I can not import dot content> 256 MB due to the string limitation in V8
(according to: https://stackoverflow.com/questions/40382268/error-tostring-failed-node-js-buffer )

My code:

fs.readFile(argv.f, 'utf8', function(err, data) {
 var graph = dot(data);

...

leads to:

dir ./graph_data/FFM/2017-05-26T09:57:20.821Z created
buffer.js:385
    throw new Error('toString failed');
    ^

Error: toString failed
    at Buffer.toString (buffer.js:385:11)
    at FSReqWrap.readFileAfterClose [as oncomplete] (fs.js:378:21)

Any chance for an additional dot(filename) approach instead of dot(string_content) with a streaming line-by-line processing parser for avoiding this buffer limitation?

Thx

Please change line

I tried using the type hierarchy in Typescript code and I think found a bug.

graph {{ }}

Consider the following graph:

digraph G {
  {}
}

According to the dot language specs the empty brackets should be considered as subgraph, since they satisfy rules:

subgraph    :   [ subgraph [ ID ] ] '{' stmt_list '}'

Square brackets [ and ] enclose optional items. Currently this parser fails to recognize it.

while using it with ngraph.fromdot i get following error:

root@master:~/allpuppet # node layout.js ./relationships.dot 
Loading graph from ./relationships.dot

/root/allpuppet/node_modules/ngraph.fromdot/node_modules/dotparser/grammar/dot.js:2853
      throw peg$buildException(null, peg$maxFailExpected, peg$maxFailPos);
            ^
SyntaxError: Expected "--", "->", ";", "<", "=", "[", "\"", "edge", "graph", "node", "subgraph", "{", "}", NUMBER, UNICODE_STRING or port but "." found.

it seems that it tries to parse node name, which is enclosed with quotes.
same with other non-text characters like slashes or dashes.

source dot file comes from puppet and is rendered correctly with graphviz.
it's here: https://gist.github.com/asquelt/785afb4249595f40c235

This is an very minor issue in an extreme corner case, but nevertheless...

This dot:

digraph {
    subgraph s1 {}
}

generates:

[
    {
        "type": "digraph",
        "children": [
            {
                "type": "subgraph",
                "id": "s1",
                "children": []
            },
            {
                "type": "subgraph",
                "children": null
            }
        ]
    }
]

So there is two subgraphs in the ast when there should be only one. The first one is correct, but the second is not.

The unnamed subgraph:

digraph {{}}

generates:

[
    {
        "type": "digraph",
        "children": [
            {
                "type": "subgraph",
                "children": null
            }
        ]
    }
]

which contains only the erroneous subgraph.

If a node is added to the named subgraph:

digraph {subgraph s1{a}}

the ast is correct:

[
    {
        "type": "digraph",
        "children": [
            {
                "type": "subgraph",
                "id": "s1",
                "children": [
                    {
                        "type": "node_stmt",
                        "node_id": {
                            "type": "node_id",
                            "id": "a"
                        },
                        "attr_list": []
                    }
                ]
            }
        ]
    }
]

I ended up copying the dotparser grammar to my own project, since I needed to add line number information and whitespace, newline and comment nodes to the ast and I didn't think that that would be generally useful. During that work I found a couple of issues that I fixed without reporting them here.

You might want to have a look at https://github.com/magjac/graphviz-visual-editor/commits/master/src/dotGrammar.pegjs to see if there are any fixes you'd like to backport to dotparser.

The following DOT source:

digraph {
    graph [label=l1;rankdir=TB]
}

gives:

SyntaxError: Expected ",", "<", """, "]", NUMBER or UNICODE_STRING but ";" found.

According to The DOT Language, semicolon is a valid separator in attribute lists:

attr_list: '[' [ a_list ] ']' [ attr_list ]
a_list: ID '=' ID [ (';' | ',') ] [ a_list ]

Hi Andrei,
I consume your very useful dotparser in a typescript project. Ideally I need the type files for it.
Can you generate them please?
Thanks,
Lev

This file is successfully processed by dot.exe. dotparser fails with the message

✖ Expected """ or "\" but "\n" found.

grammar/dot.js

519 | function peg$buildStructuredError(expected, found, location) {
ror(
| -----------^
peg$SyntaxError.buildMessage(expected, found),

test: big.gv
stack: |
peg$buildStructuredError (grammar/dot.js:520:12)
peg$parse (grammar/dot.js:2938:11)
Test. (test/index.js:9:13)
type: SyntaxError
expected:
- &a1
type: literal
text:
ignoreCase: false
- *a1
- type: literal
text: '"'
ignoreCase: false
found: "\n"
location:
start:
offset: 215
line: 5
column: 50
end:
offset: 216
line: 6
column: 1
tapCaught: testFunctionThrow.

It fails on a long string that begins like that:
struct3 [shape=record,label="hello\nworld\n
and | { Thats the reason | post-menopausal |

The graph is "big.gv" of Graphvis collection. I am not sure how to share it with you, cannot find it on the web at the moment. Let me know if it is a problem and I will share it.