uml graph creator

scripts/plantuml_parser.py

@@ -0,0 +1,81 @@
+import networkx as nx
+
+"""
+This script processes files produced by pyreverse.
+Get pyreverse by installing pylint: ``pip install pylint``.
+
+The particular files we process are generated by a command like
+    ``pyreverse  -k -d <path> -o plantuml pygsti``
+where <path> is the relative path to the directory that should contain
+the output files.
+
+Notes
+-----
+ 1. The command above will create two files: "classes.plantuml" and "packages.plantuml".
+    This script is only concerned with the first such file.
+
+ 2. The classes.plantuml file contains data for different types of class relations.
+    Most of the data concerns class inheritance. But there's also data on which classes
+    are used as members within other classes. At present, this script only tries to
+    extract inheritance structure.
+"""
+
+
+def plantuml_classes_to_networkx(plantuml_filename, prefix='pygsti'):
+    if prefix[-1] != '.':
+        prefix = prefix + '.'
+    fi = open(plantuml_filename, 'r')
+
+    def strip_readline():
+        return fi.readline().removesuffix('\n').removesuffix('\r')
+
+    def node_test(tup):
+        return len(tup) == 5 and tup[0] == 'class' and tup[2] == 'as' and tup[4] == '{'
+
+    def edge_test(tup):
+        return len(tup) == 3 and prefix in tup[0] and prefix in tup[2]
+
+    # initial file format check
+    tup = strip_readline().split(' ')
+    if len(tup) != 2 or tup[0] != '@startuml' or tup[1] != 'classes':
+        raise ValueError('Unexpected file format.')
+    fi.readline()  # toss
+
+    # read nodes
+    nodes = set()
+    while True:
+        tup = strip_readline().split(' ')
+        if node_test(tup):
+            nodes.add(tup[3].removeprefix(prefix))
+            ell = strip_readline()
+            if ell != '}':
+                raise ValueError('Unexpected file format.')
+        else:
+            if edge_test(tup):
+                break
+            else:
+                raise ValueError('Unexpected file format.')
+
+    # read inheritance edges
+    edges = set()
+    while tup[1] == '--|>':
+        edges.add((tup[0].removeprefix(prefix), tup[2].removeprefix(prefix)))
+        tup = strip_readline().split(' ')
+
+    fi.close()
+
+    G = nx.DiGraph()
+    G.add_nodes_from(nodes)
+    G.add_edges_from(edges)
+    return G
+
+
+if __name__ == '__main__':
+    # We assume this script is run from a directory that contains "classes.plantml".
+    # We save the class inheritance graph in GML format. This format is widely
+    # used, and can be read by the free software called "yEd".
+    G = plantuml_classes_to_networkx('classes.plantuml')
+    nx.write_gml(G, 'pygsti.gml')
+    # Note: once the graph is loaded into yEd it can be saved as a richer format
+    # like GraphML (that records visual positioning data). You can also delete
+    # nodes if they seem conceptually unimportant.