ooxl 0.0.1.5.0 → 0.0.1.5.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
-SHA1:
-  metadata.gz: d3da0847d61333cf06c4a9e3212604a6b6f1e1e0
-  data.tar.gz: 2b5d30b7c28f88c14adfe0fcbe4e1c0085ccfb61
+SHA256:
+  metadata.gz: 79ac42fd268e59703904fe713eb0d8369072371b96f632687dd4aac471dbd94c
+  data.tar.gz: c336a07951a0e066debbfd0ae59d080a623fc3321da0ab1050e53724e31878ec
 SHA512:
-  metadata.gz: cb1e938d50dbfa698605986ae790670f0b2c557438aeb73925baf261326fe8f79b05948d409d06a5fbf620aaa88adba26642206b3beed11ca05e05372487a0ac
-  data.tar.gz: e30534d829e6b0ba29388a950cf5743977ee0b73c942c798df26eca177f9e17a70d365515c76bf37d20d6bb42c027ddad937223a75b582143a1cf1cac7c00d82
+  metadata.gz: e9a19b166119394c6f883c2c049cae44cea55fc4d3624ad45adb48866618175ac3326ae54a768f589a2e35c18eed81dc7c4171cf01b641aa02257ce56dd809f9
+  data.tar.gz: 8a5fa75a6f2941b38ed379ca6cdf8dee5f6fa893c2e69e70bcff6db0591ce129e1dc175d9b9bf465034c00f643d74ac63bb6dcc8d937a3bbda4a41fcc6f27ed1

data/bin/console

@@ -1,7 +1,7 @@
 #!/usr/bin/env ruby
 
 require "bundler/setup"
-require "ooxml_excel"
+require "./lib/ooxl"
 
 # You can add fixtures and/or initialization code here to make experimenting
 # with your gem easier. You can also use a different console, if you like.
@@ -10,5 +10,5 @@ require "ooxml_excel"
 # require "pry"
 # Pry.start
 
-require "irb"
-IRB.start
+require "pry"
+Pry.start

data/lib/ooxl/ooxl.rb

@@ -3,20 +3,33 @@ class OOXL
   include ListHelper
   attr_reader :filename
   
-  def initialize(spreadsheet_filepath, options={})
+  def initialize(filepath = nil, contents: nil, **options)
     @workbook = nil
     @sheets = {}
     @styles = []
     @comments = {}
-    @relationships = {}
+    @workbook_relationships = nil
+    @sheet_relationships = {}
     @options = options
     @tables = []
-    @filename = File.basename(spreadsheet_filepath)
-    parse_spreadsheet_contents(spreadsheet_filepath)
+
+    @filename = filepath && File.basename(filepath)
+    if contents.present?
+      parse_spreadsheet_contents(contents)
+    elsif filepath.present?
+      parse_spreadsheet_file(filepath)
+    else
+      raise 'no file path or contents were provided'
+    end
   end
 
   def self.open(spreadsheet_filepath, options={})
-    new(spreadsheet_filepath, options)
+    new(spreadsheet_filepath, **options)
+  end
+
+  def self.parse(spreadsheet_contents, options={})
+    spreadsheet_contents.force_encoding('ASCII-8BIT') if spreadsheet_contents.respond_to?(:force_encoding)
+    new(nil, contents: spreadsheet_contents, **options)
   end
 
   def sheets(skip_hidden: false)
@@ -33,9 +46,11 @@ class OOXL
   end
 
   def sheet(sheet_name)
-    sheet_index = @workbook.sheets.index { |sheet| sheet[:name] == sheet_name}
-    raise "No #{sheet_name} in workbook." if sheet_index.nil?
-    sheet = @sheets.fetch((sheet_index+1).to_s)
+    sheet_meta = @workbook.sheets.find { |sheet| sheet[:name] == sheet_name }
+    raise "No #{sheet_name} in workbook." if sheet_meta.nil?
+
+    sheet_index = @workbook_relationships[sheet_meta[:relationship_id]].scan(/\d+/).first
+    sheet = @sheets.fetch(sheet_index)
 
     # shared variables
     sheet.name = sheet_name
@@ -73,38 +88,46 @@ class OOXL
   end
 
   def fetch_comments(sheet_index)
-    final_sheet_index = sheet_index+1
-    relationship = @relationships[final_sheet_index.to_s]
+    relationship = @sheet_relationships[sheet_index]
     @comments[relationship.comment_id] if relationship.present?
   end
 
-  def parse_spreadsheet_contents(spreadsheet)
+  def parse_spreadsheet_file(file_path)
+    Zip::File.open(file_path) { |zip| parse_zip(zip) }
+  end
+
+  def parse_spreadsheet_contents(file_contents)
+    # open_buffer works for strings and IO streams
+    Zip::File.open_buffer(file_contents) { |zip| parse_zip(zip) }
+  end
+
+  def parse_zip(spreadsheet_zip)
     shared_strings = []
-    Zip::File.open(spreadsheet) do |spreadsheet_zip|
-      spreadsheet_zip.each do |entry|
-        case entry.name
-        when /xl\/worksheets\/sheet(\d+)?\.xml/
-          sheet_id = entry.name.scan(/xl\/worksheets\/sheet(\d+)?\.xml/).flatten.first
-          @sheets[sheet_id] = OOXL::Sheet.new(entry.get_input_stream.read, shared_strings, @options)
-        when /xl\/styles\.xml/
-          @styles = OOXL::Styles.load_from_stream(entry.get_input_stream.read)
-        when /xl\/comments(\d+)?\.xml/
-          comment_id = entry.name.scan(/xl\/comments(\d+)\.xml/).flatten.first
-          @comments[comment_id] = OOXL::Comments.load_from_stream(entry.get_input_stream.read)
-        when "xl/sharedStrings.xml"
-          Nokogiri.XML(entry.get_input_stream.read).remove_namespaces!.xpath('sst/si').each do |shared_string_node|
-            shared_strings << shared_string_node.xpath('r/t|t').map { |value_node| value_node.text}.join('')
-          end
-        when /xl\/tables\/.*?/i
-          @tables << OOXL::Table.new(entry.get_input_stream.read)
-        when "xl/workbook.xml"
-          @workbook = OOXL::Workbook.load_from_stream(entry.get_input_stream.read)
-        when /xl\/worksheets\/_rels\/sheet\d+\.xml\.rels/
-          sheet_id = entry.name.scan(/sheet(\d+)/).flatten.first
-          @relationships[sheet_id] = Relationships.new(entry.get_input_stream.read)
-        else
-          # unsupported for now..
+    spreadsheet_zip.each do |entry|
+      case entry.name
+      when /xl\/worksheets\/sheet(\d+)?\.xml/
+        sheet_id = entry.name.scan(/xl\/worksheets\/sheet(\d+)?\.xml/).flatten.first
+        @sheets[sheet_id] = OOXL::Sheet.new(entry.get_input_stream.read, shared_strings, @options)
+      when /xl\/styles\.xml/
+        @styles = OOXL::Styles.load_from_stream(entry.get_input_stream.read)
+      when /xl\/comments(\d+)?\.xml/
+        comment_id = entry.name.scan(/xl\/comments(\d+)\.xml/).flatten.first
+        @comments[comment_id] = OOXL::Comments.load_from_stream(entry.get_input_stream.read)
+      when "xl/sharedStrings.xml"
+        Nokogiri.XML(entry.get_input_stream.read).remove_namespaces!.xpath('sst/si').each do |shared_string_node|
+          shared_strings << shared_string_node.xpath('r/t|t').map { |value_node| value_node.text}.join('')
         end
+      when /xl\/tables\/.*?/i
+        @tables << OOXL::Table.new(entry.get_input_stream.read)
+      when "xl/workbook.xml"
+        @workbook = OOXL::Workbook.load_from_stream(entry.get_input_stream.read)
+      when /xl\/worksheets\/_rels\/sheet\d+\.xml\.rels/
+        sheet_id = entry.name.scan(/sheet(\d+)/).flatten.first
+        @sheet_relationships[sheet_id] = Relationships.new(entry.get_input_stream.read)
+      when /xl\/_rels\/workbook\.xml\.rels/
+        @workbook_relationships = Relationships.new(entry.get_input_stream.read)
+      else
+        # unsupported for now..
       end
     end
   end

data/lib/ooxl/util.rb

@@ -1,12 +1,13 @@
-class OOXL
+class OOXL
   module Util
-    COLUMN_LETTERS = ('A'..'ZZZZ').to_a
-    def letter_equivalent(index)
-      COLUMN_LETTERS.fetch(index)
+    COLUMN_LETTERS = [nil] + ('A'..'ZZZZ').to_a
+
+    def letter_index(col_letter)
+      column_letter_to_number(col_letter) - 1
     end
 
-    def letter_index(letter)
-      COLUMN_LETTERS.index { |c_letter| c_letter == letter}
+    def letter_equivalent(col_index)
+      column_number_to_letter(col_index + 1)
     end
 
     def to_column_letter(reference)
@@ -14,13 +15,17 @@ class OOXL
     end
 
     def uniform_reference(ref)
-      ref.to_s[/[A-Z]/] ? letter_index(ref) + 1 : ref
+      ref.to_s[/[A-Z]/] ? column_letter_to_number(ref) : ref
     end
 
     def node_value_extractor(node)
       node.try(:value)
     end
 
+    def column_number_to_letter(index)
+      COLUMN_LETTERS.fetch(index)
+    end
+
     def column_letter_to_number(column_letter)
       pow = column_letter.length - 1
       result = 0

data/lib/ooxl/version.rb

@@ -1,3 +1,3 @@
 class OOXL
-  VERSION = "0.0.1.5.0"
+  VERSION = "0.0.1.5.5"
 end

data/lib/ooxl/xl_objects/relationships.rb

@@ -1,29 +1,41 @@
 class OOXL
   class Relationships
     SUPPORTED_TYPES = ['http://schemas.openxmlformats.org/officeDocument/2006/relationships/comments']
+
     def initialize(relationships_node)
-      @types = {}
+      @relationships = []
       parse_relationships(relationships_node)
     end
 
     def comment_id
-      @types['comments']
+      comment_target = by_type('comments').first
+      comment_target && extract_file_reference(comment_target)
+    end
+
+    def [](id)
+      @relationships.find { |rel| rel.id == id }&.target
+    end
+
+    def by_type(type)
+      @relationships.select { |rel| rel.type == type }.map(&:target)
     end
 
     private
+
     def parse_relationships(relationships_node)
       relationships_node = Nokogiri.XML(relationships_node).remove_namespaces!
       relationships_node.xpath('//Relationship').each do |relationship_node|
         relationship_type = relationship_node.attributes["Type"].value
         target = relationship_node.attributes["Target"].value
-        if supported_type?(relationship_type)
-          @types[extract_type(relationship_type)] = extract_file_reference(target)
-        end
+        id = extract_number(relationship_node.attributes["Id"].value)
+        type = extract_type(relationship_type)
+        target = relationship_node.attributes["Target"].value
+        @relationships << Relationship.new(id, type, target)
       end
     end
 
-    def supported_type?(type)
-      SUPPORTED_TYPES.include?(type)
+    def extract_number(str)
+      str.scan(/(\d+)/).flatten.first
     end
 
     def extract_type(type)
@@ -34,6 +46,7 @@ class OOXL
       file.scan(/(\d+)\.[\w]/).flatten.first
     end
 
+    Relationship = Struct.new(:id, :type, :target)
   end
 end
 

data/lib/ooxl/xl_objects/row.rb

@@ -45,12 +45,16 @@ class OOXL
     end
 
     def self.load_from_node(row_node, shared_strings, styles, options)
-      new(id: row_node.attributes["r"].try(:value),
+      new(id: extract_id(row_node),
           spans: row_node.attributes["spans"].try(:value),
           height: row_node.attributes["ht"].try(:value),
           cells: row_node.xpath('c').map {  |cell_node| OOXL::Cell.load_from_node(cell_node, shared_strings, styles)},
           options: options )
     end
+
+    def self.extract_id(row_node)
+      row_node.attributes["r"].try(:value)
+    end
   end
 end
 

data/lib/ooxl/xl_objects/row_cache.rb

@@ -10,6 +10,8 @@ class OOXL
     # built on-demand -- use fetch_row_by_id instead
     attr_reader :row_id_map
 
+    delegate :size, to: :row_nodes
+
     def initialize(sheet_xml, shared_strings, options = {})
       @shared_strings = shared_strings
       @sheet_xml = sheet_xml
@@ -47,11 +49,33 @@ class OOXL
       row_cache
     end
 
+    def row_range(start_index, end_index)
+      return enum_for(:row_range, start_index, end_index) unless block_given?
+
+      rows do |row|
+        row_id = row.id.to_i
+        next if row_id < start_index
+        break if row_id > end_index
+
+        yield row
+      end
+    end
+
+    def max_row_index
+      return 0 if row_nodes.empty?
+
+      if all_rows_loaded?
+        row_cache.last.id.to_i
+      else
+        Row.extract_id(row_nodes.last).to_i
+      end
+    end
+
     private
 
     def parse_more_rows
       row_nodes.drop(row_cache.count).each do |row_node|
-        row = Row.load_from_node(row_node, @shared_strings, @styles, @options)
+        row = parse_row(row_node)
         row_cache << row
         row_id_map[row.id] = row
         yield row if block_given?
@@ -62,10 +86,6 @@ class OOXL
       row_cache.count == row_nodes.count
     end
 
-    def row_nodes
-      @row_nodes ||= @sheet_xml.xpath('//sheetData/row')
-    end
-
     def fetch_row_by_id(row_id)
       row_id = row_id.to_s
       return row_id_map[row_id] if all_rows_loaded? || row_id_map.key?(row_id)
@@ -95,6 +115,14 @@ class OOXL
       end
       yielded_rows
     end
+
+    def row_nodes
+      @row_nodes ||= @sheet_xml.xpath('//sheetData/row')&.to_a
+    end
+
+    def parse_row(row_node)
+      Row.load_from_node(row_node, @shared_strings, @styles, @options)
+    end
   end
 end
 

data/lib/ooxl/xl_objects/sheet.rb

@@ -128,7 +128,7 @@ class OOXL
       # cell_range values separated by comma
       if cell_range.include?(":")
         cell_letters = cell_range.gsub(/[\d]/, '').split(':')
-        start_index, end_index = cell_range[/[A-Z]{1,}\d+/] ? cell_range.gsub(/[^\d:]/, '').split(':').map(&:to_i) : [1, rows.size]
+        start_index, end_index = cell_range[/[A-Z]{1,}\d+/] ? cell_range.gsub(/[^\d:]/, '').split(':').map(&:to_i) : [1, @row_cache.max_row_index]
         if cell_letters.uniq.size > 1
           list_values_from_rectangle(cell_letters, start_index, end_index)
         else
@@ -148,22 +148,19 @@ class OOXL
     # 3 => end_index
     # Expected output would be: [['value', 'value', 'value'], ['value', 'value', 'value'], ['value', 'value', 'value']]
     def list_values_from_rectangle(cell_letters, start_index, end_index)
-      start_index.upto(end_index).map do |row_index|
-        (letter_index(cell_letters.first)..letter_index(cell_letters.last)).map do |cell_index|
-          row = row(row_index)
-          next if row.blank?
-
-          cell_letter = letter_equivalent(cell_index)
-          row["#{cell_letter}#{row_index}"].value
+      start_col = column_letter_to_number(cell_letters.first)
+      end_col = column_letter_to_number(cell_letters.last)
+      @row_cache.row_range(start_index, end_index).map do |row|
+        (start_col..end_col).map do |col_index|
+          col_letter = column_number_to_letter(col_index)
+          row["#{col_letter}#{row.id}"].value
         end
       end
     end
 
     def list_values_from_column(column_letter, start_index, end_index)
-      (start_index..end_index).to_a.map do |row_index|
-        row = row(row_index)
-        next if row.blank?
-        row["#{column_letter}#{row_index}"].value
+      @row_cache.row_range(start_index, end_index).map do |row|
+        row["#{column_letter}#{row.id}"].value
       end
     end
 
@@ -174,19 +171,19 @@ class OOXL
       [row[cell_ref].value]
     end
 
-    def self.load_from_stream(xml_stream, shared_strings)
-      self.new(Nokogiri.XML(xml_stream).remove_namespaces!, shared_strings)
-    end
-
     def in_merged_cells?(cell_id)
       column_letter, column_index = cell_id.partition(/\d+/)
-      range = merged_cells_range.find { |column_range, index_range| column_range.cover?(column_letter) && index_range.cover?(column_index) }
+      range = merged_cells.find { |column_range, index_range| column_range.cover?(column_letter) && index_range.cover?(column_index) }
       range.present?
     end
 
+    def self.load_from_stream(xml_stream, shared_strings)
+      self.new(xml_stream, shared_strings)
+    end
+
     private
 
-    def merged_cells_range
+    def merged_cells
       @merged_cells ||= @xml.xpath('//mergeCells/mergeCell').map do |merged_cell|
         # <mergeCell ref="Q381:R381"/>
         start_reference, end_reference = merged_cell.attributes["ref"].try(:value).split(':')

data/ooxml_excel.gemspec

@@ -26,7 +26,7 @@ Gem::Specification.new do |spec|
   spec.require_paths = ["lib"]
   spec.add_dependency 'activesupport'
   spec.add_dependency 'nokogiri', '~> 1'
-  spec.add_dependency 'rubyzip', '~> 1.0', '< 2.0.0'
+  spec.add_dependency 'rubyzip', '~> 1.3.0', '< 2.0.0'
 
   spec.add_development_dependency "bundler"
   spec.add_development_dependency "pry-byebug"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ooxl
 version: !ruby/object:Gem::Version
-  version: 0.0.1.5.0
+  version: 0.0.1.5.5
 platform: ruby
 authors:
 - James Mones
-autorequire: 
+autorequire:
 bindir: exe
 cert_chain: []
-date: 2019-10-08 00:00:00.000000000 Z
+date: 2021-08-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -44,7 +44,7 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.0'
+        version: 1.3.0
     - - "<"
       - !ruby/object:Gem::Version
         version: 2.0.0
@@ -54,7 +54,7 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.0'
+        version: 1.3.0
     - - "<"
       - !ruby/object:Gem::Version
         version: 2.0.0
@@ -153,7 +153,7 @@ files:
 homepage: https://github.com/halcjames/ooxl
 licenses: []
 metadata: {}
-post_install_message: 
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -168,9 +168,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubyforge_project: 
-rubygems_version: 2.6.12
-signing_key: 
+rubygems_version: 3.2.6
+signing_key:
 specification_version: 4
 summary: OOXL Excel - Parse Excel Spreadsheets (xlsx, xlsm).
 test_files: []