cfold.c

«
 * C / C++ to HTML converter
 *
 * Converts C and C++ code to HTML with syntax hilighting
 * and folding by use of embedded comments.
 *
 * If you are viewing this file on a web page, it has already been
 * processed using itself, so you can click on the darker colored comments
 * to unfold the source code behind them.  Click on the comment again
 * (or the following source code) to re-fold it to the summary text.
 *
 * This comment is a fold, and clicking here will display an extended
 * comment section.
 »/* This display style was inspired by a combination of Inmos's folding
 * editor for OCCAM, plus a web page from Tim Wakeham which did something
 * similar but at a file rather than block level.  Tim tells me some current
 * IDEs such as SCITE have support for folded editing.  There are also
 * editors available such as 'fe'. ( http://www.moria.de/~michael/fe/ )
 *
 * The underlying HTML formatter is derived from an original by Brad Broerman.
 * (at http://www.bbroerman.net/ctohtml.html )
 *
 * The folding convention is that "/" "*" "<" starts a fold, and the contents
 * of the comment are displayed as the fold's link to be clicked on.
 * A fold is terminated by a comment starting "/" "*" ">" and the content
 * of that comment is only displayed when the fold is expanded.  If that
 * comment is completely empty (no whitespace) then it is not displayed
 * at all.
 *
 * Fold-comments are *not* escaped for HTML like other comments and can
 * therefore include HTML which will be displayed as formatted text/
 *
 * I plan an extension so that comments between "/" "*" "<" and ">" "*" "/"
 * will not generate a fold but will allow HTML, simply as a way to embed
 * formatted text and diagrams in a source file.
 *
 * I am also considering adding a "+" (eg "/" "*" "<" "+") to a fold start
 * to signify that the default state for that fold should be 'unfolded',
 * thus allowing us to post files on the web which already show the
 * critical code areas ready for viewing.
 *
 * I may change the syntax of this system later to match the more common
 * convention of "{{{" and "}}}".
 */

« Declaration section »
« Include files »#include <stdio.h>
#include <stdlib.h>
#include <ctype.h>
#include <string.h>

« Symbolic constants for colours »#define QUOTE_COLOR "#ff0000"
#define COMMENT_COLOR "#408080"
#define FOLD_COLOR "#205050"
#define PREPROC_COLOR "#008000"

« Support stuff for folding »static int next_fold = 0;
int stack[12];
int top = -1;
#ifndef FALSE
#define TRUE (0==0)
#define FALSE (0!=0)
#endif

« Support procedure section »
« Symbol table management »
int IsReservedWord (char *word)
« Is a token an identifier or a reserved keyword?»{
   static char *Reserved_Words[] = {
      « List of C and C++ keywords »#ifdef CPLUSPLUS
      "asm", "catch", "double", "cgoto", "not", "return", "throw",
      "volatile", "and", "char", "else", "if", "not_eq", "short",
      "true", "while", "and_eq", "class", "enum", "inline", "operator",
      "signed", "try", "xor", "auto", "compl", "explicit", "int", "or",
      "sizeof", "typedef", "xor_eq", "bitand", "const", "extern",
      "long", "or_eq", "static", "typename", "bitor", "continue",
      "false", "mutable", "private", "struct", "union", "bool", "default",
      "float", "namespace", "protected", "switch", "unsigned", "break",
      "delete", "for", "new", "public", "template", "virtual", "case",
      "do", "friend", "not", "register", "this", "void",
#else
      "double", "goto", "return",
      "volatile", "char", "else", "if", "short",
      "true", "while", "enum", "inline",
      "signed", "auto", "int",
      "sizeof", "typedef", "const", "extern",
      "long", "static", "continue",
      "false", "struct", "union", "bool", "default",
      "float", "switch", "unsigned", "break",
      "for", "case",
      "do", "register", "void",
#endif
      NULL 
   };

   int i;

   for (i = 0;; ++i) {
      if (Reserved_Words[i] == NULL)
	 return (0);
      if (!strcmp (word, Reserved_Words[i])) {
	 return 1;
      }
   }
}

« I/O support »static int col = 0;

int freadchar (FILE * f)
« fread, plus makes a note of the input column in a global »{
int ch;

   ch = fgetc (f);
   if (ch == EOF)
      return (EOF);
   if (ch == '\n') {
      col = 0;
      return ('\n');
   }
   if (ch == '\t') {
      col = (col + 8) & (~7);
      return ('\t');
   }
   col = col + 1;
   return ch;
}				

int fpeek (FILE * f)
« Pascal-style lookahead - leaves next character still to be read. »{
int ch;

   ch = fgetc (f);		/* don't increment col */
   ungetc (ch, f);
   return ch;
}				

static int suppress_newline_hack = FALSE; // An unfortunate hack due to the way <div> inserts an extra blank line.

void printchar (char inchar, FILE * fstream)
« quote '<' and '&' with entity references on output.  Note that fold
   descriptors like this comment do not go through this procedure, therefore
   embedded HTML is allowed!  (Useful for embedded diagrams) »{
   if (inchar == '&')
      fprintf (fstream, "&amp;");
   else if (inchar == '>')
      fprintf (fstream, "&gt;");
   else if (inchar == '<')
      fprintf (fstream, "&lt;");
   else if (inchar == '\n') {
       if (!suppress_newline_hack) fputc ('\n', fstream);
       suppress_newline_hack = FALSE;
   } else {
      fputc (inchar, fstream);
   }
}

void PrintHeader (void)
« Output Javascript and CSS to the header section »{
   « First the Javascript - comes from http://www.netlobo.com/div_hiding.html »   fprintf (stdout, "<script type=\"text/javascript\">\n");
   fprintf (stdout, "<!--\n");
   fprintf (stdout, "function toggleLayer( whichLayer, summary )\n");
   fprintf (stdout, "{\n");
   fprintf (stdout, "  var elem, vis;\n");
   fprintf (stdout, "  if( document.getElementById ) // this is the way the standards work\n");
   fprintf (stdout, "    elem = document.getElementById( whichLayer );\n");
   fprintf (stdout, "  else if( document.all ) // this is the way old msie versions work\n");
   fprintf (stdout, "      elem = document.all[whichLayer];\n");
   fprintf (stdout, "  else if( document.layers ) // this is the way nn4 works\n");
   fprintf (stdout, "    elem = document.layers[whichLayer];\n");
   fprintf (stdout, "  vis = elem.style;\n");
   fprintf (stdout, "  // if the style.display value is blank we try to figure it out here\n");
   fprintf (stdout, "  if(vis.display==''&&elem.offsetWidth!=undefined&&elem.offsetHeight!=undefined)\n");
   fprintf (stdout, "    vis.display = (elem.offsetWidth!=0&&elem.offsetHeight!=0)?'block':'none';\n");
   fprintf (stdout, "  vis.display = (vis.display==''||vis.display=='block')?'none':'block';\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "  if( document.getElementById ) // this is the way the standards work\n");
   fprintf (stdout, "    elem = document.getElementById( summary );\n");
   fprintf (stdout, "  else if( document.all ) // this is the way old msie versions work\n");
   fprintf (stdout, "      elem = document.all[summary];\n");
   fprintf (stdout, "  else if( document.layers ) // this is the way nn4 works\n");
   fprintf (stdout, "    elem = document.layers[summary];\n");
   fprintf (stdout, "  vis = elem.style;\n");
   fprintf (stdout, "  // if the style.display value is blank we try to figure it out here\n");
   fprintf (stdout, "  if(vis.display==''&&elem.offsetWidth!=undefined&&elem.offsetHeight!=undefined)\n");
   fprintf (stdout, "    vis.display = (elem.offsetWidth!=0&&elem.offsetHeight!=0)?'block':'none';\n");
   fprintf (stdout, "  vis.display = (vis.display==''||vis.display=='block')?'none':'block';\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "// -->\n");
   fprintf (stdout, "</script>\n");
   fprintf (stdout, "\n");
   
   « And then the CSS - derived from Tim Wakeham's page »   fprintf (stdout, "<style>\n");
   fprintf (stdout, ".hidecode {\n");
   fprintf (stdout, "    z-index:99;\n");
   fprintf (stdout, "    visibility:hidden;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".file {\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".file div {\n");
   fprintf (stdout, "    margin: auto;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "div p {\n");
   fprintf (stdout, "    margin-left: 15px;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "body {\n");
   fprintf (stdout, "    font-family: Calibri, Arial, Georgia, \"Times New Roman\", Times, serif;\n");
   fprintf (stdout, "    color: #000000;\n");
   fprintf (stdout, "}\n\n");
   fprintf (stdout, "h2, h3 {\n");
   fprintf (stdout, "    margin-bottom: 0px;\n   color: #000000;\n");
   fprintf (stdout, "    }\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "p {\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "h2 {\n   color: #000000;\n");
   fprintf (stdout, "    }\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "h2 .link {\n");
   fprintf (stdout, "    font-size: small;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".link a {\n");
   fprintf (stdout, "    text-decoration: none;\n    outline: none;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "h2 .link a:hover {\n");
   fprintf (stdout, "    text-decoration: none;\n");
   fprintf (stdout, "    }\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".data {\n");
   fprintf (stdout, "    display: none;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".data pre {\n");
   fprintf (stdout, "    overflow: auto;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".data p {\n");
   fprintf (stdout, "    margin-top: 0px;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".summary {\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".summary pre {\n");
   fprintf (stdout, "    overflow: auto;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".summary p {\n");
   fprintf (stdout, "    margin-top: 0px;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".summary a {\n"); /* not sure... */
   fprintf (stdout, "    text-decoration: none;\n    outline: none;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".show {\n");
   fprintf (stdout, "    text-decoration: none;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".show:hover {\n");
   fprintf (stdout, "    text-decoration: none;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".flinks {\n");
   fprintf (stdout, "    text-decoration: none;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, ".flinks:hover {\n");
   fprintf (stdout, "    text-decoration: none;\n   color: #000000;\n");
   fprintf (stdout, "}\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "\n");
   fprintf (stdout, "</style>\n");
   
}


« Four procedures to output DHTML <div>s to hide/display the nested layers »void UnexpandedStart (int seq)
{
   stack[++top] = seq;
   if (top > 0) fprintf (stdout, /* terminate link from previous block because nested links are bad! */
			 "</a>");

   fprintf (stdout,
	    "<a href=\"javascript:toggleLayer('expanded_%d','unexpanded_%d')\" class=\"show\">",
	    seq, seq);
}

void UnexpandedFinish (void)
{
   fprintf (stdout,
            "<div class=\"show\" id=\"unexpanded_%d\"></div>",
	    stack[top]);
   fprintf (stdout, "</a>");
}

void ExpandedStart (void)
{
   int seq = stack[top];

   if (top > 0) fprintf (stdout, /* terminate link from previous block because nested links are bad! */
			  "</a>"
                          "</font>");

   fprintf (stdout,
            "<div class=\"data\" id=\"expanded_%d\">"
            "<font color=\"#000000\">"
            "<a href=\"javascript:toggleLayer('expanded_%d','unexpanded_%d')\" class=\"show\">",
	    seq, seq, seq);
}

void ExpandedFinish (void)
{
   --top;
   fprintf (stdout,  "</a>"
                     "</font>"
 	             "</div>");
   if (top >= 0) {
      int seq = stack[top];
      fprintf (stdout, /* re-instate link to allow collapse of previous block */
                       "<font color=\"#000000\">"
	               "<a href=\"javascript:toggleLayer('expanded_%d','unexpanded_%d')\" class=\"show\">",
	               seq, seq);
   } else {
      fprintf (stdout, "<font color=\"#000000\">");
   }


}

int Print_HTML (FILE * fstream, char *Title)
« This is the main filter which loops over input C source and outputs HTML »{
   FILE *InFile;
   static int peekchar = -1;
   int inchar, i, newline = 1, WordIdx = 0, savecol;
   char word[280];

   word[0] = '\0';

   « Print out the HTML header. »   fprintf (fstream, "<HTML>\n<HEAD>\n<TITLE> %s </TITLE>\n", Title);
   PrintHeader ();
   fprintf (fstream, "</HEAD>\n");
   fprintf (fstream, "<BODY bgcolor=\"#ffffff\"><pre>\n");
   
   « Open the file for reading in text mode. »   InFile = fopen (Title, "r");
   if (InFile == NULL)
      return -1;
   
   /* Start reading it in, character by character. */
   inchar = freadchar (InFile);
   peekchar = fpeek (InFile);
   while (!feof (InFile)) {
      « Check for keywords (& print in bold), and simple tags (not bold) »      if (isalnum (inchar) || (inchar == '_')) { /* Is this a tag? */
	 word[WordIdx++] = inchar; word[WordIdx] = '\0';
         inchar = freadchar (InFile);
         peekchar = fpeek (InFile);
         continue;
      }
      if (WordIdx > 0) {
	  if (IsReservedWord (word)) { // flush any completed tags
	    fprintf (fstream, "<b>%s</b>", word);
	 } else {
	    fprintf (fstream, "%s", word);
	 }
	 WordIdx = 0;
      }

      if (inchar == '\n') {
	 newline = 1;
      }

      if (!isspace (inchar)) {
	 savecol = col;		/* note column position.  this is to
				   re-indent <div>s which seem to be forced
				   to start on the next line */
	 « Handle all '#' preprocessor directives. »	 if (inchar == '#') {
	    if (newline == 1) {
	       fprintf (fstream, "<font color=\"%s\">", PREPROC_COLOR);
	       for (;;) {
		  if ((inchar != '\\') && (peekchar == '\n'))
		     break;
		  printchar (inchar, fstream);
		  inchar = freadchar (InFile);
		  peekchar = fpeek (InFile);
	       }
	       printchar (inchar, fstream);
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	       fprintf (fstream, "</font>");
	       newline = 1;
	       continue;
	    }
	 }
	 
	 « Handle single line C++ style comments ('//'), and continue »	 else if ((inchar == '/') && (peekchar == '/')) {
	    fprintf (fstream, "<font color=\"%s\"><i>", COMMENT_COLOR);
	    while ((inchar != '\n') && (inchar > 0)) {
	       printchar (inchar, fstream);
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	    }
	    fprintf (fstream, "</i><!-- 1 --></font>");
	    newline = 1;
	    continue;
	 }
	 
	 « Handle traditional C-style comments and folding directives, and continue »	 else if ((inchar == '/') && (peekchar == '*')) {
	    inchar = freadchar (InFile);
	    peekchar = fpeek (InFile);
	    if (peekchar == '<') {              « start of a fold »	       next_fold += 1;
	       UnexpandedStart (next_fold);
	       inchar = freadchar (InFile);
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	       fprintf (fstream, "&laquo;<font face=\"Calibri, Arial, Georgia, \"Times New Roman\", Times, serif\" color=\"%s\" size=\"+1\">", FOLD_COLOR);
	       for (;;) {
		  fputc (inchar, fstream);
		  inchar = freadchar (InFile);
		  peekchar = fpeek (InFile);
		  if ((inchar == '*') && (peekchar == '/'))
		     break;
	       }
	       fprintf (fstream, "</font>&raquo;");
	       inchar = freadchar (InFile);
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	       UnexpandedFinish ();
	       ExpandedStart ();	
               suppress_newline_hack = TRUE;
	    } else if (peekchar == '>') {	« end of a fold »//	       fprintf (fstream, "</i></font>");
	       inchar = freadchar (InFile); // '>'
	       inchar = freadchar (InFile); // char after, may be ' ' or start of '*/'
	       peekchar = fpeek (InFile);

               if (inchar == '*' && peekchar == '/') {
		   // DO *NOT* OUTPUT COMMENT
                   inchar = freadchar (InFile); // '/'
                   inchar = freadchar (InFile); // next char to be handled
		   peekchar = fpeek (InFile);
	       } else {
	         fprintf (fstream, "<font color=\"%s\"><i>/*", FOLD_COLOR);
	         for (;;) {
		    printchar (inchar, fstream);
		    inchar = freadchar (InFile);
		    peekchar = fpeek (InFile);
		    if ((inchar == '*') && (peekchar == '/'))
		       break;
	         }
	         inchar = freadchar (InFile);
	         peekchar = fpeek (InFile);
	         inchar = freadchar (InFile);
	         peekchar = fpeek (InFile);
	         fprintf (fstream, "*/</i><!-- 3 --></font>");
 	       }
               suppress_newline_hack = TRUE;
	       ExpandedFinish ();	
	    } else {		                « regular comment »	       fprintf (fstream, "<font color=\"%s\"><i>/", COMMENT_COLOR);
	       for (;;) {
		  printchar (inchar, fstream);
		  inchar = freadchar (InFile);
		  peekchar = fpeek (InFile);
		  if ((inchar == '*') && (peekchar == '/'))
		     break;
	       }
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	       fprintf (fstream, "*/</i><!-- 4 --></font>");
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);	
	    }
	    continue;		/* Now process next char */
	 }
	 

	 « Handle double-quotes, and continue »#ifdef NEVER
	 else if (inchar == '"') {
	    fprintf (fstream, "<font color=\"%s\">\"", QUOTE_COLOR);
	    if (peekchar == '"') {
	       /* Null string bug fix */
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	       printchar (inchar, fstream);
	    } else {
	       inchar = freadchar (InFile);
	       if (inchar == EOF)
		  continue;
	       peekchar = fpeek (InFile);
	       for (;;) {
		  if ((inchar != '\\') && (peekchar == '"'))
		     break;
		  printchar (inchar, fstream);
		  inchar = freadchar (InFile);
		  if (inchar == EOF)
		     break;
		  peekchar = fpeek (InFile);
	       }
	       printchar (inchar, fstream);
	       inchar = freadchar (InFile);
	       peekchar = fpeek (InFile);
	       printchar (inchar, fstream);
	    }
	    inchar = freadchar (InFile);
	    peekchar = fpeek (InFile);
	    fprintf (fstream, "</font>");
	    newline = 0;
	    continue;
	 }

#endif
	 /* Handle Double Quotes, and continue (fixed!). */
	 else if (inchar == '"') {
	     fprintf(fstream, "<font color=\"%s\">\"", QUOTE_COLOR);
	     for (;;) { // fix by simplifying
		 inchar = fgetc(InFile);
		 if (inchar == EOF) break;
		 if (inchar == '"') break;
		 if (inchar == '\\') {
		     printchar(inchar, fstream);
		     inchar = fgetc(InFile);
		     if (inchar == EOF) break;
		 }
		 printchar(inchar, fstream);
	     }
	     inchar = fgetc(InFile);
	     peekchar = fpeek(InFile);
	     fprintf(fstream, "\"</font>");
	     newline = 0;
	     continue;
	 }

	 
         « Handle single-quotes, and continue »	 else if (inchar == '\'') {
	    fprintf (fstream, "<font color=\"%s\">'", QUOTE_COLOR);
	    inchar = freadchar (InFile);
	    if (inchar == EOF)
	       continue;
	    peekchar = fpeek (InFile);
	    for (;;) {
	       if ((inchar != '\\') && (peekchar == '\''))
		  break;
	       if ((inchar == '\\') && (peekchar == '\\')) {
		  printchar (inchar, fstream);
		  inchar = freadchar (InFile);
		  if (inchar == EOF)
		     break;
		  peekchar = fpeek (InFile);
		  if (peekchar == '\'')
		     break;
	       }
	       printchar (inchar, fstream);
	       inchar = freadchar (InFile);
	       if (inchar == EOF)
		  break;
	       peekchar = fpeek (InFile);
	    }
	    printchar (inchar, fstream);
	    inchar = freadchar (InFile);
	    peekchar = fpeek (InFile);
	    printchar (inchar, fstream);
	    inchar = freadchar (InFile);
	    peekchar = fpeek (InFile);
	    fprintf (fstream, "</font>");
	    newline = 0;
	    continue;
	 }
	 
         /* Any other characters drop through */
	 newline = 0;
      }
      /* Including white space.  I think I made an exception for it as a quick
         hack to remove initial spaces before a #define etc */

      printchar (inchar, fstream); // print normal text here.
      // move on to next character
      inchar = freadchar (InFile);
      peekchar = fpeek (InFile);
   }
   fprintf (fstream, "\n</pre></BODY></HTML>\n");
   fclose (InFile);
   return 1;
}				

int main (int argc, char *argv[])
{
   if (argc < 2) {
      printf ("Usage: %s [Filename]\n\n", argv[0]);
      return -1;
   }
   Print_HTML (stdout, argv[1]);
   return 0;
}