rdf_turtle_write.pl -- Turtle - Terse RDF Triple Language writer

This module implements the Turtle language for representing the RDF triple model as defined by Dave Beckett from the Institute for Learning and Research Technology University of Bristol in the document:

The Turtle format is designed as an RDF serialization that is easy to read and write by both machines and humans. Due to the latter property, this library goes a long way in trying to produce human-readable output.

In addition to the human-readable format, this library can write a canonical representation of RDF graphs. The canonical representation has the following properties:

Equivalent graphs result in the same document. Graphs are considered equivalent iff they contain the same set of triples, regardless of the labeling of blank nodes in the graph.
Changes to the graph are diff-friendly. This means
- Prefixes are combined in the header and thus changes to the namespaces only result in changes in the header.
- Blank nodes that are used only once (including collections) are written in-line with the object they belong to.
- For other blank nodes we to realise stable labeling that is based on property-values.

To be done: - Low-level string output takes 28% of the time. Move to C?

rdf_save_turtle(+Out, :Options) is det

Save an RDF graph as Turtle. Options processed are:

a(+Boolean): If true (default), use a for the predicate rdf:type. Otherwise use the full resource.
align_prefixes(+Boolean): Nicely align the @prefix declarations
base(+Base): Save relative to the given Base
canonize_numbers(+Boolean): If true (default false), emit numeric datatypes using Prolog's write to achieve canonical output.
comment(+Boolean): It true (default), write some informative comments between the output segments
encoding(+Encoding): Encoding used for the output stream. Default is UTF-8.
expand(:Goal): Query an alternative graph-representation. See below.
indent(+Column): Indentation for ; -lists. `0' does not indent, but writes on the same line. Default is 8.
graph(+Graph): Save only the named graph
group(+Boolean): If true (default), using P-O and O-grouping.
inline_bnodes(+Boolean): if true (default), inline bnodes that are used once.
abbreviate_literals(+Boolean): if true (default), omit the type if allowed by turtle.
only_known_prefixes(+Boolean): Only use prefix notation for known prefixes. Without, some documents produce huge amounts of prefixes.
prefixes(+List): If provided, uses exactly these prefixes. List is a list of prefix specifications, where each specification is either a term Prefix_-_URI or a prefix that is known to rdf_current_prefix/2.
silent(+Boolean): If true (default false), do not print the final informational message.
single_line_bnodes(+Bool): If true (default false), write [...] and (...) on a single line.
subject_white_lines(+Count): Extra white lines to insert between statements about a different subject. Default is 1.
tab_distance(+Tab): Distance between tab-stops. `0' forces the library to use only spaces for layout. Default is 8.
user_prefixes(+Boolean): If true (default), use prefixes from rdf_current_prefix/2.

The option expand allows for serializing alternative graph representations. It is called through call/5, where the first argument is the expand-option, followed by S,P,O,G. G is the graph-option (which is by default a variable). This notably allows for writing RDF graphs represented as rdf(S,P,O) using the following code fragment:

triple_in(RDF, S,P,O,_G) :-
    member(rdf(S,P,O), RDF).

    ...,
    rdf_save_turtle(Out, [ expand(triple_in(RDF)) ]),

Arguments:

Out - is one of stream(Stream), a stream handle, a file-URL or an atom that denotes a filename.

rdf_save_canonical_turtle(+Spec, :Options) is det

Save triples in a canonical format. This is the same as rdf_save_turtle/2, but using different defaults. In particular:

encoding(utf8),
indent(0),
tab_distance(0),
subject_white_lines(1),
align_prefixes(false),
user_prefixes(false)
comment(false),
group(false),
single_line_bnodes(true)

To be done: - Work in progress. Notably blank-node handling is incomplete.

rdf_save_ntriples(+Spec, :Options) is det

Save RDF using ntriples format. The ntriples format is a subset of Turtle, writing each triple fully qualified on its own line.

rdf_save_trig(+Spec, :Options) is det

Save multiple RDF graphs into a TriG file. Options are the same as for rdf_save_turtle/2. rdf_save_trig/2 ignores the graph(+Graph) option and instead processes one additional option:

graphs(+ListOfGraphs): List of graphs to save. When omitted, all graphs in the RDF store are stored in the TriG file.

rdf_save_canonical_trig(+Spec, :Options) is det

Save triples in a canonical format. See rdf_save_canonical_turtle/2 foir details.

trig_graphs(+State, -Graphs) is det[private]

True when Graphs is the (sorted) list of graphs we must save. If the expand argument is used and no graphs are specified, it enumerates all triples and extracts the graphs.

open_output(+Spec, +Encoding, -Stream, -Cleanup) is det[private]

Open output Spec, returning a stream using Encoding.

Arguments:

Cleanup

- is a goal that must be used to revert the side effects of open_output/4.

init_prefix_map(+State, -State) is det[private]

Set the prefix_map of State. The prefix map is list of Prefix-URI of prefixes to use for emitting the graph requested in State. If multiple prefixes are present where the one is a prefix of the other, the longer one appears first in the list.

user_prefix_map(+Prefixes, -PrefixMap) is det[private]

Convert a list of prefix specifications to a list Prefix-URI, longest URI first.

turtle_prefix(+OnlyKnown, +Where, +Prefix, +URI) is semidet[private]

Test whether we want to include the proposed prefix in the @prefix declaration.

propose_abbrev(+State, +Len, +URI, -Abbrev) is multi[private]

Propose an abbreviation for URI. Backtracking yields longer ones.

namespace_parts(+URL, -Parts)[private]

init_base(+State0, -State) is det[private]

Initialise dealing with the base URI. It sets two attributes of the state: base_root and base_path.

tw_graph(+State, +Out) is det[private]

Write an RDF graph as Turtle data.

To be done: - Write unconnected and multi-connected blank-nodes.

tw_prefix_map(+PrefixMap, +State, +Out) is det[private]

Write the @base and @prefix declarations

tw_proper_subjects(+Subjects, +State, +Out) is det[private]

Write the subjects that are not Bnodes.

tw_subject(+URI, +State, +Out) is det[private]

Write a toplevel non-bnode subject.

tw_bnodes(+Pairs, +State, +Out) is det[private]

Write the Bnodes. Pairs is a list URI-Ref, where Ref is one of written if the Bnode is already written; an integer if it is used multiple times or a variable if it has not been written. The order in which we deal with bnodes is defined as follows:

First, write the bnodes that are not referenced at all as toplevel bnodes using [ ... ] notation.
Next, write the bnodes that need written as toplevel nodes using the _:XX notation because they are referenced multiple times in the graph. Continue this process until it is exhausted.

tw_top_bnodes(+Pairs, +State, +Out, -Rest)[private]

Write the top bnodes: those that do not appear as an object anywhere.

tw_cyclic_bnodes(+Pairs, +BNode, +State, +Out, +Cycle)[private]

The rest. These are groups of bnodes that are reachable, but we cannot find a starting point, neither from a named resource, nor from an unlinked bnode. As long as we are not considering stable canonical output, we can break the cycle at any point.

tw_subject_triples(+Grouped, +Subject, +State, +Out)[private]

Save triples on Subject. Combine groups of triples with the same subject (;) and same subject+predicate (,).

Arguments:

Subject

- is either a URI or an integer. The latter is used for writing a named bnode.

tw_ungrouped_triples(+Grouped, +URI, +State, +Out)[private]

Write triples for subject URI as one line per triple. Used for canonical output.

tw_triples(+Groups, +Indent, +State, +Out) is det[private]

Triple writer that uses ; and ,- grouping

tw_object(+Value, +State, +Out) is det[private]

Write the object of a triple.

tw_bnode_object(+Value, +State, +Out) is det[private]

Write a Bnode value. There are a number of cases:

The BNode was already written. Write the same ref.
The BNode is not shared. Inline and set written
The BNode is shared. Generate a NodeID and store it
The BNode is once as object: Generate a NodeID
The BNode is more than once object: Generate a NodeID and put in table.

tw_unshared_bnode(+BNode, +State, +Out) is semidet[private]

Write a bnode if this is the only place it is used.

unshared_collection(+URI, +State, -Members) is semidet[private]

True if URI denodes an RDF list that is made up from bnodes, is linked exactly once to its context and contains no extra triples.

object_link_count(+BNode, +STate, -Count) is det[private]

Number of times BNode is used as an object in the graph

nl_indent(+Out, +State, +Indent) is det[private]

Write a newline and indent to column Indent.

subject_triples(+URI, +State, -Pairs) is det[private]

Pairs is a sorted list of P-O pairs representing all triples on the subject URI.

subjects(+State, -Subjects:ord_set) is det[private]

Subjects is a list of all subjects in the graph requested in State.

sort_bnodes(+BNodes, -Sorted, +State) is det[private]

Sort a list of blank nodes.

sort_bnode_pairs(+Pairs, -Sorted, +State) is det[private]

Sort a list of Pairs BNode-Ref

bnode_to_term(+BNode, -Term, +State)[private]

Term is a canonical representation of the graph formed by BNode. The transformation of a bnode is

bnode(p-[o1,o2,..], ..)

The arguments are alphabetically sorted on predicate (can't we leave the preds out them?) and the objects are alphabetically sorted. Sorting multiple bnode values?

next_bnode_id(+State, +BNode, -Ref) is det[private]

Generate a node-id for BNode. When writing non-canonically, we simply number the bnodes. Otherwise we want a more stable numbering. Our numbering is a hash of the content of the bnode. It is not unlikely that we find muliple copies, and therefore we number the full id is bn_<hash>_<n>, <n> counting 0...

bnode_hash(+BNode, -Hash) is det[private]

Hash is the hash-value for a bnode.

To be done: - : Hash on content.

tw_resource(+Resource, +State, +Out) is det[private]

Write a resource

tw_literal(+Literal, +State, +Out) is det[private]

Write a literal value to the stream Out.

tw_abbreviated_literal(+Type, +Value, +State, +Out) is semidet[private]

Turtle abbreviated typed literals.

To be done: - : Deal with canonical forms (or is this a task of the RDF parser?; - : What if the value is not in the lexical space of the type?

tw_quoted_string(+Atom, +State, +Out) is det[private]

Write Atom to Out as a quoted string. We only use the single-"..." representation.