using System; using System.IO; /// <summary> /// Represents a filter which removes comments and unnecessary whitespace from JavaScript files. /// </summary> public class JSMin { const int EOF = -1; StringReader sr; StringWriter sw; int theA; int theB; int theLookahead = EOF; /// <summary> /// Compresses the specified script source. /// </summary> /// <param name="source">The script source.</param> /// <returns>The compressed script content.</returns> public string Minify(string source) { using (sr = new StringReader(source)) { using (sw = new StringWriter()) { jsmin(); return sw.ToString(); } } } /* jsmin -- Copy the input to the output, deleting the characters which are insignificant to JavaScript. Comments will be removed. Tabs will be replaced with spaces. Carriage returns will be replaced with linefeeds. Most spaces and linefeeds will be removed. */ void jsmin() { theA = '\n'; action( 3 ); while( theA != EOF ) { switch( theA ) { case ' ': { if( isAlphanum( theB ) ) { action( 1 ); } else { action( 2 ); } break; } case '\n': { switch( theB ) { case '{': case '[': case '(': case '+': case '-': { action( 1 ); break; } case ' ': { action( 3 ); break; } default: { if( isAlphanum( theB ) ) { action( 1 ); } else { action( 2 ); } break; } } break; } default: { switch( theB ) { case ' ': { if( isAlphanum( theA ) ) { action( 1 ); break; } action( 3 ); break; } case '\n': { switch( theA ) { case '}': case ']': case ')': case '+': case '-': case '"': case '\'': { action( 1 ); break; } default: { if( isAlphanum( theA ) ) { action( 1 ); } else { action( 3 ); } break; } } break; } default: { action( 1 ); break; } } break; } } } } /* action -- do something! What you do is determined by the argument: 1 Output A. Copy B to A. Get the next B. 2 Copy B to A. Get the next B. (Delete A). 3 Get the next B. (Delete B). action treats a string as a single character. Wow! action recognizes a regular expression if it is preceded by ( or , or =. */ void action( int d ) { if( d <= 1 ) { put( theA ); } if( d <= 2 ) { theA = theB; if( theA == '\'' || theA == '"' ) { for( ; ; ) { put( theA ); theA = get(); if( theA == theB ) { break; } if( theA <= '\n' ) { throw new Exception( string.Format( "Error: JSMIN unterminated string literal: {0}\n", theA ) ); } if( theA == '\\' ) { put( theA ); theA = get(); } } } } if( d <= 3 ) { theB = next(); if( theB == '/' && (theA == '(' || theA == ',' || theA == '=' || theA == '[' || theA == '!' || theA == ':' || theA == '&' || theA == '|' || theA == '?' || theA == '{' || theA == '}' || theA == ';' || theA == '\n') ) { put( theA ); put( theB ); for( ; ; ) { theA = get(); if( theA == '/' ) { break; } else if( theA == '\\' ) { put( theA ); theA = get(); } else if( theA <= '\n' ) { throw new Exception( string.Format( "Error: JSMIN unterminated Regular Expression literal : {0}.\n", theA ) ); } put( theA ); } theB = next(); } } } /* next -- get the next character, excluding comments. peek() is used to see if a '/' is followed by a '/' or '*'. */ int next() { int c = get(); if( c == '/' ) { switch( peek() ) { case '/': { for( ; ; ) { c = get(); if( c <= '\n' ) { return c; } } } case '*': { get(); for( ; ; ) { switch( get() ) { case '*': { if( peek() == '/' ) { get(); return ' '; } break; } case EOF: { throw new Exception( "Error: JSMIN Unterminated comment.\n" ); } } } } default: { return c; } } } return c; } /* peek -- get the next character without getting it. */ int peek() { theLookahead = get(); return theLookahead; } /* get -- return the next character from stdin. Watch out for lookahead. If the character is a control character, translate it to a space or linefeed. */ int get() { int c = theLookahead; theLookahead = EOF; if( c == EOF ) { c = sr.Read(); } if( c >= ' ' || c == '\n' || c == EOF ) { return c; } if( c == '\r' ) { return '\n'; } return ' '; } void put( int c ) { sw.Write( (char)c ); } /* isAlphanum -- return true if the character is a letter, digit, underscore, dollar sign, or non-ASCII character. */ bool isAlphanum( int c ) { return ((c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || (c >= 'A' && c <= 'Z') || c == '_' || c == '$' || c == '\\' || c > 126); } }