Merge branch 'SU2-26' of DAAntipenko/finreport-analyzer into develop

app/services/report_parsers/base_parser.rb

@@ -96,5 +96,16 @@ module ReportParsers
            .gsub(/[«»]/, '')
            .strip
     end
+
+    def normalize_line_name(value)
+      text = value.to_s
+                 .tr("\u00A0", ' ')
+                 .gsub(/[⁰¹²³⁴⁵⁶⁷⁸⁹]/, '')
+                 .gsub(/(?<=\p{L})\d{1,2}(?=\s|\z)/u, '')
+                 .gsub(/\s+/, ' ')
+                 .strip
+
+      text.presence
+    end
   end
 end

app/services/report_parsers/form11_parser.rb

@@ -2,7 +2,6 @@ module ReportParsers
   class Form11Parser < BaseParser
     SHEET_NAME = 'Бухгалтерский баланс'
     FORM_CODE = '1.1'
-    # Add line codes or names that should be skipped during parsing.
     IGNORED_KEYS = [2].freeze
 
     def call
@@ -27,7 +26,7 @@ module ReportParsers
       end
 
       all_rows = (header_row + 1).upto(sheet.last_row).filter_map do |row|
-        line_name = read_cell(sheet, row, columns[:name_column])
+        line_name = normalize_line_name(read_cell(sheet, row, columns[:name_column]))
         line_code = read_cell(sheet, row, columns[:code_column]).to_s.strip.presence
         next if ignored_row?(line_name, line_code)
 

app/services/report_parsers/form12_parser.rb

@@ -25,7 +25,7 @@ module ReportParsers
       end
 
       all_rows = (header_row + 1).upto(sheet.last_row).filter_map do |row|
-        line_name = read_cell(sheet, row, columns[:name_column])
+        line_name = normalize_line_name(read_cell(sheet, row, columns[:name_column]))
         line_code = read_cell(sheet, row, columns[:code_column]).to_s.strip.presence
         next if ignored_row?(line_name, line_code)
 

app/services/report_parsers/form14_parser.rb

@@ -25,7 +25,7 @@ module ReportParsers
       end
 
       all_rows = (header_row + 1).upto(sheet.last_row).filter_map do |row|
-        line_name = read_cell(sheet, row, columns[:name_column])
+        line_name = normalize_line_name(read_cell(sheet, row, columns[:name_column]))
         line_code = read_cell(sheet, row, columns[:code_column]).to_s.strip.presence
         next if ignored_row?(line_name, line_code)