tables 0.0.14

data/lib/tables/table_reader.rb

@@ -0,0 +1,65 @@
+# 
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+# 
+
+module Tables
+
+  class TableReader
+
+    attr_reader :tables
+
+    def initialize
+      @tables=[]
+    end
+
+    def get_tables
+      result=[]
+      if block_given? then
+        @tables.each {|t| result<<t if yield(t)}
+      else
+        result=@tables.clone
+      end
+      result
+    end
+
+    def merge_tables
+      sortmerge={}
+      @tables.each do |t|
+        key=t.header
+        if sortmerge.has_key?(key) then
+          sortmerge[key].merge!(t)
+        else
+          sortmerge[key]=t
+        end
+      end
+      @tables=[]
+      sortmerge.values.each {|v| @tables << v}
+    end
+
+    def extract_all_tables(idx_array=[*0..(table_count-1)]) # table_count is defined by the concrete class
+      idx_array.each do |idx|
+        table = extract_table(idx)
+        unless table.nil? then
+          status = ">> Extracted table '#{table.name}'"
+          sig=">> signature - #{table.signature}"
+          if block_given? then
+            match=yield(table)
+            if match then
+              @tables << table
+            else
+              status += " - (filtered out)"
+            end
+          else
+            @tables << table
+          end
+          puts status
+          puts sig
+        end
+      end
+    end
+
+  end
+
+end

data/lib/tables/word_table_reader.rb

@@ -0,0 +1,83 @@
+# 
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+# 
+require 'win32ole'
+
+module Tables
+
+  class WordTableReader < TableReader
+
+    attr_reader :doc
+
+    def initialize(filename, opts={})
+      fso=WIN32OLE.new('Scripting.FileSystemObject')
+      path=fso.GetAbsolutePathName(filename)
+      @app=WIN32OLE.new("Word.Application")
+      @doc=@app.Documents.Open(path, opts)
+      super()
+    end
+
+    def table_count
+      @doc.tables.count
+    end
+
+    def extract_table(idx)
+      name="Table #{idx+1}"
+      table=@doc.tables(idx+1)
+      result=[]
+      jdx=0
+      table.rows.each do |row|
+        jdx+=1
+        begin
+          result << WordTableReader.extract_row(row)
+        rescue Exception=>e1
+          puts "ERROR: Error extracting row #{jdx} from table '#{name}'"
+          puts "ERROR: #{e1.message}"
+        end
+      end
+      Table.new(result, name)
+
+    rescue Exception=>e2
+      puts "ERROR: Error extracting table '#{name}'"
+      puts "ERROR: #{e2.message}"
+    end
+
+    def exit
+      @app.quit
+    end
+
+  private
+
+    def WordTableReader.extract_row(ole_row)
+      row=[]
+      ole_row.cells.each do |cell|
+        row << WordTableReader.extract_text(cell.range)
+      end
+      row
+    end
+
+    def WordTableReader.extract_text(range)
+      paragraphs=[]
+      range_text=range.Text
+      range.paragraphs.each do |p|
+        text = clean_bytes(p.range.Text)
+        list_text = clean_bytes(p.range.ListFormat.ListString)
+        list_type = p.range.ListFormat.ListType
+        p_text = text.empty? ? list_text : text   # return list_text if text is empty
+        p_bullet = list_type==0 ? "" : "\t- "
+        paragraphs << p_bullet+p_text.strip
+      end
+      paragraphs.join("\n")
+    end
+
+    def WordTableReader.clean_bytes(string)
+      new_string=""
+      string.each_char {|c| new_string<<c unless c=="\r" or c=="\a"}
+      new_string
+    end
+
+  end
+
+end

data/lib/tables/word_table_writer.rb

@@ -0,0 +1,64 @@
+#
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+#
+require 'win32ole'
+
+module Tables
+
+  class WordTableWriter
+    attr_reader :doc
+
+    def initialize
+      @app=WIN32OLE.new 'Word.Application'
+      @doc=@app.Documents.Add
+      super()
+    end
+
+    def save_as(filename)
+      fso=WIN32OLE.new('Scripting.FileSystemObject')
+      @path=fso.GetAbsolutePathName(filename)
+      @doc.SaveAs(@path)
+    end
+
+    def exit(save_changes=true)
+      code=-1
+      code=0 unless save_changes
+      @app.quit(code)
+    end
+
+    def append_table(table, caption=nil)
+      @app.Selection.EndKey(6)
+      word_table=@doc.Tables.Add(@app.Selection.Range, table.row_count, table.column_count)
+      table.each_row_with_index do |row, idx|
+        word_row=word_table.Rows(idx+1)
+        row.each_with_index do |cell_value, jdx|
+          word_row.Cells(jdx+1).Range.Text=cell_value.to_s
+        end
+      end
+
+      style_borders(word_table)
+      set_caption(word_table,caption) unless caption.nil?
+    end
+
+  private
+    def style_borders(word_table)
+      (1..6).each do |border|
+        word_table.Borders(border).LineStyle=1
+      end
+
+      (1..4).each do |border|
+        word_table.Borders(border).LineWidth=12
+      end
+      word_table.Rows(1).Borders(3).LineWidth=12
+    end
+
+    def set_caption(word_table, caption)
+      word_table.Range.InsertCaption("Table",": #{caption}",nil,0)
+    end
+
+  end
+
+end
+

data/spec/copy_spec.rb

@@ -0,0 +1,33 @@
+# 
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+# 
+
+require "rspec"
+require "tables"
+
+describe Tables::Table do
+
+  it "should do a columnwise copy of one table to another" do
+    t1=Tables::Table.new([
+      [:a,:b,:d,:f],
+      [1,1,1,1],
+      [2,2,2,2],
+      [3,3,3,3]
+      ])
+    t2=Tables::Table.new([
+      [:a,:b,:c,:d,:e,:f]
+      ])
+    t_expect=Tables::Table.new([
+      [:a,:b,:c,:d,:e,:f],
+      [1,1,nil,1,nil,1],
+      [2,2,nil,2,nil,2],
+      [3,3,nil,3,nil,3]
+      ])
+
+    t2.column_copy(t1)
+    (t2==t_expect).should==true
+  end
+
+end

data/spec/excel_table_reader_spec.rb

@@ -0,0 +1,132 @@
+# 
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+# 
+
+require 'tables'
+
+describe Tables::ExcelTableReader do
+
+  it "should read an excel document" do
+    filename=File.dirname(__FILE__)+"/../test/rtm.xlsx"
+    @xtr = Tables::ExcelTableReader.new(filename)
+    @xtr.workbook.should_not==nil
+    @xtr.exit
+  end
+
+  it "should extract a table" do
+    filename=File.dirname(__FILE__)+"/../test/rtm.xlsx"
+    @xtr = Tables::ExcelTableReader.new(filename)
+    t=@xtr.extract_table(0)
+#    t.each_row {|r| puts "#{r[0]}, #{r[1]}, #{r[2]}"}
+    t.rows.should==457
+    t[1][0].should=~/Unique ID/
+    t[2][0].should=~/CAMM\.AC\.101/
+    t[1].last.should=~/PolicyNet Doc Ref/
+  end
+
+  it "should extract multiple tables" do
+    filename=File.dirname(__FILE__)+"/../test/rtm.xlsx"
+    @xtr = Tables::ExcelTableReader.new(filename)
+    @xtr.extract_all_tables([0,1,2])
+    @xtr.tables.count.should==3
+  end
+
+  it "should clean up a table" do
+    filename=File.dirname(__FILE__)+"/../test/rtm.xlsx"
+    @xtr = Tables::ExcelTableReader.new(filename)
+    @xtr.extract_all_tables([2])
+    @xtr.tables.count.should==1
+    @xtr.clean
+    t=@xtr.tables[0]
+    t.rows.should==55
+
+    col1=[]
+    t.each_row {|row| col1<< row[0].strip }
+    col1_expected=["Unique ID", "CAMM.GE.101", "CAMM.GE.102", "CAMM.GE.103", "CAMM.GE.104", "CAMM.GE.105", "CAMM.GE.106", "CAMM.GE.107", "CAMM.GE.108", "CAMM.GE.109", "CAMM.GE.110", "CAMM.GE.111", "CAMM.GE.112", "CAMM.GE.113", "CAMM.GE.114", "CAMM.GE.115", "CAMM.SM.101", "CAMM.SM.102", "CAMM.SM.103", "CAMM.BP.101", "CAMM.BP.102", "CAMM.BP.103", "CAMM.BP.104", "CAMM.BP.105", "CAMM.BP.106", "CAMM.BP.107", "CAMM.BP.108", "CAMM.GP.101", "CAMM.SI.101", "CAMM.SI.102", "CAMM.HW.101", "CAMM.HW.102", "CAMM.HW.103", "CAMM.HW.104", "CAMM.HA.101", "CAMM.HA.102", "CAMM.HA.103", "CAMM.HA.104", "CAMM.HA.105", "CAMM.HA.106", "CAMM.HA.107", "CAMM.HA.108", "CAMM.HA.109", "CAMM.HA.110", "CAMM.HA.111", "CAMM.HA.112", "CAMM.AI.101", "CAMM.AI.102", "CAMM.AI.103", "CAMM.WG.101", "CAMM.WG.102", "CAMM.WG.103", "CAMM.WG.104", "CAMM.WG.105", "CAMM.MU.101"]
+    col1.should.eql? col1_expected
+  end
+
+  it "should write a table" do
+    t=Tables::Table.new([
+        [:a,:b,:c,:d],
+        [1,1,1,1],
+        [2,2,2,2],
+        [3,3,3,3]
+      ])
+    outfile=File.dirname(__FILE__)+"/../test/writetest.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    @xtr.write_table(t)
+    @xtr.save
+    @xtr.exit
+  end
+
+  it "should write a table to a named worksheet" do
+    t=Tables::Table.new([
+        [:a,:b,:c,:d],
+        [1,1,1,1],
+        [2,2,2,2],
+        [3,3,3,3]
+      ])
+    outfile=File.dirname(__FILE__)+"/../test/writetest1.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    @xtr.write_table(t, "Sheet2")
+    @xtr.save
+    @xtr.exit
+  end
+
+  it "should write a column" do
+    t=Tables::Table.new([
+    [:a,:b,:c,:d],
+    [1,1,1,1],
+    [2,2,2,2],
+    [3,3,3,3]
+    ])
+    outfile=File.dirname(__FILE__)+"/../test/writetest2.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    @xtr.write_column(t,:c)
+    @xtr.save
+    @xtr.exit
+  end
+
+  it "should write a column to a named worksheet" do
+    t=Tables::Table.new([
+    [:a,:b,:c,:d],
+    [1,1,1,1],
+    [2,2,2,2],
+    [3,3,3,3]
+    ])
+    outfile=File.dirname(__FILE__)+"/../test/writetest2a.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    @xtr.write_column(t,:c, "Sheet3")
+    @xtr.save
+    @xtr.exit
+  end
+
+  it "should append worksheets automatically" do
+    t=Tables::Table.new([
+    [:a,:b,:c,:d],
+    [1,1,1,1],
+    [2,2,2,2],
+    [3,3,3,3]
+    ])
+    outfile=File.dirname(__FILE__)+"/../test/writetest3.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    10.times {|idx| @xtr.write_table(t,idx)}
+    @xtr.save
+    @xtr.exit
+  end
+
+end
+

data/spec/office_copy_spec.rb

@@ -0,0 +1,55 @@
+# 
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+# 
+
+require "rspec"
+require 'tables'
+
+describe Tables::WordTableReader do
+
+  it "should copy word table to excel" do
+    filename=File.dirname(__FILE__)+"/../test/rtm.docx"
+    @wtr = Tables::WordTableReader.new(filename)
+    ntables=@wtr.table_count
+    @wtr.extract_all_tables((0..ntables-2).to_a) {|table| table[0][0]=~/Unique ID/}
+    @wtr.merge_tables
+    @wtr.tables.should have(1).item
+    table=@wtr.tables[0]
+    @wtr.exit
+
+    outfile=File.dirname(__FILE__)+"/../test/copytest.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    @xtr.write_table(table)
+    @xtr.save
+    @xtr.exit
+  end
+
+  it "should copy word table to an excel template" do
+    filename=File.dirname(__FILE__)+"/../test/rtm.docx"
+    @wtr = Tables::WordTableReader.new(filename)
+    ntables=@wtr.table_count
+    @wtr.extract_all_tables((0..ntables-2).to_a) {|table| table[0][0]=~/Unique ID/}
+    @wtr.merge_tables
+    @wtr.tables.should have(1).item
+    table=@wtr.tables[0]
+    @wtr.exit
+
+    template_file=File.dirname(__FILE__)+"/../test/rtm_template.xlsx"
+    ttr=Tables::ExcelTableReader.new(template_file)
+    template=ttr.extract_table(1)
+    template.column_copy(table)
+    ttr.exit
+
+    outfile=File.dirname(__FILE__)+"/../test/copytest2.xlsx"
+    File.delete(outfile) if File.exist?(outfile)
+    @xtr=Tables::ExcelTableReader.new()
+    @xtr.create_file(outfile)
+    @xtr.write_table(template)
+    @xtr.save
+    @xtr.exit
+  end
+end

data/spec/table_spec.rb

@@ -0,0 +1,237 @@
+# 
+#  Author::    Saul Caganoff  (mailto:scaganoff@gmail.com)
+#  Copyright:: Copyright (c) 2010, Saul Caganoff
+#  License::   Creative Commons Attribution 3.0 Australia License (http://creativecommons.org/licenses/by/3.0/au/)
+# 
+
+require 'tables'
+
+describe Tables::Table do
+  before(:each) do
+    @t1=Tables::Table.new([
+      [:a,:b,:c,:d],
+      [1,1,1,1],
+      [2,2,9,2],
+      [3,3,3,3]
+    ])
+    @t2=Tables::Table.new([
+      [:a,:b,:c,:d],
+      [4,4,4,4],
+      [5,5,5,5]
+    ])
+    @t3=Tables::Table.new([
+      [:x,:y,:z],
+      [4,4,4],
+      [5,5,5]
+    ])
+    @texpect=Tables::Table.new([
+      [:a,:b,:c,:d],
+      [1,1,1,1],
+      [2,2,9,2],
+      [3,3,3,3],
+      [4,4,4,4],
+      [5,5,5,5]
+    ])
+  end
+
+  #it "should determine equality" do
+  #  t=Tables::Table.new([
+  #      [:a,:b,:c,:d],
+  #      [1,1,1,1],
+  #      [2,2,9,2],
+  #      [3,3,3,3]
+  #   ])
+  #  (t==@t1).should==true
+  #end
+  #it "should determine similarity" do
+  #  @t1.similar?(@t2).should==true
+  #  @t1.similar?(@t3).should==false
+  #end
+  #
+  #it "should merge tables cleanly" do
+  #  @t1.merge!(@t2)
+  #  (@t1==@texpect).should==true
+  #end
+  #
+  #it "should accept a new row" do
+  #  @t1 << [4,4,4,4]
+  #  @t1 << [5,5,5,5]
+  #  (@t1==@texpect).should==true
+  #end
+  #
+  #it "should report columns" do
+  #  @t1.column_count.should==4
+  #end
+  #
+  #it "should report rows" do
+  #  @t1.row_count.should==4
+  #end
+  #
+  #it "should access cell value by column name" do
+  #  @t1.get_value(:c, 2).should==9
+  #  lambda { t.get_value(:z,2) }.should raise_error
+  #end
+  #
+  #it "should provide row as a hash" do
+  #  row=@t1.get_row(2)
+  #  row.should.is_a? Hash
+  #  row[:c].should==9
+  #  row[:a].should==2
+  #end
+  #
+  #it "should get a named column" do
+  #  c=@t1.get_column(:c)
+  #  c.should.eql? [:c,1,9,3]
+  #end
+  #
+  #it "should delete a named column" do
+  #  c=@t1.get_column(:c)
+  #  c.should.eql? [:c,1,9,3]
+  #  t=@t1.delete_column(:c)
+  #  t_expect=Tables::Table.new([
+  #    [:a,:b,:d],
+  #    [1,1,1],
+  #    [2,2,2],
+  #    [3,3,3]
+  #  ])
+  #  lambda { t.get_column(:c) }.should raise_error
+  #end
+  #
+  #it "should delete the last column" do
+  #  @t1.delete_column(@t1.header.last)
+  #  texpect=Tables::Table.new([
+  #    [:a,:b,:c],
+  #    [1,1,1],
+  #    [2,2,9],
+  #    [3,3,3]
+  #  ])
+  #  (@t1==texpect).should==true
+  #end
+  #
+  #it "should remove blank rows" do
+  #  t_blankrow=Tables::Table.new([
+  #    [:a,:b,:c,:d],
+  #    [1,1,1,1],
+  #    ["","","",""],
+  #    [3,3,3,3]
+  #  ])
+  #  t_noblankrow=Tables::Table.new([
+  #    [:a,:b,:c,:d],
+  #    [1,1,1,1],
+  #    [3,3,3,3]
+  #  ])
+  #  t_blankrow.remove_blank_rows!
+  #  t_blankrow.row_count.should==3
+  #  (t_blankrow==t_noblankrow).should==true
+  #end
+  #
+  #it "should remove head rows" do
+  #  t_blankrow=Tables::Table.new([
+  #    [:a,:b,:c,:d],
+  #    [1,1,1,1],
+  #    ["head","","",""],
+  #    [3,3,3,3]
+  #  ])
+  #  t_noblankrow=Tables::Table.new([
+  #    [:a,:b,:c,:d],
+  #    [1,1,1,1],
+  #    [3,3,3,3]
+  #  ])
+  #  t_blankrow.remove_blank_rows!(1)
+  #  (t_blankrow==t_noblankrow).should==true
+  #  t_blankrow.row_count.should==3
+  #end
+  #
+  #it "should remove repeated headers" do
+  #  t=Tables::Table.new([
+  #    [:a,:b,:c,:d],
+  #    [1,1,1,1],
+  #    [2,2,2,2],
+  #    [3,3,3,3],
+  #    [:a,:b,:c,:d],
+  #    [4,4,4,4],
+  #    [5,5,5,5]
+  #  ])
+  #  t_expect=Tables::Table.new([
+  #    [:a,:b,:c,:d],
+  #    [1,1,1,1],
+  #    [2,2,2,2],
+  #    [3,3,3,3],
+  #    [4,4,4,4],
+  #    [5,5,5,5]
+  #  ])
+  #  t.remove_repeat_headers!
+  #  (t==t_expect).should==true
+  #  t.row_count.should==6
+  #end
+  #
+  #it "should demerge on default column 0" do
+  #  t_merged=Tables::Table.new([
+  #    ["a","foo","45"],
+  #    ["","sub-foo","46"],
+  #    ["","sub-foo-2",""],
+  #    ["b","bar","47"],
+  #    ["c","baz",""]
+  #  ])
+  #  t=t_merged.demerge!
+  #  t_expect=Tables::Table.new([
+  #    ["a","foo\nsub-foo\nsub-foo-2","45\n46"],
+  #    ["b","bar","47"],
+  #    ["c","baz",""]
+  #    ])
+  #  (t==t_expect).should==true
+  #end
+  #
+  #it "should demerge on any specified column" do
+  #  t_merged=Tables::Table.new([
+  #    ["a","foo","45"],
+  #    ["","sub-foo","46"],
+  #    ["","sub-foo-2",""],
+  #    ["b","bar","47"],
+  #    ["c","baz",""]
+  #  ])
+  #  t=t_merged.demerge!(2)
+  #  t_expect=Tables::Table.new([
+  #    ["a","foo","45"],
+  #    ["","sub-foo\nsub-foo-2","46"],
+  #    ["b\nc","bar\nbaz","47"]
+  #    ])
+  #  (t==t_expect).should==true
+  #end
+  #
+  #it "should allow build Tables::Table" do
+  #  t=Tables::Table.new
+  #  t.add_row([:a,:b,:c,:d])
+  #  t.add_row([1,1,1,1])
+  #  t.add_row([2,2,9,2])
+  #  t.add_row([3,3,3,3])
+  #  (t==@t1).should==true
+  #  t.get_value(:c,2).should==9
+  #end
+  #
+  #it "should allow rename of column" do
+  #  t=Tables::Table.new
+  #  t.add_row([:a,:b,:c,:d])
+  #  t.add_row([1,1,1,1])
+  #  t.add_row([2,2,9,2])
+  #  t.add_row([3,3,3,3])
+  #  t.rename_column(:c, :z)
+  #  t.get_value(:z,2).should==9
+  #  lambda { t.get_value(:c,2) }.should raise_error
+  #end
+
+  it "should update a row via get" do
+    t_expect=Tables::Table.new([
+      [:a,:b,:c,:d],
+      [1,1,9,1],
+      [2,2,9,2],
+      [3,3,9,3]])
+    @t1.get_each_row do |row|
+      row[:c]=9
+      @t1.set_row(row)
+    end
+    (@t1==t_expect).should==true
+  end
+
+end
+