def __init__(self, stream=sys.stdout):
self.stream = stream
self.xmlparser = SAXContentWriter(stream)
self.xmlparser.startDocument()
self._add_comment()
self.xmlparser.startElement("output")
self._labels = {}
def __init__(self, stream=sys.stdout):
self.stream = stream
self.xmlparser = SAXContentWriter(stream)
self.xmlparser.startDocument()
self._add_comment()
self.xmlparser.startElement("output")
self._labels = {}
class Model2XMLConverter(object):
"""
Functionality of this class allows to convert hierarchical object model of
RRS database (module rrslib.db.model) to RRS-XML format.
If translating object with some new attributes unknown to model (in model
they are specified as type _UnknownType, the converter will omit these
attributes.
"""
def __init__(self, stream=sys.stdout):
self.stream = stream
self.xmlparser = SAXContentWriter(stream)
self.xmlparser.startDocument()
self._add_comment()
self.xmlparser.startElement("output")
self._labels = {}
def _add_comment(self):
c = "\nOrigin: This XML was generated by Model2XMLConverter\n"
c += "Format: $RRS-XML$\n"
c += "Time: $%s$\n" % time.ctime()
self.xmlparser.comment(c)
def _resolve_obj_name(self, obj):
clsname = str(obj.__class__).split("'")[1].split(".")[-1]
return clsname[3:].lower()
def _get_obj_attrs(self, obj):
a = {}
for attr in ("id", "credibility", "module"):
if obj.isset(attr):
xmlattr = attr
objattr = attr
if attr == "module":
xmlattr = "source-module"
a[xmlattr] = str(obj.get(objattr))
return a
def _get_role(self, attrstr):
role = None
tagname = attrstr
if "_" in attrstr:
# if not in tables, it has a probably a role
if attrstr not in db_tables:
sp = attrstr.split("_")
tagname = "_".join(sp[:-1])
# if it wasnt a role, but attribute with underscore
if tagname not in db_tables:
tagname = "_".join(sp[1:])
if tagname not in db_tables:
return (attrstr, None)
return (tagname, sp[0])
role = sp[-1]
if attrstr == "publisher":
return ('organization', attrstr)
return (tagname, role)
def _convert_raw_attribute(self, obj, attrstr):
# create element
if obj.isset(attrstr):
if isinstance(obj, _RRSDbEntityRelationship):
attr = obj.get_attribute(attrstr)
else:
attr = obj.get(attrstr)
val = attr
tagname, role = self._get_role(attrstr)
if type(attr) == bool:
if attr:
val = 'yes'
else:
val = 'no'
val = self._unicode(val)
d = {"value": val}
if role is not None:
d['role'] = role
self.xmlparser.startElement(tagname, d)
self.xmlparser.endElement()
def _hash(self):
while 1:
h = "".join([hex(random.randint(0,16)).lstrip("0x") for x in range(0,5)])
if h in self._labels.values(): continue
return h
def _unicode(self, string):
if isinstance(string, unicode):
return string
return unicode(str(string), encoding='utf-8')
def _conversion(self, obj, role=None, add_relation_attributes={}):
if not isinstance(obj, _RRSDatabaseEntity):
raise RRSXMLConverterError("Error while conversion model->rrs-xml. " \
"Failed to convert %s", (str(type(obj))))
name = self._resolve_obj_name(obj)
attrs = self._get_obj_attrs(obj)
if role is not None:
attrs['role'] = role
for k in add_relation_attributes:
attrs[k] = self._unicode(add_relation_attributes[k])
# create label or ref
if obj in self._labels:
attrs['ref'] = self._labels[obj]
self.xmlparser.startElement(self._unicode(name), attrs)
self.xmlparser.endElement()
return
else:
hsh = self._hash()
self._labels[obj] = hsh
attrs['label'] = hsh
self.xmlparser.startElement(self._unicode(name), attrs)
for attrstr in obj.__dict__:
if attrstr in ("__types__", "id", "credibility", "module", "lev", "_table_name"):
continue
# pass all unknown attributes
if attrstr in obj.__types__ and obj.__types__[attrstr] == _UnknownType:
continue
attr = getattr(obj, attrstr)
# handle junction objects (abstraction of N:N junc. tables)
# these are represented by RRSRelationship***** objects
if type(attr) is list:
# iterate over all Relationship objects in list
_, role = self._get_role(attrstr)
for item in attr:
raw_rel_attrs = {}
# add all attributes of the relationship
for entattrstr in item.__dict__:
if entattrstr in ("__types__", "_parent", "_entities"):
continue
entattr = getattr(item, entattrstr)
if entattr is not None:
raw_rel_attrs[entattrstr] = str(entattr)
#if isinstance(entattr, _RRSDatabaseEntity):
# self._conversion(entattr)
#else:
# self._convert_raw_attribute(item, entattrstr)
# iterate over all entities of relation (i.e Publication and Project)
for ent in item._entities:
self._conversion(ent, role=role, add_relation_attributes=raw_rel_attrs)
elif isinstance(attr, _RRSDatabaseEntity):
_ , role = self._get_role(attrstr)
self._conversion(attr, role=role)
else:
self._convert_raw_attribute(obj, attrstr)
self.xmlparser.endElement()
def convert(self, *args):
"""
Converts hierarchical object model of RRS database to rrs-xml format in
order of RRS-XML specification on
https://merlin.fit.vutbr.cz/nlp-wiki/index.php/Rrs_moduleapi
"""
self._labels = {} # clean labels
# if obj is sequence or list or some iterable, go through
for obj in args:
if isinstance(obj, list):
for o in obj:
o.compile()
self._conversion(o)
else:
obj.compile()
self._conversion(obj)
self.xmlparser.endDocument()
class Model2XMLConverter(object):
"""
Functionality of this class allows to convert hierarchical object model of
RRS database (module rrslib.db.model) to RRS-XML format.
If translating object with some new attributes unknown to model (in model
they are specified as type _UnknownType, the converter will omit these
attributes.
"""
def __init__(self, stream=sys.stdout):
self.stream = stream
self.xmlparser = SAXContentWriter(stream)
self.xmlparser.startDocument()
self._add_comment()
self.xmlparser.startElement("output")
self._labels = {}
def _add_comment(self):
c = "\nOrigin: This XML was generated by Model2XMLConverter\n"
c += "Format: $RRS-XML$\n"
c += "Time: $%s$\n" % time.ctime()
self.xmlparser.comment(c)
def _resolve_obj_name(self, obj):
clsname = str(obj.__class__).split("'")[1].split(".")[-1]
return clsname[3:].lower()
def _get_obj_attrs(self, obj):
a = {}
for attr in ("id", "credibility", "module"):
if obj.isset(attr):
xmlattr = attr
objattr = attr
if attr == "module":
xmlattr = "source-module"
a[xmlattr] = str(obj.get(objattr))
return a
def _get_role(self, attrstr):
role = None
tagname = attrstr
if "_" in attrstr:
# if not in tables, it has a probably a role
if attrstr not in db_tables:
sp = attrstr.split("_")
tagname = "_".join(sp[:-1])
# if it wasnt a role, but attribute with underscore
if tagname not in db_tables:
tagname = "_".join(sp[1:])
if tagname not in db_tables:
return (attrstr, None)
return (tagname, sp[0])
role = sp[-1]
if attrstr == "publisher":
return ('organization', attrstr)
return (tagname, role)
def _convert_raw_attribute(self, obj, attrstr):
# create element
if obj.isset(attrstr):
if isinstance(obj, _RRSDbEntityRelationship):
attr = obj.get_attribute(attrstr)
else:
attr = obj.get(attrstr)
val = attr
tagname, role = self._get_role(attrstr)
if type(attr) == bool:
if attr:
val = 'yes'
else:
val = 'no'
val = self._unicode(val)
d = {"value": val}
if role is not None:
d['role'] = role
self.xmlparser.startElement(tagname, d)
self.xmlparser.endElement()
def _hash(self):
while 1:
h = "".join(
[hex(random.randint(0, 16)).lstrip("0x") for x in range(0, 5)])
if h in self._labels.values(): continue
return h
def _unicode(self, string):
if isinstance(string, unicode):
return string
return unicode(str(string), encoding='utf-8')
def _conversion(self, obj, role=None, add_relation_attributes={}):
if not isinstance(obj, _RRSDatabaseEntity):
raise RRSXMLConverterError("Error while conversion model->rrs-xml. " \
"Failed to convert %s", (str(type(obj))))
name = self._resolve_obj_name(obj)
attrs = self._get_obj_attrs(obj)
if role is not None:
attrs['role'] = role
for k in add_relation_attributes:
attrs[k] = self._unicode(add_relation_attributes[k])
# create label or ref
if obj in self._labels:
attrs['ref'] = self._labels[obj]
self.xmlparser.startElement(self._unicode(name), attrs)
self.xmlparser.endElement()
return
else:
hsh = self._hash()
self._labels[obj] = hsh
attrs['label'] = hsh
self.xmlparser.startElement(self._unicode(name), attrs)
for attrstr in obj.__dict__:
if attrstr in ("__types__", "id", "credibility", "module", "lev",
"_table_name"):
continue
# pass all unknown attributes
if attrstr in obj.__types__ and obj.__types__[
attrstr] == _UnknownType:
continue
attr = getattr(obj, attrstr)
# handle junction objects (abstraction of N:N junc. tables)
# these are represented by RRSRelationship***** objects
if type(attr) is list:
# iterate over all Relationship objects in list
_, role = self._get_role(attrstr)
for item in attr:
raw_rel_attrs = {}
# add all attributes of the relationship
for entattrstr in item.__dict__:
if entattrstr in ("__types__", "_parent", "_entities"):
continue
entattr = getattr(item, entattrstr)
if entattr is not None:
raw_rel_attrs[entattrstr] = str(entattr)
#if isinstance(entattr, _RRSDatabaseEntity):
# self._conversion(entattr)
#else:
# self._convert_raw_attribute(item, entattrstr)
# iterate over all entities of relation (i.e Publication and Project)
for ent in item._entities:
self._conversion(ent,
role=role,
add_relation_attributes=raw_rel_attrs)
elif isinstance(attr, _RRSDatabaseEntity):
_, role = self._get_role(attrstr)
self._conversion(attr, role=role)
else:
self._convert_raw_attribute(obj, attrstr)
self.xmlparser.endElement()
def convert(self, *args):
"""
Converts hierarchical object model of RRS database to rrs-xml format in
order of RRS-XML specification on
https://merlin.fit.vutbr.cz/nlp-wiki/index.php/Rrs_moduleapi
"""
self._labels = {} # clean labels
# if obj is sequence or list or some iterable, go through
for obj in args:
if isinstance(obj, list):
for o in obj:
o.compile()
self._conversion(o)
else:
obj.compile()
self._conversion(obj)
self.xmlparser.endDocument()
