Method Format::guessFormat() doesn't support turtle files without prefixes or a base URI

[Michiel-s]

The issue

Taking Example1 from https://www.w3.org/TR/turtle/

@base <http://example.org/> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix rel: <http://www.perceive.net/schemas/relationship/> .

<#green-goblin>
    rel:enemyOf <#spiderman> ;
    a foaf:Person ;    # in the context of the Marvel universe
    foaf:name "Green Goblin" .

<#spiderman>
    rel:enemyOf <#green-goblin> ;
    a foaf:Person ;
    foaf:name "Spiderman", "Человек-паук"@ru .

Instantiating a EasyRDF Graph based on this data is ok. It parses correctly.

However, when the prefixes are not provided an exception is thrown 'Unable to parse data of an unknown format.'

E.g.

<#green-goblin>
    rel:enemyOf <#spiderman> ;
    a foaf:Person ;    # in the context of the Marvel universe
    foaf:name "Green Goblin" .

<#spiderman>
    rel:enemyOf <#green-goblin> ;
    a foaf:Person ;
    foaf:name "Spiderman", "Человек-паук"@ru .

Analysis

According to the Turtle Grammer:

• A turtleDoc is a set of statements [1]
• A statement is a directive OR triples [2]

That means that a turtle document can, but not necessarily needs to, start with @prefix or @base statements.

In the logic of method Format::guessFormat(), turtle documents are only recognized when they start with prefix or base statements (with or without the @).

...
        } elseif (preg_match('/@prefix\s|@base\s/', $short)) {
            return self::getFormat('turtle');
        } elseif (
            preg_match('/prefix\s|base\s/i', $short)
            // see FormatTest::testGuessFormatTurtleByPrefix for an example
            && false === str_contains($short, '<?xml')
        ) {
            return self::getFormat('turtle');
        } elseif (preg_match('/^\s*<.+> <.+>/m', $short)) {
            return self::getFormat('ntriples');
        } else {
            return null;
        }
...

Solution space

I propose to add a few more possibilities to recognize turtle documents.

We need to keep in mind to be able to distinguish turtle from n-triples syntax. If I'm correct, turtle is a superset of ntriples. A valid n-triples document also complies to turtle, but not the other way around.

There are specific indicators that we are dealing with a turtle document, including:

• The shorthand a is used for predicate rdf:type
• There are compact URIs used, i.e. prefix colon localname enclosed by whitespace
• There is usage of semicolon to close-continue predicateObjectLists

These are not allowed to be used in n-triples syntax so we are good here. This might result in more false positive hits for turtle when guessing the format resulting in parser errors. But I don't see a problem here, because otherwise the guessFormat method will return null which also results in an exception as described in this issue.

[k00ni]

Sorry for the late response.

I think I got your point and it seems reasonable what you suggest. To me the method name guessFormat implies that there could be a certain percentage of false positives, which is fine IMHO. Something similar is also stated in the method header Attempt to guess the document format from some content.

About the shorthand a: Would it make sense to use a regex such as ^\s+a to find instances such as the following from your example?

    a foaf:Person ;

Would you mind creating a PR and suggest a couple of changes?
Included in this could be a change of the description in the method header to better reflect the possibility of false positives.

[Michiel-s]

Thanks for your reply @k00ni.

You are right, the guessFormat has, as the name already suggests, the possibility of returning false positive matches or no match at all. This issue here is indeed to enhance/improve the method to have less false positives.

I'll see what I can do working on a PR, but that won't be in the upcoming weeks. Maybe even later.

[TallTed]

That means that a turtle document can, but not necessarily needs to, start with @prefix or @base statements.

It seems worth noting that that, if present, the @prefix or @base statements are case-sensitive.

Further, Turtle documents can also, but again are not required to, start with case-insensitive PREFIX or BASE statements.