def endElement(self, name):
SAX2DOM.endElement(self, name)
if name == 'release':
release = self.document.getElementsByTagName('release')[0]
self.restart()
#print release.toprettyxml(' ').encode('utf8')
id = int(release.getAttribute('id'))
sys.stderr.write('%d\r' % id)
if id not in self._discogs_ids:
return
catnos = []
labels = []
for node in release.getElementsByTagName('label'):
catnos.append(node.getAttribute('catno'))
labels.append(node.getAttribute('name'))
nodes = release.getElementsByTagName('country')
country = get_text(nodes[0]) if nodes else ''
nodes = release.getElementsByTagName('released')
date = get_text(nodes[0]) if nodes else ''
formats = []
for node in release.getElementsByTagName('format'):
formats.append(node.getAttribute('name'))
line = '%s\t%s\t%s\t%s\t%s\t%s' % (id, ';'.join(catnos),
';'.join(labels), country, date,
';'.join(set(formats)))
print line.encode('utf-8')
def fuck_dom(page):
page = UnicodeDammit(page).unicode_markup
tree = etree.fromstring(page,etree.HTMLParser())
#tree.docinfo.encoding = "utf-8"
handler = SAX2DOM()
sax.saxify(tree, handler)
return handler.document
def parse_lxml_dom(xml, strict_xml=True):
if strict_xml:
parse_func = lxml.etree.fromstring
else:
parse_func = lxml.html.document_fromstring
try:
tree = parse_func(xml)
except lxml.etree.XMLSyntaxError:
tree = parse_func('<body>%s</body>' % xml)
handler = SAX2DOM()
lxml.sax.saxify(tree, handler)
return handler.document
def endElement(self, name):
SAX2DOM.endElement(self, name)
if name == 'release':
release = self.document.getElementsByTagName('release')[0]
self.restart()
#print release.toprettyxml(' ').encode('utf8')
id = int(release.getAttribute('id'))
sys.stderr.write('%d\r' % id)
if id not in self._discogs_ids:
return
catnos = []
labels = []
for node in release.getElementsByTagName('label'):
catnos.append(node.getAttribute('catno'))
labels.append(node.getAttribute('name'))
nodes = release.getElementsByTagName('country')
country = get_text(nodes[0]) if nodes else ''
nodes = release.getElementsByTagName('released')
date = get_text(nodes[0]) if nodes else ''
formats = []
for node in release.getElementsByTagName('format'):
formats.append(node.getAttribute('name'))
line = '%s\t%s\t%s\t%s\t%s\t%s' % (id, ';'.join(catnos), ';'.join(labels), country, date, ';'.join(set(formats)))
print line.encode('utf-8')
def loadFromFile(cls, filename):
path_file = os.path.abspath(filename)
if not os.path.isfile(path_file):
err = "Error: '%s' does not exist or is not a file." % filename
print(err)
raise Exception(err)
# Note: parsing a file directly with dexml/minidom is supposedly slower, si I used lxml one,
# but I did not benchmark it.
tree = etree.parse(path_file)
handler = SAX2DOM()
sax.saxify(tree, handler)
dom = handler.document
# In case, you can pass the filename to parse() here to skip lxml
mdl = cls.parse(dom)
return mdl
def parse_lxml_dom(tree):
handler = SAX2DOM()
lxml.sax.saxify(tree, handler)
return handler.document
def endElement(self, name):
self._locationStack.pop()
SAX2DOM.endElement(self, name)
def startElement(self, name, attrs):
self._locationStack.append((name, self._docLocator.getLineNumber(),
self._docLocator.getColumnNumber()))
SAX2DOM.startElement(self, name, attrs)
def setDocumentLocator(self, locator):
self._docLocator = locator
SAX2DOM.setDocumentLocator(self, locator)
def __init__(self):
SAX2DOM.__init__(self)
self._locationStack = []
def endElement(self, name):
self._locationStack.pop()
SAX2DOM.endElement(self, name)
def startElement(self, name, attrs):
self._locationStack.append((name, self._docLocator.getLineNumber(), self._docLocator.getColumnNumber()))
SAX2DOM.startElement(self, name, attrs)
def setDocumentLocator(self, locator):
self._docLocator = locator
SAX2DOM.setDocumentLocator(self, locator)
def __init__(self):
SAX2DOM.__init__(self)
self._locationStack = []
def __init__(self, discogs_ids):
SAX2DOM.__init__(self)
self._discogs_ids = discogs_ids
def __init__(self, discogs_ids):
SAX2DOM.__init__(self)
self._discogs_ids = discogs_ids
