def _walkZeeRex(self, session, node):
if node.localName in ['indexInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseName = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'title':
self.title = str(flattenTexts(node))
elif node.localName == 'contact':
self.contacts.append(str(flattenTexts(node)))
elif node.localName == 'schema':
id = node.getAttribute('identifier')
location = node.getAttribute('location')
name = node.getAttribute('name')
txrid = node.getAttributeNS(self.c3Namespace, 'transformer')
if (txrid):
txr = self.get_object(session, txrid)
if (txr == None):
raise ConfigFileException("No transformer to map to for %s" % (txrid))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
self.schemaLocations[id] = location
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _handleLxmlConfigNode(self, session, node):
if node.tag in self.simpleNodes:
setattr(self, node.tag[node.tag.find("}") + 1 :], flattenTexts(node).strip())
elif node.tag in ["flags", "{%s}flags" % CONFIG_NS]:
# Extract Rights info
# <flags> <flag> <object> <value> </flag> </flags>
for c in node.iterchildren(tag=etree.Element):
if c.tag in ["flag", "{%s}flag" % CONFIG_NS]:
obj = None
flag = None
for c2 in c.iterchildren(tag=etree.Element):
if c2.tag in ["object", "{%s}object" % CONFIG_NS]:
obj = flattenTexts(c2).strip()
elif c2.tag in ["value", "{%s}value" % CONFIG_NS]:
flag = flattenTexts(c2).strip()
if flag not in self.allFlags and flag[:4] != "c3fn":
msg = "Unknown flag: %s" % flag
raise ConfigFileException(msg)
if obj is None or flag is None:
msg = "Missing object or value element for flag for " "user %s" % self.username
raise ConfigFileException()
f = self.flags.get(flag, [])
if obj:
f.append(obj)
self.flags[flag] = f
elif node.tag in ["history", "{%s}history" % CONFIG_NS]:
# Extract user history
pass
elif node.tag in ["hostmask", "{%s}hostmask" % CONFIG_NS]:
# Extract allowed hostmask list
pass
def _walkZeeRex(self, session, node):
if node.localName in ['indexInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseName = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'title':
self.title = str(flattenTexts(node))
elif node.localName == 'contact':
self.contacts.append(str(flattenTexts(node)))
elif node.localName == 'schema':
id = node.getAttribute('identifier')
location = node.getAttribute('location')
name = node.getAttribute('name')
txrid = node.getAttributeNS(self.c3Namespace, 'transformer')
if (txrid):
txr = self.get_object(session, txrid)
if (txr is None):
raise ConfigFileException(
"No transformer to map to for %s" % (txrid))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
self.schemaLocations[id] = location
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _handleLxmlConfigNode(self, session, node):
if (node.tag in self.simpleNodes):
setattr(self, node.tag, flattenTexts(node).strip())
elif (node.tag == "flags"):
# Extract Rights info
# <flags> <flag> <object> <value> </flag> </flags>
for c in node.iterchildren(tag=etree.Element):
if c.tag == "flag":
obj = None
flag = None
for c2 in c.iterchildren(tag=etree.Element):
if c2.tag == "object":
obj = flattenTexts(c2).strip()
elif c2.tag == "value":
flag = flattenTexts(c2).strip()
if (flag not in self.allFlags) and (flag[:4] != "c3fn"):
raise ConfigFileException("Unknown flag: %s" % flag)
if obj == None or flag == None:
raise ConfigFileException("Missing object or value element for flag for user %s" % self.username)
f = self.flags.get(flag, [])
if (obj):
f.append(obj)
self.flags[flag] = f
elif (node.tag == "history"):
# Extract user history
pass
elif (node.tag == "hostmask"):
# Extract allowed hostmask list
pass
def _handleLxmlConfigNode(self, session, node):
if node.tag in self.simpleNodes:
setattr(self, node.tag[node.tag.find('}') + 1:],
flattenTexts(node).strip())
elif node.tag in ["flags", '{%s}flags' % CONFIG_NS]:
# Extract Rights info
# <flags> <flag> <object> <value> </flag> </flags>
for c in node.iterchildren(tag=etree.Element):
if c.tag in ["flag", '{%s}flag' % CONFIG_NS]:
obj = None
flag = None
for c2 in c.iterchildren(tag=etree.Element):
if c2.tag in ["object", '{%s}object' % CONFIG_NS]:
obj = flattenTexts(c2).strip()
elif c2.tag in ["value", '{%s}value' % CONFIG_NS]:
flag = flattenTexts(c2).strip()
if (flag not in self.allFlags
and flag[:4] != "c3fn"):
msg = "Unknown flag: %s" % flag
raise ConfigFileException(msg)
if obj is None or flag is None:
msg = ("Missing object or value element for flag for "
"user %s" % self.username)
raise ConfigFileException()
f = self.flags.get(flag, [])
if (obj):
f.append(obj)
self.flags[flag] = f
elif node.tag in ["history", '{%s}history' % CONFIG_NS]:
# Extract user history
pass
elif node.tag in ["hostmask", '{%s}hostmask' % CONFIG_NS]:
# Extract allowed hostmask list
pass
def _handleLxmlConfigNode(self, session, node):
if (node.tag == "cluster"):
maps = []
for child in node.iterchildren(tag=etree.Element):
if (child.tag == "map"):
t = child.attrib.get('type', '')
map = []
for xpchild in child.iterchildren(tag=etree.Element):
if (xpchild.tag == "xpath"):
map.append(flattenTexts(xpchild).strip())
elif (xpchild.tag == "process"):
# turn xpath chain to workflow
ref = xpchild.attrib.get('ref', None)
if ref is not None:
process = self.get_object(session, ref)
else:
xpchild.tag = 'workflow'
process = CachingWorkflow(
session, xpchild, self)
process._handleLxmlConfigNode(session, xpchild)
map.append(process)
#vxp = [map[0]]
if (len(map) < 3):
# default ExactExtractor
map.append([['extractor', 'SimpleExtractor']])
if (t == u'key'):
self.keyMap = [map[0], map[1], map[2]]
else:
maps.append([map[0], map[1], map[2]])
self.maps = maps
def _handleLxmlConfigNode(self, session, node):
if (node.tag == "cluster"):
maps = []
for child in node.iterchildren(tag=etree.Element):
if (child.tag == "map"):
t = child.attrib.get('type', '')
map = []
for xpchild in child.iterchildren(tag=etree.Element):
if (xpchild.tag == "xpath"):
map.append(flattenTexts(xpchild).strip())
elif (xpchild.tag == "process"):
# turn xpath chain to workflow
ref = xpchild.attrib.get('ref', None)
if ref is not None:
process = self.get_object(session, ref)
else:
xpchild.tag = 'workflow'
process = CachingWorkflow(session, xpchild, self)
process._handleLxmlConfigNode(session, xpchild)
map.append(process)
#vxp = [map[0]]
if (len(map) < 3):
# default ExactExtractor
map.append([['extractor', 'SimpleExtractor']])
if (t == u'key'):
self.keyMap = [map[0], map[1], map[2]]
else:
maps.append([map[0], map[1], map[2]])
self.maps = maps
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseUrl = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseUrl = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _handleLxmlConfigNode(self, session, node):
if node.tag == 'relations':
self.relations = {}
for rel in node.iterchildren(tag=etree.Element):
if rel.tag == 'relation':
relName = rel.attrib.get('name', None)
if relName is None:
raise ConfigFileException('Name not supplied for relation')
fields = []
for fld in rel.iterchildren(tag=etree.Element):
if fld.tag == 'object':
oid = flattenTexts(fld)
fields.append([oid, 'VARCHAR', oid])
elif fld.tag == 'field':
fname = fld.attrib.get('name', None)
if fname is None:
ConfigFileException('Name not supplied for field')
ftype = flattenTexts(fld)
fields.append([fname, ftype, ''])
self.relations[relName] = fields
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo', 'indexInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseUrl = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'schema':
id = node.getAttribute('identifier')
name = node.getAttribute('name')
xsl = node.getAttributeNS(self.c3Namespace, 'transformer')
if (xsl):
txr = self.get_object(session, xsl)
if (txr == None):
raise ConfigFileException("No transformer to map to for %s" % (xsl))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
elif node.localName == "supports":
stype = node.getAttribute('type')
data = flattenTexts(node)
if (stype == 'operation'):
wflw = node.getAttributeNS(self.c3Namespace, 'workflow')
if (wflw):
flow = self.get_object(session, wflw)
if (flow == None):
raise ConfigFileException("No workflow to map to for %s" % wflw)
self.workflowHash[data] = self.get_object(session, wflw)
elif (node.localName == 'default'):
dtype = node.getAttribute('type')
pname = "default" + dtype[0].capitalize() + dtype[1:]
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, pname, data)
elif (node.localName =='setting'):
dtype = node.getAttribute('type')
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, dtype, data)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _walkZeeRex(self, session, node):
if node.localName in ["databaseInfo", "metaInfo", "indexInfo"]:
# Ignore
return
elif node.localName == "serverInfo":
self.version = node.getAttribute("version")
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == "database":
self.databaseUrl = str(flattenTexts(node))
elif node.localName == "host":
self.host = str(flattenTexts(node))
elif node.localName == "port":
self.port = int(flattenTexts(node))
elif node.localName == "schema":
id = node.getAttribute("identifier")
name = node.getAttribute("name")
xsl = node.getAttributeNS(self.c3Namespace, "transformer")
if xsl:
txr = self.get_object(session, xsl)
if txr is None:
raise ConfigFileException("No transformer to map to for " "%s" % (xsl))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
elif node.localName == "supports":
stype = node.getAttribute("type")
data = flattenTexts(node)
if stype == "operation":
wflw = node.getAttributeNS(self.c3Namespace, "workflow")
if wflw:
flow = self.get_object(session, wflw)
if flow is None:
raise ConfigFileException("No workflow to map to for " "%s" % wflw)
self.workflowHash[data] = self.get_object(session, wflw)
elif node.localName == "default":
dtype = node.getAttribute("type")
pname = "default" + dtype[0].capitalize() + dtype[1:]
data = flattenTexts(node)
if data.isdigit():
data = int(data)
elif data == "false":
data = 0
elif data == "true":
data = 1
setattr(self, pname, data)
elif node.localName == "setting":
dtype = node.getAttribute("type")
data = flattenTexts(node)
if data.isdigit():
data = int(data)
elif data == "false":
data = 0
elif data == "true":
data = 1
setattr(self, dtype, data)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _handleLxmlConfigNode(self, session, node):
if node.tag in ['relations', '{%s}relations' % CONFIG_NS]:
self.relations = {}
for rel in node.iterchildren(tag=etree.Element):
if rel.tag in ['relation', '{%s}relation' % CONFIG_NS]:
relName = rel.attrib.get('name', None)
if relName is None:
raise ConfigFileException('Name not supplied for '
'relation')
fields = []
for fld in rel.iterchildren(tag=etree.Element):
if fld.tag in ['object', '{%s}object' % CONFIG_NS]:
oid = flattenTexts(fld)
fields.append([oid, 'VARCHAR', oid])
elif fld.tag in ['field', '{%s}field' % CONFIG_NS]:
fname = fld.attrib.get('name', None)
if fname is None:
ConfigFileException('Name not supplied for '
'field')
ftype = flattenTexts(fld)
fields.append([fname, ftype, ''])
self.relations[relName] = fields
def dataFromRecordXPaths(session, rec, xps, nTerms=1, joiner=u'; '):
"""Extract data from ``rec`` return a single unicode object.
Extract data from ``rec`` using multiple XPaths ``xps`` in priority order.
Return a maximum of ``nTerms`` matches, joining any multiple values with
`joiner``.
"""
global namespaceUriHash
data = []
for xp in xps:
data.extend(rec.process_xpath(session, xp, namespaceUriHash))
if len(data) >= nTerms:
break
return joiner.join([flattenTexts(d) for d in data[:nTerms]])
def __init__(self, data, xml="", docId=None, wordCount=0, byteCount=0):
self.dom = data
self.xml = xml
self.id = docId
self.parent = ('', '', -1)
self.context = None
self.metadata = {}
if wordCount:
self.wordCount = wordCount
else:
try:
# Sometimes this blows up
self.wordCount = len(flattenTexts(data).split())
except:
self.wordCount = 0
self.byteCount = byteCount
def __init__(self, data, xml="", docId=None, wordCount=0, byteCount=0):
self.dom = data
self.xml = xml
self.id = docId
self.parent = ('', '', -1)
self.context = None
self.metadata = {}
if wordCount:
self.wordCount = wordCount
else:
try:
# Sometimes this blows up
self.wordCount = len(flattenTexts(data).split())
except:
self.wordCount = 0
self.byteCount = byteCount
def _backwalkTitles(rec, xpath):
titles = []
xpathParts = xpath.split('/')
while xpathParts[-1] != 'dsc':
try:
tn = rec.process_xpath(session, '/'.join(xpathParts) + '/did/unittitle')[0]
t = flattenTexts(tn)
titles.append(t.strip())
except IndexError:
print etree.dump(rec.process_xpath(session, '/'.join(xpathParts) + '/did')[0])
raise
xpathParts.pop(-1)
titles.reverse()
return titles
def _handleLxmlConfigNode(self, session, node):
if (node.tag == "transform"):
self.tagset = node.attrib.get('tagset', '')
maps = []
for child in node.iterchildren(tag=etree.Element):
if child.tag == 'map':
map = []
for xpchild in child.iterchildren(tag=etree.Element):
if xpchild.tag == "xpath":
map.append(flattenTexts(xpchild))
if map[0][0] != "#":
vxp = [map[0]]
else:
# special case to process
vxp = [map[0]]
maps.append([vxp[0], map[1]])
self.maps = maps
def _handleLxmlConfigNode(self,session, node):
if node.tag in ["transform", '{%s}transform' % CONFIG_NS]:
self.tagset = node.attrib.get('tagset', '')
maps = []
for child in node.iterchildren(tag=etree.Element):
if child.tag in ['map', '{%s}map' % CONFIG_NS]:
map = []
for xpchild in child.iterchildren(tag=etree.Element):
if xpchild.tag in ["xpath", '{%s}xpath' % CONFIG_NS]:
map.append(flattenTexts(xpchild))
if map[0][0] != "#":
vxp = [map[0]]
else:
# special case to process
vxp = [map[0]]
maps.append([vxp[0], map[1]])
self.maps = maps
def _handleLxmlLog(self, node):
text = flattenTexts(node)
if text[0] != '"':
text = repr(text)
ref = node.attrib.get('ref', '')
lvl = node.attrib.get('level', '')
if (ref):
self.objrefs.add(ref)
obj = "self.objcache[%s]" % ref
else:
obj = "self.defaultLogger"
if lvl:
if lvl.isdigit():
return ["%s.log_lvl(session, %s, str(%s).strip())" % (obj, lvl, text)]
else:
return ["%s.log_%s(session, str(%s).strip())" % (obj, lvl, text)]
else:
return ["%s.log(session, str(%s).strip())" % (obj, text)]
def _handleConfigNode(self,session, node):
if (node.localName == "transform"):
self.tagset = node.getAttributeNS(None, 'tagset')
maps = []
for child in node.childNodes:
if (child.nodeType == elementType and child.localName == "map"):
map = []
for xpchild in child.childNodes:
if (xpchild.nodeType == elementType and xpchild.localName == "xpath"):
map.append(flattenTexts(xpchild))
if map[0][0] != "#":
# vxp = verifyXPaths([map[0]])
vxp = [map[0]]
else:
# special case to process
vxp = [map[0]]
maps.append([vxp[0], map[1]])
self.maps = maps
def _handleConfigNode(self, session, node):
if (node.localName == "cluster"):
maps = []
for child in node.childNodes:
if (child.nodeType == elementType
and child.localName == "map"):
t = child.getAttributeNS(None, 'type')
map = []
for xpchild in child.childNodes:
if (xpchild.nodeType == elementType
and xpchild.localName == "xpath"):
map.append(flattenTexts(xpchild))
elif (xpchild.nodeType == elementType
and xpchild.localName == "process"):
# turn xpath chain to workflow
ref = xpchild.getAttributeNS(None, 'ref')
if ref:
process = self.get_object(session, ref)
else:
try:
xpchild.localName = 'workflow'
except:
# 4suite dom sets read only
newTop = xpchild.ownerDocument.createElementNS(
None, 'workflow')
for kid in xpchild.childNodes:
newTop.appendChild(kid)
xpchild = newTop
process = CachingWorkflow(
session, xpchild, self)
process._handleConfigNode(session, xpchild)
map.append(process)
# XXX FIX ME
# vxp = verifyXPaths([map[0]])
vxp = [map[0]]
if (len(map) < 3):
# default ExactExtractor
map.append([['extractor', 'SimpleExtractor']])
if (t == u'key'):
self.keyMap = [vxp[0], map[1], map[2]]
else:
maps.append([vxp[0], map[1], map[2]])
self.maps = maps
def _handleLxmlLog(self, node):
code = []
ref = node.attrib.get('ref', '')
if ref:
code.append("object = db.get_object(session, '%s')" % ref)
else:
code.append("object = db.get_path(session, 'defaultLogger')")
text = flattenTexts(node)
if text[0] != '"':
text = repr(text)
lvl = node.attrib.get('level', '')
if (lvl):
if lvl.isdigit():
code.append("object.log_lvl(session, %s, str(%s).strip())" % (int(lvl), text))
else:
code.append("object.log_%s(session, str(%s).strip())" % (lvl, text))
else:
code.append("object.log(session, str(%s).strip())" % text)
return code
def _handleLog(self, node):
text = flattenTexts(node)
if not text.startswith('"'):
text = repr(text)
ref = node.getAttributeNS(None, 'ref')
lvl = node.getAttributeNS(None, 'level')
if (ref):
self.objrefs.add(ref)
obj = "self.objcache[%s]" % ref
else:
obj = "self.defaultLogger"
if lvl:
if lvl.isdigit():
return ["%s.log_lvl(session, %s, "
"str(%s).strip())" % (obj, lvl, text)]
else:
return ["%s.log_%s(session, "
"str(%s).strip())" % (obj, lvl, text)]
else:
return ["%s.log(session, str(%s).strip())" % (obj, text)]
def _handleLog(self, node):
code = []
ref = node.getAttributeNS(None, 'ref')
if (ref):
code.append("object = db.get_object(session, '%s')" % ref)
else:
code.append("object = db.get_path(session, 'defaultLogger')")
text = flattenTexts(node)
if not text.startswith('"'):
text = repr(text)
lvl = node.getAttributeNS(None, 'level')
if (lvl):
if lvl.isdigit():
code.append("object.log_lvl(session, %s, "
"str(%s).strip())" % (int(lvl), text))
else:
code.append("object.log_%s(session, "
"str(%s).strip())" % (lvl, text))
else:
code.append("object.log(session, str(%s).strip())" % text)
return code
def _handleLog(self, node):
code = []
ref = node.getAttributeNS(None, 'ref')
if (ref):
code.append("object = db.get_object(session, '%s')" % ref)
else:
code.append("object = db.get_path(session, 'defaultLogger')")
text = flattenTexts(node)
if not text.startswith('"'):
text = repr(text)
lvl = node.getAttributeNS(None, 'level')
if (lvl):
if lvl.isdigit():
code.append("object.log_lvl(session, %s, "
"str(%s).strip())" % (int(lvl), text))
else:
code.append("object.log_%s(session, "
"str(%s).strip())" % (lvl, text))
else:
code.append("object.log(session, str(%s).strip())" % text)
return code
def _handleConfigNode(self, session, node):
if (node.localName == "cluster"):
maps = []
for child in node.childNodes:
if (child.nodeType == elementType and child.localName == "map"):
t = child.getAttributeNS(None, 'type')
map = []
for xpchild in child.childNodes:
if (xpchild.nodeType == elementType and xpchild.localName == "xpath"):
map.append(flattenTexts(xpchild))
elif (xpchild.nodeType == elementType and xpchild.localName == "process"):
# turn xpath chain to workflow
ref = xpchild.getAttributeNS(None, 'ref')
if ref:
process = self.get_object(session, ref)
else:
try:
xpchild.localName = 'workflow'
except:
# 4suite dom sets read only
newTop = xpchild.ownerDocument.createElementNS(None, 'workflow')
for kid in xpchild.childNodes:
newTop.appendChild(kid)
xpchild = newTop
process = CachingWorkflow(session, xpchild, self)
process._handleConfigNode(session, xpchild)
map.append(process)
# XXX FIX ME
# vxp = verifyXPaths([map[0]])
vxp = [map[0]]
if (len(map) < 3):
# default ExactExtractor
map.append([['extractor', 'SimpleExtractor']])
if (t == u'key'):
self.keyMap = [vxp[0], map[1], map[2]]
else:
maps.append([vxp[0], map[1], map[2]])
self.maps = maps
def _handleLog(self, node):
text = flattenTexts(node)
if not text.startswith('"'):
text = repr(text)
ref = node.getAttributeNS(None, 'ref')
lvl = node.getAttributeNS(None, 'level')
if (ref):
self.objrefs.add(ref)
obj = "self.objcache[%s]" % ref
else:
obj = "self.defaultLogger"
if lvl:
if lvl.isdigit():
return [
"%s.log_lvl(session, %s, "
"str(%s).strip())" % (obj, lvl, text)
]
else:
return [
"%s.log_%s(session, "
"str(%s).strip())" % (obj, lvl, text)
]
else:
return ["%s.log(session, str(%s).strip())" % (obj, text)]
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseName = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'recordSyntax':
id = node.getAttribute('identifier')
# id is string dotted OID
thash = {}
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'elementSet'):
name = c.getAttribute('name')
xsl = c.getAttributeNS(self.c3Namespace, 'transformer')
if (xsl):
txr = self.get_object(session, xsl)
if (txr == None):
raise ConfigFileException("[%s] No transformer to map to for %s" % (self.id, xsl))
else:
txr = None
thash[name.lower()] = txr
self.transformerHash[id] = thash
elif node.localName == 'set':
name = node.getAttribute('name')
name = name.lower()
uri = node.getAttribute('identifier')
if not name or not uri:
raise ConfigFileException('Missing name or identifier for attribute set mappting.')
if (name in self.prefixes and uri != self.prefixes[name]):
raise(ConfigFileException('Multiple OIDs bound to same short name: %s -> %s' % (name, uri)))
self.prefixes[str(name)] = str(uri)
elif node.localName == 'index':
# Process indexes
idxName = node.getAttributeNS(self.c3Namespace, 'index')
indexObject = self.get_object(session, idxName)
if indexObject == None:
raise(ConfigFileException("Could not find Index object %s" % (idxName)))
maps = []
defaults = []
supports = []
for c in node.childNodes:
if c.nodeType == elementType:
if c.localName == 'map':
maps.append(self._walkZeeRex(session, c))
elif c.localName == 'configInfo':
for c2 in c.childNodes:
if c2.nodeType == elementType:
if c2.localName == "default":
# Get default attributes
for c3 in c2.childNodes:
if c3.nodeType == elementType and c3.localName == 'map':
defaults = self._walkZeeRex(session, c3)
elif c2.localName == "supports":
# Get other supported attributes
for c3 in c2.childNodes:
if c3.nodeType == elementType and c3.localName == 'map':
# Can't use c3:index to redirect here Too complicated
data = self._walkZeeRex(session, c3)
supports.append(data)
# FIXME: This is wrong. It doesn't respect combinations.
for s in supports:
defaults.extend(s)
for m in maps:
curr = self.indexHash.get(tuple(m), [])
curr.append((defaults, indexObject))
self.indexHash[tuple(m)] = curr
elif (node.localName == 'map'):
attrs = []
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'attr'):
short = c.getAttribute('set')
short = short.lower()
if not short:
short = 'bib1'
oid = self.prefixes.get(short, None)
if not oid:
raise ConfigFileException('No mapping for attribute set %s in Zeerex file' % short)
type = c.getAttribute('type')
if not type:
raise ConfigFileException('No type attribute for Z39.50 attr mapping in Zeerex')
type = int(type)
if type < 1:
raise ConfigFileException('Invalid type attribute for Z39.50 attr mapping in Zeerex: %s' % type)
attrVal = flattenTexts(c).strip()
if not attrVal:
raise ConfigFileException('No value given for attribute in Z39.50 mapping')
if attrVal.isdigit():
attrVal = int(attrVal)
else:
attrVal = attrVal.lower()
attrs.append((oid, type, attrVal))
return attrs
elif (node.localName == 'default'):
dtype = node.getAttribute('type')
pname = "default" + dtype[0].capitalize() + dtype[1:]
if dtype == 'attribute':
# Get map instead of text
for c in node.childNodes:
if c.nodeType == elementType and c.localName == 'map':
data = self._walkZeeRex(session, c)
else:
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
elif data.lower() in self.prefixes.keys():
data = self.prefixes[data.lower()]
setattr(self, pname, data)
elif (node.localName =='setting'):
dtype = node.getAttribute('type')
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, dtype, data)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseUrl = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'schema':
id = node.getAttribute('identifier')
name = node.getAttribute('name')
xsl = node.getAttributeNS(self.c3Namespace, 'transformer')
if (xsl):
txr = self.get_object(session, xsl)
if (txr is None):
raise ConfigFileException("No transformer to map to for "
"%s" % (xsl))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
elif node.localName == 'set':
name = node.getAttribute('name')
uri = node.getAttribute('identifier')
if (name in self.prefixes and uri != self.prefixes[name]):
raise ConfigFileException('Multiple URIs bound to same short '
'name: %s -> %s' % (name, uri))
self.prefixes[str(name)] = str(uri)
elif node.localName == 'index':
# Process indexes
idxName = node.getAttributeNS(self.c3Namespace, 'index')
indexObject = self.get_object(session, idxName)
if indexObject is None:
raise(ConfigFileException("[%s] No Index to map to for "
"%s" % (self.id, idxName)))
maps = []
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'map'):
maps.append(self._walkZeeRex(session, c))
for m in maps:
self.indexHash[m] = indexObject
# Need to generate all relations and modifiers
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'configInfo'):
for c2 in c.childNodes:
if (
c2.nodeType == elementType and
c2.localName == 'supports'
):
idxName2 = c2.getAttributeNS(self.c3Namespace,
'index')
if (not idxName2):
indexObject2 = indexObject
else:
indexObject2 = self.get_object(session,
idxName2)
if indexObject2 is None:
raise ConfigFileException(
"[%s] No Index to map to for "
"%s" % (self.id, idxName2)
)
st = str(c2.getAttribute('type'))
val = str(flattenTexts(c2))
for m in maps:
self.indexHash[(m[0],
m[1],
(st, val))] = indexObject2
elif (node.localName == 'map'):
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'name'):
short = c.getAttribute('set')
index = flattenTexts(c)
index = index.lower()
uri = self.resolvePrefix(short)
if (not uri):
raise ConfigFileException("No mapping for %s in "
"Zeerex" % (short))
return (str(uri), str(index))
elif (node.localName == 'default'):
dtype = node.getAttribute('type')
# XXX: would dtype.title() be nicer!?
pname = "default" + dtype[0].capitalize() + dtype[1:]
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, pname, data)
elif (node.localName == 'setting'):
dtype = node.getAttribute('type')
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, dtype, data)
elif (node.localName == 'supports'):
stype = node.getAttribute('type')
if stype in ['extraData', 'extraSearchData', 'extraScanData',
'extraExplainData', 'extension']:
# function, xnType, sruName
xn = node.getAttributeNS(self.c3Namespace, 'type')
if (not xn in ['record', 'term', 'searchRetrieve', 'scan',
'explain', 'response']):
raise ConfigFileException('Unknown extension type %s' % xn)
sru = node.getAttributeNS(self.c3Namespace, 'sruName')
fn = node.getAttributeNS(self.c3Namespace, 'function')
data = flattenTexts(node)
data = data.strip()
data = tuple(data.split(' '))
if fn.find('.') > -1:
# new version
try:
fn = dynamic.importObject(session, fn)
except ImportError:
raise ConfigFileException("Cannot find handler "
"function %s (in %s)" %
(fn, repr(sys.path)))
self.sruExtensionMap[sru] = (xn, fn, data)
else:
if (hasattr(srwExtensions, fn)):
fn = getattr(srwExtensions, fn)
else:
raise ConfigFileException('Cannot find handler '
'function %s in '
'srwExtensions.' % fn)
xform = node.getAttributeNS(self.c3Namespace,
'sruFunction')
if not xform:
sruFunction = srwExtensions.simpleRequestXform
elif hasattr(srwExtensions, xform):
sruFunction = getattr(srwExtensions, xform)
else:
raise ConfigFileException('Cannot find transformation '
'function %s in '
'srwExtensions.' % xform)
hashAttr = xn + "ExtensionHash"
curr = getattr(self, hashAttr)
curr[data] = fn
setattr(self, hashAttr, curr)
self.sruExtensionMap[sru] = (data[0], data[1], sruFunction)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo', 'indexInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseUrl = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'schema':
id = node.getAttribute('identifier')
name = node.getAttribute('name')
xsl = node.getAttributeNS(self.c3Namespace, 'transformer')
if (xsl):
txr = self.get_object(session, xsl)
if (txr is None):
raise ConfigFileException("No transformer to map to for "
"%s" % (xsl))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
elif node.localName == "supports":
stype = node.getAttribute('type')
data = flattenTexts(node)
if (stype == 'operation'):
wflw = node.getAttributeNS(self.c3Namespace, 'workflow')
if (wflw):
flow = self.get_object(session, wflw)
if (flow is None):
raise ConfigFileException("No workflow to map to for "
"%s" % wflw)
self.workflowHash[data] = self.get_object(session, wflw)
elif (node.localName == 'default'):
dtype = node.getAttribute('type')
pname = "default" + dtype[0].capitalize() + dtype[1:]
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, pname, data)
elif (node.localName == 'setting'):
dtype = node.getAttribute('type')
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, dtype, data)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _walkZeeRex(self, session, node):
if node.localName in ['databaseInfo', 'metaInfo']:
# Ignore
return
elif node.localName == 'serverInfo':
self.version = node.getAttribute('version')
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == 'database':
self.databaseName = str(flattenTexts(node))
elif node.localName == 'host':
self.host = str(flattenTexts(node))
elif node.localName == 'port':
self.port = int(flattenTexts(node))
elif node.localName == 'recordSyntax':
id = node.getAttribute('identifier')
# id is string dotted OID
thash = {}
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'elementSet'):
name = c.getAttribute('name')
xsl = c.getAttributeNS(self.c3Namespace, 'transformer')
if (xsl):
txr = self.get_object(session, xsl)
if (txr is None):
raise ConfigFileException(
"[%s] No transformer to map to for %s" %
(self.id, xsl))
else:
txr = None
thash[name.lower()] = txr
self.transformerHash[id] = thash
elif node.localName == 'set':
name = node.getAttribute('name')
name = name.lower()
uri = node.getAttribute('identifier')
if not name or not uri:
raise ConfigFileException(
'Missing name or identifier for attribute set mappting.')
if (name in self.prefixes and uri != self.prefixes[name]):
raise (ConfigFileException(
'Multiple OIDs bound to same short name: %s -> %s' %
(name, uri)))
self.prefixes[str(name)] = str(uri)
elif node.localName == 'index':
# Process indexes
idxName = node.getAttributeNS(self.c3Namespace, 'index')
indexObject = self.get_object(session, idxName)
if indexObject is None:
raise (ConfigFileException("Could not find Index object %s" %
(idxName)))
maps = []
defaults = []
supports = []
for c in node.childNodes:
if c.nodeType == elementType:
if c.localName == 'map':
maps.append(self._walkZeeRex(session, c))
elif c.localName == 'configInfo':
for c2 in c.childNodes:
if c2.nodeType == elementType:
if c2.localName == "default":
# Get default attributes
for c3 in c2.childNodes:
if c3.nodeType == elementType and c3.localName == 'map':
defaults = self._walkZeeRex(
session, c3)
elif c2.localName == "supports":
# Get other supported attributes
for c3 in c2.childNodes:
if c3.nodeType == elementType and c3.localName == 'map':
# Can't use c3:index to redirect here Too complicated
data = self._walkZeeRex(
session, c3)
supports.append(data)
# FIXME: This is wrong. It doesn't respect combinations.
for s in supports:
defaults.extend(s)
for m in maps:
curr = self.indexHash.get(tuple(m), [])
curr.append((defaults, indexObject))
self.indexHash[tuple(m)] = curr
elif (node.localName == 'map'):
attrs = []
for c in node.childNodes:
if (c.nodeType == elementType and c.localName == 'attr'):
short = c.getAttribute('set')
short = short.lower()
if not short:
short = 'bib1'
oid = self.prefixes.get(short, None)
if not oid:
raise ConfigFileException(
'No mapping for attribute set %s in Zeerex file' %
short)
type = c.getAttribute('type')
if not type:
raise ConfigFileException(
'No type attribute for Z39.50 attr mapping in Zeerex'
)
type = int(type)
if type < 1:
raise ConfigFileException(
'Invalid type attribute for Z39.50 attr mapping in Zeerex: %s'
% type)
attrVal = flattenTexts(c).strip()
if not attrVal:
raise ConfigFileException(
'No value given for attribute in Z39.50 mapping')
if attrVal.isdigit():
attrVal = int(attrVal)
else:
attrVal = attrVal.lower()
attrs.append((oid, type, attrVal))
return attrs
elif (node.localName == 'default'):
dtype = node.getAttribute('type')
pname = "default" + dtype[0].capitalize() + dtype[1:]
if dtype == 'attribute':
# Get map instead of text
for c in node.childNodes:
if c.nodeType == elementType and c.localName == 'map':
data = self._walkZeeRex(session, c)
else:
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
elif data.lower() in self.prefixes.keys():
data = self.prefixes[data.lower()]
setattr(self, pname, data)
elif (node.localName == 'setting'):
dtype = node.getAttribute('type')
data = flattenTexts(node)
if (data.isdigit()):
data = int(data)
elif data == 'false':
data = 0
elif data == 'true':
data = 1
setattr(self, dtype, data)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
def _walkZeeRex(self, session, node):
if node.localName in ["databaseInfo", "metaInfo"]:
# Ignore
return
elif node.localName == "serverInfo":
self.version = node.getAttribute("version")
for c in node.childNodes:
self._walkZeeRex(session, c)
elif node.localName == "database":
self.databaseUrl = str(flattenTexts(node))
elif node.localName == "host":
self.host = str(flattenTexts(node))
elif node.localName == "port":
self.port = int(flattenTexts(node))
elif node.localName == "schema":
id = node.getAttribute("identifier")
name = node.getAttribute("name")
xsl = node.getAttributeNS(self.c3Namespace, "transformer")
if xsl:
txr = self.get_object(session, xsl)
if txr is None:
raise ConfigFileException("No transformer to map to for " "%s" % (xsl))
self.transformerHash[id] = txr
self.recordNamespaces[name] = id
elif node.localName == "set":
name = node.getAttribute("name")
uri = node.getAttribute("identifier")
if name in self.prefixes and uri != self.prefixes[name]:
raise ConfigFileException("Multiple URIs bound to same short " "name: %s -> %s" % (name, uri))
self.prefixes[str(name)] = str(uri)
elif node.localName == "index":
# Process indexes
idxName = node.getAttributeNS(self.c3Namespace, "index")
indexObject = self.get_object(session, idxName)
if indexObject is None:
raise (ConfigFileException("[%s] No Index to map to for " "%s" % (self.id, idxName)))
maps = []
for c in node.childNodes:
if c.nodeType == elementType and c.localName == "map":
maps.append(self._walkZeeRex(session, c))
for m in maps:
self.indexHash[m] = indexObject
# Need to generate all relations and modifiers
for c in node.childNodes:
if c.nodeType == elementType and c.localName == "configInfo":
for c2 in c.childNodes:
if c2.nodeType == elementType and c2.localName == "supports":
idxName2 = c2.getAttributeNS(self.c3Namespace, "index")
if not idxName2:
indexObject2 = indexObject
else:
indexObject2 = self.get_object(session, idxName2)
if indexObject2 is None:
raise ConfigFileException("[%s] No Index to map to for " "%s" % (self.id, idxName2))
st = str(c2.getAttribute("type"))
val = str(flattenTexts(c2))
for m in maps:
self.indexHash[(m[0], m[1], (st, val))] = indexObject2
elif node.localName == "map":
for c in node.childNodes:
if c.nodeType == elementType and c.localName == "name":
short = c.getAttribute("set")
index = flattenTexts(c)
index = index.lower()
uri = self.resolvePrefix(short)
if not uri:
raise ConfigFileException("No mapping for %s in " "Zeerex" % (short))
return (str(uri), str(index))
elif node.localName == "default":
dtype = node.getAttribute("type")
# XXX: would dtype.title() be nicer!?
pname = "default" + dtype[0].capitalize() + dtype[1:]
data = flattenTexts(node)
if data.isdigit():
data = int(data)
elif data == "false":
data = 0
elif data == "true":
data = 1
setattr(self, pname, data)
elif node.localName == "setting":
dtype = node.getAttribute("type")
data = flattenTexts(node)
if data.isdigit():
data = int(data)
elif data == "false":
data = 0
elif data == "true":
data = 1
setattr(self, dtype, data)
elif node.localName == "supports":
stype = node.getAttribute("type")
if stype in ["extraData", "extraSearchData", "extraScanData", "extraExplainData", "extension"]:
# function, xnType, sruName
xn = node.getAttributeNS(self.c3Namespace, "type")
if not xn in ["record", "term", "searchRetrieve", "scan", "explain", "response"]:
raise ConfigFileException("Unknown extension type %s" % xn)
sru = node.getAttributeNS(self.c3Namespace, "sruName")
fn = node.getAttributeNS(self.c3Namespace, "function")
data = flattenTexts(node)
data = data.strip()
data = tuple(data.split(" "))
if fn.find(".") > -1:
# new version
try:
fn = dynamic.importObject(session, fn)
except ImportError:
raise ConfigFileException("Cannot find handler " "function %s (in %s)" % (fn, repr(sys.path)))
self.sruExtensionMap[sru] = (xn, fn, data)
else:
if hasattr(srwExtensions, fn):
fn = getattr(srwExtensions, fn)
else:
raise ConfigFileException("Cannot find handler " "function %s in " "srwExtensions." % fn)
xform = node.getAttributeNS(self.c3Namespace, "sruFunction")
if not xform:
sruFunction = srwExtensions.simpleRequestXform
elif hasattr(srwExtensions, xform):
sruFunction = getattr(srwExtensions, xform)
else:
raise ConfigFileException(
"Cannot find transformation " "function %s in " "srwExtensions." % xform
)
hashAttr = xn + "ExtensionHash"
curr = getattr(self, hashAttr)
curr[data] = fn
setattr(self, hashAttr, curr)
self.sruExtensionMap[sru] = (data[0], data[1], sruFunction)
else:
for c in node.childNodes:
if c.nodeType == elementType:
self._walkZeeRex(session, c)
