def findEntry(self, journal, volume, page):
from papers.pdfglobals import PDFGetGlobals
for label in self.records:
try:
record = self.records[label]
volcheck = str(record.get("volume"))
if not volume == volcheck:
continue
pagecheck = str(record.get("pages")).split("-")[0].strip()
if not page == pagecheck:
continue
jobj = PDFGetGlobals.get_journal(journal)
jrec = PDFGetGlobals.get_journal(str(record.get("journal")))
if jobj and jobj.name != jrec.name:
continue
return record
except Exception, error:
continue
def parse_reference(text):
from papers.pdfglobals import PDFGetGlobals as pdfglobals
repls = {
".," : ".",
" and" : ", ",
".-" : ". ",
}
for entry in repls:
text = text.replace(entry, repls[entry])
match = re.compile("^(.*?)\n\s*[\(]\d+[\)]", re.DOTALL).search(text)
if match:
text = match.groups()[0]
journal = pdfglobals.find_journal_in_entry(text)
if not journal:
sys.stderr.write("Could not find properly formatted journal\n")
return None
match = re.compile("^.*?(\d{4}[,]\s*\d+[,]\s*\d+[.]?)", re.DOTALL).search(text) #for some reason spaces are not matching here
if not match:
sys.stderr.write("Could not find properly formatted volume, page, year\n")
return None
numbers = match.groups()[0]
match = re.compile("(\d{4})[,].*?(\d+)[,].*?(\d+)", re.DOTALL).search(numbers)
if not match:
sys.stderr.write("Could not find properly formatted volume, page, year\n")
return None
year, volume, page = match.groups()
#throw out the journal
new_text = text.replace(journal, "").replace(numbers, "")
match = re.compile("[\(]\d+[\)](.*)", re.DOTALL).search(new_text)
if match:
new_text = match.groups()[0]
#get the authors
matches = new_text.split(";")
if not matches:
sys.stderr.write("Could not find properly formatted authors\n")
return None
authors = []
for author in matches:
lastname, initials = author.split(",")
initials = map(lambda x: x.strip().strip("."), initials)
initials = "".join(initials)
authors.append("%s, %s" % (lastname, initials))
vals = {}
vals["volume"] = int(volume)
vals["page"] = Page(page)
vals["year"] = int(year)
vals["authors"] = authors
vals["journal"] = journal
return vals
def __init__(self, value):
from papers.pdfglobals import PDFGetGlobals as globals
jobj = globals.get_journal(value)
WOKField.__init__(self, "%s*" % jobj.name)
