Re: Two questions

-- Trey <trey(a)veggie.stu.wesleyan.edu&gt; spake thusly: I've been doing a lot of work with this one lately, so I may be able to help here. `parse-partial-sexp' is the primary tool used for parsing syntax in (X)Emacs. It's actually useful for more than just Lisp syntax. It can parse over any well-formed balanced expressions (C, C++, etc.), and is used by such things as font-lock to do syntax parsing. It works like this: (parse-partial-sexp from to &optional targetdepth stopbefore oldstate commentstop buffer) It parses a file from <from> and stops at <to> (or possibly before). It returns a list which indicates the state of the file at the point it stopped at (ie, point). I won't detail each element of the returned state (do '(describe-function "parse-partial-sexp")' for details), but the basic information is: - The depth of parentheses-like expressions (determined by (),{},[], etc.) as defined by the buffer's syntax table. - Whether or not point is inside a comment or string. - The address of various things (start of innermost balanced expression, start of containing comment/string if relevant). You can also get parse-partial-sexp to stop at specific places (other than <stop>) if desired by setting the inputs: - targetdepth (intgeger): cause parsing to stop as soon as that depth of balances expressions is reached - stopbefore (t): stop before the beginning of a paren-like opener. - commentstop (t): stop at the start of a comment - commentstop ('syntax-table): stop at the start of a comment or string, or after the end of a comment or string If you want to continue parsing, the proper thing to do is call `parse-partial-sexp' again from the point it stopped and pass the previously returned state as <oldstate>. As for how to use it, it depends on what you're trying to get. You can get the syntax state of a particular position in a buffer by calling "(parse-partial-sexp (point-min) (point))". The returned list would tell you how deep that point is, and whether it's in a comment or string. For instance, the beginning-of-defun that I just rewrote does something like: (let* ((state (parse-partial-sexp (point-min) (point))) (depth (nth 0 state)) (pt (if (or (nth 3 state) (nth 4 state)) (nth 8 state) (point)))) (if (> depth 0) (goto-char (scan-lists pt -1 depth))) This gets the current syntax state, figures out the beginning of the enclosing comment or string if there is one, and then uses scan-lists to jump to the beginning of the outermost enclosing sexp (balanced parentheses). `font-lock-fontify-syntactically-region' uses `parse-partial-sexp' to scan the region, stopping at the beginning and end of each comment and string, and uses the information returned to fontify said comments and strings. I also suggest reading up on syntax tables: M-: (Info-goto-node "(lispref)Syntax Tables") `parse-partial-sexp' uses these to determine syntax of characters when parsing. Understanding syntax tables is important for understanding how `parse-partial-sexp' works. Hope this helps. Let me know if you have any further questions.