This is an API function but if you use it you risk duplicating queries + processing. This patch adds static caching.
function taxonomy_get_vocabularies($type = NULL) {
+ static $vocabularies_cache = array();
+ static $vocabularies_null;
+
if ($type) {
+ if (is_array($vocabularies_cache[$type])) {
+ return $vocabularies_cache[$type];
+ }
$result = db_query(db_rewrite_sql("SELECT v.vid, v.*, n.type FROM {vocabulary} v LEFT JOIN {vocabulary_node_types} n ON v.vid = n.vid WHERE n.type = '%s' ORDER BY v.weight, v.name", 'v', 'vid'), $type);
}
else {
+ if (is_array($vocabularies_null)) {
+ return $vocabularies_null;
+ }
$result = db_query(db_rewrite_sql('SELECT v.*, n.type FROM {vocabulary} v LEFT JOIN {vocabulary_node_types} n ON v.vid = n.vid ORDER BY v.weight, v.name', 'v', 'vid'));
}
@@ -695,7 +704,15 @@ function taxonomy_get_vocabularies($type
$vocabularies[$voc->vid] = $voc;
}
+ if ($type) {
+ $vocabularies_cache[$type] = $vocabularies;
+ }
+ else {
+ $vocabularies_null = $vocabularies;
+ }
+
return $vocabularies;
+
}
| Comment | File | Size | Author |
|---|---|---|---|
| #10 | taxonomy-163018.patch | 2.15 KB | catch |
| #9 | taxonomy-163018.patch | 2.15 KB | catch |
| #4 | taxonomy_get_vocabularies_cache2.patch.txt | 1.39 KB | robertdouglass |
| #1 | taxonomy_get_vocabularies_cache2.patch | 1.47 KB | bennybobw |
| taxonomy_get_vocabularies_cache.patch | 1.48 KB | robertdouglass |
Comments
Comment #1
bennybobw commentedI looked through this patch and everything looks good. I agree this should be cached. The only thing I wasn't sure of was the variable name $vocabularies_null, which as I understand it, gets set if a type isn't specified.
What about $vocabularies_all or something like that? Maybe someone else with a weightier opinion can give an opinion and change the patch code status.
Attaching a patch rolled against latest HEAD without offsets.
Comment #2
catchpatch still applies with offset...
Comment #3
moshe weitzman commentedIs there a or two modules out thee that calls this function repeatedly? Just want some justification for the cache.
Comment #4
robertdouglass commentedthat's on the D5 contrib repository. Rerolled patch to apply without offsets.
Comment #5
catchStill valid in D7.
We should go the whole hog and change this to taxonomy_vocabulary_load_multiple() so that hook_taxonomy_vocabulary_load() works with it and we can share the static between taxonomy_get_vocabulary() too.
Comment #6
catchtaxonomy_vocabulary_load_multiple() is in.
Comment #7
catchThis no longer applies to Drupal 7, but it's still valid for D6.
Comment #9
catchHere's a re-roll that also turns taxonomy_vocabulary_load() into a wrapper around taxonomy_get_vocabularies() - this means more vocabularies loaded if you were to only try to load one, but that is not often the case.
Comment #10
catchFixed a notice in taxonomy_get_vocabularies().
Comment #11
mr.j commentedPatch is good. It saves over 150 queries on loading the home page of a site I am working on.
However I don't think its a good idea to change taxonomy_vocabulary_load() because it has a reset flag. Assuming it is used somewhere, every time you call the function with $reset = TRUE, it will just grab the cached taxonomies from taxonomy_get_vocabularies().
Comment #12
xanoThis is now done by the entity API.
Comment #13
xanoOops. Didn't see this was D6.
Comment #15
xanoComment #17
djbobbydrake commentedI used this patch in our environment, and it appears that patching the taxonomy_vocabulary_load() function breaks two simpletests: PathautoTaxonomyTokenTestCase and TokenTaxonomyTestCase. It seems like we should pull the patch to taxonomy_vocabulary_load(), but leave the patch to taxonomy_get_vocabularies().
The setup method for TokenTaxonomyTestCase has taxonomy_vocabulary_load(0, TRUE), which return different values b/n the patched vs non-patched taxonomy_vocabulary_load() versions.