019, + "publisher": "microPublication Biology", + "title": "Sample" +} diff --git a/python/tests/import_datacite.py b/python/tests/import_datacite.py index 15650375..c9210ea4 100644 --- a/python/tests/import_datacite.py +++ b/python/tests/import_datacite.py @@ -287,7 +287,7 @@ def test_datacite_conversions(datacite_importer): for now. """ datacite_importer.debug = True - for i in range(31): + for i in range(33): src = 'tests/files/datacite/datacite_doc_{0:02d}.json'.format(i) dst = 'tests/files/datacite/datacite_result_{0:02d}.json'.format(i) with open(src, 'r') as f: |