def read_pgf(pgf_file, graph=None, pickle_graph=False):
"""
Import graph from Graph Python Format file
PGF format is the modules own file format consisting of a serialized
graph data, nodes and edges dictionaries. Import either as plain text
serialized dictionary or pickled graph object.
The format is feature rich with good performance but is not portable.
:param pgf_file: PGF data to parse
:type pgf_file: File, string, stream or URL
:param graph: Graph object to import to or Graph by default
:type graph: :graphit:Graph
:param pickle_graph: PGF format is a pickled graph
:type pickle_graph: :py:bool
:return: Graph instance
:rtype: :graphit:Graph
"""
# Unpickle pickled PGF format
if pickle_graph:
pgf_file = open_anything(pgf_file, mode='rb')
pgraph = pickle.load(pgf_file)
# Transfer data from unpickled graph to graph if defined
if graph:
graph.origin.nodes, graph.origin.edges, graph.origin.adjacency, graph.origin.data = graph.storagedriver(
pgraph.nodes, pgraph.edges, pgraph.data)
return graph
return pgraph
pgf_file = open_anything(pgf_file)
# Import graph from serialized Graph Python Format
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
pgf_eval = ast.literal_eval(pgf_file.read())
if not isinstance(pgf_eval, dict):
raise GraphitException('Invalid PGF file format')
missing_data = [d for d in pgf_file if d not in ('data', 'nodes', 'edges')]
if missing_data:
raise GraphitException(
'Invalid PGF file format, missing required attributes: {0}'.format(
','.join(missing_data)))
graph.origin.nodes, graph.origin.edges, graph.origin.adjacency, graph.origin.data = graph.storagedriver(
pgf_eval['nodes'], pgf_eval['edges'], pgf_eval['data'])
return graph
def read_json(json_file, graph=None, **kwargs):
"""
Parse (hierarchical) JSON data structure to a graph
Use the default Python json parser to parse the JSON file to a dictionary
followed by io_dict_format.read_pydata to parse to a graph structure.
Additional keyword arguments (kwargs) are passed to `read_pydata`
:param json_file: json data to parse
:type json_file: File, string, stream or URL
:param graph: Graph object to import dictionary data in
:type graph: :graphit:Graph
:return: GraphAxis object
:rtype: :graphit:GraphAxis
"""
# Try parsing the string using default Python json parser
json_file = open_anything(json_file)
try:
json_file = json.load(json_file)
except IOError:
logger.error('Unable to decode JSON string')
return
return read_pydata(json_file, graph=graph, **kwargs)
def read_xml(xml_file, graph=None):
"""
Parse hierarchical XML data structure to a graph
Uses the Python build-in etree cElementTree parser to parse the XML
document and convert the elements into nodes.
The XML element tag becomes the node key, XML text becomes the node
value and XML attributes are added to the node as additional attributes.
:param xml_file: XML data to parse
:type xml_file: File, string, stream or URL
:param graph: Graph object to import dictionary data in
:type graph: :graphit:Graph
:return: GraphAxis object
:rtype: :graphit:GraphAxis
"""
# User defined or default GraphAxis object
if graph is None:
graph = GraphAxis()
if not isinstance(graph, GraphAxis):
raise TypeError('Unsupported graph type {0}'.format(type(graph)))
# Try parsing the string using default Python cElementTree parser
xml_file = open_anything(xml_file)
try:
tree = et.fromstring(xml_file.read())
except et.ParseError as error:
logging.error(
'Unable to parse XML file. cElementTree error: {0}'.format(error))
return
def walk_element_tree(element, parent=None):
for child in element:
child_data = child.attrib
if child.text and len(child.text.strip()):
child_data[graph.value_tag] = child.text.strip()
nid = graph.add_node(child.tag, **child_data)
graph.add_edge(parent, nid)
walk_element_tree(child, parent=nid)
is_empty = graph.empty()
# Add root element
element_data = tree.attrib
if tree.text and len(tree.text.strip()):
element_data[graph.value_tag] = tree.text.strip()
rid = graph.add_node(tree.tag, **element_data)
if is_empty:
graph.root = rid
# Recursive add XML elements as nodes
walk_element_tree(tree, parent=graph.root)
return graph
def read_gml(gml, graph=None):
"""
Read graph in GML format
:param gml: GML graph data.
:type gml: File, string, stream or URL
:param graph: Graph object to import GML data in
:type graph: :graphit:Graph
:return: Graph object
:rtype: :graphit:Graph
"""
# User defined or default Graph object
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(type(graph)))
# Parse GML into nested structure of Record class instances
gml_stream = StreamReader(open_anything(gml))
records = Record(gml_stream, name='root')
gml_graphs = [g for g in records if g.name == 'graph']
if len(gml_graphs) > 1:
logging.warning("GML file contains {0} 'graph' objects. Only parse first".format(len(gml_graphs)))
gml_graph_record = gml_graphs[0]
# GML node and edge labels are unique, turn off auto_nid
graph.data['auto_nid'] = False
# Set graph meta-data and attributes
graph_attr = gml_graph_record.to_dict({})
graph.directed = True
if 'directed' in graph_attr:
directed = graph_attr.pop('directed')
graph.directed = True if directed == 1 else False
graph.data['directed'] = graph.directed
graph.data.update(graph_attr)
# Build graph from records
build_nodes(graph, gml_graph_record)
build_edges(graph, gml_graph_record)
return graph
def read_yaml(yaml_file, graph=None, **kwargs):
"""
Parse (hierarchical) YAML data structure to a graph
Additional keyword arguments (kwargs) are passed to `read_pydata`
:param yaml_file: yaml data to parse
:type yaml_file: File, string, stream or URL
:param graph: Graph object to import dictionary data in
:type graph: :graphit:Graph
:return: GraphAxis object
:rtype: :graphit:GraphAxis
"""
# Try parsing the string using default Python yaml parser
yaml_file = open_anything(yaml_file)
try:
yaml_file = yaml.safe_load(yaml_file)
except IOError:
logger.error('Unable to decode YAML string')
return
if not isinstance(yaml_file, list):
yaml_file = [yaml_file]
base_graph = read_pydata(yaml_file[0], graph=graph, **kwargs)
for yaml_object in yaml_file[1:]:
sub_graph = read_pydata(yaml_object)
# If sub-graph root is of type 'root', connect children to base_graph
root = sub_graph.getnodes(sub_graph.root)
if root[sub_graph.key_tag] == 'root':
links = [(base_graph.root, child)
for child in root.children(return_nids=True)]
else:
links = [(base_graph.root, sub_graph.root)]
graph_join(base_graph, sub_graph, links=links)
return base_graph
def read_adl(adl_file, graph=None):
"""
Construct a graph from a adjacency list (ADL)
.. note:: the directionality of the graph is not defined explicitly
in the adjacency list and thus depends on the graph.directional
attribute that is False (undirectional) by default.
:param adl_file: ADL graph data.
:type adl_file: File, string, stream or URL
:param graph: Graph object to import ADL data in
:type graph: :graphit:Graph
:return: Graph object
:rtype: :graphit:Graph
"""
adl_file = open_anything(adl_file)
# User defined or default Graph object
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(type(graph)))
# ADL node labels are unique, turn off auto_nid
graph.data['auto_nid'] = False
for line in adl_file.readlines():
# Ignore comments (# ..)
line = line.split('#')[0].strip()
if line:
nodes = line.split()
graph.add_nodes(nodes)
if len(nodes) > 1:
graph.add_edges([(nodes[0], n) for n in nodes[1:]])
return graph
def read_tgf(tgf, graph=None, key_tag=None):
"""
Read graph in Trivial Graph Format
TGF format dictates that nodes to be listed in the file first with each
node on a new line. A '#' character signals the end of the node list and
the start of the edge list.
Node and edge ID's can be integers, float or strings. They are parsed
automatically to their most likely format.
Simple node and edge labels are supported in TGF as all characters that
follow the node or edge ID's. They are parsed and stored in the Graph
node and edge data stores using the graphs default or custom 'key_tag'.
TGF data is imported into a default Graph object if no custom Graph
instance is provided. The graph behaviour and the data import process is
influenced and can be controlled using a (custom) Graph class.
.. note:: TGF format always defines edges in a directed fashion.
This is enforced even for custom graphs.
:param tgf: TGF graph data.
:type tgf: File, string, stream or URL
:param graph: Graph object to import TGF data in
:type graph: :graphit:Graph
:param key_tag: Data key to use for parsed node/edge labels.
:type key_tag: :py:str
:return: Graph object
:rtype: :graphit:Graph
"""
tgf_file = open_anything(tgf)
if not isinstance(graph, Graph):
graph = Graph()
# Define node/edge data labels
if key_tag:
graph.key_tag = key_tag
# TGF defines edges in a directed fashion. Enforce but restore later
default_directionality = graph.directed
graph.directed = True
# TGF node and edge labels are unique, turn off auto_nid
graph.auto_nid = False
# Start parsing. First extract nodes
nodes = True
node_dict = {}
for line in tgf_file.readlines():
line = line.strip()
if len(line):
# Reading '#' character means switching from node
# definition to edges
if line.startswith('#'):
nodes = False
continue
# Coarse string to types
line = [coarse_type(n) for n in line.split()]
# Parse nodes
if nodes:
attr = {}
# Has node data
if len(line) > 1:
attr = {graph.key_tag: ' '.join(line[1:])}
nid = graph.add_node(line[0], **attr)
node_dict[line[0]] = nid
# Parse edges
else:
e1 = node_dict[line[0]]
e2 = node_dict[line[1]]
attr = {}
# Has edge data
if len(line) > 2:
attr = {graph.key_tag: ' '.join(line[2:])}
graph.add_edge(e1, e2, **attr)
tgf_file.close()
# Restore directionality
graph.directed = default_directionality
return graph
def read_jgf(jgf_format, graph=None):
"""
Read JSON graph format (.jgf)
This is a propitiatory format in which the graph meta-data, the nodes,
edges and their data dictionaries are stored in JSON format.
Format description. Primary key/value pairs:
* graph: Graph class meta-data. Serializes all class attributes of type
int, float, bool, long, str or unicode.
* nodes: Graph node identifiers (keys) and attributes (values)
* edges: Graph enumerated edge identifiers
* edge_attr: Graph edge attributes
:param jgf_format: JSON encoded graph data to parse
:type jgf_format: :py:str
:param graph: Graph object to import TGF data in
:type graph: :graphit:Graph
:return: Graph object
:rtype: Graph or GraphAxis object
"""
# Try parsing the string using default Python json parser
if isinstance(jgf_format, dict):
parsed = jgf_format
else:
jgf_format = open_anything(jgf_format)
try:
parsed = json.load(jgf_format)
except IOError:
logger.error('Unable to decode JSON string')
return
# Check graphit version and format validity
if not check_graphit_version(parsed['data'].get('graphit_version')):
return
keywords = ['graph', 'data', 'nodes', 'edges', 'edge_attr']
if not set(keywords).issubset(set(parsed.keys())):
logger.error('JSON format does not contain required graph data')
return
# User defined or default Graph object
if graph is None:
if parsed['graph'].get('root') is not None:
graph = GraphAxis(data=parsed['data'])
else:
graph = Graph(data=parsed['data'])
elif not isinstance(graph, (Graph, GraphAxis)):
raise GraphitException('Unsupported graph type {0}'.format(type(graph)))
# Init graph meta-data attributes
for key, value in parsed['graph'].items():
setattr(graph, key, value)
# Init graph nodes
for node_key, node_value in parsed['nodes'].items():
# JSON objects don't accept integers as dictionary keys
# If graph.auto_nid equals True, course node_key to integer
if graph.data.auto_nid:
node_key = int(node_key)
graph.nodes[node_key] = node_value
# Init graph edges
for edge_key, edge_value in parsed['edges'].items():
edge_value = tuple(edge_value)
graph.edges[edge_value] = parsed['edge_attr'].get(edge_key, {})
# Set auto nid
graph._set_auto_nid()
return graph
def read_json_schema(schema, graph=None, exclude_args=None, resolve_ref=True):
"""
Import hierarchical data structures defined in a JSON schema format
:param schema: JSON Schema data format to import
:type schema: dict, file, string, stream or URL
:param graph: graph object to import TGF data in
:type graph: :graphit:Graph
:param exclude_args: JSON schema arguments to exclude from import
:type exclude_args: :py:list
:param resolve_ref: Parse JSON schema 'definitions'
:type resolve_ref: :py:bool
:return: Graph object
:rtype: :graphit:Graph
"""
json_schema = schema
if not isinstance(schema, dict):
# Try parsing the string using default Python json parser
json_schema = open_anything(schema)
try:
json_schema = json.load(json_schema)
except (IOError, ValueError) as error:
logger.error('Unable to decode JSON string: {0}'.format(error))
return
# User defined or default Graph object
if graph is None:
graph = GraphAxis()
elif not isinstance(graph, GraphAxis):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
if graph.empty():
rid = graph.add_node('root')
graph.root = rid
# Build JSON schema parser ORM with format specific conversion classes
graph.node_tools = JSONSchemaValidatorDraft07
graph.orm = JSONSchemaORMDraft07
# What data-blocks to parse, properties by default, definitions if required
datablock = ['properties']
if resolve_ref:
datablock.append('definitions')
if exclude_args is None:
exclude_args = []
def walk_schema(schema_block, parent=None):
# Get all JSON schema definitions for this data instance
attributes = dict([(k, v) for k, v in schema_block.items()
if not isinstance(v, dict) and k not in exclude_args
])
node = graph.getnodes(parent)
node.update(attributes)
# Get 'required' attribute
required = schema_block.get('required', [])
if not isinstance(required, list):
required = []
# Store default data or None
if attributes.get('default') is not None:
node.set(graph.data.value_tag, attributes.get('default'))
# For all child elements in datablock, make new node
# and parse using recursive calls to parse_schema
for block in schema_block.keys():
if block in datablock:
for child, attr in schema_block[block].items():
nid = graph.add_node(child)
# Register block_name in child attributes
attr['schema_label'] = block
# Register 'required' elements
if child in required:
attr['required'] = True
graph.add_edge(parent, nid)
walk_schema(attr, parent=nid)
walk_schema(json_schema, graph.root)
# Parse schema meta data
document_path = ''
if isinstance(schema, PY_STRING):
document_path = os.path.abspath(schema)
root = graph.get_root()
root.set('document_path', document_path)
parse_schema_meta_data(root)
# Resolve JSON Schema $ref
if resolve_ref:
resolve_json_ref(graph, exclude_args=exclude_args)
return graph
def read_web(web,
graph=None,
orm_data_tag='haddock_type',
auto_parse_format=True):
"""
Import hierarchical data structures defined in the Spider .web format
The data block type identifiers used in the .web format are stored in
the nodes using the `orm_data_tag` attribute. These can be used by the
Graph ORM mapper for custom data exchange in the graph.
:param web: Spider .web data format to import
:type web: file, string, stream or URL
:param graph: graph object to import TGF data in
:type graph: :graphit:Graph
:param orm_data_tag: data key to use for .web data identifier
:type orm_data_tag: :py:str
:param auto_parse_format: automatically detect basic format types using JSON decoding
:type auto_parse_format: :py:bool
:return: Graph object
:rtype: :graphit:Graph
"""
web_file = open_anything(web)
if graph is None:
graph = GraphAxis()
elif not isinstance(graph, GraphAxis):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
# Build .web parser ORM with format specific conversion classes
weborm = GraphORM()
weborm.node_mapping.add(
RestraintsInterface,
lambda x: x.get(graph.data.key_tag) == 'activereslist')
weborm.node_mapping.add(
RestraintsInterface,
lambda x: x.get(graph.data.key_tag) == 'passivereslist')
# Set current ORM aside and register parser ORM.
curr_orm = graph.orm
graph.orm = weborm
curr_obj_nid = None
object_open_tags = 0
object_close_tags = 0
array_key_counter = 1
array_store = []
for i, line in enumerate(web_file.readlines()):
line = line.strip()
if len(line):
# Detect start of new object definition
if line.endswith('('):
# Process data
meta_data = [n.strip() for n in line.strip('(').split('=', 1)]
ddict = {orm_data_tag: meta_data[-1], 'is_array': False}
if len(meta_data) > 1:
node_key = meta_data[0]
else:
node_key = 'item{0}'.format(array_key_counter)
ddict['is_array'] = True
array_key_counter += 1
# Clear the array store
array_store = []
# First object defines graph root
if graph.empty():
curr_obj_nid = graph.add_node(node_key, **ddict)
graph.root = curr_obj_nid
# Add new object as child of current object
else:
child_obj_nid = graph.add_node(node_key, **ddict)
graph.add_edge(curr_obj_nid, child_obj_nid)
curr_obj_nid = child_obj_nid
object_open_tags += 1
# Detect end of object definition
elif line.startswith(')'):
# If there is data in the array store, add it to node
if len(array_store):
array_node = graph.getnodes(curr_obj_nid)
array_node.is_array = True
array_node.set(graph.data.value_tag, array_store)
# Reset array key counter
array_key_counter = 1
# Move one level up the object three
curr_obj_nid = node_parent(graph, curr_obj_nid,
graph.root) or graph.root
object_close_tags += 1
# Parse object parameters
else:
# Parse key,value pairs and add as leaf node
params = [n.strip() for n in line.rstrip(',').split('=', 1)]
if '=' in line and len(params) == 2:
leaf_nid = graph.add_node(params[0])
graph.add_edge(curr_obj_nid, leaf_nid)
value = params[1]
if auto_parse_format:
value = json_decode_params(params[1])
leaf_node = graph.getnodes(leaf_nid)
leaf_node.set(graph.data.value_tag, value)
# Parse single values as array data
elif len(params) == 1:
value = params[0]
if auto_parse_format:
value = json_decode_params(params[0])
# Store array items as nodes
array_store.append(value)
else:
logger.warning(
'Unknown .web data formatting on line: {0}, {1}'.
format(i, line))
web_file.close()
# Object blocks opening '(' and closing ')' tag count should be balanced
if object_open_tags != object_close_tags:
raise AssertionError(
'Unbalanced object block, something is wrong with the file format')
# Restore original ORM
graph.orm = curr_orm
# Root is of type 'array', rename key from 'item1' to 'project'
root = graph.getnodes(graph.root)
root.key = 'project'
return graph
def read_dot(dot, graph=None):
"""
Read graph in DOT format
:param dot: DOT graph data.
:type dot: File, string, stream or URL
:param graph: Graph object to import DOT data in
:type graph: :graphit:Graph
:return: Graph object
:rtype: :graphit:Graph
"""
dot_stream = StreamReader(open_anything(dot))
# User defined or default Graph object
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(type(graph)))
block = None
node_attr = {}
edges = []
nodes = []
parse = True
init_graph = False
while parse or dot_stream.has_more:
# Parse start graph block
if not init_graph:
graph = parse_graph_type(dot_stream, graph)
if graph is not None:
init_graph = True
# Parse up to DOT reserved chars
line, char = dot_stream.read_upto_char(('\n', ';', '[', ']', '}'))
line = line.strip() if line else ''
if line:
# Comment line
if line[0] in ('/', '#'):
logging.info('Skip DOT comment line: "{0}"'.format(line))
# Read till end of line
if char != '\n':
dot_stream.read_upto_char('\n')
# Grouping not supported
elif line[0] == '{':
nxt_line, nxt_char = dot_stream.read_upto_char('}')
logging.info('Skip group: {0}{1}{2}{3}'.format(line, char, nxt_line, nxt_char))
# Subgraphs
elif 'subgraph' in line:
block = 'subgraph'
node_attr[block] = shlex.split(line)[1]
# Node attribute block
elif 'node' in line:
block = 'node'
node_attr = {}
# Parse edges
elif '--' in line or '->' in line:
attr = {}
if char == '[':
attr = parse_attributes(dot_stream.read_upto_char(']')[0])
edges.extend(parse_edge(line, graph, attr=attr))
else:
if '=' in line:
if block in ('subgraph', 'node'):
node_attr.update(parse_attributes(line))
else:
graph.data.update(parse_attributes(line))
else:
nodes.extend(parse_nodes(line, graph))
elif (char == '}' and block == 'subgraph') or block == 'node':
logging.info('Stop parsing {0} group at position: {1}'.format(block, dot_stream.block_pos[1]))
nodes.extend(list(set(sum(edges, ()))))
for node in nodes:
graph.nodes[node].update(node_attr)
node_attr = {}
edges = []
nodes = []
block = None
else:
parse = False
return graph
def read_lgf(lgf, graph=None):
"""
Read graph in LEMON Graph Format (LGF)
:param lgf: LGF graph data.
:type lgf: File, string, stream or URL
:param graph: Graph object to import LGF data in
:type graph: :graphit:Graph
:return: Graph object
:rtype: :graphit:Graph
"""
lgf_file = open_anything(lgf)
# User defined or default Graph object
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
# LGF node and edge labels are unique, turn off auto_nid
graph.data['auto_nid'] = False
parser = None
header = None
did_parse_nodes = False
is_directed = False
for line in lgf_file.readlines():
line = line.strip()
# Skip empty lines and comment lines
if not len(line) or line.startswith('#'):
parser = None
continue
# Define parser
if line.startswith('@') or parser is None:
if 'nodes' in line:
parser = parse_nodes
did_parse_nodes = True
elif line.startswith('@edges'):
parser = parse_edges
elif line.startswith('@arcs'):
parser = parse_arcs
is_directed = True
elif line.startswith('@attributes'):
logging.warning(
'Not importing LGF @attributes. Graph attributes not supported by graphit'
)
header = None
continue
# Immediately after parser definition, parse table column headers
if header is None:
header = split_line(line)
continue
parser(line, header, graph, did_parse_nodes=did_parse_nodes)
# Set graph to 'directed' if arcs where parsed
if is_directed:
graph.directed = True
return graph
def read_p2g(p2g_file, graph=None):
"""
Read graph in P2G format
:param p2g_file: P2G data to parse
:type p2g_file: File, string, stream or URL
:param graph: Graph object to import to or Graph by default
:type graph: :graphit:Graph
:return: Graph instance
:rtype: :graphit:Graph
"""
p2g_file = open_anything(p2g_file)
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
# P2G graphs are directed
graph.directed = True
graph_name = None
graph_layout = None
curr_node = None
nodes = {}
for i, line in enumerate(p2g_file.readlines()):
line = line.strip()
if line:
# Parse p2g graph name (first line)
sline = line.split()
if not graph_name:
graph_name = line
continue
# Parse number of nodes and edges (second line)
elif not graph_layout:
try:
graph_layout = map(int, sline)
except ValueError:
raise GraphitException(
'P2G import error: line {0} - {1}'.format(i, line))
continue
# Parse nodes and edges
if len(sline) == 1:
nodes[line] = []
curr_node = line
elif len(sline) == 2:
try:
nodes[curr_node] = map(int, sline)
except ValueError:
raise GraphitException(
'P2G import error: malformed edge on line {0} - {1}'.
format(i, line))
else:
raise GraphitException(
'P2G import error: line {0} - {1}'.format(i, line))
graph.data['name'] = graph_name
# Add nodes
mapped_nodes = graph.add_nodes(nodes.keys())
# Add edges
for i, nid in enumerate(nodes.keys()):
for e in nodes[nid]:
if e < len(mapped_nodes):
graph.add_edge(mapped_nodes[i], mapped_nodes[e])
else:
raise GraphitException(
'P2G import error: edge node index {0} not in graph'.
format(e))
if len(nodes) != graph_layout[0] or (len(graph.edges)) != graph_layout[1]:
logging.warning(
'P2G import warning: declared number of nodes and edges {0}-{1} does not match {2}-{3}'
.format(graph_layout[0], graph_layout[1], len(nodes),
len(graph.edges)))
return graph
def read_lgr(lgr, graph=None, edge_label='label'):
"""
Read graph in LEDA format
Nodes are added to the graph using a unique ID or with the node data
as label depending if the graph.data.auto_nid is True or False.
Edge data is added to the edge attributes using `edge_label` as key.
The data types for both nodes and edges is set according to the
specifications in the LEDA header as either string, int, float or bool.
:param lgr: LEDA graph data.
:type lgr: File, string, stream or URL
:param graph: Graph object to import LEDA data in
:type graph: :graphit:Graph
:param edge_label: edge data label name
:type edge_label: :py:str
:return: Graph object
:rtype: :graphit:Graph
:raises: TypeError if node/edge type conversion failed
GraphitException in case of malformed LEDA file
"""
# User defined or default Graph object
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
# Parse LEDA file
lgr_file = open_anything(lgr)
header = []
nodes = []
edges = []
container = header
for line in lgr_file.readlines():
line = line.strip()
if line:
if line.startswith('#header'):
container = header
continue
if line.startswith('#nodes'):
container = nodes
continue
if line.startswith('#edges'):
container = edges
continue
container.append(line)
# Parse LEDA header
if not header[0] == 'LEDA.GRAPH':
raise GraphitException('File is not a valid LEDA graph format')
# Node and edge data types and graph directionality
node_type = data_types.get(header[1])
edge_type = data_types.get(header[2])
graph.directed = int(header[3]) == -1
# Parse LEDA nodes
node_mapping = {}
for i, node in enumerate(nodes[1:], start=1):
data = node.strip('|{}|') or None
if node_type and data:
data = node_type(data)
nid = graph.add_node(data)
node_mapping[i] = nid
# Parse LEDA edges
for edge in edges[1:]:
try:
source, target, reversal, label = edge.split()
except ValueError:
raise GraphitException(
'Too few fields in LEDA edge {0}'.format(edge))
attr = {edge_label: label.strip('|{}|') or None}
if edge_type and attr[edge_label]:
attr[edge_label] = edge_type(attr[edge_label])
graph.add_edge(node_mapping[int(source)], node_mapping[int(target)],
**attr)
return graph
def read_gexf(gexf_file, graph=None):
"""
Read graphs in GEXF format
Uses the Python build-in etree cElementTree parser to parse the XML
document and convert the elements into nodes.
The XML element tag becomes the node key, XML text becomes the node
value and XML attributes are added to the node as additional attributes.
:param gexf_file: XML data to parse
:type gexf_file: File, string, stream or URL
:param graph: Graph object to import dictionary data in
:type graph: :graphit:Graph
:return: GraphAxis object
:rtype: :graphit:GraphAxis
"""
gexf_file = open_anything(gexf_file)
# User defined or default Graph object
if graph is None:
graph = Graph()
elif not isinstance(graph, Graph):
raise GraphitException('Unsupported graph type {0}'.format(
type(graph)))
# Try parsing the string using default Python cElementTree parser
try:
tree = et.fromstring(gexf_file.read())
except et.ParseError as error:
logging.error(
'Unable to parse GEXF file. cElementTree error: {0}'.format(error))
return
# Get XMLNS namespace from root
xmlns = None
for elem in tree.iter():
if elem.tag.endswith('gexf'):
xmlns = elem.tag.split('}')[0] + '}'
break
if xmlns is None:
raise GraphitException(
'Invalid GEXF file format, "gexf" tag not found')
# Add graph meta-data and XMLNS namespace
for meta in tree.iter('{0}meta'.format(xmlns)):
graph.data.update(meta.attrib)
for meta_data in meta:
tag = meta_data.tag.split('}')[1]
graph.data[tag] = meta_data.text
# GEXF node and edge labels are unique, turn off auto_nid
graph.data['auto_nid'] = False
graph_tag = tree.find('{0}graph'.format(xmlns))
graph.directed = graph_tag.get('defaultedgetype', 'directed') == 'directed'
graph.data.update(graph_tag.attrib)
# Parse all nodes
nodes = tree.findall('.//{0}node'.format(xmlns))
if not len(nodes):
raise GraphitException('GEXF file containes no "node" elements')
for node in nodes:
attr = node.attrib
attr = parse_attvalue_elements(node, attr, xmlns=xmlns)
graph.add_node(attr['id'],
**dict([n for n in attr.items() if n[0] != 'id']))
# Parse all edges
edges = tree.findall('.//{0}edge'.format(xmlns))
for edge in edges:
attr = edge.attrib
# Edge direction differs from global graph directionality
edge_directed = graph.directed
if 'type' in attr:
edge_directed = attr['type'] == 'directed'
attr = parse_attvalue_elements(edge, attr, xmlns=xmlns)
graph.add_edge(attr['source'],
attr['target'],
directed=edge_directed,
**dict([
n for n in attr.items()
if n[0] not in ('source', 'target')
]))
logger.info('Import graph in GEXF format. XMLNS: {0}'.format(xmlns))
return graph
