minor optimization/refactoring in the grammar reader

3 files changed, 137 insertions, 176 deletions

diff --git a/src/runtime/c/pgf/linearizer.c b/src/runtime/c/pgf/linearizer.c
index 93fbb452c..f73c650d8 100644
--- a/src/runtime/c/pgf/linearizer.c
+++ b/src/runtime/c/pgf/linearizer.c
@@ -35,60 +35,34 @@ pgf_lzr_add_overl_entry(PgfCncOverloadMap* overl_table,
 	gu_buf_push(entries, void*, entry);
 }
 
-typedef struct {
-	GuMapItor fn;
-	PgfConcr* concr;
-	GuPool* pool;
-} PgfLzrIndexFn;
-
-static void
-pgf_lzr_index_itor(GuMapItor* fn, const void* key, void* value, GuExn* err)
+void
+pgf_lzr_index(PgfConcr* concr, 
+              PgfCCat* ccat, PgfProduction prod,
+              GuPool *pool)
 {
-	(void) (key && err);
-
-	PgfLzrIndexFn* clo = (PgfLzrIndexFn*) fn;
-    PgfCCat* ccat = *((PgfCCat**) value);
-    PgfConcr *concr = clo->concr;
-    GuPool *pool = clo->pool;
-
-	if (gu_seq_is_null(ccat->prods))
-		return;
-
-	size_t n_prods = gu_seq_length(ccat->prods);
-	for (size_t i = 0; i < n_prods; i++) {
-		PgfProduction prod = gu_seq_get(ccat->prods, PgfProduction, i);
-
-		void* data = gu_variant_data(prod);
-		switch (gu_variant_tag(prod)) {
-		case PGF_PRODUCTION_APPLY: {
-			PgfProductionApply* papply = data;
-			PgfCncOverloadMap* overl_table =
-				gu_map_get(concr->fun_indices, &papply->fun->absfun->name,
-					PgfCncOverloadMap*);
-			if (!overl_table) {
-				overl_table = gu_map_type_new(PgfCncOverloadMap, pool);
-				gu_map_put(concr->fun_indices,
-					&papply->fun->absfun->name, PgfCncOverloadMap*, overl_table);
-			}
-			pgf_lzr_add_overl_entry(overl_table, ccat, papply, pool);
-			break;
-		}
-		case PGF_PRODUCTION_COERCE: {
-			PgfProductionCoerce* pcoerce = data;
-			pgf_lzr_add_overl_entry(concr->coerce_idx, ccat, pcoerce, pool);
-			break;
-		}
-		default:
-			gu_impossible();
+	void* data = gu_variant_data(prod);
+	switch (gu_variant_tag(prod)) {
+	case PGF_PRODUCTION_APPLY: {
+		PgfProductionApply* papply = data;
+		PgfCncOverloadMap* overl_table =
+			gu_map_get(concr->fun_indices, &papply->fun->absfun->name,
+				PgfCncOverloadMap*);
+		if (!overl_table) {
+			overl_table = gu_map_type_new(PgfCncOverloadMap, pool);
+			gu_map_put(concr->fun_indices,
+				&papply->fun->absfun->name, PgfCncOverloadMap*, overl_table);
 		}
+		pgf_lzr_add_overl_entry(overl_table, ccat, papply, pool);
+		break;
+	}
+	case PGF_PRODUCTION_COERCE: {
+		PgfProductionCoerce* pcoerce = data;
+		pgf_lzr_add_overl_entry(concr->coerce_idx, ccat, pcoerce, pool);
+		break;
+	}
+	default:
+		gu_impossible();
 	}
-}
-
-void
-pgf_lzr_index(PgfConcr* concr, GuPool *pool)
-{
-	PgfLzrIndexFn clo = { { pgf_lzr_index_itor }, concr, pool };
-	gu_map_iter(concr->ccats, &clo.fn, NULL);
 }
 
 typedef struct PgfLzn PgfLzn;
diff --git a/src/runtime/c/pgf/parser.c b/src/runtime/c/pgf/parser.c
index 6a3fa0bbb..fb930704c 100644
--- a/src/runtime/c/pgf/parser.c
+++ b/src/runtime/c/pgf/parser.c
@@ -2561,96 +2561,71 @@ pgf_parser_index_epsilon(PgfConcr* concr,
 	gu_buf_push(gu_seq_buf(prods), PgfProduction, prod);
 }
 
-typedef struct {
-	GuMapItor fn;
-	PgfConcr* concr;
-	GuPool* pool;
-} PgfLeftcornerFn;
-
-static void
-pgf_parser_index_cats(GuMapItor* fn, const void* key, void* value, GuExn* err)
+void
+pgf_parser_index(PgfConcr* concr, 
+                 PgfCCat* ccat, PgfProduction prod,
+                 GuPool *pool)
 {
-	(void) (key && err);
-
-	PgfLeftcornerFn* clo = (PgfLeftcornerFn*) fn;
-    PgfCCat* ccat = *((PgfCCat**) value);
-
-	if (gu_seq_is_null(ccat->prods))
-		return;
-
 	for (size_t lin_idx = 0; lin_idx < ccat->cnccat->n_lins; lin_idx++) {
-		size_t n_prods = gu_seq_length(ccat->prods);
-		for (size_t i = 0; i < n_prods; i++) {
-			PgfProduction prod = gu_seq_get(ccat->prods, PgfProduction, i);
+		GuVariantInfo i = gu_variant_open(prod);
+		switch (i.tag) {
+		case PGF_PRODUCTION_APPLY: {
+			PgfProductionApply* papp = i.data;
 
-			GuVariantInfo i = gu_variant_open(prod);
-			switch (i.tag) {
-			case PGF_PRODUCTION_APPLY: {
-				PgfProductionApply* papp = i.data;
+			if (gu_seq_length(papp->args) > 0)
+				break;
 
-				if (gu_seq_length(papp->args) > 0)
+			PgfSequence seq = papp->fun->lins[lin_idx];
+			if (gu_seq_length(seq) > 0) {
+				GuVariantInfo i = gu_variant_open(gu_seq_get(seq, PgfSymbol, 0));
+				switch (i.tag) {
+				case PGF_SYMBOL_KS: {
+					PgfSymbolKS* sks = i.data;
+					pgf_parser_index_token(concr, 
+										   sks->tokens,
+										   ccat, lin_idx, prod,
+										   pool);
 					break;
-
-				PgfSequence seq = papp->fun->lins[lin_idx];
-				if (gu_seq_length(seq) > 0) {
-					GuVariantInfo i = gu_variant_open(gu_seq_get(seq, PgfSymbol, 0));
-					switch (i.tag) {
-					case PGF_SYMBOL_KS: {
-						PgfSymbolKS* sks = i.data;
-						pgf_parser_index_token(clo->concr, 
-						                       sks->tokens,
-						                       ccat, lin_idx, prod,
-						                       clo->pool);
-						break;
-					}
-					case PGF_SYMBOL_KP: {
-						PgfSymbolKP* skp = i.data;
-						pgf_parser_index_token(clo->concr, 
-						                       skp->default_form,
-						                       ccat, lin_idx, prod,
-						                       clo->pool);
-						for (size_t i = 0; i < skp->n_forms; i++) {
-							pgf_parser_index_token(clo->concr, 
-							                       skp->forms[i].form,
-							                       ccat, lin_idx, prod,
-							                       clo->pool);
-						}
-						break;
-					}
-					case PGF_SYMBOL_CAT:
-					case PGF_SYMBOL_LIT:
-					case PGF_SYMBOL_NE:
-					case PGF_SYMBOL_VAR:
-						// Nothing to be done here
-						break;
-					default:
-						gu_impossible();
+				}
+				case PGF_SYMBOL_KP: {
+					PgfSymbolKP* skp = i.data;
+					pgf_parser_index_token(concr, 
+										   skp->default_form,
+										   ccat, lin_idx, prod,
+										   pool);
+					for (size_t i = 0; i < skp->n_forms; i++) {
+						pgf_parser_index_token(concr, 
+											   skp->forms[i].form,
+											   ccat, lin_idx, prod,
+											   pool);
 					}
-				} else {
-					pgf_parser_index_epsilon(clo->concr,
-					                         ccat, lin_idx, prod,
-					                         clo->pool);
+					break;
+				}
+				case PGF_SYMBOL_CAT:
+				case PGF_SYMBOL_LIT:
+				case PGF_SYMBOL_NE:
+				case PGF_SYMBOL_VAR:
+					// Nothing to be done here
+					break;
+				default:
+					gu_impossible();
 				}
+			} else {
+				pgf_parser_index_epsilon(concr,
+										 ccat, lin_idx, prod,
+										 pool);
 			}
+		}
+		break;
+		case PGF_PRODUCTION_COERCE:
+			// Nothing to be done here
 			break;
-			case PGF_PRODUCTION_COERCE:
-				// Nothing to be done here
-				break;
-			default:
-				gu_impossible();
-			}
+		default:
+			gu_impossible();
 		}
 	}
 }
 
-void
-pgf_parser_index(PgfConcr* concr, GuPool *pool)
-{
-	PgfLeftcornerFn clo1 = { { pgf_parser_index_cats }, 
-	                         concr, pool };
-	gu_map_iter(concr->ccats, &clo1.fn, NULL);
-}
-
 prob_t
 pgf_ccat_set_viterbi_prob(PgfCCat* ccat) {
 	if (ccat->fid < 0)
diff --git a/src/runtime/c/pgf/reader.c b/src/runtime/c/pgf/reader.c
index 9b2844e34..abfbfe025 100644
--- a/src/runtime/c/pgf/reader.c
+++ b/src/runtime/c/pgf/reader.c
@@ -1024,65 +1024,80 @@ pgf_read_cnccats(PgfReader* rdr, PgfAbstr* abstr, PgfConcr* concr)
 	return cnccats;
 }
 
-typedef struct {
-	GuMapItor fn;
-	PgfReader* rdr;
-} PgfIndexFn;
-
-static PgfCncCat*
-pgf_ccat_set_cnccat(PgfCCat* ccat)
+static void
+pgf_ccat_set_cnccat(PgfCCat* ccat, PgfProduction prod)
 {
-	if (!ccat->cnccat) {
-		size_t n_prods = gu_seq_length(ccat->prods);
-		for (size_t i = 0; i < n_prods; i++) {
-			PgfProduction prod = 
-				gu_seq_get(ccat->prods, PgfProduction, i);
-			GuVariantInfo i = gu_variant_open(prod);
-			switch (i.tag) {
-			case PGF_PRODUCTION_COERCE: {
-				PgfProductionCoerce* pcoerce = i.data;
-				PgfCncCat* cnccat = 
-					pgf_ccat_set_cnccat(pcoerce->coerce);
-				if (!ccat->cnccat) {
-					ccat->cnccat = cnccat;
-				} else if (ccat->cnccat != cnccat) {
-					// XXX: real error
-					gu_impossible();
-				}
- 				break;
-			}
-			case PGF_PRODUCTION_APPLY:
-				// Shouldn't happen with current PGF.
-				// XXX: real error
-				gu_impossible();
-				break;
-			default:
-				gu_impossible();
-			}
+	GuVariantInfo i = gu_variant_open(prod);
+	switch (i.tag) {
+	case PGF_PRODUCTION_COERCE: {
+		PgfProductionCoerce* pcoerce = i.data;
+		PgfCncCat* cnccat = pcoerce->coerce->cnccat;
+		if (!ccat->cnccat) {
+			ccat->cnccat = cnccat;
+		} else if (ccat->cnccat != cnccat) {
+			// XXX: real error
+			gu_impossible();
 		}
+		break;
+	}
+	case PGF_PRODUCTION_APPLY:
+		// Shouldn't happen with current PGF.
+		// XXX: real error
+		gu_impossible();
+		break;
+	default:
+		gu_impossible();
 	}
-	return ccat->cnccat;
 }
 
-extern float
+extern prob_t
 pgf_ccat_set_viterbi_prob(PgfCCat* ccat);
 
+typedef struct {
+	GuMapItor fn;
+	PgfConcr* concr;
+	GuPool* pool;
+} PgfIndexFn;
+
+extern void
+pgf_parser_index(PgfConcr* concr, 
+                 PgfCCat* ccat, PgfProduction prod,
+                 GuPool *pool);
+
+void
+pgf_lzr_index(PgfConcr* concr, 
+              PgfCCat* ccat, PgfProduction prod,
+              GuPool *pool);
+
 static void
 pgf_read_ccat_cb(GuMapItor* fn, const void* key, void* value, GuExn* err)
 {
 	(void) (key && err);
-	PgfCCat* ccat = *((PgfCCat**) value);
+	
+	PgfIndexFn* clo = (PgfIndexFn*) fn;
+    PgfCCat* ccat = *((PgfCCat**) value);
+    PgfConcr *concr = clo->concr;
+    GuPool *pool = clo->pool;
+
+	if (gu_seq_is_null(ccat->prods))
+		return;
+
+	size_t n_prods = gu_seq_length(ccat->prods);
+	for (size_t i = 0; i < n_prods; i++) {
+		PgfProduction prod = 
+			gu_seq_get(ccat->prods, PgfProduction, i);
+			
+		if (!ccat->cnccat) {
+			pgf_ccat_set_cnccat(ccat, prod);
+		}
+
+		pgf_parser_index(concr, ccat, prod, pool);
+		pgf_lzr_index(concr, ccat, prod, pool);
+	}
 
-	pgf_ccat_set_cnccat(ccat);
 //	pgf_ccat_set_viterbi_prob(ccat);
 }
 
-void
-pgf_parser_index(PgfConcr* concr, GuPool *pool);
-
-void
-pgf_lzr_index(PgfConcr* concr, GuPool *pool);
-
 static PgfConcr*
 pgf_read_concrete(PgfReader* rdr, PgfAbstr* abstr)
 {
@@ -1121,12 +1136,9 @@ pgf_read_concrete(PgfReader* rdr, PgfAbstr* abstr)
 	concr->callbacks = pgf_new_callbacks_map(concr, rdr->opool); 
 	concr->total_cats = pgf_read_int(rdr);
 
-	PgfIndexFn clo1 = { { pgf_read_ccat_cb }, rdr };
+	PgfIndexFn clo1 = { { pgf_read_ccat_cb }, concr, rdr->opool };
 	gu_map_iter(concr->ccats, &clo1.fn, NULL);
 
-	pgf_parser_index(concr, rdr->opool);
-	pgf_lzr_index(concr, rdr->opool);
-
 	return concr;
 }