mie: implement caching and emitting string data

mie/builder.c

@@ -66,43 +66,39 @@ void mie_builder_put_record(
 	b_queue_push_back(&builder->b_module->m_records, &rec->r_base.v_entry);
 }
 
+#if 0
 void mie_builder_put_data(struct mie_builder *builder, struct mie_data *data)
 {
 	if (!builder->b_module) {
 		return;
 	}
 
-	b_queue_iterator it = {};
-	b_queue_foreach (&it, &builder->b_module->m_data) {
-		struct mie_value *cur
-			= b_unbox(struct mie_value, it.entry, v_entry);
-
-		if (!strcmp(cur->v_name.n_str, data->d_base.v_name.n_str)) {
+	const char *data_name = data->d_base.v_name.n_str;
+	struct mie_data *v = mie_module_get_data(builder->b_module, data_name);
+	if (v) {
 		return;
 	}
-	}
 
-	b_queue_push_back(&builder->b_module->m_data, &data->d_base.v_entry);
+	mie_module_put_data(builder->b_module, data);
 }
+#endif
 
+#if 0
 void mie_builder_put_type(struct mie_builder *builder, struct mie_type *type)
 {
 	if (!builder->b_module) {
 		return;
 	}
 
-	b_queue_iterator it = {};
-	b_queue_foreach (&it, &builder->b_module->m_types) {
-		struct mie_value *cur
-			= b_unbox(struct mie_value, it.entry, v_entry);
-
-		if (!strcmp(cur->v_name.n_str, type->t_base.v_name.n_str)) {
+	const char *type_name = type->t_base.v_name.n_str;
+	struct mie_data *v = mie_module_get_data(builder->b_module, type_name);
+	if (v) {
 		return;
 	}
-	}
 
-	b_queue_push_back(&builder->b_module->m_data, &type->t_base.v_entry);
+	mie_module_put_data(builder->b_module, MIE_VALUE);
 }
+#endif
 
 void mie_builder_set_insert_point(
 	struct mie_builder *builder, struct mie_block *block)
@@ -127,7 +123,32 @@ struct mie_value *mie_builder_get_data_ptr(
 			return NULL;
 		}
 
-		mie_module_put_data(builder->b_module, data);
+		mie_module_put_data(builder->b_module, data, data_ident);
+	}
+
+	return MIE_VALUE(data);
+}
+
+struct mie_value *mie_builder_get_string_ptr(
+	struct mie_builder *builder, const char *s)
+{
+	if (!builder->b_module) {
+		return NULL;
+	}
+
+	struct mie_data *data = mie_module_get_string_ptr(builder->b_module, s);
+	if (!data) {
+		struct mie_type *str
+			= mie_ctx_get_type(builder->b_ctx, MIE_TYPE_STR);
+		struct mie_value *value = mie_ctx_get_string(builder->b_ctx, s);
+
+		data = mie_data_create_const((struct mie_const *)value);
+
+		if (!data) {
+			return NULL;
+		}
+
+		mie_module_put_data(builder->b_module, data, ".str");
 	}
 
 	return MIE_VALUE(data);

mie/const.c

@@ -2,6 +2,22 @@
 #include <stdlib.h>
 #include <string.h>
 
+struct mie_const *mie_const_create(struct mie_type *type)
+{
+	struct mie_const *c = malloc(sizeof *c);
+	if (!c) {
+		return NULL;
+	}
+
+	memset(c, 0x0, sizeof *c);
+
+	mie_value_init(&c->c_base, MIE_VALUE_CONST);
+
+	c->c_type = type;
+
+	return c;
+}
+
 static struct mie_type *get_type(struct mie_value *v)
 {
 	struct mie_const *c = MIE_CONST(v);

mie/convert/text-write.c

@@ -1,5 +1,6 @@
 #include "convert.h"
 
+#include <blue/object/hashmap.h>
 #include <inttypes.h>
 #include <mie/alloca.h>
 #include <mie/arg.h>
@@ -153,7 +154,7 @@ static b_status write_operand_const(
 		break;
 	case MIE_TYPE_STR:
 	case MIE_TYPE_ATOM:
-		write_string(converter, c->c_v.v_str);
+		write_string_f(converter, "\"%s\"", c->c_v.v_str);
 		break;
 	case MIE_TYPE_CLASS:
 		write_string_f(converter, "@%s", value->v_name.n_str);
@@ -266,9 +267,10 @@ static b_status write_module(
 		write_char(converter, '\n');
 	}
 
-	b_queue_foreach (&it, &mod->m_data) {
-		struct mie_value *data
-			= b_unbox(struct mie_value, it.entry, v_entry);
+	b_hashmap_iterator it2;
+	b_hashmap_foreach(&it2, mod->m_data)
+	{
+		struct mie_value *data = it2.value->value_data;
 		status = write_value_to_text(converter, data);
 
 		if (!B_OK(status)) {
@@ -276,7 +278,7 @@ static b_status write_module(
 		}
 	}
 
-	if (!b_queue_empty(&mod->m_data)) {
+	if (!b_hashmap_is_empty(mod->m_data)) {
 		write_char(converter, '\n');
 	}
 
@@ -552,7 +554,8 @@ static b_status write_data(
 		break;
 	case MIE_DATA_CONST:
 		write_operand_const(
-			converter, data->d_const.c_value, F_INCLUDE_TYPE);
+			converter, MIE_VALUE(data->d_const.c_value),
+			F_INCLUDE_TYPE);
 		break;
 	default:
 		write_string(converter, "<unknown>");

mie/ctx.c

@@ -43,6 +43,7 @@ struct mie_ctx *mie_ctx_create(void)
 	out->ctx_false = MIE_CONST(mie_ctx_get_int(out, 0, 1));
 
 	out->ctx_sel_cache = b_hashmap_create(free, free);
+	out->ctx_string_cache = b_hashmap_create(free, free);
 
 	return out;
 }
@@ -214,3 +215,38 @@ struct mie_value *mie_ctx_get_selector(struct mie_ctx *ctx, const char *sel)
 
 	return MIE_VALUE(sel_value);
 }
+
+struct mie_value *mie_ctx_get_string(struct mie_ctx *ctx, const char *s)
+{
+	b_hashmap_key key = {
+		.key_data = s,
+		.key_size = strlen(s),
+	};
+
+	const b_hashmap_value *cache_entry
+		= b_hashmap_get(ctx->ctx_string_cache, &key);
+
+	if (cache_entry) {
+		return cache_entry->value_data;
+	}
+
+	struct mie_const *string_value = malloc(sizeof *string_value);
+	if (!string_value) {
+		return NULL;
+	}
+
+	mie_value_init(&string_value->c_base, MIE_VALUE_CONST);
+	string_value->c_type = mie_ctx_get_type(ctx, MIE_TYPE_STR);
+	string_value->c_v.v_str = b_strdup(s);
+
+	key.key_data = string_value->c_v.v_str;
+
+	b_hashmap_value hashmap_value = {
+		.value_data = string_value,
+		.value_size = sizeof *string_value,
+	};
+
+	b_hashmap_put(ctx->ctx_string_cache, &key, &hashmap_value);
+
+	return MIE_VALUE(string_value);
+}

mie/data.c

@@ -33,6 +33,23 @@ struct mie_data *mie_data_create_extern_global(
 	return data;
 }
 
+struct mie_data *mie_data_create_const(struct mie_const *value)
+{
+	struct mie_data *data = malloc(sizeof *data);
+	if (!data) {
+		return NULL;
+	}
+
+	memset(data, 0x0, sizeof *data);
+
+	mie_value_init(&data->d_base, MIE_VALUE_DATA);
+
+	data->d_type = MIE_DATA_CONST;
+	data->d_const.c_value = value;
+
+	return data;
+}
+
 static struct mie_type *get_type(struct mie_value *v)
 {
 	if (!ptr_type.t_id) {

mie/include/mie/builder.h

@@ -27,18 +27,6 @@ enum mie_builder_flags {
 	MIE_BUILDER_IGNORE_RESULT = 0x01u,
 };
 
-extern struct mie_ctx *mie_ctx_create(void);
-extern void mie_ctx_destroy(struct mie_ctx *ctx);
-
-extern struct mie_type *mie_ctx_get_type(
-	struct mie_ctx *ctx, enum mie_type_id type_id);
-extern struct mie_type *mie_ctx_get_int_type(
-	struct mie_ctx *ctx, unsigned int nr_bits);
-extern struct mie_value *mie_ctx_get_bool(struct mie_ctx *ctx, bool val);
-extern struct mie_value *mie_ctx_get_int(
-	struct mie_ctx *ctx, long long val, unsigned int nr_bits);
-extern struct mie_value *mie_ctx_get_selector(struct mie_ctx *ctx, const char *sel);
-
 extern struct mie_builder *mie_builder_create(
 	struct mie_ctx *ctx, struct mie_module *mod);
 extern void mie_builder_destroy(struct mie_builder *builder);
@@ -59,6 +47,8 @@ extern void mie_builder_set_insert_point(
 
 extern struct mie_value *mie_builder_get_data_ptr(
 	struct mie_builder *builder, const char *data_ident);
+extern struct mie_value *mie_builder_get_string_ptr(
+	struct mie_builder *builder, const char *s);
 
 extern struct mie_value *mie_builder_ret(
 	struct mie_builder *builder, struct mie_value *val);

mie/include/mie/const.h

@@ -18,4 +18,6 @@ struct mie_const {
 	} c_v;
 };
 
+extern struct mie_const *mie_const_create(struct mie_type *type);
+
 #endif

mie/include/mie/ctx.h

@@ -9,6 +9,7 @@ struct mie_ctx {
 	struct mie_type *ctx_types[__MIE_TYPE_COUNT];
 	b_btree ctx_int_cache;
 	struct b_hashmap *ctx_sel_cache;
+	struct b_hashmap *ctx_string_cache;
 };
 
 extern struct mie_ctx *mie_ctx_create(void);
@@ -21,6 +22,7 @@ extern struct mie_type *mie_ctx_get_int_type(
 extern struct mie_value *mie_ctx_get_bool(struct mie_ctx *ctx, bool val);
 extern struct mie_value *mie_ctx_get_int(
 	struct mie_ctx *ctx, long long val, unsigned int nr_bits);
+extern struct mie_value *mie_ctx_get_string(struct mie_ctx *ctx, const char *s);
 extern struct mie_value *mie_ctx_get_selector(struct mie_ctx *ctx, const char *sel);
 
 #endif

mie/include/mie/data.h

@@ -16,7 +16,7 @@ struct mie_data {
 	enum mie_data_type d_type;
 	union {
 		struct {
-			struct mie_value *c_value;
+			struct mie_const *c_value;
 		} d_const;
 
 		struct {
@@ -27,5 +27,6 @@ struct mie_data {
 
 extern struct mie_data *mie_data_create_extern_global(
 	struct mie_type *type, const char *ident);
+extern struct mie_data *mie_data_create_const(struct mie_const *value);
 
 #endif

mie/include/mie/module.h

@@ -4,23 +4,33 @@
 #define MIE_MODULE(p) ((struct mie_module *)(p))
 
 #include <blue/core/queue.h>
+#include <mie/name.h>
 #include <mie/value.h>
 
 struct mie_func;
+struct b_hashmap;
 
 struct mie_module {
 	struct mie_value m_base;
+	struct mie_name_map *m_names;
 	b_queue m_records;
-	b_queue m_data;
 	b_queue m_types;
 	b_queue m_func;
+
+	struct b_hashmap *m_data;
+	struct b_hashmap *m_data_strings;
 };
 
 extern struct mie_module *mie_module_create(void);
 extern void mie_module_add_function(struct mie_module *mod, struct mie_func *func);
+extern struct mie_data *mie_module_get_string_ptr(
+	struct mie_module *mod, const char *s);
 extern struct mie_data *mie_module_get_data(
 	struct mie_module *mod, const char *name);
 extern enum b_status mie_module_put_data(
-	struct mie_module *mod, struct mie_data *data);
+	struct mie_module *mod, struct mie_data *data, const char *name);
+
+extern struct mie_value *mie_module_generate_value_name(
+	struct mie_module *mod, struct mie_value *val, const char *hint);
 
 #endif

mie/module.c

@@ -1,6 +1,9 @@
+#include <blue/object/hashmap.h>
+#include <mie/const.h>
 #include <mie/data.h>
 #include <mie/func.h>
 #include <mie/module.h>
+#include <mie/type.h>
 #include <stdlib.h>
 #include <string.h>
 
@@ -15,6 +18,10 @@ struct mie_module *mie_module_create(void)
 
 	mie_value_init(&out->m_base, MIE_VALUE_MODULE);
 
+	out->m_names = mie_name_map_create();
+	out->m_data = b_hashmap_create(NULL, NULL);
+	out->m_data_strings = b_hashmap_create(NULL, NULL);
+
 	return out;
 }
 
@@ -23,25 +30,77 @@ void mie_module_add_function(struct mie_module *mod, struct mie_func *func)
 	b_queue_push_back(&mod->m_func, &func->f_base.v_entry);
 }
 
-struct mie_data *mie_module_get_data(struct mie_module *mod, const char *name)
+struct mie_data *mie_module_get_string_ptr(struct mie_module *mod, const char *s)
 {
-	b_queue_iterator it;
-	b_queue_foreach (&it, &mod->m_data) {
-		struct mie_data *data = (struct mie_data *)b_unbox(
-			struct mie_value, it.entry, v_entry);
-		if (!strcmp(data->d_base.v_name.n_str, name)) {
-			return data;
-		}
+	b_hashmap_key key = {
+		.key_data = s,
+		.key_size = strlen(s),
+	};
+
+	const b_hashmap_value *entry = b_hashmap_get(mod->m_data_strings, &key);
+	if (entry) {
+		return entry->value_data;
 	}
 
 	return NULL;
 }
 
-enum b_status mie_module_put_data(struct mie_module *mod, struct mie_data *data)
+struct mie_data *mie_module_get_data(struct mie_module *mod, const char *name)
 {
-	/* TODO enforce unique names */
-	b_queue_push_back(&mod->m_data, &data->d_base.v_entry);
+	b_hashmap_key key = {
+		.key_data = name,
+		.key_size = strlen(name),
+	};
+
+	const b_hashmap_value *entry = b_hashmap_get(mod->m_data, &key);
+	if (entry) {
+		return entry->value_data;
+	}
+
+	/* don't search m_data_strings, as it is keyed by the string contents
+	 * rather than the data entry name, and its entries are duplicated in
+	 * m_data anyway */
+
+	return NULL;
+}
+
+enum b_status mie_module_put_data(
+	struct mie_module *mod, struct mie_data *data, const char *name)
+{
+	struct mie_value *v
+		= mie_module_generate_value_name(mod, MIE_VALUE(data), name);
+
+	b_hashmap_key key = {
+		.key_data = v->v_name.n_str,
+		.key_size = strlen(name),
+	};
+
+	b_hashmap_value value = {
+		.value_data = data,
+		.value_size = sizeof *data,
+	};
+
+	b_hashmap_put(mod->m_data, &key, &value);
+
+	if (data->d_type != MIE_DATA_CONST) {
 		return B_SUCCESS;
+	}
+
+	struct mie_const *const_data = data->d_const.c_value;
+	if (const_data->c_type->t_id == MIE_TYPE_STR) {
+		key.key_data = const_data->c_v.v_str;
+		key.key_size = strlen(key.key_data);
+		b_hashmap_put(mod->m_data_strings, &key, &value);
+	}
+
+	return B_SUCCESS;
+}
+
+struct mie_value *mie_module_generate_value_name(
+	struct mie_module *mod, struct mie_value *val, const char *hint)
+{
+	struct mie_name *name = mie_name_map_put(mod->m_names, &val->v_name, hint);
+	return b_unbox(struct mie_value, name, v_name);
 }
 
 static void cleanup(struct mie_value *value)
@@ -56,13 +115,6 @@ static void cleanup(struct mie_value *value)
 		mie_value_destroy(v);
 	}
 
-	b_queue_iterator_begin(&module->m_data, &it);
-	while (b_queue_iterator_is_valid(&it)) {
-		struct mie_value *v = b_unbox(struct mie_value, it.entry, v_entry);
-		b_queue_iterator_erase(&it);
-		mie_value_destroy(v);
-	}
-
 	b_queue_iterator_begin(&module->m_types, &it);
 	while (b_queue_iterator_is_valid(&it)) {
 		struct mie_value *v = b_unbox(struct mie_value, it.entry, v_entry);
@@ -76,6 +128,9 @@ static void cleanup(struct mie_value *value)
 		b_queue_iterator_erase(&it);
 		mie_value_destroy(v);
 	}
+
+	b_hashmap_release(module->m_data_strings);
+	b_hashmap_release(module->m_data);
 }
 
 const struct mie_value_type module_value_type = {