refactoring

Author: ginberg

R/runTimeSeries.R

@@ -14,52 +14,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-getTimeSeries <- function(
-    connection = NULL,
-    tempEmulationSchema = NULL,
-    cdmDatabaseSchema,
-    cohortDatabaseSchema = cdmDatabaseSchema,
-    cohortTable = "cohort",
-    runCohortTimeSeries = TRUE,
-    runDataSourceTimeSeries = FALSE,
-    timeSeriesMinDate = as.Date("1980-01-01"),
-    timeSeriesMaxDate = as.Date(Sys.Date()),
-    stratifyByGender = TRUE,
-    stratifyByAgeGroup = TRUE,
-    cohortIds = NULL) {
-  
-  if (all(!runCohortTimeSeries, !runDataSourceTimeSeries)) {
-    warning(
-      " - Both Cohort Time Series and Data Source Time Series are set to FALSE. Exiting time series diagnostics."
-    )
-    return(NULL)
-  }
-  start <- Sys.time()
-
-  ParallelLogger::logTrace(" - Creating Andromeda object to collect results")
-  resultsInAndromeda <- Andromeda::andromeda()
-
-  if (runCohortTimeSeries) {
-    sqlCount <-
-      " SELECT cohort_definition_id, COUNT(*) count
-      FROM @cohort_database_schema.@cohort_table
-      {@cohort_ids != ''} ? { where cohort_definition_id IN (@cohort_ids)}
-      GROUP BY cohort_definition_id;"
-    resultsInAndromeda$cohortCount <- renderTranslateQuerySql(
-      connection = connection,
-      sql = sqlCount,
-      cohort_database_schema = cohortDatabaseSchema,
-      cohort_ids = cohortIds,
-      cohort_table = cohortTable
-    )
-    if (resultsInAndromeda$cohortCount %>%
-      dplyr::summarise(n = dplyr::n()) %>%
-      dplyr::pull(.data$n) == 0) {
-      warning("Please check if cohorts are instantiated. Exiting cohort time series.")
-      return(NULL)
-    }
-  }
-  ## Calendar period----
+createCalendarPeriodsTable <- function(connection, tempEmulationSchema, timeSeriesMinDate, timeSeriesMaxDate) {
   ParallelLogger::logTrace(" - Preparing calendar table for time series computation.")
   # note calendar span is created based on all dates in observation period table,
   # with 1980 cut off/left censor (arbitrary choice)
@@ -70,7 +25,7 @@ getTimeSeries <- function(
     ) %>% as.integer())
   maxYear <-
     clock::get_year(timeSeriesMaxDate) %>% as.integer()
-
+  
   calendarQuarter <-
     dplyr::tibble(
       periodBegin = clock::date_seq(
@@ -81,7 +36,7 @@ getTimeSeries <- function(
     ) %>%
     dplyr::mutate(periodEnd = clock::add_months(x = .data$periodBegin, n = 3) - 1) %>%
     dplyr::mutate(calendarInterval = "q")
-
+  
   calendarMonth <-
     dplyr::tibble(
       periodBegin = clock::date_seq(
@@ -92,7 +47,7 @@ getTimeSeries <- function(
     ) %>%
     dplyr::mutate(periodEnd = clock::add_months(x = .data$periodBegin, n = 1) - 1) %>%
     dplyr::mutate(calendarInterval = "m")
-
+  
   calendarYear <-
     dplyr::tibble(
       periodBegin = clock::date_seq(
@@ -103,13 +58,13 @@ getTimeSeries <- function(
     ) %>%
     dplyr::mutate(periodEnd = clock::add_years(x = .data$periodBegin, n = 1) - 1) %>%
     dplyr::mutate(calendarInterval = "y")
-
+  
   timeSeriesDateRange <- dplyr::tibble(
     periodBegin = timeSeriesMinDate,
     periodEnd = timeSeriesMaxDate,
     calendarInterval = "c"
   )
-
+  
   calendarPeriods <-
     dplyr::bind_rows(
       calendarMonth,
@@ -120,7 +75,7 @@ getTimeSeries <- function(
     dplyr::distinct() %>%
     dplyr::arrange(.data$periodBegin, .data$periodEnd, .data$calendarInterval) %>%
     dplyr::mutate(timeId = dplyr::row_number())
-
+  
   ParallelLogger::logTrace(" - Inserting calendar periods")
   DatabaseConnector::insertTable(
     connection = connection,
@@ -133,7 +88,60 @@ getTimeSeries <- function(
     tempEmulationSchema = tempEmulationSchema,
     camelCaseToSnakeCase = TRUE
   )
+}
+
+getTimeSeries <- function(
+    connection = NULL,
+    tempEmulationSchema = NULL,
+    cdmDatabaseSchema,
+    cohortDatabaseSchema = cdmDatabaseSchema,
+    cohortTable = "cohort",
+    runCohortTimeSeries = TRUE,
+    runDataSourceTimeSeries = FALSE,
+    timeSeriesMinDate = as.Date("1980-01-01"),
+    timeSeriesMaxDate = as.Date(Sys.Date()),
+    stratifyByGender = TRUE,
+    stratifyByAgeGroup = TRUE,
+    cohortIds = NULL) {
+  
+  if (all(!runCohortTimeSeries, !runDataSourceTimeSeries)) {
+    warning(
+      " - Both Cohort Time Series and Data Source Time Series are set to FALSE. Exiting time series diagnostics."
+    )
+    return(NULL)
+  }
+  start <- Sys.time()
+
+  ParallelLogger::logTrace(" - Creating Andromeda object to collect results")
+  resultsInAndromeda <- Andromeda::andromeda()
 
+  if (runCohortTimeSeries) {
+    sqlCount <-
+      " SELECT cohort_definition_id, COUNT(*) count
+      FROM @cohort_database_schema.@cohort_table
+      {@cohort_ids != ''} ? { where cohort_definition_id IN (@cohort_ids)}
+      GROUP BY cohort_definition_id;"
+    resultsInAndromeda$cohortCount <- renderTranslateQuerySql(
+      connection = connection,
+      sql = sqlCount,
+      cohort_database_schema = cohortDatabaseSchema,
+      cohort_ids = cohortIds,
+      cohort_table = cohortTable
+    )
+    if (resultsInAndromeda$cohortCount %>%
+      dplyr::summarise(n = dplyr::n()) %>%
+      dplyr::pull(.data$n) == 0) {
+      warning("Please check if cohorts are instantiated. Exiting cohort time series.")
+      return(NULL)
+    }
+  }
+  
+  ## Create calendar periods table
+  createCalendarPeriodsTable(connection, 
+                             tempEmulationSchema, 
+                             timeSeriesMinDate, 
+                             timeSeriesMaxDate)
+  
   tsSetUpSql <- "-- #time_series
                 DROP TABLE IF EXISTS #time_series;
                 DROP TABLE IF EXISTS #c_time_series1;