# Reads an XML file and returns XML elements.
# Based on inputs.fileinput.XmlElementStreamFileInput.
#
# Author: Frank Steggink
#
from copy import deepcopy
from stetl.component import Config
from stetl.filter import Filter
from stetl.util import Util, etree
from stetl.packet import FORMAT
log = Util.get_log('xmlelementreader')
[docs]class XmlElementReader(Filter):
"""
Extracts XML elements from a file, outputs each feature element in Packet.
Parsing is streaming (no internal DOM buildup) so any file size can be handled.
Use this class for your big GML files!
consumes=FORMAT.string, produces=FORMAT.etree_element
"""
# Start attribute config meta
[docs] @Config(ptype=bool, default=False, required=False)
def strip_namespaces(self):
"""
should namespaces be removed from the input document and thus not be present in the output element stream?
"""
pass
# End attribute config meta
# Constructor
def __init__(self, configdict, section):
Filter.__init__(self, configdict, section, consumes=FORMAT.string, produces=FORMAT.etree_element)
self.context = None
self.root = None
self.cur_file_path = None
self.elem_count = 0
log.info("Element tags to be matched: %s" % self.element_tags)
[docs] def invoke(self, packet):
if packet.data is None:
log.info("No XML file given")
return packet
if self.cur_file_path is None:
self.cur_file_path = packet.data
event = None
packet.data = None
if self.context is None:
# Open file
fd = open(self.cur_file_path, 'rb')
self.elem_count = 0
log.info("file opened : %s" % self.cur_file_path)
self.context = etree.iterparse(fd, events=("start", "end"))
self.context = iter(self.context)
event, self.root = next(self.context)
packet = self.process_xml(packet)
return packet
def process_xml(self, packet):
while self.context is not None:
# while not packet.is_end_of_doc():
try:
event, elem = next(self.context)
except (etree.XMLSyntaxError, StopIteration):
# workaround for etree.XMLSyntaxError https://bugs.launchpad.net/lxml/+bug/1185701
self.context = None
if self.context is None:
# Always end of doc
# TODO: is this still useful for a non-input component?
packet.set_end_of_doc()
log.info("End of doc: %s elem_count=%d" % (self.cur_file_path, self.elem_count))
return packet
# Filter out Namespace from the tag
# this is the easiest way to go for now
tag = elem.tag.split('}')
if len(tag) == 2:
# Namespaced tag: 2nd is tag
tag = tag[1]
else:
# Non-namespaced tag: first
tag = tag[0]
if tag in self.element_tags:
if event == "start":
pass
elif event == "end":
packet.data = deepcopy(elem)
self.elem_count += 1
if self.strip_namespaces:
packet.data = Util.stripNamespaces(elem).getroot()
# Clear the element which has been read. Don't clear the root document,
# since the last element hasn't been processed yet.
elem.clear()
# If there is a next component, let it process
if self.next:
# Hand-over data (line, doc whatever) to the next component
packet.format = self._output_format
packet = self.next.process(packet)
return packet