mirror of
https://github.com/ganelson/inform.git
synced 2024-07-05 16:44:21 +03:00
78 lines
2.6 KiB
OpenEdge ABL
78 lines
2.6 KiB
OpenEdge ABL
[Diagramming::] Diagramming.
|
|
|
|
Turning a file of natural language into a syntax tree.
|
|
|
|
@ Everything is now set up to read in a text file, break it into sentences,
|
|
and then hand each one to //linguistics// in turn to construct syntax diagrams.
|
|
|
|
We need to tell //linguistics// which parts of a verb we will allow in these
|
|
sentences -- the answer being, all of them. (Inform is more restrictive.)
|
|
|
|
@d ALLOW_VERB_IN_ASSERTIONS_LINGUISTICS_CALLBACK Diagramming::allow_in_assertions
|
|
|
|
=
|
|
int Diagramming::allow_in_assertions(verb_conjugation *vc, int tense, int sense, int person) {
|
|
return TRUE;
|
|
}
|
|
|
|
@ And so here goes.
|
|
|
|
=
|
|
parse_node_tree *syntax_tree = NULL;
|
|
parse_node_tree *Diagramming::test_diagrams(text_stream *arg, int raw) {
|
|
syntax_tree = SyntaxTree::new();
|
|
@<Turn the file into a syntax tree@>;
|
|
@<Use the linguistics module on each sentence@>;
|
|
return syntax_tree;
|
|
}
|
|
|
|
@<Turn the file into a syntax tree@> =
|
|
filename *F = Filenames::from_text(arg);
|
|
feed_t FD = Feeds::begin();
|
|
source_file *sf = TextFromFiles::feed_into_lexer(F, NULL_GENERAL_POINTER);
|
|
wording W = Feeds::end(FD);
|
|
if (sf == NULL) { PRINT("File has failed to open\n"); return NULL; }
|
|
Sentences::break(syntax_tree, W);
|
|
|
|
@<Use the linguistics module on each sentence@> =
|
|
SyntaxTree::traverse(syntax_tree, Diagramming::diagram);
|
|
if (raw == FALSE) Diagramming::parse_noun_phrases(syntax_tree->root_node);
|
|
|
|
@ The work of the //words// and //syntax// modules means that we now have a
|
|
rudimentary syntax tree, in which each sentence is just a single |SENTENCE_NT|
|
|
node without children. We look for these, and apply <sentence>, the most
|
|
powerful nonterminal from the //linguistics// module, to them. All being well
|
|
(i.e., if any sentence structure can be found), this returns a subtree of
|
|
further nodes, which we graft below the |SENTENCE_NT|.
|
|
|
|
=
|
|
void Diagramming::diagram(parse_node *p) {
|
|
if (Node::get_type(p) == SENTENCE_NT) {
|
|
wording W = Node::get_text(p);
|
|
if (<sentence>(W)) {
|
|
parse_node *n = <<rp>>;
|
|
switch (Annotations::read_int(p, linguistic_error_here_ANNOT)) {
|
|
case TwoLikelihoods_LINERROR:
|
|
Errors::nowhere("sentence has two certainties");
|
|
break;
|
|
default:
|
|
SyntaxTree::graft(syntax_tree, n, p);
|
|
break;
|
|
}
|
|
} else {
|
|
Errors::nowhere("sentence has no primary verb");
|
|
}
|
|
}
|
|
}
|
|
|
|
@ That sorts out the verbs and prepositions, but the noun phrases are not
|
|
by default parsed: they are simply left as |UNPARSED_NOUN_NT| nodes.
|
|
|
|
=
|
|
void Diagramming::parse_noun_phrases(parse_node *p) {
|
|
for (; p; p = p->next) {
|
|
if (Node::get_type(p) == UNPARSED_NOUN_NT) Nouns::recognise(p);
|
|
Diagramming::parse_noun_phrases(p->down);
|
|
}
|
|
}
|