creation of the prep_mdb_table_extract helper function for the prep_m…

…db_merge function

DESCRIPTION

@@ -37,16 +37,17 @@ Imports:
   future,
   purrr,
   archive,
-  doParallel
+  doParallel,
+  RODBC
 Suggests: 
-  testthat,
-  progressr,
-  janitor,
-  withr,
-  stringdist,
-  epiCleanr,
-  crayon,
-  glue
+    testthat,
+    progressr,
+    janitor,
+    withr,
+    stringdist,
+    epiCleanr,
+    crayon,
+    glue
 Config/testthat/edition: 3
 Encoding: UTF-8
 LazyData: true

NAMESPACE

@@ -8,6 +8,7 @@ export(get_updated_ona_data)
 export(prep_geonames)
 export(prep_match_datatypes)
 export(prep_match_names)
+export(prep_mdb_table_extract)
 export(read)
 export(save)
 import(doFuture)

R/prep_mdb_merge.R

@@ -0,0 +1,93 @@
+################################################################################
+#                                                                              #
+#               FUNCTION FOR MERGING DIFFERENT MDB FILES                       #
+#                                                                              #
+################################################################################
+
+#' Function to extract tables from multiple Microsoft Access databases.
+#'
+#' @param mdb_folder Path to the folder containing the .mdb files
+#' @param target_table The table to be extracted from each mdb file. This should
+#'                     unique name accross all the .mdb files.
+#'
+#' @return Returns a list all all the extracted tables indexed by their names.
+#' @export
+#'
+#' @examples
+#' 
+#' # How to run the function
+#' prep_mdb_table_extract(mdb_folder, target_table)
+#' 
+#' 
+
+prep_mdb_table_extract <- function(mdb_folder, target_table){
+  # Get the start time of the function
+  start_time <- Sys.time()
+  tryCatch({
+    # a) check if the foldername exist
+    if(!dir.exists(mdb_folder)){
+      cli::cli_alert_danger(paste("Folder path", mdb_folder, "does not exist."))
+    }
+
+    # b) get the list of all the files with mdb extension from the folder
+    mdb_files <-
+      dir(path = mdb_folder,
+          pattern = "\\.mdb",
+          full.names = T)
+
+    if(length(mdb_files > 0)){
+      # c) loop through each .mdb file and look for the target table
+      afp_list <- list()
+
+      cli::cli_progress_bar("Extracting data", total = 100)
+      for(mdb_file in mdb_files) {
+
+        #print(path)
+        ch1 <-
+          RODBC::odbcDriverConnect(paste0(
+            "DRIVER={Microsoft Access Driver (*.mdb, *.accdb)};DBQ=",
+            mdb_file
+          ))
+        # Get list of all tables in the mdb
+        tables <- RODBC::sqlTables(ch1)
+        # Get the only tables with type TABLE
+        tables <- tables$TABLE_NAME[tables$TABLE_TYPE == "TABLE"]
+        # Check if the target table exist in list of tables in an .mdb
+        if(target_table %in% tables){
+          query <- RODBC::sqlQuery(ch1, query = paste("select * from ", target_table))
+          afp_list[[paste0(basename(mdb_file), "_", target_table)]] <- query
+        }else{
+          cli::cli_text(paste("No table named : ", target_table, " in ", basename(mdb_file)))
+        }
+        # close db connection
+        RODBC::odbcCloseAll()
+        cli::cli_progress_update(total = length(mdb_files))
+      }
+
+      # print the time taken to complete the process
+      elapsed_time <- Sys.time() - start_time
+      cli::cli_alert_success(
+        paste("Process completed sucessfully in ", 
+              elapsed_time, "Sec"
+        ))
+      # return the list of tables
+      return(afp_list)
+    } else{
+      # print the time taken to complete the process
+      cli::cli_alert_warning(
+        paste("No mdb file to extract table from"))
+    }
+
+  }, error = function(e){
+    # print out the error message
+    cli::cli_alert_danger(e$message)
+  }, warning = function(w){
+    # print out the warning message
+    cli::cli_alert_warning(w$message)
+  })
+}
+
+
+#prep_mdb_table_extract(mdb_folder, target_table)
+
+

tests/testthat/test-prep_mdb_merge.R

@@ -0,0 +1,15 @@
+test_that("Expect No Error test", {
+  # no error
+  fake_folder <- "output/mdb_files"
+  target_tab <- "polio"
+  expect_no_error(object = prep_mdb_table_extract(fake_folder, target_tab))
+
+})
+
+
+
+
+
+
+
+