Querying CaosDB
You should have the web interface of a CaosDB instance at hand. If you do not have one, you can visit https://demo.indiscale.com
Introduction
The semantic data model of CaosDB allows efficient data access. The CaosDB Query Language (CQL) is used to search data. Queries can be entered in the webinterface under the respective menu entry.
Let’s start with a simple one:
FIND RECORD MusicalInstrument
Most queries simply start with the FIND
keyword and describe what we are
looking for behind that. The RECORD
keyword denotes that we are only looking
for Records (and not Files, Properties or RecordTypes). Finally, we provided
a RecordType name: MusicalInstrument. This means that we will get all Records
that have this RecordType as parent. Try it out!
Let’s look at:
FIND Guitar
When we leave out the RECORD
keyword, we will get every entity that is a
Guitar. When you submit this query you should find also a RecordType Guitar
in the results. Using FIND RecordType Guitar
would restrict the result to
only that RecordType.
Note, that you cannot only provide RecordType names after the FIND
, but names
in general: FIND RECORD Nice Guitar
. This will give you a Record with the
name “Nice Guitar” (if one exists… and there should be one in the demo instance).
While it does not matter whether you use capital letters or not, the names have to
be exact. There are two features that make it easy to use names for querying
in spite of this:
- You can use “*” to match any string. E.g. FIND RECORD Nice*
- After typing three letters, names that start with those three are
suggested by the auto completion.
Note
Train yourself by trying to guess what the result will be before actually executing the query.
Searching Data Using Properties
Looking for entities with certain names or such that have certain parents is nice. However, the queries become really useful if we can impose further conditions on the results. Let’s start with an example again:
FIND Guitar with price > 10000
This should list expensive guitars where are in the demo instance. Thus, we are using a property (the price) of the Guitar Records to restrict the result set. In general this looks like:
FIND <Name> <Property Filter>
Typically, the filter has the form <Property> <Operator> <Value>
,
for example length >= 0.7mm
.
There are many filters available. You can check the specification for a comprehensive description of
those. Here, we will only look at the most common examples.
If you only want to assure that Records have a certain Property, without imposing constrains on the value, you can use:
FIND RECORD MusicalInstrument WITH Manufacturer
Similarly, to what we saw above when using incomplete names, you can use a “*” to match parts of text properties:
FIND RECORD WITH serialNumber like KN*
There is large number of operators that can be used together with dates or timestamps. One of the most useful is probably:
FIND RECORD WITH date in 2019
A lot of valuable information is often stored in the relations among data, i.e. in the references of entities. So how can we use those?:
FIND RECORD WHICH REFERENCES A Guitar
This should be pretty self explanatory. And it is also possible to check for references in the other direction:
FIND RECORD WHICH IS REFERENCED BY A Analysis
You can also simply provide the ID of the entity:
FIND RECORD WHICH IS REFERENCED BY 123``
Using Multiple Filters
Often, one condition is not sufficient. Thus multiple filters/conditions can be combined. This can for example be done using the following structure:
FIND <Name> <Property Filter> (AND|OR) <Property Filter>
An example would be:
FIND Guitar WITH price>48 AND electric=TRUE
Furthermore, reference conditions can be nested:
FIND <Name> WHICH REFERENCES <Name> WHICH REFERENCES <Name>
For example:
FIND Manufacturer WHICH IS REFERENCED BY Guitar WHICH IS REFERENCED BY Analysis
Restricting Result Information
Using COUNT
instead of FIND
will only return the number of
entities in the result set.
Note
This is often useful when experimenting with queries.
Using SELECT ... FROM
instead of FIND
returns specific
information in a table. A comma separated list of Property names can be provided behind the
SELECT
keyword:
SELECT price, electric FROM Guitar
Or:
SELECT quality_factor, report, date FROM Analysis WHICH REFERENCES A Guitar WITH electric=TRUE