1#!/usr/bin/env python3.8
2
3""" Convert a grammar into a dot-file suitable for use with GraphViz
4
5    For example:
6        Generate the GraphViz file:
7        # scripts/grammar_grapher.py data/python.gram > python.gv
8
9        Then generate the graph...
10
11        # twopi python.gv -Tpng > python_twopi.png
12
13        or
14
15        # dot python.gv -Tpng > python_dot.png
16
17        NOTE: The _dot_ and _twopi_ tools seem to produce the most useful results.
18              The _circo_ tool is the worst of the bunch. Don't even bother.
19"""
20
21import argparse
22import sys
23
24from typing import Any, List
25
26sys.path.insert(0, ".")
27
28from pegen.build import build_parser
29from pegen.grammar import (
30    Alt,
31    Cut,
32    Forced,
33    Grammar,
34    Group,
35    Leaf,
36    Lookahead,
37    Rule,
38    NameLeaf,
39    NamedItem,
40    Opt,
41    Repeat,
42    Rhs,
43)
44
45argparser = argparse.ArgumentParser(
46    prog="graph_grammar",
47    description="Graph a grammar tree",
48)
49argparser.add_argument(
50    "-s",
51    "--start",
52    choices=["exec", "eval", "single"],
53    default="exec",
54    help="Choose the grammar's start rule (exec, eval or single)",
55)
56argparser.add_argument("grammar_file", help="The grammar file to graph")
57
58
59def references_for_item(item: Any) -> List[Any]:
60    if isinstance(item, Alt):
61        return [_ref for _item in item.items for _ref in references_for_item(_item)]
62    elif isinstance(item, Cut):
63        return []
64    elif isinstance(item, Forced):
65        return references_for_item(item.node)
66    elif isinstance(item, Group):
67        return references_for_item(item.rhs)
68    elif isinstance(item, Lookahead):
69        return references_for_item(item.node)
70    elif isinstance(item, NamedItem):
71        return references_for_item(item.item)
72
73    # NOTE NameLeaf must be before Leaf
74    elif isinstance(item, NameLeaf):
75        if item.value == "ENDMARKER":
76            return []
77        return [item.value]
78    elif isinstance(item, Leaf):
79        return []
80
81    elif isinstance(item, Opt):
82        return references_for_item(item.node)
83    elif isinstance(item, Repeat):
84        return references_for_item(item.node)
85    elif isinstance(item, Rhs):
86        return [_ref for alt in item.alts for _ref in references_for_item(alt)]
87    elif isinstance(item, Rule):
88        return references_for_item(item.rhs)
89    else:
90        raise RuntimeError(f"Unknown item: {type(item)}")
91
92
93def main() -> None:
94    args = argparser.parse_args()
95
96    try:
97        grammar, parser, tokenizer = build_parser(args.grammar_file)
98    except Exception as err:
99        print("ERROR: Failed to parse grammar file", file=sys.stderr)
100        sys.exit(1)
101
102    references = {}
103    for name, rule in grammar.rules.items():
104        references[name] = set(references_for_item(rule))
105
106    # Flatten the start node if has only a single reference
107    root_node = {"exec": "file", "eval": "eval", "single": "interactive"}[args.start]
108
109    print("digraph g1 {")
110    print('\toverlap="scale";')  # Force twopi to scale the graph to avoid overlaps
111    print(f'\troot="{root_node}";')
112    print(f"\t{root_node} [color=green, shape=circle];")
113    for name, refs in references.items():
114        for ref in refs:
115            print(f"\t{name} -> {ref};")
116    print("}")
117
118
119if __name__ == "__main__":
120    main()
121