Hierarchical Directories to Dot Graph – Python

Creating a hierarchical dot graph from a list of paths.

The Function

def dir2dot(str_dir_file):
    
    # split lines
    list_dir_lines = str_dir_file.split('\n')
    
    # find the number of layers
    list_line_layers = []
    for str_each_line in list_dir_lines:
        list_l_layer=str_each_line.split('_')
        list_line_layers.append(len(list_l_layer))
    int_total_layers = max(list_line_layers)

    # dictionary of layers
    dict_layers={}
    for int_layer in range(0,int_total_layers):
        dict_layers[int_layer] = set()

    # place stuff in ther dicts
    for str_each_line in list_dir_lines:
        list_line_comps = str_each_line.split('_')
        int_line_layers = len(list_line_comps)

        for int_layer in range(0,int_line_layers):
            if len(list_line_comps[int_layer]) > 0:
                dict_layers[int_layer].add(list_line_comps[int_layer])

    # normalise nodes to add color
    int_node_width = 1.0/int_total_layers

    # all the Nodes section
    list_dot_nodes = []
    for int_layer in range(0,int_total_layers):
        list_layer_nodes = dict_layers[int_layer]
        str_layer = '"\n\t\t"'.join(list_layer_nodes)
        str_dot_format = '\t{node[style=filled,color=grey%d]\n\t\t"%s"\n\t}\n'
        str_node_layer = str_dot_format % (100*(int_node_width*int_layer),
                                            str_layer)
        list_dot_nodes.append(str_node_layer)


    # sort out the edges
    set_dot_edges = set()

    for str_each_line in list_dir_lines:
        list_line_comps = str_each_line.split('_')
        for int_level in range(1,len(list_line_comps)):
            str_from_node = list_line_comps[int_level-1]
            str_to_node = list_line_comps[int_level]
            str_dot_edge = '"%s" -- "%s"' % (str_from_node,
                                            str_to_node)
            set_dot_edges.add(str_dot_edge)

    # output
    str_dot_nodes = '\n'.join(list_dot_nodes)
    str_dot_edges = '\n'.join(set_dot_edges)
    str_dot_format = 'graph{\n%s\n%s\n}' % (str_dot_nodes,str_dot_edges)

    file_w_output = open('cm_dot.dot','w')
    file_w_output.write(str_dot_format)
    file_w_output.close()

Running the Function

# first open the file and read the contents
# as a string
with open('raw_tree_data.txt','r') as file_r_dir_file:
    str_dir_data = file_r_dir_file.read()

# Then feed that string to the function.
dir2dot(str_dir_data)