ngty · May 21, 2010 01:52
diff --git a/power_xpath_for_matching_table_by_th_and_td b/power_xpath_for_matching_table_by_th_and_td
 # This script shows how a single xpath can be used for matching a specific row in an html
 # table, taking into account the the header text <th/> and cell text <td/>. It is able to
 # handle case-sensitivity & nested inner text.
 # 
 # Pls note that in order to test out case-sensitivity, u need to tweak IS_CASE_SENSITIVE &
 # the html string.
 #
 # BTW, let me know how the generated xpath can be more elegant than it currently is.
 #
 require 'rubygems'                                                                                                             
 require 'nokogiri'                                                                                                             
                                                                                                                               
 UPPERCASE_CHARS = ('A'..'Z').to_a.join('')                                                                                     
 LOWERCASE_CHARS = ('a'..'z').to_a.join('')                                                                                     
 IS_CASE_SENSITIVE = false                                                                                                      
                                                                                                                               
 def t(string)                                                                                                                  
  IS_CASE_SENSITIVE ? string : %\translate(#{string},"#{UPPERCASE_CHARS}","#{LOWERCASE_CHARS}")\                               
 end                                                                                                                            
                                                                                                                               
 def qt(string)                                                                                                                 
  t(%\"#{string}"\)                                                                                                            
 end                                                                                                                            
                                                                                                                               
 begin                                                                                                                          
  puts \                                                                                                                       
 "---------------------------------------------------------\n" +                                                                
 "Expected Result: " + Nokogiri::HTML(<<-EOL                                                                                    
  <html>                                                                                                                       
    <body>                                                                                                                     
      <table>                                                                                                                  
        <tr><th><span>#</span></th><th><span>Na<br/>me</span></th><th>Gender</th></tr>                                         
        <tr><td>1</td><td>LL</td><td>Fem</br>ale</td></tr>                                                                     
        <tr><td>2</td><td>TY</td><td>Male</td></tr>                                                                            
      </table>                                                                                                                 
    </body>                                                                                                                    
  </html>                                                                                                                      
  EOL                                                                                                                          
  ).xpath(
    # NOTE: This is the part that generates the xpath !!                                                                                                                   
    @path = '//table[.' +                                                                                                      
      [%w{# 1}, %w{Name LL}, %w{Gender Female}].inject([]) do |memo, arg|                                                      
        field, value = arg                                                                                                     
        pos_cond = %\count(./ancestor::table[1]//th[#{t('normalize-space(.)')}=#{qt(field)}][1]/preceding-sibling::th)+1\      
        txt_cond = %\#{t('normalize-space(.)')}=#{qt(value)}\                                                                  
        memo << %\//td[#{pos_cond}][#{txt_cond}]\                                                                              
      end.join('/ancestor::tr[1]') + ']'                                                                                       
  ).text.split(/\s+/).to_s                                                                                                     
                                                                                                                               
  puts [                                                                                                                       
    "---------------------------------------------------------",                                                               
    "Hurray !! We have a powerful XPATH builder that builds:",                                                                 
    "---------------------------------------------------------",                                                               
    @path                                                                                                                      
  ].join("\n")                                                                                                                 
                                                                                                                               
 rescue                                                                                                                         
  puts([                                                                                                                       
    "---------------------------------------------------------",                                                               
    "Yukes !! Something wrong with built XPATH:",                                                                              
    "---------------------------------------------------------",                                                               
    @path                                                                                                                      
  ].join("\n"))                                                                                                                
 end
	# This script shows how a single xpath can be used for matching a specific row in an html
	# table, taking into account the the header text <th/> and cell text <td/>. It is able to
	# handle case-sensitivity & nested inner text.
	#
	# Pls note that in order to test out case-sensitivity, u need to tweak IS_CASE_SENSITIVE &
	# the html string.
	#
	# BTW, let me know how the generated xpath can be more elegant than it currently is.
	#
	require 'rubygems'
	require 'nokogiri'

	UPPERCASE_CHARS = ('A'..'Z').to_a.join('')
	LOWERCASE_CHARS = ('a'..'z').to_a.join('')
	IS_CASE_SENSITIVE = false

	def t(string)
	IS_CASE_SENSITIVE ? string : %\translate(#{string},"#{UPPERCASE_CHARS}","#{LOWERCASE_CHARS}")\
	end

	def qt(string)
	t(%\"#{string}"\)
	end

	begin
	puts \
	"---------------------------------------------------------\n" +
	"Expected Result: " + Nokogiri::HTML(<<-EOL
	<html>
	<body>
	<table>
	<tr><th><span>#</span></th><th><span>Na<br/>me</span></th><th>Gender</th></tr>
	<tr><td>1</td><td>LL</td><td>Fem</br>ale</td></tr>
	<tr><td>2</td><td>TY</td><td>Male</td></tr>
	</table>
	</body>
	</html>
	EOL
	).xpath(
	# NOTE: This is the part that generates the xpath !!
	@path = '//table[.' +
	[%w{# 1}, %w{Name LL}, %w{Gender Female}].inject([]) do \|memo, arg\|
	field, value = arg
	pos_cond = %\count(./ancestor::table[1]//th[#{t('normalize-space(.)')}=#{qt(field)}][1]/preceding-sibling::th)+1\
	txt_cond = %\#{t('normalize-space(.)')}=#{qt(value)}\
	memo << %\//td[#{pos_cond}][#{txt_cond}]\
	end.join('/ancestor::tr[1]') + ']'
	).text.split(/\s+/).to_s

	puts [
	"---------------------------------------------------------",
	"Hurray !! We have a powerful XPATH builder that builds:",
	"---------------------------------------------------------",
	@path
	].join("\n")

	rescue
	puts([
	"---------------------------------------------------------",
	"Yukes !! Something wrong with built XPATH:",
	"---------------------------------------------------------",
	@path
	].join("\n"))
	end