core@ review workflow and job data models and their indexes

opencb · Sep 22, 2024 · f625346 · f625346
1 parent c248a37
commit f625346
Show file tree

Hide file tree

Showing 7 changed files with 87 additions and 140 deletions.
diff --git a/opencga-catalog/src/main/resources/catalog-indexes.txt b/opencga-catalog/src/main/resources/catalog-indexes.txt
@@ -1,21 +1,10 @@
-{"collections": ["note", "note_archive"], "fields": {"id": 1, "studyUid": 1, "version": 1}, "options": {"unique": true}}
-{"collections": ["note", "note_archive"], "fields": {"uid": 1, "version": 1}, "options": {"unique": true}}
-{"collections": ["note", "note_archive"], "fields": {"uuid": 1, "version": 1}, "options": {"unique": true}}
-{"collections": ["note", "note_archive"], "fields": {"scope": 1, "studyUid": 1}, "options": {}}
-{"collections": ["note", "note_archive"], "fields": {"studyUid": 1}, "options": {}}
-{"collections": ["note", "note_archive"], "fields": {"visibility": 1, "studyUid": 1}, "options": {}}
-{"collections": ["note", "note_archive"], "fields": {"userId": 1, "studyUid": 1}, "options": {}}
-{"collections": ["note", "note_archive"], "fields": {"tags": 1, "studyUid": 1}, "options": {}}
-{"collections": ["note", "note_archive"], "fields": {"_creationDate": 1, "studyUid": 1}, "options": {}}
-{"collections": ["note", "note_archive"], "fields": {"_modificationDate": 1, "studyUid": 1}, "options": {}}
+{"collections": ["user"], "fields": {"id": 1}, "options": {"unique": true}}
 
 {"collections": ["project"], "fields": {"id": 1}, "options": {"unique": true}}
 {"collections": ["project"], "fields": {"uid": 1}, "options": {"unique": true}}
 {"collections": ["project"], "fields": {"uuid": 1}, "options": {"unique": true}}
 {"collections": ["project"], "fields": {"fqn": 1}, "options": {"unique": true}}
 
-{"collections": ["user"], "fields": {"id": 1}, "options": {"unique": true}}
-
 {"collections": ["study"], "fields": {"uid": 1}, "options": {"unique": true}}
 {"collections": ["study"], "fields": {"uuid": 1}, "options": {"unique": true}}
 {"collections": ["study"], "fields": {"fqn": 1}, "options": {"unique": true}}
@@ -31,14 +20,20 @@
 {"collections": ["job"], "fields": {"id": 1, "studyUid": 1}, "options": {"unique": true}}
 {"collections": ["job"], "fields": {"tool.id": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"tool.type": 1, "studyUid": 1}, "options": {}}
+{"collections": ["job"], "fields": {"tool.externalExecutor.id": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"userId": 1, "studyUid": 1}, "options": {}}
+{"collections": ["job"], "fields": {"parentId": 1, "studyUid": 1}, "options": {}}
+{"collections": ["job"], "fields": {"dependsOn.id": 1, "studyUid": 1}, "options": {}}
+{"collections": ["job"], "fields": {"outDir.uid": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"input.uid": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"output.uid": 1, "studyUid": 1}, "options": {}}
-{"collections": ["job"], "fields": {"outDir.uid": 1, "studyUid": 1}, "options": {}}
+{"collections": ["job"], "fields": {"priority": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"tags": 1, "studyUid": 1}, "options": {}}
+{"collections": ["job"], "fields": {"dryRun: 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"visited": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"internal.status.id": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"internal.status.id": 1, "_priority": 1, "_creationDate": 1}, "options": {}}
+{"collections": ["job"], "fields": {"internal.killJobRequested": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"_priority": 1, "_creationDate": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"_priority": 1, "studyUid": 1}, "options": {}}
 {"collections": ["job"], "fields": {"_creationDate": 1, "studyUid": 1}, "options": {}}
@@ -261,14 +256,26 @@
 {"collections": ["audit"], "fields": {"status.name": 1, "studyUuid": 1}, "options": {}}
 {"collections": ["audit"], "fields": {"date": 1, "studyUuid": 1}, "options": {}}
 
+{"collections": ["note", "note_archive"], "fields": {"id": 1, "studyUid": 1, "version": 1}, "options": {"unique": true}}
+{"collections": ["note", "note_archive"], "fields": {"uid": 1, "version": 1}, "options": {"unique": true}}
+{"collections": ["note", "note_archive"], "fields": {"uuid": 1, "version": 1}, "options": {"unique": true}}
+{"collections": ["note", "note_archive"], "fields": {"scope": 1, "studyUid": 1}, "options": {}}
+{"collections": ["note", "note_archive"], "fields": {"studyUid": 1}, "options": {}}
+{"collections": ["note", "note_archive"], "fields": {"visibility": 1, "studyUid": 1}, "options": {}}
+{"collections": ["note", "note_archive"], "fields": {"userId": 1, "studyUid": 1}, "options": {}}
+{"collections": ["note", "note_archive"], "fields": {"tags": 1, "studyUid": 1}, "options": {}}
+{"collections": ["note", "note_archive"], "fields": {"_creationDate": 1, "studyUid": 1}, "options": {}}
+{"collections": ["note", "note_archive"], "fields": {"_modificationDate": 1, "studyUid": 1}, "options": {}}
+
 {"collections": ["workflow", "workflow_archive"], "fields": {"uid": 1, "version": 1}, "options": {"unique": true}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"id": 1, "studyUid": 1, "version": 1}, "options": {"unique": true}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"uuid": 1, "version": 1}, "options": {"unique": true}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"name": 1, "studyUid": 1}, "options": {"unique": true}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"type": 1, "studyUid": 1}, "options": {}}
+{"collections": ["workflow", "workflow_archive"], "fields": {"manager.id": 1, "studyUid": 1}, "options": {}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"tags": 1, "studyUid": 1}, "options": {}}
-{"collections": ["workflow", "workflow_archive"], "fields": {"scripts.fileName": 1, "studyUid": 1}, "options": {}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"draft": 1, "studyUid": 1}, "options": {}}
+{"collections": ["workflow", "workflow_archive"], "fields": {"scripts.fileName": 1, "studyUid": 1}, "options": {}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"internal.registrationUserId": 1, "studyUid": 1}, "options": {}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"_creationDate": 1, "studyUid": 1}, "options": {}}
 {"collections": ["workflow", "workflow_archive"], "fields": {"_modificationDate": 1, "studyUid": 1}, "options": {}}
diff --git a/opencga-client/src/main/javascript/Workflow.js b/opencga-client/src/main/javascript/Workflow.js
@@ -59,7 +59,7 @@ export default class Workflow extends OpenCGAParentClass {
     create(data, params) {
         return this._post("workflows", null, null, null, "create", data, params);
     }
-#
+
     /** Workflow distinct method
     * @param {String} field - Comma separated list of fields for which to obtain the distinct values.
     * @param {Object} [params] - The Object containing the following optional parameters:

diff --git a/opencga-core/src/main/java/org/opencb/opencga/core/models/job/Job.java b/opencga-core/src/main/java/org/opencb/opencga/core/models/job/Job.java
@@ -28,150 +28,91 @@
 import java.util.List;
 import java.util.Map;
 
-/**
- * Created by jacobo on 11/09/14.
- */
-@DataClass(id = "Job", since = "1.0",
-        description = "Job data model hosts information about any job.")
+
+@DataClass(id = "Job", since = "1.0.0", description = "Job data model hosts information about any job.")
 public class Job extends PrivateStudyUid {
 
     public static final String OPENCGA_PARENTS = "OPENCGA_PARENTS";
-    /**
-     * Job ID is a mandatory parameter when creating a new sample, this ID cannot be changed at the moment.
-     *
-     * @apiNote Required, Immutable, Unique
-     */
+
     @DataField(id = "id", required = true, indexed = true, unique = true, immutable = true,
             description = FieldConstants.GENERIC_ID_DESCRIPTION)
     private String id;
-    /**
-     * Global unique ID at the whole OpenCGA installation. This is automatically created during the sample creation and cannot be changed.
-     *
-     * @apiNote Internal, Unique, Immutable
-     */
 
     @DataField(id = "uuid", managed = true, indexed = true, unique = true, immutable = true,
             description = FieldConstants.GENERIC_UUID_DESCRIPTION)
     private String uuid;
-    /**
-     * An string to describe the properties of the Job.
-     *
-     * @apiNote
-     */
-    @DataField(id = "description", defaultValue = "No description available",
-            description = FieldConstants.GENERIC_DESCRIPTION_DESCRIPTION)
+
+    @DataField(id = "description", defaultValue = "", description = FieldConstants.GENERIC_DESCRIPTION_DESCRIPTION)
     private String description;
 
-    @DataField(id = "tool", indexed = true,
-            description = FieldConstants.JOB_TOOL)
+    @DataField(id = "tool", indexed = true, description = FieldConstants.JOB_TOOL)
     private ToolInfo tool;
 
-    @DataField(id = "userId", indexed = true,
-            description = FieldConstants.JOB_USER_ID)
+    @DataField(id = "study", indexed = true, description = FieldConstants.JOB_STUDY)
+    private JobStudyParam study;
+
+    @DataField(id = "userId", indexed = true, description = FieldConstants.JOB_USER_ID)
     private String userId;
 
-    @DataField(id = "commandLine", indexed = true,
-            description = FieldConstants.JOB_COMMAND_LINE)
+    @DataField(id = "commandLine", indexed = true, description = FieldConstants.JOB_COMMAND_LINE)
     private String commandLine;
 
-    @DataField(id = "params", indexed = true,
-            description = FieldConstants.JOB_PARAMS)
+    @DataField(id = "params", indexed = true, description = FieldConstants.JOB_PARAMS)
     private Map<String, Object> params;
-    /**
-     * String representing when the Cohort was created, this is automatically set by OpenCGA.
-     *
-     * @apiNote Internal
-     */
-    @DataField(id = "creationDate", indexed = true,
-            description = FieldConstants.GENERIC_CREATION_DATE_DESCRIPTION)
+
+    @DataField(id = "creationDate", indexed = true, description = FieldConstants.GENERIC_CREATION_DATE_DESCRIPTION)
     private String creationDate;
-    /**
-     * String representing when was the last time the Cohort was modified, this is automatically set by OpenCGA.
-     *
-     * @apiNote Internal
-     */
-    @DataField(id = "modificationDate", indexed = true, since = "1.0",
-            description = FieldConstants.GENERIC_MODIFICATION_DATE_DESCRIPTION)
-    private String modificationDate;
 
+    @DataField(id = "modificationDate", indexed = true, since = "1.0", description = FieldConstants.GENERIC_MODIFICATION_DATE_DESCRIPTION)
+    private String modificationDate;
 
-    @DataField(id = "priority", indexed = true,
-            description = FieldConstants.JOB_PRIORITY_DESCRIPTION)
+    @DataField(id = "priority", indexed = true, description = FieldConstants.JOB_PRIORITY_DESCRIPTION)
     private Enums.Priority priority;
 
-    /**
-     * An object describing the internal information of the Sample. This is managed by OpenCGA.
-     *
-     * @apiNote Internal
-     */
-
-    @DataField(id = "release", indexed = true,
-            description = FieldConstants.GENERIC_RELEASE_DESCRIPTION)
+    @DataField(id = "release", indexed = true, description = FieldConstants.GENERIC_RELEASE_DESCRIPTION)
     private JobInternal internal;
 
-    @DataField(id = "outDir", indexed = true,
-            description = FieldConstants.JOB_OUT_DIR_DESCRIPTION)
+    @DataField(id = "outDir", indexed = true, description = FieldConstants.JOB_OUT_DIR_DESCRIPTION)
     private File outDir;
 
-    @DataField(id = "input", indexed = true,
-            description = FieldConstants.JOB_INPUT_DESCRIPTION)
-    private List<File> input;    // input files to this job
+    @DataField(id = "input", indexed = true, description = FieldConstants.JOB_INPUT_DESCRIPTION)
+    private List<File> input;
 
-    @DataField(id = "output", indexed = true,
-            description = FieldConstants.JOB_OUTPUT_DESCRIPTION)
-    private List<File> output;   // output files of this job
+    @DataField(id = "output", indexed = true, description = FieldConstants.JOB_OUTPUT_DESCRIPTION)
+    private List<File> output;
 
-    @DataField(id = "tags", indexed = true,
-            description = FieldConstants.JOB_TAGS_DESCRIPTION)
+    @DataField(id = "tags", indexed = true, description = FieldConstants.JOB_TAGS_DESCRIPTION)
     private List<String> tags;
 
-    @DataField(id = "dependsOn", indexed = true,
-            description = FieldConstants.JOB_DEPENDS_ON_DESCRIPTION)
+    @DataField(id = "dependsOn", indexed = true, description = FieldConstants.JOB_DEPENDS_ON_DESCRIPTION)
     private List<Job> dependsOn;
 
-    @DataField(id = FieldConstants.JOB_PARENT_ID, indexed = false, since = "3.2.0",
-            description = FieldConstants.JOB_PARENT_ID_DESCRIPTION)
+    @DataField(id = FieldConstants.JOB_PARENT_ID, indexed = true, since = "3.2.0", description = FieldConstants.JOB_PARENT_ID_DESCRIPTION)
     private String parentId;
 
-    @DataField(id = FieldConstants.JOB_SCHEDULED_START_TIME, indexed = false, since = "3.2.0",
+    @DataField(id = FieldConstants.JOB_SCHEDULED_START_TIME, since = "3.2.0",
             description = FieldConstants.JOB_SCHEDULED_START_TIME_DESCRIPTION)
     private String scheduledStartTime;
 
     @DataField(id = ParamConstants.JOB_DRY_RUN, since = "3.2.0", description = ParamConstants.JOB_DRY_RUN_DESCRIPTION)
     private boolean dryRun;
 
-    @DataField(id = "execution", indexed = true,
-            description = FieldConstants.JOB_EXECUTION_DESCRIPTION)
+    @DataField(id = "execution", indexed = true, description = FieldConstants.JOB_EXECUTION_DESCRIPTION)
     private ExecutionResult execution;
 
-    @DataField(id = "stdout", indexed = true,
-            description = FieldConstants.JOB_STDOUT_DESCRIPTION)
+    @DataField(id = "stdout", indexed = true, description = FieldConstants.JOB_STDOUT_DESCRIPTION)
     private File stdout;
 
-    @DataField(id = "stderr", indexed = true,
-            description = FieldConstants.JOB_STDERR_DESCRIPTION)
+    @DataField(id = "stderr", indexed = true, description = FieldConstants.JOB_STDERR_DESCRIPTION)
     private File stderr;
 
-
-    @DataField(id = "visited", indexed = true,
-            description = FieldConstants.JOB_VISITED)
+    @DataField(id = "visited", indexed = true, description = FieldConstants.JOB_VISITED)
     private boolean visited;
-    /**
-     * An integer describing the current data release.
-     *
-     * @apiNote Internal
-     */
-
-    @DataField(id = "release", indexed = true,
-            description = FieldConstants.GENERIC_RELEASE_DESCRIPTION)
-    private int release;
 
-    @DataField(id = "study", indexed = true,
-            description = FieldConstants.JOB_STUDY)
-    private JobStudyParam study;
+    @DataField(id = "release", indexed = true, description = FieldConstants.GENERIC_RELEASE_DESCRIPTION)
+    private int release;
 
-    @DataField(id = "attributes", indexed = true,
-            description = FieldConstants.GENERIC_ATTRIBUTES_DESCRIPTION)
+    @DataField(id = "attributes", indexed = true, description = FieldConstants.GENERIC_ATTRIBUTES_DESCRIPTION)
     private Map<String, Object> attributes;
 
     public Job() {

diff --git a/opencga-core/src/main/java/org/opencb/opencga/core/models/job/JobInternal.java b/opencga-core/src/main/java/org/opencb/opencga/core/models/job/JobInternal.java
@@ -29,17 +29,14 @@
 
 public class JobInternal extends Internal implements Cloneable {
 
-    @DataField(id = "status", indexed = true,
-            description = FieldConstants.JOB_INTERNAL_STATUS_DESCRIPTION)
+    @DataField(id = "status", indexed = true, description = FieldConstants.JOB_INTERNAL_STATUS_DESCRIPTION)
     private Enums.ExecutionStatus status;
 
-    @DataField(id = "webhook", indexed = true,
-            description = FieldConstants.JOB_INTERNAL_WEBHOOK_DESCRIPTION)
+    @DataField(id = "webhook", indexed = true, description = FieldConstants.JOB_INTERNAL_WEBHOOK_DESCRIPTION)
     private JobInternalWebhook webhook;
 
     //TODO add tags to commons-datastore-core
-    @DataField(id = "events", indexed = true, uncommentedClasses = {"Event"},
-            description = FieldConstants.JOB_INTERNAL_EVENTS_DESCRIPTION)
+    @DataField(id = "events", indexed = true, uncommentedClasses = {"Event"}, description = FieldConstants.JOB_INTERNAL_EVENTS_DESCRIPTION)
     private List<Event> events;
 
     @DataField(id = "killJobRequested", indexed = false, since = "3.2.0",

diff --git a/opencga-core/src/main/java/org/opencb/opencga/core/models/job/ToolInfo.java b/opencga-core/src/main/java/org/opencb/opencga/core/models/job/ToolInfo.java
@@ -41,7 +41,7 @@ public class ToolInfo {
     @DataField(id = "resource", description = FieldConstants.TOOL_INFO_RESOURCE_DESCRIPTION)
     private Enums.Resource resource;
 
-    @DataField(id = "resource", description = FieldConstants.TOOL_INFO_EXTERNAL_EXECUTOR_DESCRIPTION)
+    @DataField(id = "externalExecutor", since = "3.3.0", description = FieldConstants.TOOL_INFO_EXTERNAL_EXECUTOR_DESCRIPTION)
     private ToolInfoExecutor externalExecutor;