Merge pull request #275 from OHDSI/geom_dep

Move geom_dependency_uuid from data_source to variable_source table

R/createIndices.R

@@ -14,6 +14,7 @@
 
 createIndices  <-  function(connectionDetails) {
   dataSourceTable <- getDataSourceTable(connectionDetails = connectionDetails)
+  variableSourceTable <- getVariableSourceTable(connectionDetails = connectionDetails)
   uuids <- dataSourceTable$data_source_uuid
   geomIndexTable <- getGeomIndexTable(connectionDetails = connectionDetails)
   attrIndexTable <- getAttrIndexTable(connectionDetails = connectionDetails)
@@ -29,19 +30,28 @@ createIndices  <-  function(connectionDetails) {
                                               dataSourceRecord = record)
     }
     # IF attr AND not in aidsid check dependency
-    if (record$has_attributes == 1 & !id %in% attrIndexTable$data_source_id) {
+    # Removed  "& !id %in% attrIndexTable$data_source_id", need to reincorporate at variable_source level
+    if (record$has_attributes == 1) {
       ## IF geom dependency AND dependency not in gidsid then create geom index record AND insert into db
-      if (!is.na(record$geom_dependency_uuid) &
-          !record$geom_dependency_uuid %in% geomIndexTable$data_source_id &
-          record$geom_dependency_uuid != record$data_source_uuid) {
-        geomDependencyDataSourceRecord <- dplyr::filter(dataSourceTable,
-                                                        data_source_uuid == record$geom_dependency_uuid)
-        geomIndexTable <- appendGeomIndexRecord(geomIndexTable = geomIndexTable,
-                                                dataSourceRecord = geomDependencyDataSourceRecord)
+      # TODO for every variable_source$data_source_uuid == id
+      for (vid in variableSourceTable[variableSourceTable$data_source_uuid == id,]$variable_source_id) {
+        if (!vid %in% attrIndexTable$variable_source_id) {
+          variableRecord <- variableSourceTable[variableSourceTable$variable_source_id == vid,]
+        # TODO switch record$geom_dependency_uuid for varRecord$geom_dependency_uuid
+          if (!is.na(variableRecord$geom_dependency_uuid) &
+              !variableRecord$geom_dependency_uuid %in% geomIndexTable$data_source_id &
+              variableRecord$geom_dependency_uuid != variableRecord$data_source_uuid) {
+            geomDependencyDataSourceRecord <- dplyr::filter(dataSourceTable,
+                                                            data_source_uuid == variableRecord$geom_dependency_uuid)
+            geomIndexTable <- appendGeomIndexRecord(geomIndexTable = geomIndexTable,
+                                                    dataSourceRecord = geomDependencyDataSourceRecord)
+          }
+          attrIndexTable <- appendAttrIndexRecord(attrIndexTable = attrIndexTable,
+                                                  geomIndexTable = geomIndexTable,
+                                                  dataSourceRecord = record,
+                                                  variableSourceRecord = variableRecord)
+        }
       }
-      attrIndexTable <- appendAttrIndexRecord(attrIndexTable = attrIndexTable,
-                                              geomIndexTable = geomIndexTable,
-                                              dataSourceRecord = record)
     }
   }
 
@@ -90,16 +100,18 @@ appendGeomIndexRecord <- function(geomIndexTable, dataSourceRecord) {
 #' @return (data.frame) updated attrIndexTable
 #'
 
-appendAttrIndexRecord <- function(attrIndexTable, geomIndexTable, dataSourceRecord) {
+# TODO incorporate varSourceRecord
+appendAttrIndexRecord <- function(attrIndexTable, geomIndexTable, dataSourceRecord, variableSourceRecord) {
   if (nrow(attrIndexTable) == 0) {
     attr_index_id <- 1
   } else {
     attr_index_id <- max(attrIndexTable$attr_index_id) + 1
   }
   indexRecord <- dplyr::tibble(
     attr_index_id = attr_index_id,
+    variable_source_id = variableSourceRecord$variable_source_id,
     attr_of_geom_index_id = getAttrOfGeomIndexId(geomIndexTable = geomIndexTable,
-                                                 dataSourceUuid = dataSourceRecord$geom_dependency_uuid),
+                                                 dataSourceUuid = variableSourceRecord$geom_dependency_uuid),
     database_schema = createSchemaString(dataSourceRecord),
     table_name = createNameString(name = dataSourceRecord$dataset_name),
     data_source_id = dataSourceRecord$data_source_uuid)
@@ -110,7 +122,7 @@ appendAttrIndexRecord <- function(attrIndexTable, geomIndexTable, dataSourceReco
 #' Get foreign key for attr_of_geom_index_id
 #'
 #' @param geomIndexTable (data.frame) the existing geomIndexTable
-#' @param dataSourceUuid (UUID) The UUID for the data source that is registered in the backbone.data_source table
+#' @param dataSourceUuid (UUID) The UUID for the geometry data source that is registered in the backbone.data_source table
 #'
 #' @return (integer) Identifier for the corresponding backbone.geom_index entry
 #'

R/dbUtils.R

@@ -39,13 +39,13 @@ getVariableSourceSummaryTable <- function(connectionDetails) {
   getVariableSourceSummaryQuery <- paste0(
     "select vs.variable_source_id, ds.data_source_uuid as data_source_id, ",
     "vs.variable_name, vs.variable_desc, ds.org_id, ds.org_set_id, ds.documentation_url, ",
-    "ds.dataset_name, ds.dataset_version, ds.boundary_type, ds.geom_dependency_uuid, ",
+    "ds.dataset_name, ds.dataset_version, ds.boundary_type, vs.geom_dependency_uuid, ",
     "ds2.dataset_name as geom_dependency_name, ds2.geom_type ",
     "from backbone.variable_source vs ",
     "join backbone.data_source ds ",
     "on vs.data_source_uuid=ds.data_source_uuid ",
     "join backbone.data_source ds2 ",
-    "on ds.geom_dependency_uuid = ds2.data_source_uuid")
+    "on vs.geom_dependency_uuid = ds2.data_source_uuid")
   summaryTable <- DatabaseConnector::querySql(conn, getVariableSourceSummaryQuery)
   ids <- loadedVariableSourceRecordIds$VARIABLE_SOURCE_RECORD_ID
   summaryTable <- dplyr::mutate(summaryTable,
@@ -75,17 +75,10 @@ getGeomNameFromVariableSourceId <- function(connectionDetails, variableSourceId)
   on.exit(DatabaseConnector::disconnect(conn))
   DatabaseConnector::dbGetQuery(conn, paste0(
     "select concat(database_schema, '.geom_', table_name)
-    from backbone.geom_index gi 
-    where data_source_id in (
-        select geom_dependency_uuid
-        from backbone.data_source ds 
-        where data_source_uuid in (
-            select data_source_uuid
-            from backbone.variable_source vs 
-            where variable_source_id = ", variableSourceId,"
-        )
-    )"
-  )
+    from backbone.geom_index gi
+    inner join backbone.variable_source vs
+    on vs.geom_dependency_uuid = gi.data_source_id 
+    and vs.variable_source_id = ", variableSourceId)
   )[[1]]  
 }
 

R/loadVariable.R

@@ -46,13 +46,13 @@ loadVariable <- function(connectionDetails, variableSourceId){
     return(message("Variable already exists in the database."))
   }
 
+  geomIndexRecord <- getGeomIndexRecord(connectionDetails = connectionDetails,
+                                        dataSourceUuid = variableSourceRecord$geom_dependency_uuid)
+
   # get data_source_record
   dataSourceRecord <- getDataSourceRecord(connectionDetails = connectionDetails,
                                           dataSourceUuid = variableSourceRecord$data_source_uuid)
-
-  geomIndexRecord <- getGeomIndexRecord(connectionDetails = connectionDetails,
-                                        dataSourceUuid = dataSourceRecord$geom_dependency_uuid)
-
+
   # get stage data
   staged <- getStaged(dataSourceRecord)
 
@@ -76,7 +76,7 @@ loadVariable <- function(connectionDetails, variableSourceId){
   if (!geomTableExists) {
     message("Loading geom table dependency")
     loadGeometry(connectionDetails = connectionDetails,
-                 dataSourceUuid = dataSourceRecord$geom_dependency_uuid)
+                 dataSourceUuid = variableSourceRecord$geom_dependency_uuid)
   }
 
   # get mapping values from geom table