bail("Cannot find a title in that article")
print "type\tREP"
metaheaders.print_item("title","citation_title")
authors = metaheaders.get_multi_item("citation_author")
if authors:
for a in authors:
print "author\t%s" % a
journal = metaheaders.get_item("citation_technical_report_institution")
if journal:
print "journal\t%s" % journal
issn = metaheaders.get_item("citation_issn")
if issn:
print "issn\t%s" % issn
abstract = metaheaders.get_item("description")
if abstract:
print "abstract\t%s" % abstract
start_page = metaheaders.get_item("citation_firstpage")
if start_page:
print "start_page\t%s" % start_page
metaheaders.print_date("citation_date")
print "end_tsv"
print "status\tok"
sys.exit(0)
doi = doi.replace("doi:", "")
print "begin_tsv"
print "linkout\tDOI\t\t%s\t\t" % (doi)
print "type\tJOUR"
print "doi\t" + doi
for f in key_map.keys():
k = key_map[f]
v = metaheaders.get_item(k)
if not v:
continue
v = v.strip()
print "%s\t%s" % (f, v)
authors = metaheaders.get_multi_item("dc.creator")
if not authors:
authors = metaheaders.get_multi_item("citation_author")
if authors:
for a in authors:
print "author\t%s" % a
if metaheaders.get_item("citation_date"):
metaheaders.print_date("citation_date")
elif metaheaders.get_item("dc.date"):
metaheaders.print_date("dc.date")
print "end_tsv"
print "status\tok"
print "type\tJOUR"
print "doi\t" + doi
for f in key_map.keys():
k = key_map[f]
v = metaheaders.get_item(k)
if not v:
continue
v = v.strip()
print "%s\t%s" % (f, v)
authors = metaheaders.get_multi_item("DC.Contributor")
if authors:
for a in authors:
print "author\t%s" % a
metaheaders.print_date("DC.Date")
# Hmmm. there are sometimes 2 issns, one empty
issn = metaheaders.get_multi_item("citation_issn")
if issn:
for i in issn:
if i != "":
print "issn\t%s" % i
break
root = metaheaders.root
abs = []
for p in root.cssselect("#abstract p"):
abs.append(p.xpath("string()"))
if len(abs) > 0:
abstract = ' '.join(abs)
print "begin_tsv"
print "linkout\tDOI\t\t%s\t\t" % (doi)
print "type\tJOUR"
print "doi\t" + doi
for f in key_map.keys():
k = key_map[f]
v = metaheaders.get_item(k)
if not v:
continue
v = v.strip()
print "%s\t%s" % (f, v)
authors = metaheaders.get_multi_item("citation_author")
if authors:
for a in authors:
print "author\t%s" % a
metaheaders.print_date("citation_publication_date")
# Hmmm. there are sometimes 2 issns, one empty
issn = metaheaders.get_multi_item("citation_issn")
if issn:
for i in issn:
if i != "":
print "issn\t%s" % i
break
print "end_tsv"
print "status\tok"
print "type\tJOUR"
print "doi\t" + doi
for f in key_map.keys():
k = key_map[f]
v = metaheaders.get_item(k)
if not v:
continue
v = v.strip()
print "%s\t%s" % (f,v)
authors = metaheaders.get_multi_item("DC.Contributor")
if authors:
for a in authors:
print "author\t%s" % a
metaheaders.print_date("DC.Date")
# Hmmm. there are sometimes 2 issns, one empty
issn = metaheaders.get_multi_item("citation_issn")
if issn:
for i in issn:
if i != "":
print "issn\t%s" % i
break
root = metaheaders.root
abs = []
for p in root.cssselect("#abstract p"):
abs.append(p.xpath("string()"))
if len(abs) > 0:
abstract = ' '.join(abs)
doi_match = re.search(r'doi:(10\.[^/]+/[^\s]+)', doi_str, re.IGNORECASE)
if doi_match:
doi = doi_match.group(1)
if not doi:
bail("Couldn't find a DOI")
if not metaheaders.get_item("DC.title"):
bail("Unable to find the article title")
print "begin_tsv"
print "publisher\tDryad Digital Repository"
print "type\tGEN"
metaheaders.print_item("title","DC.title")
authors = metaheaders.get_multi_item("DC.creator")
if authors:
for a in authors:
print "author\t%s" % a
metaheaders.print_date("DCTERMS.issued")
abstract = metaheaders.get_item("DC.description");
if abstract:
print "abstract\t%s" % abstract
print "linkout\tDRYAD\t\t%s\t\t" % (doi)
print "linkout\tDOI\t\t%s\t\t" % (doi)
print "end_tsv"
print "status\tok"
if abstract:
print "abstract\t%s" % abstract.encode('utf-8')
volume = metaheaders.get_item("prism.volume")
if volume:
print "volume\t%s" % volume
issue = metaheaders.get_item("prism.number")
if issue:
print "issue\t%s" % issue
start_page = metaheaders.get_item("prism.startingPage")
if start_page:
print "start_page\t%s" % start_page
end_page = metaheaders.get_item("prism.endingPage")
if end_page:
print "end_page\t%s" % end_page
publisher = metaheaders.get_item("DC.publisher")
if publisher:
print "publisher\t%s" % publisher
if metaheaders.get_item("citation_date"):
metaheaders.print_date("citation_date")
else:
metaheaders.print_date("prism.coverDate")
print "end_tsv"
print "status\tok"