From a77dc568bb334bd9ca1530475a2ae894db3fd85c Mon Sep 17 00:00:00 2001
From: "kr.angelov" <kr.angelov@gmail.com>
Date: Fri, 7 Mar 2014 21:24:20 +0000
Subject: [PATCH] added pgf_lookup_word_prefix which makes it possible to do
 simple word prediction

---
 src/runtime/c/pgf/parser.c | 37 ++++++++++++++++++++++++++++++++++++-
 src/runtime/c/pgf/pgf.h    |  4 ++++
 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/src/runtime/c/pgf/parser.c b/src/runtime/c/pgf/parser.c
index 8f58026ed..74e3fb989 100644
--- a/src/runtime/c/pgf/parser.c
+++ b/src/runtime/c/pgf/parser.c
@@ -2581,6 +2581,7 @@ pgf_lookup_morpho(PgfConcr *concr, GuString sentence,
 typedef struct {
 	GuEnum en;
 	PgfSequences* sequences;
+	GuString prefix;
 	size_t seq_idx;
 } PgfFullFormState;
 
@@ -2600,6 +2601,12 @@ gu_fullform_enum_next(GuEnum* self, void* to, GuPool* pool)
 		while (st->seq_idx < n_seqs) {
 			PgfSymbols* syms = gu_seq_index(st->sequences, PgfSequence, st->seq_idx)->syms;
 			GuString tokens = pgf_get_tokens(syms, 0, pool);
+			
+			if (!gu_string_is_prefix(st->prefix, tokens)) {
+				st->seq_idx = n_seqs;
+				break;
+			}
+				
 			if (strlen(tokens) > 0 &&
 				gu_seq_index(st->sequences, PgfSequence, st->seq_idx)->idx != NULL) {
 				entry = gu_new(PgfFullFormEntry, pool);
@@ -2609,7 +2616,7 @@ gu_fullform_enum_next(GuEnum* self, void* to, GuPool* pool)
 				st->seq_idx++;
 				break;
 			}
-			
+
 			st->seq_idx++;
 		}
 	}
@@ -2623,6 +2630,7 @@ pgf_fullform_lexicon(PgfConcr *concr, GuPool* pool)
 	PgfFullFormState* st = gu_new(PgfFullFormState, pool);
 	st->en.next   = gu_fullform_enum_next;
 	st->sequences = concr->sequences;
+	st->prefix    = "";
 	st->seq_idx   = 0;
 	return &st->en;
 }
@@ -2640,6 +2648,33 @@ pgf_fullform_get_analyses(PgfFullFormEntry* entry,
 	pgf_morpho_iter(entry->idx, callback, err);
 }
 
+GuEnum*
+pgf_lookup_word_prefix(PgfConcr *concr, GuString prefix,
+                       GuPool* pool, GuExn* err)
+{
+	if (concr->sequences == NULL) {
+		GuExnData* err_data = gu_raise(err, PgfExn);
+		if (err_data) {
+			err_data->data = "The concrete syntax is not loaded";
+			return NULL;
+		}
+	}
+
+	PgfFullFormState* state = gu_new(PgfFullFormState, pool);
+	state->en.next   = gu_fullform_enum_next;
+	state->sequences = concr->sequences;
+	state->prefix    = prefix;
+	state->seq_idx   = 0;
+
+	if (!gu_seq_binsearch_index(concr->sequences, pgf_sequence_order,
+	                            PgfSequence, (void*) prefix, 
+	                            &state->seq_idx)) {
+		state->seq_idx++;
+	}
+
+	return &state->en;
+}
+
 // The 'pre' construction needs a special handling since
 // it cannot be sorted alphabetically (a single pre contains 
 // many alternative tokens).
diff --git a/src/runtime/c/pgf/pgf.h b/src/runtime/c/pgf/pgf.h
index 20e6b1fa8..87ca3ae28 100644
--- a/src/runtime/c/pgf/pgf.h
+++ b/src/runtime/c/pgf/pgf.h
@@ -131,6 +131,10 @@ void
 pgf_fullform_get_analyses(PgfFullFormEntry* entry,
                           PgfMorphoCallback* callback, GuExn* err);
 
+GuEnum*
+pgf_lookup_word_prefix(PgfConcr *concr, GuString prefix,
+                       GuPool* pool, GuExn* err);
+
 PgfExprEnum*
 pgf_parse_with_heuristics(PgfConcr* concr, PgfCId cat, 
                           GuString sentence, double heuristics,