dot attribute parser defects

[SaculRennorb]

Expected Behavior

dot files get loaded properly

Current Behavior

they do not

Steps to Reproduce

save this as text.dot or similar and open it in gephi:

digraph test {
    Skill_38d2a09ab424df40b347da44c3b7c192 [label="a"] [color="green"];
    Skill_7bf98ed70a7d964b9b3771e26bc44124 [label="b"] [color="green"];
    Skill_38d2a09ab424df40b347da44c3b7c192 -> Skill_7bf98ed70a7d964b9b3771e26bc44124 [label=""] [color="azure"];
}

Your Environment

• Version used: Gephi 0.10.1
• Operating System: Win10x64

It seems like the importer only looks for the label attribute.

This:

digraph test {
    n1 [label="a"];
    n2 [label="b"];
}

gets loaded properly, however this

digraph test {
    n1 [label="l1"] [color="green"];
    n2 [label="l2"] [color="green"];
}

only processes the label attribute and ignores color.
Setting only the color attribute also gets ignored.
edit: I was confused about the color attribute, as it doesn't show up as a separate attribute in the data lab. It does however seem to get picked up, even if it only seems to work for specific colors.

It seems like using [atr1=".." attr2=".."] works.

It also looks like the importer completely falls apart once you introduce multiple attributes, as seen by the test data resulting in

From having a quick look at the importer it seems like you assume there is only ever one attribute per node, which is not correct.

gephi/modules/ImportPlugin/src/main/java/org/gephi/io/importer/plugin/file/ImporterDOT.java

Lines 181 to 192 in 760df9a

	if (streamTokenizer.ttype == '-' || (streamTokenizer.ttype == StreamTokenizer.TT_WORD &&
	(streamTokenizer.sval.equalsIgnoreCase("-") || streamTokenizer.sval.equalsIgnoreCase("--")))) {
	NodeDraft nodeDraft = getOrCreateNode(nodeId);
	edgeStructure(streamTokenizer, nodeDraft);
	} else if (streamTokenizer.ttype == '[') {
	NodeDraft nodeDraft = getOrCreateNode(nodeId);
	nodeAttributes(streamTokenizer, nodeDraft);
	} else {
	getOrCreateNode(nodeId);
	streamTokenizer.pushBack();
	}
	}

Should maybe loop until ; or EOL, but looping the whole section is likely going to mess up edge parsing, so this is just a rough suggestion. also I don't really know the code, I just had a quick gander.

I wasn't sure where exactly to put this, so ill just link possibly related issues here: #320 #1644 #2780