parsing_c/parsing_hacks.mli

   1 open Common
   2
   3 (* This module tries to detect some cpp idioms so that we can parse as-is
   4  * files by adjusting or commenting some tokens. Parsing hack style.
   5  * Sometime we use some indentation information,
   6  * sometimes we do some kind of lalr(k) by finding patterns. Often try to
   7  * work on better token representation, like ifdef-paren-ized, brace-ized,
   8  * paren-ized, so can do easier pattern matching to more easily match
   9  * complex cpp idiom pattern. Also try to get context info such as
  10  * whether the token is in a initializer as some common patterns have different
  11  * use depending on context.
  12  *
  13  *
  14  * Example of cpp idioms:
  15  *  - if 0 for commenting stuff (not always code, sometimes just real comments)
  16  *  - ifdef old version
  17  *  - ifdef funheader
  18  *  - ifdef statements, ifdef expression, ifdef-mid
  19  *  - macro toplevel (with or without ptvirg)
  20  *  - macro foreach
  21  *  - macro higher order
  22  *  - macro declare
  23  *  - macro debug
  24  *  - macro no ptvirg
  25  *  - macro string, and macro function string taking param and ##
  26  *  - macro attribute
  27  *
  28  * Cf the TMacroXxx in parser_c.mly and MacroXxx in ast_c.ml
  29  *
  30  * Also try infer typedef.
  31  *
  32  * Also do other stuff involving cpp like expanding some macros,
  33  * or try parse well define body by finding the end of define virtual
  34  * end-of-line token.
  35  *)
  36
  37 (* corresponds to what is in the yacfe configuration file (e.g. standard.h) *)
  38 type define_def = string * define_param * define_body
  39  and define_param =
  40    | NoParam
  41    | Params of string list
  42  and define_body =
  43    | DefineBody of Parser_c.token list
  44    | DefineHint of parsinghack_hint
  45
  46    (* strongly corresponds to the TMacroXxx in the grammar and lexer and the
  47     * MacroXxx in the ast.
  48     *)
  49    and parsinghack_hint =
  50      | HintIterator
  51      | HintDeclarator
  52      | HintMacroString
  53      | HintMacroStatement
  54      | HintAttribute
  55      | HintMacroIdentBuilder
  56
  57 val regexp_macro: Str.regexp
  58 val regexp_annot: Str.regexp
  59 val regexp_declare: Str.regexp
  60 val regexp_foreach: Str.regexp
  61 val regexp_typedef: Str.regexp
  62
  63 val _defs : (string, define_def) Hashtbl.t ref
  64
  65 (* can reset it *)
  66 val ifdef_paren_cnt: int ref
  67
  68 val fix_tokens_define : Parser_c.token list -> Parser_c.token list
  69 val extract_cpp_define : Parser_c.token list -> (string, define_def) assoc
  70
  71
  72 val fix_tokens_cpp : Parser_c.token list -> Parser_c.token list
  73
  74 (* next stream tokens -> passed stream tokens -> final next token *)
  75 val lookahead :
  76   pass:int ->
  77   Parser_c.token list -> Parser_c.token list -> Parser_c.token
  78
  79