Re: Source code for Syntax Coloring... (Jacob Navia)
20 Dec 1996 16:34:33 -0500

          From comp.compilers

Related articles
Source code for Syntax Coloring... (micky) (1996-12-14)
Re: Source code for Syntax Coloring... (1996-12-20)
| List of all articles for this month |

From: (Jacob Navia)
Newsgroups: comp.compilers
Date: 20 Dec 1996 16:34:33 -0500
Organization: Compilers Central
Keywords: analysis
Reference: 96-12-090

> Is there any public domain source code available for syntax coloring...

I have built an integrated development environment for a version of
lcc under windows. My IDE has full syntax coloring in real time,
i.e. the colors change instantly when the keywords/comments of the
text change. The languages supported are C, Pascal and Assembly.

Algorithm syntax coloring:
1. Case keywords coloring.
      After you have eliminated strings and comments text, just scan each word
      to see if is a keyword. If it is, paint it as you like. This is very easy
      Be careful to treat the end of line as white space, and be sure.
2. Case comments coloring.
      This is much more tricky. I use following algorithm:
      1. When you load the file into the editor, scan all the comments and
            build a list associated with the file. Each member should contain:
            The starting line of the comment.
            The starting column of the comment
            The end line of the comment
            The end column of the comment.
      2. When you paint your text, see if each line painted is in the list.
            If it is, use the information to paint the comment. Be aware that
            that a line could be part of a comment without being explicitely
            in the list: I do not keep a pointer to all lines in a comment,
            just to the start/ending line. The middle lines are implicit.
      3. At each character typed, look if it can modify a nearby comment.
            All characters should be tested, since, for instance adding a space
            between the '/' and the '*' could change the status of a line from
            being a comment to being normal text.
      4. Whenever text is erased, test if the deletion affects the comments

3. Optimizations:
      The user can not modified anything that is not in the portion of the file
      being displayed in the screen. I maintain a pointer to the comment just
      before the screen text. Instead of scanning the whole comments list at
      each line, I start with that comment, since all comments above cannot
      be modified by the user.
      Block move operations (Cut, Copy and Paste commands) are VERY tricky.
      I just gave up handling all cases. I destroy the comments list and
      rebuild it from scratch after doing the block operation.
4. Performance.
      My IDE will run without noticable slow down in a 486-66.
5. Bugs.
      After having introduced this more than 2 years ago, there are still
      situations where the display is not coherent. This algorithm is easy
      to describe, but VERY difficult to implement. To see a proof, just use
      Microsoft's IDE for a while... Sometimes you will caught it with comments
      being confused as normal text.
      Watch for situations of
          Comments within comments.
          Erasing a nested comment.
          Specification ambiguities:
          int i; // this is a // comment, but is followed by a /*
                        // and here we do not know if the comment ends or not */

Jacob Navia Logiciels/Informatique
41 rue Maurice Ravel Tel (1)
93430 Villetaneuse Fax (1)

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.