Search in sources :

Example 1 with QueryContext

use of edu.uci.ics.texera.dataflow.plangen.QueryContext in project textdb by TextDB.

the class QueryPlanResource method executeQueryPlan.

/**
 * This is the edu.uci.ics.texera.web.request handler for the execution of a Query Plan.
 * @param logicalPlanJson, the json representation of the logical plan
 * @return - Generic GenericWebResponse object
 */
@POST
@Path("/execute")
public // TODO: investigate how to use LogicalPlan directly
JsonNode executeQueryPlan(@Session HttpSession session, String logicalPlanJson) {
    try {
        UserResource.User user = UserResource.getUser(session);
        QueryContext ctx = new QueryContext();
        if (user != null) {
            ctx.setProjectOwnerID(user.userID.toString());
        }
        LogicalPlan logicalPlan = new ObjectMapper().readValue(logicalPlanJson, LogicalPlan.class);
        logicalPlan.setContext(ctx);
        Plan plan = logicalPlan.buildQueryPlan();
        return executeMutipleSinkPlan(plan);
    } catch (IOException | TexeraException e) {
        throw new TexeraWebException(e.getMessage());
    }
}
Also used : LogicalPlan(edu.uci.ics.texera.dataflow.plangen.LogicalPlan) QueryContext(edu.uci.ics.texera.dataflow.plangen.QueryContext) IOException(java.io.IOException) Plan(edu.uci.ics.texera.api.engine.Plan) LogicalPlan(edu.uci.ics.texera.dataflow.plangen.LogicalPlan) TexeraException(edu.uci.ics.texera.api.exception.TexeraException) TexeraWebException(edu.uci.ics.texera.web.TexeraWebException) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Path(javax.ws.rs.Path) POST(javax.ws.rs.POST)

Example 2 with QueryContext

use of edu.uci.ics.texera.dataflow.plangen.QueryContext in project textdb by TextDB.

the class QueryPlanResource method suggestAutocompleteSchema.

/**
 * This is the edu.uci.ics.texera.web.request handler for the execution of a Query Plan.
 * @param logicalPlanJson, the json representation of the logical plan
 * @return - Generic GenericWebResponse object
 */
/* EG of using /autocomplete end point (how this inline update method works):

    1. At the beginning of creating a graph, (for example) when a scan source and a keyword search
        operators are initailized (dragged in the flow-chart) but unlinked, the graph looks like this:

         ___________________                             ___________________
        |                   |                           |                   |
        |                   |                           |  Keyword Search   |
        |  Source: Scan     |                           |  Attributes: N/A  |
        |  TableName: N/A   |                           |  (Other inputs...)|
        |                   |                           |                   |
        |                   |                           |                   |
        |                   |                           |                   |
        |___________________|                           |___________________|

    2. Then, you can feel free to link these two operators together, or go ahead and select a
        table as the source first. Let's link them together first.

         ___________________                             ___________________
        |                   |                           |                   |
        |                   |                           |  Keyword Search   |
        |  Source: Scan     |                           |  Attributes: N/A  |
        |  TableName: N/A   | ========================> |  (Other inputs...)|
        |                   |                           |                   |
        |                   |                           |                   |
        |                   |                           |                   |
        |___________________|                           |___________________|

    3. At this moment, the Keyword Search operator still does NOT have any available options for
        its Attributes field because of the lack of the source. Therefore, we can select a table
        name as the source next (let's use table "plan" as an example here)

         ___________________                             ___________________
        |                   |                           |                   |
        |                   |                           |  Keyword Search   |
        |  Source: Scan     |                           |  Attributes: N/A  |
        |  TableName: plan  | ========================> |  (Other inputs...)|
        |                   |                           |                   |
        |                   |                           |                   |
        |                   |                           |                   |
        |___________________|                           |___________________|

    4. After select table "plan" as the source, now you can see the options list of Attributes in
        the Keyword Search operator becomes available. you should see 4 options in the list: name,
        description, logicPlan, payload. Feel free to choose whichever you need for your desired result.

         ___________________                             ___________________
        |                   |                           |                   |
        |                   |                           |  Keyword Search   |
        |  Source: Scan     |                           |  Attributes: name |
        |  TableName: plan  | ========================> |  (Other inputs...)|
        |                   |                           |                   |
        |                   |                           |                   |
        |                   |                           |                   |
        |___________________|                           |___________________|

    5. Basically, the method supports that whenever you connect a source (with a valid table name)
        to a regular search operator, the later operator is able to recognize the metadata of its
        input operator (which is the source), and then updates its attribute options in the drop-down
        list. To illustrate how powerful this functionality is, you can add a new (Scan) Source and
        pick another table which is different than table "plan" we have already created. The graph
        now should be looked like the following:

         ___________________                             ___________________
        |                   |                           |                   |
        |                   |                           |  Keyword Search   |
        |  Source: Scan     |                           |  Attributes: name |
        |  TableName: plan  | ========================> |  (Other inputs...)|
        |                   |                           |                   |
        |                   |                           |                   |
        |                   |                           |                   |
        |___________________|                           |___________________|


         _______________________
        |                       |
        |                       |
        | Source: Scan          |
        | TableName: dictionary |
        |                       |
        |                       |
        |                       |
        |_______________________|

    6. Then, connect "dictionary" to the Keyword Search operator. The original link between "plan"
        and Keyword Search will automatically disappear.


         ___________________                             ___________________
        |                   |                           |                   |
        |                   |                           |  Keyword Search   |
        |  Source: Scan     |                           |  Attributes: N/A  |
        |  TableName: plan  |                           |  (Other inputs...)|
        |                   |            /============> |                   |
        |                   |           //              |                   |
        |                   |          //               |                   |
        |___________________|         //                |___________________|
                                     //
                                    //
         _______________________   //
        |                       | //
        |                       |//
        | Source: Scan          |/
        | TableName: dictionary |
        |                       |
        |                       |
        |                       |
        |_______________________|

    7. After the new link generated, the Attributes field of the Keyword Search will be empty again. When
        you try to check its drop-down list, the options are all updated to dictionary's attributes, which
        are name and payload. The options from "plan" are all gone.

    */
@POST
@Path("/autocomplete")
public JsonNode suggestAutocompleteSchema(@Session HttpSession session, String logicalPlanJson) {
    try {
        UserResource.User user = UserResource.getUser(session);
        QueryContext ctx = new QueryContext();
        if (user != null) {
            ctx.setProjectOwnerID(user.userID.toString());
        }
        JsonNode logicalPlanNode = new ObjectMapper().readTree(logicalPlanJson);
        ArrayNode operators = (ArrayNode) logicalPlanNode.get(PropertyNameConstants.OPERATOR_LIST);
        ArrayNode links = (ArrayNode) logicalPlanNode.get(PropertyNameConstants.OPERATOR_LINK_LIST);
        ArrayNode validOperators = new ObjectMapper().createArrayNode();
        ArrayNode validLinks = new ObjectMapper().createArrayNode();
        ArrayNode linksEndWithInvalidDest = new ObjectMapper().createArrayNode();
        Set<String> validOperatorsId = new HashSet<>();
        getValidOperatorsAndLinks(operators, links, validOperators, validLinks, linksEndWithInvalidDest, validOperatorsId);
        ObjectNode validLogicalPlanNode = new ObjectMapper().createObjectNode();
        (validLogicalPlanNode).putArray(PropertyNameConstants.OPERATOR_LIST).addAll(validOperators);
        (validLogicalPlanNode).putArray(PropertyNameConstants.OPERATOR_LINK_LIST).addAll(validLinks);
        LogicalPlan logicalPlan = new ObjectMapper().treeToValue(validLogicalPlanNode, LogicalPlan.class);
        logicalPlan.setContext(ctx);
        // Get all input schema for valid operator with valid links
        Map<String, List<Schema>> inputSchema = logicalPlan.retrieveAllOperatorInputSchema();
        // Get all input schema for invalid operator with valid input operator
        for (JsonNode linkNode : linksEndWithInvalidDest) {
            String origin = linkNode.get(PropertyNameConstants.ORIGIN_OPERATOR_ID).textValue();
            String dest = linkNode.get(PropertyNameConstants.DESTINATION_OPERATOR_ID).textValue();
            Optional<Schema> schema = logicalPlan.getOperatorOutputSchema(origin, inputSchema);
            if (schema.isPresent()) {
                if (inputSchema.containsKey(dest)) {
                    inputSchema.get(dest).add(schema.get());
                } else {
                    inputSchema.put(dest, new ArrayList<>(Arrays.asList(schema.get())));
                }
            }
        }
        ObjectNode result = new ObjectMapper().createObjectNode();
        for (Map.Entry<String, List<Schema>> entry : inputSchema.entrySet()) {
            Set<String> attributes = new HashSet<>();
            for (Schema schema : entry.getValue()) {
                attributes.addAll(schema.getAttributeNames());
            }
            ArrayNode currentSchemaNode = result.putArray(entry.getKey());
            for (String attrName : attributes) {
                currentSchemaNode.add(attrName);
            }
        }
        ObjectNode response = new ObjectMapper().createObjectNode();
        response.put("code", 0);
        response.set("result", result);
        return response;
    } catch (JsonMappingException je) {
        ObjectNode response = new ObjectMapper().createObjectNode();
        response.put("code", -1);
        response.put("message", "Json Mapping Exception would not be handled for auto plan. " + je.getMessage());
        return response;
    } catch (IOException | TexeraException e) {
        if (e.getMessage().contains("does not exist in the schema:")) {
            ObjectNode response = new ObjectMapper().createObjectNode();
            response.put("code", -1);
            response.put("message", "Attribute Not Exist Exception would not be handled for auto plan. " + e.getMessage());
            return response;
        }
        throw new TexeraWebException(e.getMessage());
    }
}
Also used : ObjectNode(com.fasterxml.jackson.databind.node.ObjectNode) Schema(edu.uci.ics.texera.api.schema.Schema) JsonNode(com.fasterxml.jackson.databind.JsonNode) QueryContext(edu.uci.ics.texera.dataflow.plangen.QueryContext) IOException(java.io.IOException) TexeraException(edu.uci.ics.texera.api.exception.TexeraException) JsonMappingException(com.fasterxml.jackson.databind.JsonMappingException) LogicalPlan(edu.uci.ics.texera.dataflow.plangen.LogicalPlan) ArrayNode(com.fasterxml.jackson.databind.node.ArrayNode) TexeraWebException(edu.uci.ics.texera.web.TexeraWebException) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Path(javax.ws.rs.Path) POST(javax.ws.rs.POST)

Aggregations

ObjectMapper (com.fasterxml.jackson.databind.ObjectMapper)2 TexeraException (edu.uci.ics.texera.api.exception.TexeraException)2 LogicalPlan (edu.uci.ics.texera.dataflow.plangen.LogicalPlan)2 QueryContext (edu.uci.ics.texera.dataflow.plangen.QueryContext)2 TexeraWebException (edu.uci.ics.texera.web.TexeraWebException)2 IOException (java.io.IOException)2 POST (javax.ws.rs.POST)2 Path (javax.ws.rs.Path)2 JsonMappingException (com.fasterxml.jackson.databind.JsonMappingException)1 JsonNode (com.fasterxml.jackson.databind.JsonNode)1 ArrayNode (com.fasterxml.jackson.databind.node.ArrayNode)1 ObjectNode (com.fasterxml.jackson.databind.node.ObjectNode)1 Plan (edu.uci.ics.texera.api.engine.Plan)1 Schema (edu.uci.ics.texera.api.schema.Schema)1