Ruby on Rails Diff Text to HTML <ins> and <del>
This code is perfect if you have 2 text objects in your Rails application and you want to compare their differences in one of your HTML views. It’s 99% pure Ruby too, so if you alter the first line, you can use it for other purposes.
Only one thing to note: you must have diff installed. I’m using: diff (GNU diffutils) 2.8.1.
-
#set up some variables to reference later
-
temporary_directory = File.join(Rails.root, "tmp")
-
max_lines = 9999999 #needs to be larger than the most lines you'll consider
-
diff_header_length = 3
-
-
# text_old and text_new should be the values of the string objects to compare
-
# these are just example strings to show it works
-
text_old = "line1\ndeleted line2\nline3\n\nline4\nline5"
-
text_new = "line1\ninserted line2\nline3\n\nline4\nline5"
-
-
# since we're using diff on the file system, we'll save the text we want to compare
-
# and then run diff against the two files
-
file_old_name = File.join(temporary_directory,"file_old"+rand(1000000).to_s)
-
file_new_name = File.join(temporary_directory,"file_new"+rand(1000000).to_s)
-
file_old = File.new(file_old_name, "w+")
-
file_new = File.new(file_new_name, "w+")
-
file_old.write(text_old+"\n")
-
file_new.write(text_new+"\n")
-
file_old.close
-
file_new.close
-
-
# diff will give provide a string showing insertions and deletions. We will
-
# split this string out by newlines if there are difference, and mark it up
-
# accordingly with html
-
lines = %x(diff --unified=#{max_lines} #{file_old_name} #{file_new_name})
-
if lines.empty?
-
lines = text_new.split(/\n/)
-
else
-
lines = lines.split(/\n/)[diff_header_length..max_lines].
-
collect do |i|
-
if i.empty?
-
""
-
else
-
case i[0,1]
-
when "+"; then "<ins>"+i[1..i.length-1]+"</ins>"
-
when "-"; then "<del>"+i[1..i.length-1]+"</del>"
-
else; i[1..i.length-1]
-
end
-
end
-
end
-
end
-
-
#clean up the temporary diff files we created
-
File.delete(file_new_name)
-
File.delete(file_old_name)
-
-
#return marked up text
-
lines.join("\n")
If you fire up RAILS_ROOT/script/console and paste that code in, it will return a nicely marked up string like this:
-
line1
-
<del>deleted line2</del>
-
<ins>inserted line2</ins>
-
line3
-
-
line4
-
line5
Use CSS to make your ins and del tags render however you like.
November 26th, 2008 at 1:50 am
Looks pretty cool. If it works as advertised, this is will fantastic. In combination with the plugin that lets you save multiple variations, it will be very useful. Thank you. I’ll turn it into a plugin when I get to using it in my plugin, unless you’ve already done that… have you?
November 26th, 2008 at 2:55 am
No, I’ve not made it a plugin. Please drop a link to yours though if you make it publicly available. Glad this helped!
February 19th, 2009 at 4:10 pm
Good idea. It might be better to append hash of the file name + current date time than to rely on randomly generated file names. Plus, on my system, a unified diff requires two hyphens
diff –unified instead of diff -unified
February 20th, 2009 at 12:52 am
@tkramar, thanks. It is two hyphens … that’s just WordPress thinking for me and deciding that when I type two hyphens that I really want a really long hyphen. I corrected it.
And yeah, the file name could be anything. I’m just a random sort of guy. And I generally avoid time-stamping files that don’t need to be moved around, archived, etc., as the OS does that for me, i.e., it’s redundant info.
March 16th, 2009 at 8:09 am
Hi Mark,
Just a message to say thanks! Works like a charm
Fabrice.
May 31st, 2009 at 3:22 pm
can it be modified to show visual diff like this image
http://tiny.cc/fnbiz
May 31st, 2009 at 8:58 pm
hi, how do this differ from the differ gem on github?
http://github.com/pvande/differ
June 1st, 2009 at 4:08 am
Simpler than i[1..i.length-1] is i[1..-1]
Also: it is safer to use the block form of File so that in the case of exceptions, the file is closed. Even better would be to use Tempfile, and/or to put File.delete in an ‘ensure’ block. Tempfile would also reduce the possibility of the same name being used twice concurrently.