replace every occurrence of 'line 2' with line_2 with regex - ruby

Question

I'm parsing some text from an XML file which has sentences like
"Subtract line 4 from line 1.", "Enter the amount from line 5"
i want to replace all occurrences of line with line_
eg. Subtract line 4 from line 1 --> Subtract line_4 from line_1
Also, there are sentences like "Are the amounts on lines 4 and 8 the same?" and "Skip lines 9 through 12; go to line 13."
I want to process these sentences to become
"Are the amounts on line_4 and line_8 the same?"
and
"Skip line_9 through line_12; go to line_13."

Answer 1

Here's a working implementation with rspec test. You call it like this: output = LineIdentifier[input]. To test, spec file.rb after installing rspec gem.
require 'spec'
class LineIdentifier
def self.[](input)
output = input.gsub /line (\d+)/, 'line_\1'
output.gsub /lines (\d+) (and|from|through) (line )?(\d+)/, 'line_\1 \2 line_\4'
end
end
describe "LineIdentifier" do
it "should identify line mentions" do
examples = {
#Input Output
'Subtract line 4 from line 1.' => 'Subtract line_4 from line_1.',
'Enter the amount from line 5' => 'Enter the amount from line_5',
'Subtract line 4 from line 1' => 'Subtract line_4 from line_1',
}
examples.each do |input, output|
LineIdentifier[input].should == output
end
end
it "should identify line ranges" do
examples = {
#Input Output
'Are the amounts on lines 4 and 8 the same?' => 'Are the amounts on line_4 and line_8 the same?',
'Skip lines 9 through 12; go to line 13.' => 'Skip line_9 through line_12; go to line_13.',
}
examples.each do |input, output|
LineIdentifier[input].should == output
end
end
end

Answer 2

This works for the specific examples including the ones in the OP comments. As is often the case when using regex to do parsing, it becomes a hodge-podge of additional cases and tests to handle ever-increasing known inputs. This handles the lists of line numbers using a while loop with a non-greedy match. As written, it is simply processing an input line-by-line. To get series of line numbers across line boundaries, it would need to be changed to process it as one chunk with matching across lines.
open( ARGV[0], "r" ) do |file|
while ( line = file.gets )
# replace both "line ddd" and "lines ddd" with line_ddd
line.gsub!( /(lines?\s)(\d+)/, 'line_\2' )
# Now replace the known sequences with a non-greedy match
while line.gsub!( /(line_\d+[a-z]?,?)(\sand\s|\sthrough\s|,\s)(\d+)/, '\1\2line_\3' )
end
puts line
end
end
Sample Data: For this input:
Subtract line 4 from line 1.
Enter the amount from line 5
on lines 4 and 8 the same?
Skip lines 9 through 12; go to line 13.
... on line 10 Form 1040A, lines 7, 8a, 9a, 10, 11b, 12b, and 13
Add lines 2, 3, and 4
It produces this output:
Subtract line_4 from line_1.
Enter the amount from line_5
on line_4 and line_8 the same?
Skip line_9 through line_12; go to line_13.
... on line_10 Form 1040A, line_7, line_8a, line_9a, line_10, line_11b, line_12b, and line_13
Add line_2, line_3, and line_4

Answer 3

sed is your friend:
lines.sed:
#!/bin/sed -rf
s/lines? ([0-9]+)/line_\1/g
s/\b([0-9]+[a-z]?)\b/line_\1/g
lines.txt:
Subtract line 4 from line 1.
Enter the amount from line 5
Are the amounts on lines 4 and 8 the same?
Skip lines 9 through 12; go to line 13.
Enter the total of the amounts from Form 1040A, lines 7, 8a, 9a, 10, 11b, 12b, and 13
Add lines 2, 3, and 4
demo:
$ cat lines.txt | ./lines.sed
Subtract line_4 from line_1.
Enter the amount from line_5
Are the amounts on line_4 and line_8 the same?
Skip line_9 through line_12; go to line_13.
Enter the total of the amounts from Form 1040A, line_7, line_8a, line_9a, line_10, line_11b, line_12b, and line_13
Add line_2, line_3, and line_4
You can also make this into a sed one-liner if you prefer, although the file is more maintainable.