sys.exit(1)
outfile = 'oup.tag'
parser = OUPJATSParser()
files = glob(DIR+'/TagTextFiles/*.xml')
documents = []
for f in files:
try:
with open(f,'rU') as fp:
doc = parser.parse(fp)
documents.append(doc)
except Exception as e:
print("Error in OUP parser:", f, e)
# print(documents)
# Write everything out in Classic tagged format
fo = open(outfile, 'a')
serializer = Tagged()
ref_handler = ReferenceWriter()
for d in documents:
#print(d)
serializer.write(d,fo)
ref_handler.writeref(d,'oup')
fo.close()
outfile = 'pnas.tag'
fo = open(outfile, 'a')
for k, v in PNAS_RSS_URLS.items():
feed = feedparser.parse(v)
# print "feed:",k
for _item in feed['entries']:
try:
record = {}
absURL = _item['link']
volno = _item['prism_volume'].zfill(4)
ident = _item['dc_identifier']
ident = ident.replace('hwp:master-id:pnas;', '')
# print absURL,volno,ident
pnas = PNASParser()
output = pnas.parse(absURL)
except Exception as err:
print("Error in parser:", err)
else:
try:
serializer = Tagged()
serializer.write(output, fo)
except Exception as err:
print("Error in serializer:", err)
try:
ref_handler = ReferenceWriter()
ref_handler.writeref(output, 'pnas')
except Exception as err:
print("Error in writeref:", err)
fo.close()
abs_source = urllib.urlopen(absURL).read()
open(archive_file, 'w').write(abs_source)
pnas = PNASParser()
output = pnas.parse(abs_source)
records.append(output)
except Exception, err:
print("Error parsing %s: %s:" % (absURL, err))
if records:
try:
fo = open(outfile, 'a')
for rec in records:
try:
serializer = Tagged()
serializer.write(rec, fo)
except Exception, err:
print("Error in serializer: %s" % err)
try:
ref_handler = ReferenceWriter()
ref_handler.writeref(rec, 'pnas')
except Exception, err:
print("Error in writeref: %s" % err)
print("New PNAS records available in %s" % outfile)
fo.close()
except:
print("Error writing PNAS records: %s" % err)
else:
print("No new PNAS records available.")
print("End PNAS harvest.")
i = i - 1
v = vols[i]
papers = glob(v + '/*.xml')
print "VEE:", v
# Try the parser
documents = []
for p in papers:
try:
with open(p, 'rU') as fp:
doc = parser.parse(fp)
documents.append(doc)
except Exception as e:
print("Error in IOP parser:", p, e)
# Write everything out in Classic tagged format
fo = open(outfile, 'a')
serializer = Tagged()
refwriter = ReferenceWriter()
refwriter.refsource = '.jats.iopft.xml'
for d in documents:
serializer.write(d, fo)
try:
refwriter.writeref(d)
except Exception, err:
print("Error in refwriter: %s" % err)
fo.close()