[tracker/wip/carlosg/anonymous-nodes: 12/24] libtracker-data: Deal with triple subjects as int64 ROWIDs
- From: Carlos Garnacho <carlosg src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [tracker/wip/carlosg/anonymous-nodes: 12/24] libtracker-data: Deal with triple subjects as int64 ROWIDs
- Date: Fri, 8 Oct 2021 09:40:54 +0000 (UTC)
commit 3cb76b3da8f7eb125c7fb27b1daca9c5e49a4504
Author: Carlos Garnacho <carlosg gnome org>
Date: Sat Sep 25 12:52:29 2021 +0200
libtracker-data: Deal with triple subjects as int64 ROWIDs
Make the deepmost tracker_data_insert/delete_statement()
(and it's ugly tracker_data_update_statement() brother) handle
triple subjects as ROWIDs, instead of URI strings.
This makes the caller responsible for transforming these URI
strings to ROWIDs (maybe creating a new entry in the Resource
table if necessary), but makes this internal machinery agnostic
of the resource URIs themselves.
This will be useful to minimize URI<->ID transformations, and
deal with truly blank nodes.
src/libtracker-data/tracker-data-manager.c | 44 +++++++++--
src/libtracker-data/tracker-data-update.c | 123 ++++++++++++-----------------
src/libtracker-data/tracker-data-update.h | 6 +-
src/libtracker-data/tracker-sparql.c | 16 +++-
4 files changed, 104 insertions(+), 85 deletions(-)
---
diff --git a/src/libtracker-data/tracker-data-manager.c b/src/libtracker-data/tracker-data-manager.c
index 2eeb144c1..d94e8bb7e 100644
--- a/src/libtracker-data/tracker-data-manager.c
+++ b/src/libtracker-data/tracker-data-manager.c
@@ -568,6 +568,7 @@ update_property_value (TrackerDataManager *manager,
if (!unsup_onto_err) {
GValue value = G_VALUE_INIT;
+ gint64 subject_id = 0;
tracker_data_query_string_to_value (manager,
str, NULL,
@@ -575,7 +576,12 @@ update_property_value (TrackerDataManager *manager,
&value, &error);
if (!error)
- tracker_data_delete_statement (manager->data_update, NULL,
subject, pred, &value, &error);
+ subject_id = tracker_data_update_ensure_resource
(manager->data_update,
+ subject,
+ &error);
+
+ if (!error)
+ tracker_data_delete_statement (manager->data_update, NULL,
subject_id, pred, &value, &error);
g_value_unset (&value);
if (!error)
@@ -601,6 +607,7 @@ update_property_value (TrackerDataManager *manager,
if (!error && needed && object) {
GValue value = G_VALUE_INIT;
+ gint64 subject_id = 0;
tracker_data_query_string_to_value (manager,
object, NULL,
@@ -608,7 +615,12 @@ update_property_value (TrackerDataManager *manager,
&value, &error);
if (!error)
- tracker_data_insert_statement (manager->data_update, NULL, subject,
+ subject_id = tracker_data_update_ensure_resource (manager->data_update,
+ subject,
+ &error);
+
+ if (!error)
+ tracker_data_insert_statement (manager->data_update, NULL, subject_id,
pred, &value,
&error);
g_value_unset (&value);
@@ -1537,6 +1549,7 @@ check_for_deleted_domain_index (TrackerDataManager *manager,
TrackerProperty *prop = l->data;
const gchar *uri;
GValue value = G_VALUE_INIT;
+ gint64 class_id = 0;
TRACKER_NOTE (ONTOLOGY_CHANGES,
g_message ("Ontology change: deleting nrl:domainIndex: %s",
@@ -1550,8 +1563,14 @@ check_for_deleted_domain_index (TrackerDataManager *manager,
uri, NULL,
tracker_property_get_data_type (property),
&value, error)) {
+ class_id = tracker_data_update_ensure_resource (manager->data_update,
+ tracker_class_get_uri (class),
+ error);
+ }
+
+ if (class_id != 0) {
tracker_data_delete_statement (manager->data_update, NULL,
- tracker_class_get_uri (class),
+ class_id,
property,
&value,
error);
@@ -1721,6 +1740,7 @@ check_for_deleted_super_properties (TrackerDataManager *manager,
const gchar *object = tracker_property_get_uri (prop_to_remove);
const gchar *subject = tracker_property_get_uri (property);
GValue value = G_VALUE_INIT;
+ gint64 subject_id;
property = tracker_ontologies_get_property_by_uri (ontologies,
TRACKER_PREFIX_RDFS
"subPropertyOf");
@@ -1740,7 +1760,12 @@ check_for_deleted_super_properties (TrackerDataManager *manager,
&value, &n_error);
if (!n_error)
- tracker_data_delete_statement (manager->data_update, NULL, subject,
+ subject_id = tracker_data_update_ensure_resource (manager->data_update,
+ subject,
+ &n_error);
+
+ if (!n_error)
+ tracker_data_delete_statement (manager->data_update, NULL, subject_id,
property, &value, &n_error);
g_value_unset (&value);
@@ -2194,6 +2219,7 @@ tracker_data_ontology_process_statement (TrackerDataManager *manager,
{
TrackerProperty *property;
GValue value = G_VALUE_INIT;
+ gint64 subject_id = 0;
if (g_strcmp0 (predicate, RDF_TYPE) == 0) {
if (g_strcmp0 (object, RDFS_CLASS) == 0) {
@@ -2284,14 +2310,20 @@ tracker_data_ontology_process_statement (TrackerDataManager *manager,
&value, error))
goto out;
+ subject_id = tracker_data_update_ensure_resource (manager->data_update,
+ subject,
+ error);
+ if (subject_id == 0)
+ goto out;
+
if (tracker_property_get_is_new (property) ||
tracker_property_get_multiple_values (property)) {
tracker_data_insert_statement (manager->data_update, NULL,
- subject, property, &value,
+ subject_id, property, &value,
error);
} else {
tracker_data_update_statement (manager->data_update, NULL,
- subject, property, &value,
+ subject_id, property, &value,
error);
}
diff --git a/src/libtracker-data/tracker-data-update.c b/src/libtracker-data/tracker-data-update.c
index 99d422d64..9bf38e806 100644
--- a/src/libtracker-data/tracker-data-update.c
+++ b/src/libtracker-data/tracker-data-update.c
@@ -69,7 +69,6 @@ struct _TrackerDataUpdateBufferGraph {
struct _TrackerDataUpdateBufferResource {
const TrackerDataUpdateBufferGraph *graph;
- const gchar *subject;
gint64 id;
gboolean create;
gboolean modified;
@@ -153,10 +152,6 @@ static gboolean delete_metadata_decomposed (TrackerData *data,
TrackerProperty *property,
const GValue *object,
GError **error);
-static gboolean resource_buffer_switch (TrackerData *data,
- const gchar *graph,
- const gchar *subject,
- GError **error);
static gboolean update_resource_single (TrackerData *data,
const gchar *graph,
TrackerResource *resource,
@@ -166,13 +161,13 @@ static gboolean update_resource_single (TrackerData *data,
void tracker_data_insert_statement_with_uri (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error);
void tracker_data_insert_statement_with_string (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error);
@@ -1218,7 +1213,6 @@ static void resource_buffer_free (TrackerDataUpdateBufferResource *resource)
{
g_hash_table_unref (resource->predicates);
g_hash_table_unref (resource->tables);
- resource->subject = NULL;
g_ptr_array_free (resource->types, TRUE);
resource->types = NULL;
@@ -1676,7 +1670,7 @@ get_old_property_values (TrackerData *data,
return old_values;
}
-static const gchar *
+static gint64
get_bnode_for_resource (GHashTable *bnodes,
TrackerData *data,
TrackerResource *resource,
@@ -1689,16 +1683,16 @@ get_bnode_for_resource (GHashTable *bnodes,
identifier = tracker_resource_get_identifier (resource);
bnode = g_hash_table_lookup (bnodes, identifier);
if (bnode)
- return bnode;
+ return query_resource_id (data, bnode, error);
iface = tracker_data_manager_get_writable_db_interface (data->manager);
bnode = tracker_data_update_ensure_new_bnode (data, iface, error);
if (!bnode)
- return NULL;
+ return 0;
g_hash_table_insert (bnodes, g_strdup (identifier), bnode);
- return bnode;
+ return query_resource_id (data, bnode, error);
}
static gboolean
@@ -2123,7 +2117,7 @@ ensure_graph_buffer (TrackerDataUpdateBuffer *buffer,
}
graph_buffer->resources =
- g_hash_table_new_full (g_str_hash, g_str_equal, g_free,
+ g_hash_table_new_full (g_int64_hash, g_int64_equal, NULL,
(GDestroyNotify) resource_buffer_free);
g_ptr_array_add (buffer->graphs, graph_buffer);
@@ -2133,7 +2127,7 @@ ensure_graph_buffer (TrackerDataUpdateBuffer *buffer,
static gboolean
resource_buffer_switch (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
GError **error)
{
TrackerDataUpdateBufferGraph *graph_buffer;
@@ -2141,7 +2135,7 @@ resource_buffer_switch (TrackerData *data,
if (data->resource_buffer != NULL &&
g_strcmp0 (data->resource_buffer->graph->graph, graph) == 0 &&
- strcmp (data->resource_buffer->subject, subject) == 0) {
+ data->resource_buffer->id == subject) {
/* Resource buffer stays the same */
return TRUE;
}
@@ -2160,28 +2154,19 @@ resource_buffer_switch (TrackerData *data,
return FALSE;
data->resource_buffer =
- g_hash_table_lookup (graph_buffer->resources, subject);
+ g_hash_table_lookup (graph_buffer->resources, &subject);
if (data->resource_buffer == NULL) {
TrackerDataUpdateBufferResource *resource_buffer;
- gchar *subject_dup;
- gint64 resource_id;
gboolean create;
GPtrArray *rdf_types;
- /* subject not yet in cache, retrieve or create ID */
- resource_id = tracker_data_update_ensure_resource (data,
- subject,
- error);
- if (resource_id == 0)
- return FALSE;
-
create = g_hash_table_contains (data->update_buffer.new_resources,
- subject);
+ &subject);
if (!create) {
rdf_types = tracker_data_query_rdf_type (data->manager,
graph,
- resource_id,
+ subject,
&inner_error);
if (!rdf_types) {
g_propagate_error (error, inner_error);
@@ -2190,9 +2175,7 @@ resource_buffer_switch (TrackerData *data,
}
resource_buffer = g_slice_new0 (TrackerDataUpdateBufferResource);
- resource_buffer->id = resource_id;
- subject_dup = g_strdup (subject);
- resource_buffer->subject = subject_dup;
+ resource_buffer->id = subject;
resource_buffer->create = create;
resource_buffer->fts_updated = FALSE;
@@ -2205,7 +2188,7 @@ resource_buffer_switch (TrackerData *data,
resource_buffer->tables = g_hash_table_new_full (g_str_hash, g_str_equal, g_free,
(GDestroyNotify) cache_table_free);
resource_buffer->graph = graph_buffer;
- g_hash_table_insert (graph_buffer->resources, subject_dup, resource_buffer);
+ g_hash_table_insert (graph_buffer->resources, &resource_buffer->id, resource_buffer);
data->resource_buffer = resource_buffer;
}
@@ -2216,28 +2199,20 @@ resource_buffer_switch (TrackerData *data,
void
tracker_data_delete_statement (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error)
{
TrackerClass *class;
- gint64 subject_id = 0;
gboolean change = FALSE;
TrackerOntologies *ontologies;
- g_return_if_fail (subject != NULL);
+ g_return_if_fail (subject != 0);
g_return_if_fail (predicate != NULL);
g_return_if_fail (object != NULL);
g_return_if_fail (data->in_transaction);
- subject_id = query_resource_id (data, subject, error);
-
- if (subject_id == 0) {
- /* subject not in database */
- return;
- }
-
if (!resource_buffer_switch (data, graph, subject, error))
return;
@@ -2275,7 +2250,7 @@ tracker_data_delete_statement (TrackerData *data,
static gboolean
delete_all_helper (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *subproperty,
TrackerProperty *property,
GArray *old_values,
@@ -2350,27 +2325,19 @@ delete_all_helper (TrackerData *data,
static gboolean
tracker_data_delete_all (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
const gchar *predicate,
GError **error)
{
- gint64 subject_id = 0;
TrackerOntologies *ontologies;
TrackerProperty *property;
GArray *old_values;
GError *inner_error = NULL;
- g_return_val_if_fail (subject != NULL, FALSE);
+ g_return_val_if_fail (subject != 0, FALSE);
g_return_val_if_fail (predicate != NULL, FALSE);
g_return_val_if_fail (data->in_transaction, FALSE);
- subject_id = query_resource_id (data, subject, error);
-
- if (subject_id == 0) {
- /* subject not in database */
- return FALSE;
- }
-
if (!resource_buffer_switch (data, graph, subject, error))
return FALSE;
@@ -2393,7 +2360,7 @@ tracker_data_delete_all (TrackerData *data,
static gboolean
delete_single_valued (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
gboolean super_is_single_valued,
GError **error)
@@ -2451,12 +2418,12 @@ delete_single_valued (TrackerData *data,
void
tracker_data_insert_statement (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error)
{
- g_return_if_fail (subject != NULL);
+ g_return_if_fail (subject != 0);
g_return_if_fail (predicate != NULL);
g_return_if_fail (object != NULL);
g_return_if_fail (data->in_transaction);
@@ -2471,7 +2438,7 @@ tracker_data_insert_statement (TrackerData *data,
void
tracker_data_insert_statement_with_uri (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error)
@@ -2482,7 +2449,7 @@ tracker_data_insert_statement_with_uri (TrackerData *data,
gboolean change = FALSE;
TrackerOntologies *ontologies;
- g_return_if_fail (subject != NULL);
+ g_return_if_fail (subject != 0);
g_return_if_fail (predicate != NULL);
g_return_if_fail (object != NULL);
g_return_if_fail (data->in_transaction);
@@ -2544,7 +2511,7 @@ tracker_data_insert_statement_with_uri (TrackerData *data,
void
tracker_data_insert_statement_with_string (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error)
@@ -2553,7 +2520,7 @@ tracker_data_insert_statement_with_string (TrackerData *data,
gboolean change;
gint64 pred_id = 0;
- g_return_if_fail (subject != NULL);
+ g_return_if_fail (subject != 0);
g_return_if_fail (predicate != NULL);
g_return_if_fail (object != NULL);
g_return_if_fail (data->in_transaction);
@@ -2585,7 +2552,7 @@ tracker_data_insert_statement_with_string (TrackerData *data,
void
tracker_data_update_statement (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error)
@@ -2593,7 +2560,7 @@ tracker_data_update_statement (TrackerData *data,
TrackerOntologies *ontologies;
GError *new_error = NULL;
- g_return_if_fail (subject != NULL);
+ g_return_if_fail (subject != 0);
g_return_if_fail (predicate != NULL);
g_return_if_fail (data->in_transaction);
@@ -2852,7 +2819,7 @@ tracker_data_load_turtle_file (TrackerData *data,
TrackerTurtleReader *reader = NULL;
TrackerOntologies *ontologies;
GError *inner_error = NULL;
- const gchar *subject, *predicate_str, *object_str, *langtag;
+ const gchar *subject_str, *predicate_str, *object_str, *langtag;
gboolean object_is_uri;
goffset last_parsed_line_no, last_parsed_column_no;
gchar *ontology_uri = g_file_get_uri (file);
@@ -2864,7 +2831,7 @@ tracker_data_load_turtle_file (TrackerData *data,
ontologies = tracker_data_manager_get_ontologies (data->manager);
while (tracker_turtle_reader_next (reader,
- &subject,
+ &subject_str,
&predicate_str,
&object_str,
&langtag,
@@ -2874,6 +2841,7 @@ tracker_data_load_turtle_file (TrackerData *data,
&inner_error)) {
TrackerProperty *predicate;
GValue object = G_VALUE_INIT;
+ gint64 subject;
predicate = tracker_ontologies_get_property_by_uri (ontologies, predicate_str);
if (predicate == NULL) {
@@ -2892,6 +2860,12 @@ tracker_data_load_turtle_file (TrackerData *data,
&inner_error))
goto failed;
+ subject = tracker_data_update_ensure_resource (data,
+ subject_str,
+ &inner_error);
+ if (inner_error)
+ goto failed;
+
if (object_is_uri) {
tracker_data_insert_statement_with_uri (data, graph,
subject, predicate, &object,
@@ -2987,23 +2961,23 @@ static gboolean
resource_maybe_reset_property (TrackerData *data,
const gchar *graph,
TrackerResource *resource,
- const gchar *subject_uri,
+ gint64 subject,
const gchar *property_uri,
GHashTable *bnodes,
GError **error)
{
GError *inner_error = NULL;
- const gchar *subject;
+ const gchar *subject_name;
/* If the subject is a blank node, this is a whole new insertion.
* We don't need deleting anything then.
*/
- subject = tracker_resource_get_identifier (resource);
- if (g_str_has_prefix (subject, "_:"))
+ subject_name = tracker_resource_get_identifier (resource);
+ if (g_str_has_prefix (subject_name, "_:"))
return TRUE;
if (!tracker_data_delete_all (data,
- graph, subject_uri, property_uri,
+ graph, subject, property_uri,
&inner_error)) {
if (inner_error)
g_propagate_error (error, inner_error);
@@ -3018,7 +2992,7 @@ static gboolean
update_resource_property (TrackerData *data,
const gchar *graph_uri,
TrackerResource *resource,
- const gchar *subject,
+ gint64 subject,
const gchar *property,
GHashTable *visited,
GHashTable *bnodes,
@@ -3087,7 +3061,8 @@ update_resource_single (TrackerData *data,
{
GList *properties, *l;
GError *inner_error = NULL;
- const gchar *subject;
+ const gchar *subject_str;
+ gint64 subject;
gchar *graph_uri = NULL;
gboolean is_bnode = FALSE;
@@ -3098,12 +3073,16 @@ update_resource_single (TrackerData *data,
properties = tracker_resource_get_properties (resource);
- subject = tracker_resource_get_identifier (resource);
- if (!subject || g_str_has_prefix (subject, "_:")) {
+ subject_str = tracker_resource_get_identifier (resource);
+ if (!subject_str || g_str_has_prefix (subject_str, "_:")) {
is_bnode = TRUE;
subject = get_bnode_for_resource (bnodes, data, resource, error);
if (!subject)
return FALSE;
+ } else {
+ subject = tracker_data_update_ensure_resource (data, subject_str, error);
+ if (subject == 0)
+ return FALSE;
}
if (graph) {
diff --git a/src/libtracker-data/tracker-data-update.h b/src/libtracker-data/tracker-data-update.h
index 97428b5a6..123dbf532 100644
--- a/src/libtracker-data/tracker-data-update.h
+++ b/src/libtracker-data/tracker-data-update.h
@@ -59,19 +59,19 @@ GQuark tracker_data_error_quark (void);
/* Metadata */
void tracker_data_delete_statement (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error);
void tracker_data_insert_statement (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error);
void tracker_data_update_statement (TrackerData *data,
const gchar *graph,
- const gchar *subject,
+ gint64 subject,
TrackerProperty *predicate,
const GValue *object,
GError **error);
diff --git a/src/libtracker-data/tracker-sparql.c b/src/libtracker-data/tracker-sparql.c
index d3e2b0bb5..3a3a06c04 100644
--- a/src/libtracker-data/tracker-sparql.c
+++ b/src/libtracker-data/tracker-sparql.c
@@ -2072,6 +2072,7 @@ tracker_sparql_apply_quad (TrackerSparql *sparql,
GError *inner_error = NULL;
TrackerProperty *predicate = NULL;
GValue object = G_VALUE_INIT;
+ gint64 subject = 0;
if ((tracker_token_is_empty (&sparql->current_state->graph) &&
sparql->policy.filter_unnamed_graph) ||
@@ -2086,7 +2087,7 @@ tracker_sparql_apply_quad (TrackerSparql *sparql,
sparql->current_state->type == TRACKER_SPARQL_TYPE_DELETE ||
sparql->current_state->type == TRACKER_SPARQL_TYPE_UPDATE) {
TrackerOntologies *ontologies;
- const gchar *property;
+ const gchar *property, *subject_str;
ontologies = tracker_data_manager_get_ontologies (sparql->data_manager);
property = tracker_token_get_idstring (&sparql->current_state->predicate);
@@ -2109,6 +2110,13 @@ tracker_sparql_apply_quad (TrackerSparql *sparql,
g_propagate_error (error, inner_error);
return FALSE;
}
+
+ subject_str = tracker_token_get_idstring (&sparql->current_state->subject);
+ subject = tracker_data_update_ensure_resource (tracker_data_manager_get_data
(sparql->data_manager),
+ subject_str,
+ error);
+ if (subject == 0)
+ return FALSE;
}
switch (sparql->current_state->type) {
@@ -2131,7 +2139,7 @@ tracker_sparql_apply_quad (TrackerSparql *sparql,
case TRACKER_SPARQL_TYPE_INSERT:
tracker_data_insert_statement (tracker_data_manager_get_data (sparql->data_manager),
tracker_token_get_idstring (&sparql->current_state->graph),
- tracker_token_get_idstring (&sparql->current_state->subject),
+ subject,
predicate,
&object,
&inner_error);
@@ -2139,7 +2147,7 @@ tracker_sparql_apply_quad (TrackerSparql *sparql,
case TRACKER_SPARQL_TYPE_DELETE:
tracker_data_delete_statement (tracker_data_manager_get_data (sparql->data_manager),
tracker_token_get_idstring (&sparql->current_state->graph),
- tracker_token_get_idstring (&sparql->current_state->subject),
+ subject,
predicate,
&object,
&inner_error);
@@ -2147,7 +2155,7 @@ tracker_sparql_apply_quad (TrackerSparql *sparql,
case TRACKER_SPARQL_TYPE_UPDATE:
tracker_data_update_statement (tracker_data_manager_get_data (sparql->data_manager),
tracker_token_get_idstring (&sparql->current_state->graph),
- tracker_token_get_idstring (&sparql->current_state->subject),
+ subject,
predicate,
&object,
&inner_error);
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]
