lib/remi/data_subjects/gsheet.rb in remi-0.3.1 vs lib/remi/data_subjects/gsheet.rb in remi-0.3.2

@@ -44,22 +44,23 @@
 
     def service_list_files(service, folder_id)
       service.list_files(q: "'#{folder_id}' in parents", page_size: 10, order_by: 'createdTime desc', fields: 'nextPageToken, files(id, name, createdTime, mimeType)')
     end
 
-    def get_spreadsheet_vals(service, spreadsheet_id)
-      service.get_spreadsheet_values(spreadsheet_id, 'Sheet1')
+    def get_spreadsheet_vals(service, spreadsheet_id, sheet_name = 'Sheet1')
+      service.get_spreadsheet_values(spreadsheet_id, sheet_name)
     end
 
     def extract
       service                                 = Google::Apis::SheetsV4::SheetsService.new
       service.client_options.application_name = @application_name
       service.authorization                   = authorize()
       @data                                   = []
 
       entries.each do |file|
-        response = get_spreadsheet_vals(service, file.raw)
+        logger.info "Extracting Google Sheet data from #{file.pathname}, with sheet name : #{@sheet_name}"
+        response = get_spreadsheet_vals(service, file.raw, @sheet_name)
         data.push(response)
       end
 
       self
     end
@@ -83,12 +84,13 @@
       end
     end
 
     private
 
-    def init_gsheet_extractor(*args, credentials:, folder_id:, **kargs)
+    def init_gsheet_extractor(*args, credentials:, folder_id:, sheet_name: 'Sheet1', **kargs)
       @default_folder_id   = folder_id
+      @sheet_name          = sheet_name
       @oob_uri             = 'urn:ietf:wg:oauth:2.0:oob'
       @application_name    = credentials.fetch(:application_name)
 
       @client_secrets_path = File.join(
         Dir.home,
@@ -109,28 +111,26 @@
   end
 
   class Parser::Gsheet < Parser
 
     def parse(gs_extract)
-      google_vals = gs_extract.data
       return_hash = nil
-      google_vals.each do |google_val|
+      gs_extract.data.each do |gs_data|
 
         if return_hash.nil?
           return_hash = Hash.new
-          google_val.values[0].each do |header|
+          gs_data.values[0].each do |header|
             return_hash[field_symbolizer.call(header)] = []
           end
         end
 
-        keys_temp = return_hash.keys
+        headers = return_hash.keys
+        header_idx = headers.each_with_index.to_h
 
-        google_val.values[1..-1].each do |rows|
-          col_num = 0
-
-          rows.each do |value|
-            return_hash[keys_temp[col_num]] << value
-            col_num +=1
+        gs_data.values[1..-1].each do |row|
+          headers.each do |header|
+            idx = header_idx[header]
+            return_hash[header] << (idx < row.size ? row[idx] : nil)
           end
         end
       end
       Remi::DataFrame.create(:daru, return_hash, order: return_hash.keys)
     end