fancier align assignments command for textmate

0-tm-align-assignments.rb

#!/usr/bin/env ruby1.9
# encoding: UTF-8
# Copyright © 2009 Caio Chassot
# Licensed under the WTFPL

def alternation(*s); s.map(&Regexp.method(:escape)).join("|") end

# All input is read here.
LINES                = $stdin.readlines
IS_SELECTION         = ENV.key?("TM_SELECTED_TEXT")
CURRENT_LINE         = ENV["TM_LINE_NUMBER"].to_i.pred

# Magic happens here.
# ASSIGNMENT_OPERATORS: There's love for C, ruby, perl, python, javascript, java, even pascal. Hash/Dict assignment included too.
# TROUBLEMAKERS: non-assignment operators that could match as assigment.
ASSIGNMENT_OPERATORS = %w( = -= += /= //= %= *= **= ^= |= &= ||= &&= <<= >>= >>>= .= x= := ::= => )
TROUBLEMAKERS        = %w( <= >= <=> == === != =~ )
TROUBLEMAKERS_BEFORE = TROUBLEMAKERS.map { |s| s[/^(.+)=/, 1] }.compact.uniq
TROUBLEMAKERS_AFTER  = TROUBLEMAKERS.map { |s| s[/=(.+)$/, 1] }.compact.uniq
RX_ASSIGNMENT_LINE   = %r[
  (^.*?)                                          # capture 1 — everything before assignment
  ( \s*                                           # capture 2 — assignment operator with surrounding spaces
    (?<!  #{alternation(*TROUBLEMAKERS_BEFORE)} )
    ( (?: #{alternation(*ASSIGNMENT_OPERATORS)} ) # capture 3 — assignment operator
    | :(?!\w|:)                                 ) #   special handling for the ':' assignment op (yaml, javascript, etc)
    (?!   #{alternation(*TROUBLEMAKERS_AFTER )} )
    \s* )
]x

l0, lf = [0, LINES.length.pred]
l0, lf = begin                                                                                                          # try to look up assignment block since we're not in a selection
  ix0  = ixf = %w[ to_i succ pred ].map { |m| CURRENT_LINE.send(m) }.find { |ix| LINES[ix] =~ RX_ASSIGNMENT_LINE }.to_i # assignment must exist in either this line, next, or previous
  ix0  = ix0.pred while ix0 > l0 && LINES[ix0.pred] =~ RX_ASSIGNMENT_LINE                                               # extend scope to assignments immediately above
  ixf  = ixf.succ while ix0 < lf && LINES[ixf.succ] =~ RX_ASSIGNMENT_LINE                                               # extend scope to assignments immediately below
  [ix0, ixf]
end unless IS_SELECTION

match_lines  = LINES[l0..lf].map { |s| [s.match(RX_ASSIGNMENT_LINE), s] }.select { |m, s| m }
len_leftside = match_lines.map   { |m, s| m.begin(2)  }.max
len_operator = match_lines.map   { |m, s| m[3].length }.max

match_lines.each { |m, s| s.replace [m[1].ljust(len_leftside), m[3].rjust(len_operator), m.post_match].join(" ") }
puts LINES

1-tm-align-assignments-original.rb

#!/usr/bin/env ruby
#
# Assignment block tidier, version 0.1.
#
# Copyright Chris Poirier 2006.
# Licensed under the Academic Free License version 3.0.
#
# This script can be used as a command for TextMate to align all
# of the equal signs within a block of text.  When using it with
# TextMate, set the command input to "Selected Text" or "Document",
# and the output to "Replace Selected Text".  Map it to a key
# equivalent, and any time you want to tidy up a block, either
# select it, or put your cursor somewhere within it; then hit the
# key equivalent.  Voila.
#
# Note that this is the first version of the script, and it hasn't
# been heavily tested.  You might encounter a bug or two.
#
# Per the license, use of this script is ENTIRELY at your own risk.
# See the license for full details (they override anything I've
# said here).

lines = STDIN.readlines()
selected_text = ENV.member?("TM_SELECTED_TEXT")

relevant_line_pattern = /^[^=]+[^-+<>=!%\/|&*^]=(?!=|~)/
column_search_pattern = /[\t ]*=/


#
# If called on a selection, every assignment statement
# is in the block.  If called on the document, we start on the
# current line and look up and down for the start and end of the
# block.

if selected_text then
   block_top    = 1
   block_bottom = lines.length
else

   #
   # We start looking on the current line.  However, if the
   # current line doesn't match the pattern, we may be just
   # after or just before a block, and we should check.  If
   # neither, we are done.

   start_on      = ENV["TM_LINE_NUMBER"].to_i
   block_top     = lines.length + 1
   block_bottom  = 0
   search_top    = 1
   search_bottom = lines.length
   search_failed = false

   if lines[start_on - 1] !~ relevant_line_pattern then
      if lines[start_on - 2] =~ relevant_line_pattern then
         search_bottom = start_on = start_on - 1
      elsif lines[start_on] =~ relevant_line_pattern then
         search_top = start_on = start_on
      else
         search_failed = true
      end
   end

   #
   # Now with the search boundaries set, start looking for
   # the block top and bottom.

   unless search_failed
      start_on.downto(search_top) do |number|
         if lines[number-1] =~ relevant_line_pattern then
            block_top = number
         else
            break
         end
      end

      start_on.upto(search_bottom) do |number|
         if lines[number-1] =~ relevant_line_pattern then
            block_bottom = number
         else
            break
         end
      end
   end
end


#
# Now, iterate over the block and find the best column number
# for the = sign.  The pattern will tell us the position of the
# first bit of whitespace before the equal sign.  We put the
# equals sign to the right of the furthest-right one.  Note that
# we cannot assume every line in the block is relevant.

best_column = 0
block_top.upto(block_bottom) do |number|
   line = lines[number - 1]
   if line =~ relevant_line_pattern then
      m = column_search_pattern.match(line)
      best_column = m.begin(0) if m.begin(0) > best_column
   end
end


#
# Reformat the block.  Again, we cannot assume all lines in the
# block are relevant.

block_top.upto(block_bottom) do |number|
   if lines[number-1] =~ relevant_line_pattern then
      before, after = lines[number-1].split(/[\t ]*=[\t ]*/, 2)
      lines[number-1] = [before.ljust(best_column), after].join(after[0,1] == '>' ? " =" : " = ")
   end
end


#
# Output the replacement text

lines.each do |line|
   puts line
end