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_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_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