Add complex text XLSX test, improve text extraction, update version t…

…o 3.0.8

.rubocop.yml

@@ -111,5 +111,8 @@ Lint/MissingSuper:
 Lint/ConstantDefinitionInBlock:
  Enabled: false
 
+Lint/SuppressedException:
+ Enabled: false
+
 Style/SingleArgumentDig:
  Enabled: false

lib/simple_text_extract/extract.rb

@@ -87,13 +87,14 @@ def xlsx_extract
  spreadsheet = Roo::Spreadsheet.open(file, only_visible_sheets: true)
 
  text = []
-
  spreadsheet.sheets.each_with_index do |name, i|
  text << "# Sheet Index: #{i}"
  text << "# Sheet Name: #{name}"
 
  spreadsheet.sheet(name)&.each_row_streaming do |row|
- text << row.map(&:to_s).join(" ")
+ text << row.map do |cell|
+ cell.value.to_s
+ end.join(" ")
  end
  end
 

lib/simple_text_extract/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module SimpleTextExtract
- VERSION = "3.0.7"
+ VERSION = "3.0.8"
 end

simple_text_extract.gemspec

@@ -27,6 +27,6 @@ Gem::Specification.new do |spec|
  spec.requirements << "pdftotext/poppler"
 
  spec.add_dependency "roo", "~> 2.10.0"
- spec.add_dependency "spreadsheet", "~> 1.3.0"
  spec.add_dependency "rubyzip", "~> 2.3.2"
+ spec.add_dependency "spreadsheet", "~> 1.3.0"
 end

test/simple_text_extract_test.rb

@@ -112,6 +112,11 @@ def test_it_parses_xlsx_files_to_text_from_raw_excludes_hidden
  assert_equal "# Sheet Index: 0\n# Sheet Name: Sheet1\nruby 25\n# Sheet Index: 1\n# Sheet Name: Sheet2\njs 35", result
  end
 
+ def test_it_parses_xlsx_files_to_text_with_complex_text
+ result = SimpleTextExtract.extract(filename: "test_xlsx.xlsx", raw: File.read("test/fixtures/test_file_with_complex_text.xlsx"))
+ assert result.include?("Pricing Template")
+ end
+
  def test_nil_to_integer
  result = SimpleTextExtract.extract(filename: "roo_bad_link.xlsx", raw: File.read("test/fixtures/roo_bad_link.xlsx"))