(* * This file is part of Coccinelle, lincensed under the terms of the GPL v2. * See copyright.txt in the Coccinelle source code for more information. * The Coccinelle source code can be obtained at http://coccinelle.lip6.fr *) (* Lexer/parser for spgen config file *) { (* ------------------------------------------------------------------------- *) (* TYPES *) (* the config syntax has pretty low complexity and is very lenient, so we just * interpret the tokens directly and skip the lexical analysis. *) type attribute = | Org of string * string list | Report of string * string list type part = | Description of string | Limitations of string list | Keywords of string | Confidence of string | Comments of string | Options of string | Authors of string list (* author, affiliation, license in one *) | Url of string | Rule of (string * string option) * attribute list (* the table is for line numbers to produce more meaningful error messages *) let table = ref (Array.make 0 (0,0)) exception Eof let get_start_line lexbuf = match !table.(Lexing.lexeme_start lexbuf) with (l,_) -> l let get_column lexbuf = match !table.(Lexing.lexeme_start lexbuf) with (_,c) -> c let get_position_str lexbuf = "around line " ^ (string_of_int (get_start_line lexbuf)) ^ ", column " ^ (string_of_int (get_column lexbuf)) let error prob lexbuf = let pos = get_position_str lexbuf in failwith ("Config error: " ^ prob ^ pos) let illegal = error "Illegal syntax " let split_list delim = Str.split (Str.regexp (" *"^delim^" *")) } (* ------------------------------------------------------------------------- *) (* LEXING RULES *) let sp = [' ' '\t'] let ws = ['\n' '\r' '\012' '\013' ' ' '\t']* let notws = [^ '\n' '\r' '\012' '\013' ' ' '\t']* let notnl = [^ '\n' '\r' '\012' '\013']* let equal = (sp* "=" sp*) let colon = (sp* ":" sp*) let percent = (sp* "%" sp*) let letter = ['A'-'Z' 'a'-'z' '_'] let number = ['0'-'9']* let cname = letter (letter | number)* rule token = parse | ("description"|"d") equal (notnl as d) { Description d } | ("keywords" |"k") equal (notnl as v) { Keywords v } | ("confidence" |"c") equal (notws as c) { Confidence c } | ("comments" |"m") equal (notnl as m) { Comments m } | ("options" |"o") equal (notnl as o) { Options o } | ("url" |"u") equal (notnl as u) { Url u } | ("limitations"|"l") equal (notnl as l) { Limitations (split_list "|" l) } | ("author"|"authors"|"a") equal (notnl as a) { Authors (split_list "|" a) } (* for naming unnamed rules. e.g. 8:name = ... for the rule on line 8. *) | (number as oldrule) colon (cname as newrule) equal { Rule ((oldrule, Some newrule), cocci_rule lexbuf) } (* standard rules with org and report messages *) | (cname as rulenm) equal { Rule ((rulenm, None), cocci_rule lexbuf) } (* comments and whitespace are skipped *) | ws | "//" [^ '\n']* { token lexbuf } | "/*" { comment lexbuf } | eof { raise Eof} | _ { illegal lexbuf } and comment = parse | "*/" {token lexbuf} | eof { error "unclosed comment, needs \"*/\"! " lexbuf} | _ {comment lexbuf } and cocci_rule = parse | ws* "org" colon { let (a,b) = format_string lexbuf in Org (a,b) :: (cocci_rule lexbuf) } | ws* "report" colon { let (a,b) = format_string lexbuf in Report (a,b) :: (cocci_rule lexbuf) } | _ { [] } (* possibly formatted org and report messages *) and format_string = parse | "\"" (notnl as msg) "\"" percent "(" (notnl as lst) ")" { (msg, split_list "," lst) } | "\"" (notnl as msg) "\"" percent (notnl as lst) { (msg,[lst]) } | "\"" (notnl as msg) "\"" { (msg, []) } | _ { illegal lexbuf }