How to store k-v relationship in Lucene? - elasticsearch

Currently, I use indice to store a key-value relationship, which helps to do 'join' query between indices.
Minimum example may be like:
Indice A: field id, field xxx
Indice B: field id, field xxx
Indice C: field a, id of a; field b, id of b
I want to do query: Query of A and Query of B and a join condition that a.id = b.id.
Indice C help me to organize the query and optimize the query. After get the result of Query of A, I will look up the indice c, and then search the docs that with eligible id.
I tried to implements this join relationship with LongPoint.
Document doc = new Document();
doc.add(new LongPoint("a", idOfA));
doc.add(new LongPoint("b", idOfB));
This will contruct inverted indice to implement the relationship.
Actually, this may be not a good choice to implements k-v relationship of <Long, List<Long>>.
Is there any structure or anther option so that I can get a better performance of storage and queries.
Thank you.

Related

How to search on a many-to-many field

Let's assume I have the below structs
type Job struct {
ID string `sql:"type:uuid;primary_key;"`
Title string `json:"title,omitempty"`
Skills []*skill.Skill `json:"skills,omitempty"gorm:"many2many:job_skill;"`
}
type Skill struct {
Name string `json:"name" gorm:"primary_key"`
}
To query all the jobs I do:
jobs := &[]Job{}
gorm.DB.Preload("Skills").Find(&jobs)
How do I search for a Job that contains a certain skill? I have tried the below but it says the column does not exist.
s := "golang"
jobs := &[]Job{}
gorm.DB.Preload("Skills").Where("skill = ?", s).Find(&jobs)
I can see my issues, = doesn't seem to be the correct operator as it needs to search in a list. And it also isn't loading the join table as I assumed it would
Debug output
pq: column "skill" does not exist
SELECT * FROM "jobs" WHERE skill = 'golang'
The Preload method and the Associations feature help you load your fields by constructing basic SQL queries based on the relationships you have created. Queries like loading all skills for a specific job (or jobs). But it doesn't go any more complex than that.
Rather, think of go-gorm as a way to construct your SQL queries and load the data into your models.
Having that in mind, one solution would be to use Joins to include the skill table.
gorm.DB.Preload("Skills")
.Joins("INNER JOIN job_skill js ON js.job_id = jobs.id").
.Joins("INNER JOIN skill s ON s.id = js.skill_id").
.Where("s.name = ?", s).Find(&jobs)

Criteria Query to retrieve data from DB using specification and predicate

I have two tables (user, vehicles) and i want to write criteria query to retrieve data from db using criteria query specification and predicate to both Join Tables.
select ur.id, count (ur.vehicle_FK) from user so
inner join VEHICLE vhe on vhe.user_id_FK = ur."ID"
group by ur.id, vhe.user_id_FK;
How to implement it using criteria query ??
Try something like this :
Criteria criteria = session.createCriteria(User.class, "user");
criteria.createAlias("user.vehicle_FK", "vehicle", Criteria.INNER_JOIN);
criteria.setProjection(
Projections.projectionList().add(Projections.groupProperty("user.id"))
.add(Projections.countDistinct("user.id")));
Parameters in Criteria Queries
The following query string represents a JPQL query with a parameter:
SELECT c FROM Country c WHERE c.population > :p
An equivalent query can be built using the JPA criteria API as follows:
CriteriaBuilder cb = em.getCriteriaBuilder();
CriteriaQuery<Country> q = cb.createQuery(Country.class);
Root<Country> c = q.from(Country.class);
ParameterExpression<Integer> p = cb.parameter(Integer.class);
q.select(c).where(cb.gt(c.get("population"), p));
The ParameterExpression instance, p, is created to represent the query parameter. The where method sets the WHERE clause. As shown above, The CriteriaQuery interface supports method chaining. See the links in the next sections of this page for detailed explanations on how to set criteria query clauses and build criteria expressions.
You can find more examples here
https://www.objectdb.com/java/jpa/query/criteria
Remember to post an answer when you find one! :)

Drupal 7 | Query through multiple node references

Let me start with structure first:
[main_node]->field_reference_to_sub_node->[sub_node]->field_ref_to_sub_sub_node->[sub_sub_node]
[sub_sub_node]->field_type = ['wrong_type', 'right_type']
How to efficiently query all [sub_sub_node] ids with right_type, referenced by main_node (which is current opened node)?
Doing node_load on foreach seems a bit of overkill for this. Anybody has some better solutions? Greatly appreciated!
If you want to directly query the table of the fields:
$query = db_select('node', 'n')->fields('n_sub_subnode', array('nid'));
$query->innerJoin('table_for_field_reference_to_sub_node', 'subnode', "n.nid = subnode.entity_id AND subnode.entity_type='node'");
$query->innerJoin('node', 'n_subnode', 'subnode.subnode_target_id = n_subnode.nid');
$query->innerJoin('table_for_field_ref_to_sub_sub_node', 'sub_subnode', "n_subnode.nid = sub_subnode.entity_id AND sub_subnode.entity_type='node'");
$query->innerJoin('node', 'n_sub_subnode', 'sub_subnode.sub_subnode_target_id = n_sub_subnode.nid');
$query->innerJoin('table_for_field_type', 'field_type', "n_sub_subnode.nid = field_type.entity_id AND field_type.entity_type='node'");
$query->condition('n.nid', 'your_main_node_nid');
$query->condition('field_type.field_type_value', 'right_type');
Here is the explanation of each line:
$query = db_select('node', 'n')->fields('n_sub_subnode', array('nid'));
We start by querying the base node table, with the alias 'n'. This is the table used for the 'main_node'. The node ids which will be returned will be however from another alias (n_sub_subnode), you will see it below.
$query->innerJoin('table_for_field_reference_to_sub_node', 'subnode', "n.nid = subnode.entity_id AND subnode.entity_type='node'");
The first join is with the table of the field_reference_to_sub_node field, so you have to replace this with the actual name of the table. This is how we will get the references to the subnodes.
$query->innerJoin('node', 'n_subnode', 'subnode.subnode_target_id = n_subnode.nid');
A join back to the node table for the subnodes. You have to replace the 'subnode_target_id' with the actual field for the target id from the field_reference_to_sub_node table. The main purpose of this join is to make sure there are valid nodes in the subnode field.
$query->innerJoin('table_for_field_ref_to_sub_sub_node', 'sub_subnode', "n_subnode.nid = sub_subnode.entity_id AND sub_subnode.entity_type='node'");
The join to the table that contains references to the sub_sub_node, so you have to replace the 'table_for_field_ref_to_sub_sub_node' with the actual name of the table. This is how we get the references to the sub_sub_nodes.
$query->innerJoin('node', 'n_sub_subnode', 'sub_subnode.sub_subnode_target_id = n_sub_subnode.nid');
The join back to the node table for the sub_sub_nodes, to make sure we have valid references. You have to replace the 'sub_subnode_target_id' with the actual field for the target id from the 'field_ref_to_sub_sub_node' table.
$query->innerJoin('table_for_field_type', 'field_type', "n_sub_subnode.nid = field_type.entity_id AND field_type.entity_type='node'");
And we can now finally join the table with the field_type information. You have to replace the 'table_for_field_type' with the actual name of the table.
$query->condition('n.nid', 'your_main_node_nid');
You can put now a condition for the main node id if you want.
$query->condition('field_type.field_type_value', 'right_type');
And the condition for the field type. You have to replace the 'field_type_value' with the actual name of the table field for the value.
Of course, if you are really sure that you always have valid references, you can skip the joins to the node table and directly join the field tables using the target id and the entity_id fields (basically the target_id from on field table has to be the entity_id for the next one).
I really hope I do not have typos, so please check the queries carefully.

Obtain documents by multiple ids

I have a MongoDB (and for queries use Java Spring) database of persons which has unique ids. Let say that I have two ids and I want to obtain their persons. Something like this.
Query q = new Query();
Criteria c = new Criteria().orOperator(Criteria.where("id").is("1"),
Criteria.where("id").is("2"));
q.addCriteria(c);
operations.find(q, Person.class);
But this does not work becase I use "id" field two times. Any idea?
ops.find(query(where("id").in([1, 2]));

How do I improve the performance of this simple LINQ?

I have two tables, one parent "Point" and one child "PointValue", connected by a single foreign key "PointID", making a one-to-many relation in SQL Server 2005.
I have a LINQ query:
var points = from p in ContextDB.Points
//join v in ContextDB.PointValues on p.PointID equals v.PointID
where p.InstanceID == instanceId
orderby p.PointInTime descending
select new
{
Point = p,
Values = p.PointValues.Take(16).ToList()
};
As you can see from the commented out join and the "Values" assignment, the "Point" table has a relation to "PointValue" (called "Points" and "PointValues" by LINQ).
When iterating through the "var points" IQueryable (say, when binding it to a GridView, etc.) the initial query is very fast, however iterating through the "Values" property is very slow. SQL Profiler shows me that for each value in the "points" IQueryable another query is executed.
How do I get this to be one query?
Interestingly, the initial query becomes very slow when the join is uncommented.
I think you want to use the DataLoadOptions.LoadWith method, described here:
http://msdn.microsoft.com/en-us/library/system.data.linq.dataloadoptions.loadwith.aspx
In your case you would do something like the following, when creating your DataContext:
DataLoadOptions options = new DataLoadOptions();
ContextDB.LoadOptions = options;
options.LoadWith((Point p) => p.PointValues);
You should make sure that the PointValues table has an index on the PointID column.
See also this SO question: Does Foreign Key improve query performance?

Resources