Visual Query Builder

The graph patterns section lists all the visual patterns you have created for the current graph.

Other than the visual view of the graph pattern, you can optionally add a description for the pattern. This can help other users to understand what problem you are trying to solve.

Remember to save your changes after adding a description.

If you added some parameters in your visual pattern, you need to provide the values to the parameters when running the pattern.

If you enter editing mode of a vertex pattern and expand the Basic Info section, you can edit its name, decide whether or not to put it into result, change its vertex type, provide optional matching conditions by giving a list of ids, or provide a parameter name.

The name supplied in the panel will be used as the vertex’s JSON key when producing output and when the pattern is saved as a GSQL query.

The dropdown under Vertex Type allows you to add or drop IDs or add an input parameter.

If you enter editing mode of an edge pattern and expand the Basic Info section, you can edit its name, decide whether or not to put it into result, or change its edge type. You can also specify this edge pattern as a regex match by providing Repeat as least (a non-negative integer) and Repeat at most (a positive integer). Due to current GSQL limitations, if you decide to change these numbers, you cannot give the edge pattern a name or put it into result.

The name supplied in the panel will be used as the edge’s JSON key when producing output and when the pattern is saved as a GSQL query.

When finished editing, you need to confirm the change. You can also cancel the change if you made a mistake:

If you expand the Filter section, you can add/edit/delete filters for the selected vertex or edge pattern:

Click besides the filter expression and enter editing mode of the filter:

You can add multiple filters for each vertex and edge pattern, and they are AND relationship when executing the pattern. Building the filter is similar with building attribute filter in the Explore Graph page.

One thing special here is that you can refer to attributes on other vertex patterns and edge patterns. Choose Attribute of vertex or edge as expression type, then choose the name of the vertex or edge pattern whose attribute you want to refer to (see above why we need give a name to vertex or edge patterns), then choose the attribute you want to refer to.

When finished with editing, you need confirm or cancel the change:

If you expand the Attachment section, you can add/edit/delete attachments for the selected vertex pattern (attachment on edge patterns is not supported):

Click beside one attachment and enter editing mode for the attachment. You can edit both the attachment expression and attachment name:

When finished with editing, you need to confirm or cancel the change:

In attachments, you can either attach a single value or attach the results of an aggregation function, to all the matched vertex entity results.

Attaching a single value means attaching the given value to all vertex entities within the matched result. If for example, you choose to attach the integer value 1, then that value will be attached to all the matched vertex entity results where the attachment was applied to. If the attachment is applied to an entity that is involved in multiple matches, then the attached value is non-deterministic.

Attaching the results of an aggregation function means grouping all the matching results by the vertex entity which match the vertex pattern, and then aggregated based on the expression.

Take this example:

Consider we have the following matching result:

We have the following attachment result table for city vertex san jose:

As you can see above the values of attaching a single value for datetime, vertex and edge are stored in a list. This is because, we currently do not have an accumulator to store these data types directly yet.

The data types supported for single value attachments are: integer, real, string, bool, datetime, vertex and edge.

If you expand the Order section, you can add/edit/delete ordering for the selected vertex pattern (ordering on edge patterns is not supported):

Click beside the ordering and enter editing mode for the ordering. You can edit both the ordering expression and whether results are in ascending or descending order:

When finished with editing, you need to confirm the change.

You can refer to attachments in ordering expression. You can add multiple orderings, which follow the multi-key ordering rule (upper ordering dominates). We support ordering by data of types: integer, real, string and datetime. However, single value attachments with the type datetime are stored in a list, thus, we currently do not support ordering based on single value attachments that have the datetime data type.

If you only want a subset of your matching result, you can use limit. Only the top limit results will be returned based on your ordering settings. If you don’t have orderings, the result will be randomly picked from all matchings.

If you expand the Limit section, you can add/edit/delete limit for the selected vertex pattern (limit on edge patterns is not supported):

Toggle use limit checkbox to enable/disable limit. You can also edit the limit number:

When finished editing, you need confirm or cancel the change:

Viewing graph schema makes it more convenient for developers to refer to the schema topology logic and easier to construct the visual pattern.

If the pattern execution result contains a graph structure, the result will be visualized in this panel as a graph. The panel is the same as the Explore Graph panel. The only difference is that each time you run a pattern, the previous result will be erased. In Explore Graph, the results are added incrementally.

You can switch to the JSON Result panel to see the result in JSON format.

You can see the JSON response of running the visual pattern from this tab:

If the visual pattern contains errors, you can see them here and debug:

Click , then click two vertex patterns of the same vertex type:

And an intersection widget is added to the visual pattern:

The output vertex pattern means matching all company vertices located at redwood city, and belongs to big data industry.

The output company vertex pattern can be part of another larger pattern. You can think of the input vertex patterns of the widget as constraints of the output vertex pattern.

Click , then click two vertex patterns of the same vertex type. A union widget is added to the visual pattern:

The output vertex pattern means matching all company vertices located at redwood city, or belongs to big data industry.

Click , then click two vertex patterns of the same vertex type. A subtract widget is added to the visual pattern:

The output vertex pattern means matching all company vertices located at redwood city, but not belong to big data industry.

The Within widget allows you to create a vertex pattern whose matching result is constrained within the matching result of another vertex pattern.

For example, say you have a vertex pattern for all people who currently work at company A, and you want to select from them everyone who has bought product B. You can use the Within widget to select from the vertex pattern with all person vertices that work at company A, and create another pattern for people who have bought product B. The output from the second pattern will be a subset of the matching result of the first pattern - that is, the output from the second pattern will be people who work at company A, who bought product B.

Click , and then select a vertex pattern whose result you want to select from. This creates a within widget on the first vertex pattern, and you will see a second vertex pattern added to the graph pattern. Below are two visual examples to select results within a vertex pattern.

In the above example, the graph pattern outputs two bottom patterns which are both within the top pattern. The output vertex pattern means matching all company vertices or all city vertices that a person who attends san jose university is working at or was born in. In other words, a person vertex that attends san jose university needs either a work_at edge connected to a company vertex, or a born_in edge connected to a city vertex to be included in the result.

The output vertex pattern means matching all company vertices and the city vertices that a person who attends san jose university is working at and was born in. In other words, a person vertex needs to have both a word_at edge connected to a company and a born_in edge connected to a city to be included in the result.