BBEdit & TextWrangler text clean-up script for the Morning Star newspaper

CopyCleaner.applescript

-- By Rob Wells for the Morning Star

on open theStories
    repeat with aStory in theStories
        tell application "TextWrangler"
            open aStory
            tell the front text document
                set encoding to "Unicode (UTF-8)"
                
                educate quotes with replacing target
                
                -- Remove hard-wraps (not perfect)
                my grepRep("([-,—:;[:alnum:]]) *\\r *([-,—:;[:alnum:]])", "\\1 \\2")
                -- Important thing to note is that it only finds non-terminal punctuation.
                -- It hits properly formatted bylines, but the byline replace fixes it later.
                
                my grepRep(" {2,}", " ") -- Multiple spaces to single space
                my grepRep("^ ", "") -- Remove spaces at the start of lines
                my grepRep("\\t+", "") -- Remove tabs
                
                -- Superscript numbers to quote marks
                my litRep("¹", "’") -- Apostrophe
                my litRep("²", "”") -- Right double
                my litRep("³", "“") -- Left double
                
                my grepRep(" [–-] ", " — ") -- En-dashes & hyphens to em-dashes
                my grepRep("^• *|^n ", "n") -- Blob-pars (nHeady heady)
                my grepRep("([.0-9]*) *(%|percent)(?!age)", "\\1 per cent") -- "per cent"
                my litRep("...", "…") -- Ellipses
                
                -- Break byline after name and ensure lower-case 'b'
                -- Check for (rare) 3-word bylines, else handle 2-word bylines
                my grepRep("^(by Our (?:Foreign|News|Sports) Desk|by Morning Star Reporter)(?:[ ,]*)(.*)\\r+", "\\l\\1\\r\\2\\r")
                if the result is 0 then -- 3-word byline not found
                    -- Break two-word byline
                    my grepRep("^(by [-[:alpha:]]+ [-[:alpha:]]+)(?:[ ,]*)(.*)\\r+", "\\l\\1\\r\\2\\r")
                end if
                
                my grepRep("\\r+\\z", "") -- Delete empty lines at end
                my grepRep("”([[:punct:]])", "\\1”") -- Transpose rdquo with punctuation.
                
            end tell
        end tell
    end repeat
end open


-- Convenience wrapper functions

on litRep(searchString, replaceString) -- Literal search and replace
    tell application "TextWrangler"
        tell text 1 of text window 1
            replace searchString using replaceString options {search mode:literal, starting at top:true}
        end tell
    end tell
end litRep

on grepRep(searchString, replaceString) -- Grep search and replace
    tell application "TextWrangler"
        tell text 1 of text window 1
            replace searchString using replaceString options {search mode:grep, starting at top:true}
        end tell
    end tell
end grepRep