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"
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"
bail("Couldn't find an DOI")
docType = metapropsheaders.get_item("og:type");
if not docType:
bail("Cannot determine the publication type")
if docType != "article":
bail("Only supports journal papers ('article', 'JOUR') at this moment, but found " + docType)
if not metaheaders.get_item("citation_title"):
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
bail('Unable to find a DOI')
sys.exit(0)
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.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
bail('Unable to find a DOI')
sys.exit(0)
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"
bail('Unable to find a DOI')
sys.exit(0)
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.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
#
url = sys.stdin.readline().strip()
#
# Fetch the page
#
try:
page = urllib2.urlopen(url).read().strip()
except:
bail("Couldn't fetch page (" + url + ")")
#
# DOI is in the page
#
metaheaders = metaheaders.MetaHeaders(page=page)
dois = metaheaders.get_multi_item("DC.identifier")
doi = None
if dois:
for doi_str in dois:
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"
if doi:
print "linkout\tIUCR\t\t%s\t\t" % key
print "linkout\tDOI\t\t%s\t\t" % doi
print "url\thttp://dx.doi.org/" + doi
print "doi\t" + doi
else:
bail("Couldn't find a DOI")
if not metaheaders.get_item("DC.title"):
bail("Cannot find a title in that article")
title = metaheaders.get_item("DC.title")
if title:
print "title\t%s" % title.encode('utf-8')
authors = metaheaders.get_multi_item("DC.creator")
firstAuthorSurname = None
if authors:
for a in authors:
if not firstAuthorSurname:
firstAuthorSurname = ""
match = re.search(r'^([^\s|^,]+)', a.encode('utf-8'), re.IGNORECASE)
if match:
firstAuthorSurname = match.group(1)
print "author\t%s" % a.encode('utf-8')
shortCode = None
if metaheaders.get_item("DC.link"):
link = metaheaders.get_item("DC.link")
match = re.search(r'^http://scripts.iucr.org/cgi-bin/paper\?(.+)', link, re.IGNORECASE)
if match: