Macworld/IDG Markdown to HTML AppleScript

gistfile1.applescript

--Macworld Markdown to HTML script
--by Jason Snell <jsnell@macworld.com>
--Markdown and SmartyPants by John Gruber <http://daringfireball.net/projects>

--Version 10: Added Euro support and forced Unicode for output file
--Version 11: Output now opens as new document in source window.
--Version 12: Makes sure the output document is set to source language of HTML
--Version 13: Support for [[jump]] tag
--Version 14: Rewrite iLife ‘09 style tags to proper ’09 style.
--Version 15: Support pound symbol
--Version 16: Embedded SmartyPants and Markdown inside the app bundle. 
--Verison 17: Removed Convert to ASCII step to allow accented characters
--Version 18: Fixes for BBEdit 9.5
--Version 19: Closes up emdashes
--Version 20: Fixes for BBEdit 10
--Version 21: ebook and email
--Version 22: edit-staging removed
--Verison 23: Moved to BBEdit package. SmartyPants and Markdown moved next door to Text Filters.
--Version 25: New CMS
--Version 26: More changes for new CMS, multisite

--The script needs to know where it is on your system
--This block figures out its path
set thePath to (path to me as text)
set AppleScript's text item delimiters to ":"
set thePathItems to text items of thePath
set AppleScript's text item delimiters to {""}
set parsedPath to items 1 thru ((count of items of thePathItems) - 2) of thePathItems
set AppleScript's text item delimiters to ":"
set parentPath to parsedPath as string
set AppleScript's text item delimiters to {""}

--Remember where SmartyPants and Markdown are, next door in Text Filters
set SmartyPants to parentPath & ":Text Filters:SmartyPants.pl"
set Markdown to parentPath & ":Text Filters:Markdown.pl"

--Copy existing document to new window for conversion
tell application "BBEdit"
  set theItem to (contents of window 1 as text)
	
	set theDocument to (make new text document of window 1)
	set theVersion to version
	set encoding of theDocument to "Unicode (UTF-8)"
	set text of theDocument to theItem
	set name of theDocument to "HTML output"
	set source language of theDocument to "HTML"
end tell

--style check!
replacemeliteral("web site", "website")
replacemeliteral("e-mail", "email")
replacemeliteral("e-book", "ebook")
replacemeliteral("Ethernet", "ethernet")
replacemeliteral(" target=\"_blank\"", "")
replacemeliteral("web page", "webpage")
replacemeliteral("home page", "homepage")
replacemeliteral("NVIDIA", "Nvidia")
replaceme("(Command|Control|Option|Shift)\\+(Command|Control|Option|Shift)\\+(.)", "\\1-\\2-\\3")
replaceme("(Command|Control|Option|Shift)\\+(.)", "\\1-\\2")

--hide the pound sign
replacemeliteral("£", "[poundsymbol]")

--edit staging check
replacemeliteral("preview.www.", "www.")

-- Convert Markdown to HTML
tell application "BBEdit"
	run unix filter Markdown with replacing selection
end tell


-- Use SmartyPants to convert quotes, etc. into entities
tell application "BBEdit"
	run unix filter SmartyPants with replacing selection
end tell


--The great emdasherator - All hail Dan Frakes!
replaceme(" *-- *", "—")

--hides greater-thans and less-thans
replacemeliteral("&gt;", "[greaterthansymbol]")
replacemeliteral("&lt;", "[lessthansymbol]")


-- Use SmartyPants to convert quotes, etc. into entities
tell application "BBEdit"
	run unix filter SmartyPants with replacing selection
end tell

tell application "BBEdit"
	-- converts HTML entities back into 8-bit characters
	translate html to text entity conversion true ¬
		create new document false ¬
		without selection only
end tell

--returns greater-thans and less-thans
replacemeliteral("[greaterthansymbol]", "&gt;")
replacemeliteral("[lessthansymbol]", "&lt;")

--Kicking it Eurostyle, like MC Dan Frakes tells us to
replacemeliteral("[euro]", "€")
replacemeliteral("[poundsymbol]", "£")

-- replaces stray ampersands with the ampersand entity to satisfy TULIPS and its cruel masters
-- but doesn't swallow ampersands that are part of the < > entities
-- replaceme("&(?!(lt|gt);)", "&amp;")

--Expand [[4.0 mice]] style into proper IMG tag
replaceme("\\[\\[([0-5])\\.([05]) (mice)\\]\\]", "<img src=\"http://images.macworld.com/images/layout/bluemouse\\1\\2.gif\" border=\"0\" alt=\"2.5-mouse rating\"/>")

--TULIPS-ify images. default is medium, align right.
--replaceme("<img src=\"([^\"]+)\" alt=\"([^\"]+)\" title=\"([^\"]+)\" />", "<img src=\"http://images.macworld.com/images/reviews/graphics/AID-\\1\" align=\"right\" alt=\"\\2\" title=\"\\3\" class=\"medium\" />")


--replace jump tag with actual jump tag
replaceme("<p>\\[\\[jump\\]\\]</p>", "<a href=\"/jump\">Read more...</a>")

--close up emdashes
replacemeliteral(" — ", "—")

--Properly orient curly tags for iLife ’09 and its ilk
replaceme("‘(?=[0-9]{2}[^0-9])", "’")

--and we're done.

(*

Below here are the subroutines that do the search-and-replace action.

*)


--replace a string in BBEdit using grep
on replaceme(theFind, theReplace)
	tell application "BBEdit"
		replace theFind using theReplace searching in text of text window 1 options {search mode:grep, starting at top:true, wrap around:true, backwards:false, case sensitive:false, match words:false, extend selection:false}
	end tell
end replaceme

--replace a string in BBEdit without using grep
on replacemeliteral(theFind, theReplace)
	tell application "BBEdit"
		replace theFind using theReplace searching in text of text window 1 options {search mode:literal, starting at top:true, wrap around:true, backwards:false, case sensitive:false, match words:false, extend selection:false}
	end tell
end replacemeliteral

Apologies to real programmers. This is ugly but it's helpful.