diff options
| author | Martin Czygan <martin.czygan@gmail.com> | 2019-12-26 17:36:18 +0100 |
|---|---|---|
| committer | Martin Czygan <martin.czygan@gmail.com> | 2019-12-28 23:07:32 +0100 |
| commit | 1f7bbc5a582db45fcd6034800959e158d35a2297 (patch) | |
| tree | a0b584353ad6b638f641f35638d87d70cd28463a /python | |
| parent | a4cd65ed4897987e70520d81c7caa27cd28ed5a3 (diff) | |
| download | fatcat-1f7bbc5a582db45fcd6034800959e158d35a2297.tar.gz fatcat-1f7bbc5a582db45fcd6034800959e158d35a2297.zip | |
datacite: include doi in error messages
Diffstat (limited to 'python')
| -rw-r--r-- | python/fatcat_tools/importers/datacite.py | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py index 66f812e2..a4a3ef8b 100644 --- a/python/fatcat_tools/importers/datacite.py +++ b/python/fatcat_tools/importers/datacite.py @@ -264,6 +264,7 @@ class DataciteImporter(EntityImporter): return None attributes = obj['attributes'] + doi = attributes.get('doi', '').lower() # Contributors. Many nameIdentifierSchemes, we do not use (yet): # "attributes.creators[].nameIdentifiers[].nameIdentifierScheme": @@ -313,7 +314,7 @@ class DataciteImporter(EntityImporter): contribs.append(fatcat_openapi_client.ReleaseContrib( index=i, extra=extra)) else: - print('unknown name type: {}'.format(nameType), file=sys.stderr) + print('[{}] unknown name type: {}'.format(doi, nameType), file=sys.stderr) # Title, may come with "attributes.titles[].titleType", like # "AlternativeTitle", "Other", "Subtitle", "TranslatedTitle" @@ -322,7 +323,7 @@ class DataciteImporter(EntityImporter): titles) if not title: - print('skipping record w/o title: {}'.format(obj), file=sys.stderr) + print('[{}] skipping record w/o title: {}'.format(doi, obj), file=sys.stderr) return False if not subtitle: @@ -370,7 +371,7 @@ class DataciteImporter(EntityImporter): container_title = container.get('title') if isinstance(container_title, list): if len(container_title) > 0: - print('too many container titles: {}'.format( + print('[{}] too many container titles: {}'.format(doi, len(container_title))) container_title = container_title[0] assert isinstance(container_title, str) @@ -398,7 +399,7 @@ class DataciteImporter(EntityImporter): int(first_page) < int(last_page) pages = '{}-{}'.format(first_page, last_page) except ValueError as err: - print(err, file=sys.stderr) + print('[{}] {}'.format(doi, err), file=sys.stderr) pass if not pages and first_page: @@ -426,7 +427,7 @@ class DataciteImporter(EntityImporter): break if release_type is None: - print("no mapped type: {}".format(value), file=sys.stderr) + print("[{}] no mapped type: {}".format(doi, value), file=sys.stderr) # Language values are varied ("ger", "es", "English", "ENG", "en-us", # "other", ...). Try to crush it with langcodes: "It may sound to you @@ -439,7 +440,7 @@ class DataciteImporter(EntityImporter): try: language = pycountry.languages.lookup(value).alpha_2 except (LookupError, AttributeError) as err: - print('language lookup miss for {}: {}'.format(value, err), file=sys.stderr) + print('[{}] language lookup miss for {}: {}'.format(doi, value, err), file=sys.stderr) # Abstracts appear in "attributes.descriptions[].descriptionType", some # of the observed values: "Methods", "TechnicalInfo", @@ -461,7 +462,7 @@ class DataciteImporter(EntityImporter): try: lang = langdetect.detect(text) except langdetect.lang_detect_exception.LangDetectException as err: - print('language detection failed: {}'.format(err), + print('[{}] language detection failed: {}'.format(doi, err), file=sys.stderr) abstracts.append( fatcat_openapi_client.ReleaseAbstract( @@ -534,7 +535,6 @@ class DataciteImporter(EntityImporter): if extra_datacite: extra['datacite'] = extra_datacite - doi = attributes.get('doi', '').lower() extids = self.lookup_ext_ids(doi=doi) # Assemble release. |