#!/bin/bash
## jsminify -- remove comments and some other whitespace from a javascript file -- (c) Eugene Reimer 2007-August.
##
## I was hoping to use Douglas Crockford's jsmin.  However it only promises to work for files that pass his jslint.
## And since his jslint outlaws several programming practices I'm unwilling to give up (for example, writing a 
## trivial for-loop on one line without braces), I wrote this bash script as a simpler alternative to his jsmin.
## Mine can be safely used on any javascript.  No jslint needed!!
## The space-saving is much the same as with his:  my 110KB testcase shrinks to 36.6KB with his, or with mine.
## 
## USAGE:  jsminify  [string]...   --reads from stdin, writes minified javascript to stdout with the cmdline args as the only comments.
##
## LIMITATIONS:
## Only handles  //...  comments,  not the  /*...*/  kind.
## Requires that the input be in ASCII or UTF-8 or ISO-8859-1 or any charset that has ASCII as a subset.
## Uses Unix sed, and needs a bit of trickery to handle things not easily parsed with regular-expressions;  this will 
##	be safe provided your javascript is free from hex-01, hex-02...hex-05 characters  (not to worry, it will be).
##
## My first attempt was foiled by sed's behaviour on a command such as:  s|\(...\)*...|\1|
## when that didn't work, the sed solution got so ugly that any reasonable person would've switched to something else.
##
## Approach WRT string-literals:
## break line at first quote|stroph, and break again at next occurrence of what it starts with (quote|stroph);
## repeat the preceding N times to handle up to N string-literals per line;
##    (do comment-removal at the same time:  at the first quote|stroph|SLASHSLASH...)
## can now do space-removal, leaving string-literals unchanged;
## then remove the introduced linebreaks  (possible due to having added something to mark those introduced NLs).
##
## Please send bug-reports, comments, or suggestions to ereimer@shaw.ca.

while [ $# -ne 0 ]; do echo "//$1"; shift; done		##emit leading comments for the cmdline-args
sed $'
s|\x0d$||;						##remove trailing CR, in case input is MS-DOS style
s|^[ \t][ \t]*||;					##remove leading whitespace
s|[ \t][ \t]*$||;					##remove trailing whitespace
/^\\/\\//d;						##remove line containing nothing but comment
/^$/d;							##remove empty line
s|\\\\"|\x02|g;						##replace backslash+quote  with hex02  (the bash dollar-stroph undoubles backslashes, sed undoubles again)
s|\\\\\x27|\x03|g;					##replace backslash+stroph with hex03
s|//|\x04|g;						##replace slash+slash      with hex04

/[^"\x27\x04]*\x04/{					##on a line where the first quote|stroph|SLASHSLASH is a SLASHSLASH:
  s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|; s|[ \t][ \t]*$||};##  remove comment, everything after that SLASHSLASH, and redo trailing-whitespace removal
s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|;	##put hex05+hex01 before  opening-quote thats nonfirst char;  hex05s needed to undo...
s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|;##put hex05+hex01 before opening-stroph thats nonfirst
' |tr $'\x01' '\n' |sed $'				##hex01->NL, introducing linebreaks...
s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|;		##put hex05+hex01 after  closing-quote thats nonlast
s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g;##put hex05+hex01 after closing-stroph thats nonlast
' |tr $'\x01' '\n' |sed $'				##hex01->NL, introducing linebreaks...
##do those same steps about another 10 or so times:
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'
/[^"\x27\x04]*\x04/{s|^\\([^"\x27\x04]*\\)\x04.*$|\\1|;s|[ \t][ \t]*$||}; s|^\\([^"\x27\x04][^"\x27\x04]*\\)"|\\1\x05\x01"|; s|^\\([^"\x27\x04][^"\x27\x04]*\\)\x27|\\1\x05\x01\x27|'|tr $'\x01' '\n'|sed $' s|^\\("[^"]*"\\)\\([^\x05]\\)|\\1\x05\x01\\2|; s|^\\(\x27[^\x27]*\x27\\)\\([^\x05]\\)|\\1\x05\x01\\2|g'|tr $'\x01' '\n'|sed $'

/["\x27]/!{						##on lines other than string-literals do:  (each string-literal is now on a line by itself)
  s|\t| |g;						##TAB->SPACE
  s|  *| |g;						##destutter spaces
  s|\\([][(){},;<=>&|!*/%?:]\\) |\\1|g;			##remove space after  bracket|paren|brace|comma|semicolon|operator(most)
  s| \\([][(){},;<=>&|!*/%?:]\\)|\\1|g;			##remove space before bracket|paren|brace|comma|semicolon|operator(most)
  s|\\([+-]\\) \\([^+-]\\)|\\1\\2|g;			##remove space after  plus|minus only if not followed by plus|minus
  s|\\([^+-]\\) \\([+-]\\)|\\1\\2|g;			##remove space before plus|minus only if not preceded by plus|minus
  s|^ +|+|;						##remove leading space (can only happen after a string-literal)
  s|+ $|+|;						##remove ending space (can only happen before a string-literal)
}
' |tr '\n' $'\x01' |sed $'				##NL->hex01  (so can remove selected NLs, and undo the added linebreaks)
s|\x05\x01||g;						##remove the hex05+hex01 that were added earlier, before and after string-literals
s|{\x01|{|g;						##remove NL after opening-brace  (can yank this to avoid long lines)
s|;\x01|;|g;						##remove NL after semicolon      (can yank this to avoid long lines)
s|\\([[(,<=>&|!*/%?:]\\)\x01|\\1|g;			##remove NL after  opening-bracket|opening-paren|comma|operator(most)
s|\x01\\([]),<=>&|!*/%?:]\\)|\\1|g;			##remove NL before closing-bracket|closing-paren|comma|operator(most)
s|\\([^+]+\\)\x01\\([^+]\\)|\\1\\2|g;			##remove NL after plus, but not after plusplus, nor before plus
s|\\([^-]-\\)\x01\\([^-]\\)|\\1\\2|g;			##remove NL after minus, but not after minusminus, nor before minus
' |tr $'\x01' '\n' |sed $'				##restore remaining hex01->NL

s|\x02|\\\\"|g;						##restore hex02->backslash+quote
s|\x03|\\\\\x27|g;					##restore hex03->backslash+stroph
s|\x04|//|g;						##restore hex04->slash+slash
'