Добавлено парсирование формы 14 и сохранение финансовых отчетов в базу данных; добавлена миграция для начальных данных типов документов

2026-03-25 09:48:22 +03:00
parent 207e4b45a7
commit 7ee21d5e5d
5 changed files with 151 additions and 37 deletions
@@ -11,8 +11,11 @@ class ReportParser

      form_11_data = ReportParsers::Form11Parser.new(@yearly_file, spreadsheet: spreadsheet).call
      form_12_data = ReportParsers::Form12Parser.new(@yearly_file, spreadsheet: spreadsheet).call
+      form_14_data = ReportParsers::Form14Parser.new(@yearly_file, spreadsheet: spreadsheet).call

-      merge_financial_statements(form_11_data, form_12_data)
+      save_form_statements(form_11_data, 1) # 1 - Бух. баланс
+      save_form_statements(form_12_data, 2) # 2 - Отчет о фин. результатах
+      save_form_statements(form_14_data, 4) # 4 - Отчет о движении денежных средств
    end
  end

@@ -28,35 +31,21 @@ class ReportParser
    end
  end

-  def merge_financial_statements(form_11_data, form_12_data)
-    document_type = DocumentType.find_or_create_by!(name: 'Финансовые данные')
+  def save_form_statements(form_data, form_id)
+    document_type = DocumentType.find_by(id: form_id)
+    raise "DocumentType with id=#{form_id} not found!" unless document_type

-    all_years = (form_11_data.keys + form_12_data.keys).uniq
+    form_data.each do |year, data|
+      rows = data[:rows] || data['rows'] || []
+      next if rows.blank?

-    all_years.each do |year|
      statement = find_or_build_statement(document_type, year)
-
-      all_rows = []
-
-      if form_11_data[year].present?
-        form_11_rows = form_11_data[year][:rows] || form_11_data[year]['rows'] || []
-        all_rows.concat(form_11_rows)
-      end
-
-      if form_12_data[year].present?
-        form_12_rows = form_12_data[year][:rows] || form_12_data[year]['rows'] || []
-        all_rows.concat(form_12_rows)
-      end
-
-      if all_rows.present?
-        payload = {
-          data_year: year,
-          rows: all_rows
-        }
-
-        statement.assign_attributes(data: payload)
-        statement.save!
-      end
+      payload = {
+        data_year: year,
+        rows: rows
+      }
+      statement.assign_attributes(data: payload)
+      statement.save!
    end
  end

@@ -98,21 +98,16 @@ module ReportParsers

      value_columns = columns.keys.select do |col|
        text = columns[col]
-        text.include?('за ') ||
-        text.include?('на ') ||
-        text.match?(/\b20\d{2}\b/) ||
-        (col > code_column && text.include?('года')) ||
-        (col > code_column && text.match?(/\d+/))
+        next false if text.include?('поясн')
+        text.match?(/20\d{2}/) || text.include?('за ') || text.include?('на ')
      end

-      if value_columns.empty?
-        value_columns = columns.keys.select { |col| col > code_column && columns[col].present? }
-      end
+      value_columns = value_columns.sort.last(2)

      {
        name_column: name_column,
        code_column: code_column,
-        value_columns: value_columns.sort.first(3)
+        value_columns: value_columns
      }
    end
  end
@@ -0,0 +1,112 @@
+        def extract_year_from_header(header_text)
+          year_match = header_text[/\b(20\d{2})\b/]
+          year_match.to_i if year_match
+        end
+    def find_header_row(sheet)
+      1.upto(sheet.last_row) do |row|
+        normalized_cells = 1.upto(sheet.last_column).map { |col| normalize_label(sheet.cell(row, col)) }
+        has_name = normalized_cells.any? { |text| text.include?("наименование показателя") }
+        has_code = normalized_cells.any? { |text| text == "код" || text.start_with?("код ") || text.start_with?("код:") }
+        return row if has_name && has_code
+      end
+
+      1
+    end
+
+    def detect_columns(sheet, header_row)
+      columns = 1.upto(sheet.last_column).each_with_object({}) do |col, hash|
+        hash[col] = normalize_label(sheet.cell(header_row, col))
+      end
+
+      name_column = columns.key(columns.values.find { |text| text.include?("наименование показателя") }) || 1
+
+      code_column = columns.key(columns.values.find { |text|
+        text == "код строки" || text.start_with?("код строки ") ||
+        (text == "код" && columns[name_column + 1] == text)
+      }) || columns.key(columns.values.find { |text| text == "код" || text.start_with?("код ") }) || [name_column + 1, sheet.last_column].min
+
+      value_columns = columns.keys.select do |col|
+        text = columns[col]
+        next false if text.include?("поясн")
+        text.match?(/20\d{2}/) || text.include?("за ") || text.include?("на ")
+      end
+      value_columns = value_columns.sort.last(2)
+
+      {
+        name_column: name_column,
+        code_column: code_column,
+        value_columns: value_columns
+      }
+    end
+
+    def normalize_label(text)
+      text.to_s.downcase.gsub(/[[:space:]]+/, ' ').strip
+    end
+module ReportParsers
+  class Form14Parser < BaseParser
+    SHEET_NAME = 'Отчет о движении денежных средс'
+    FORM_CODE = '1.4'
+
+    def call
+      return {} unless sheet_exists?(SHEET_NAME)
+
+      sheet = spreadsheet.sheet(SHEET_NAME)
+      build_payloads(sheet)
+    end
+
+    private
+
+    def build_payloads(sheet)
+      header_row = find_header_row(sheet)
+      columns = detect_columns(sheet, header_row)
+
+      value_headers = columns[:value_columns].each_with_object({}) do |col, hash|
+        header_text = read_cell(sheet, header_row, col).to_s
+        year = extract_year_from_header(header_text)
+        hash[col] = year
+      end
+
+      all_rows = (header_row + 1).upto(sheet.last_row).filter_map do |row|
+        line_name = read_cell(sheet, row, columns[:name_column])
+        line_code = read_cell(sheet, row, columns[:code_column]).to_s.strip.presence
+
+        values = columns[:value_columns].each_with_object({}) do |col, hash|
+          value = read_cell(sheet, row, col)
+          year = value_headers[col]
+          hash[year.to_s] = value if value.present? && year.present?
+        end.compact_blank
+
+        next if line_name.blank? && line_code.blank? && values.blank?
+
+        {
+          line_name: line_name,
+          line_code: line_code,
+          values: values
+        }
+      end
+
+      payloads_by_year = {}
+      value_headers.each do |col, year|
+        next unless year.present?
+
+        rows_for_year = all_rows.map do |row|
+          value = row[:values][year.to_s]
+          next unless value.present?
+
+          {
+            line_name: row[:line_name],
+            line_code: row[:line_code],
+            value: value
+          }
+        end.compact
+
+        payloads_by_year[year] = {
+          form_code: FORM_CODE,
+          rows: rows_for_year
+        }
+      end
+
+      payloads_by_year
+    end
+  end
+end
@@ -0,0 +1,18 @@
+class SeedDocumentTypes < ActiveRecord::Migration[7.0]
+  def up
+    execute <<-SQL
+      INSERT INTO document_types (id, name, created_at, updated_at) VALUES
+        (1, 'Бухгалтерский баланс', NOW(), NOW()),
+        (2, 'Отчет о финансовых результатах', NOW(), NOW()),
+        (3, 'Отчет об изменениях капитала', NOW(), NOW()),
+        (4, 'Отчет о движении денежных средств', NOW(), NOW())
+      ON CONFLICT (id) DO NOTHING;
+    SQL
+  end
+
+  def down
+    execute <<-SQL
+      DELETE FROM document_types WHERE id IN (1,2,3,4);
+    SQL
+  end
+end
@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.

-ActiveRecord::Schema[7.1].define(version: 2026_03_21_000006) do
+ActiveRecord::Schema[7.1].define(version: 2026_03_25_063000) do
  # These are extensions that must be enabled in order to support this database
  enable_extension "plpgsql"