def __init__(self, name):
self.archive = Archive(name)
self.master = MasterArchive()
self.lib = Library()
class SavedRecordParser:
def __init__(self, name):
self.archive = Archive(name)
self.master = MasterArchive()
self.lib = Library()
def __iter__(self):
return iter(self.archive)
def __getitem__(self, index):
return self.archive.__getitem__(index)
def add_pdf(self, path):
self.archive.add_pdf(path)
def get_text(self, text, start, stop):
regexp = "\n%s (.*?)\n%s " % (start.upper(), stop.upper())
match = re.compile(regexp, re.DOTALL).search(text)
if not match:
return None
return match.groups()[0].strip()
def exclude_entry(self, entry, exclude):
for exc in exclude:
if exc in entry:
return True
return False
def get_entry(self, attr, method=None, default=None, require=True, exclude=(), entries=()):
set = getattr(self.article, "set_%s" % attr)
str_arr = []
for start, stop in entries:
str_arr.append("%s->%s" % (start, stop))
entry = self.get_text(self.block, start, stop)
if entry and not self.exclude_entry(entry, exclude):
if method:
entry = method(entry)
set(entry)
return
if not default == None:
set(default)
return
if require:
sys.stderr.write("ERROR: %s\n" % self.block)
msg = "no %s for tags\n" % attr
msg += "\n".join(str_arr)
raise ISIError(msg)
def feed(self, text, notes):
journals = {}
blocks = re.compile("PT\sJ(.*?)\nER", re.DOTALL).findall(text)
for block in blocks:
try:
self.block = block
self.article = self.archive.create_article()
get_number = lambda x: re.compile("(\d+)").search(x).groups()[0]
get_page = lambda x: Page(get_number(x))
#clean_title = lambda x: clean_line(clean_entry(x))
clean_title = Cleanup.clean_title
self.get_entry("journal", entries=(("so", "la"), ("so", "ab"), ("so", "sn")) )
self.get_entry("volume", method=int, entries=(("vl", "is"), ("vl", "bp")) )
self.get_entry("issue", method=lambda x: int(get_number(x)), require=False, entries=(("is", "bp"),) )
self.get_entry("start_page", method=get_page, exclude=("art. no.",), entries=(("bp", "ep"), ("bp", "ut"), ("ar", "di"), ("ar", "ut")) )
self.get_entry("end_page", method=get_page, require=False, entries=(("ep", "di"), ("ep", "ut")) )
self.get_entry("authors", method=lambda x: get_authors(x, "\n", ","), entries=(("af", "ti"), ("au", "ti"), ("au", "so")))
self.get_entry("title", method=clean_title, entries=(("ti", "so"),) )
self.get_entry("abstract", method=clean_entry, require=False, entries=(("ab", "sn"),) )
self.get_entry("year", method=int, entries=(("py", "vl"), ("py", "tc") ) )
self.get_entry("doi", require=False, entries=(("di", "pg"), ("di", "ut"),("di", "er")) )
self.article.set_notes(notes)
journal = ISIArticle.get_journal(self.article.get_journal())
volume = self.article.get_volume()
page = self.article.get_page()
name = "%s %d %s" % (journal, volume, page)
if not self.master.has(self.article):
self.archive.test_and_add(self.article)
else:
println("%s exists in archive\n" % name)
continue
except Exception, error:
sys.stderr.write("ERROR: %s\n%s\n" % (error, block))