parsing_c/comment_annotater_c.ml

   1 (* Yoann Padioleau
   2  *
   3  * Copyright (C) 2009 University of Urbana Champaign
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License (GPL)
   7  * version 2 as published by the Free Software Foundation.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * file license.txt for more details.
  13  *)
  14
  15 open Common
  16
  17 module T = Token_c
  18
  19
  20 (*****************************************************************************)
  21 (* Prelude *)
  22 (*****************************************************************************)
  23
  24 (* A trimmed down version of my comment_annotater of CComment. In CComment
  25  * I was also trying to associate the comment to the relevant entity, not
  26  * just the closest token (e.g. a function comment is not placed next to the
  27  * identifier of the function but before its return type or storage).
  28  *)
  29
  30
  31 (*****************************************************************************)
  32 (* Helpers *)
  33 (*****************************************************************************)
  34
  35 let is_comment_or_space_or_stuff tok =
  36   Token_helpers.is_not_in_ast tok && Token_helpers.is_origin tok
  37
  38 (* coupling with token_helpers.is_not_in_ast, and of course with tokens_c.ml *)
  39 let convert_relevant_tokens x =
  40   assert (Token_helpers.is_origin x);
  41
  42   match x with
  43   | Parser_c.TCommentSpace info ->
  44       Token_c.TCommentSpace, (Ast_c.parse_info_of_info info)
  45   | Parser_c.TCommentNewline info ->
  46       Token_c.TCommentNewline, (Ast_c.parse_info_of_info info)
  47
  48   | Parser_c.TComment info ->
  49       Token_c.TComment, (Ast_c.parse_info_of_info info)
  50
  51   (* the passed tokens because of our limited handling of cpp *)
  52   | Parser_c.TCommentCpp(cppcommentkind, info) ->
  53       Token_c.TCommentCpp cppcommentkind, (Ast_c.parse_info_of_info info)
  54
  55   | _ -> raise Impossible
  56
  57
  58 (*****************************************************************************)
  59 (* Main entry *)
  60 (*****************************************************************************)
  61
  62 (* right now we just add comment-like and origin-tok tokens,
  63  * as explained in token_c.ml.
  64  *
  65  * This simplified comment_annotater (compared to CComment) is really
  66  * simple as the tokens and the Ast_c.info in the asts actually share
  67  * the same refs.
  68  * So, modifying fields in the tokens will also modify the info in
  69  * the ast. Sometimes side effects simplify programming ...
  70  * We use similar tricks in unparse_c.ml. So really the asts argument
  71  * is not needed.
  72  *
  73  * ex: C1 C2 T1 T2 C3 C4 T3 C5 T4.
  74  *  => infoT1(-C1C2,+), infoT2(-,+C3C4), infoT3(-C3C4,+C5), infoT4(-C5,+)
  75  *)
  76
  77 (*
  78 let (agglomerate_either:
  79  ('a, 'a) Common.either list -> ('a list, 'a list) Common.either list) = fun xs ->
  80   raise Todo
  81
  82 let (span_and_pack:
  83  ('a -> ('a, 'a) Common.either) -> 'a list ->
  84       ('a list, 'a list) Common.either list) = fun f_either xs ->
  85   let xs' = List.map f_either xs in
  86   agglomerate_either xs'
  87 *)
  88
  89
  90 (* the asts is not really used, we do all via side effect on the tokens,
  91  * which share the info reference with the elements in the ast.
  92  *)
  93 let annotate_program toks asts =
  94    (* Common.exclude_but_keep_attached gather all comments before a
  95     * token and then associates to this token those comments. Note that
  96     * if reverse the list of tokens then this function can also be used
  97     * to gather all the comments after a token :)
  98     *)
  99
 100    (* before phase *)
 101    let toks_with_before =
 102      Common.exclude_but_keep_attached is_comment_or_space_or_stuff
 103        toks
 104    in
 105
 106   (* after phase. trick: reverse the tokens and reuse previous func *)
 107    let toks_with_after =
 108      List.rev
 109        (List.map
 110           (function (x,l) -> (x,List.rev l))
 111           (Common.exclude_but_keep_attached is_comment_or_space_or_stuff
 112              (List.rev toks)))
 113    in
 114
 115   (* merge *)
 116    assert(List.length toks_with_after =|= List.length toks_with_before);
 117
 118    Common.zip toks_with_before toks_with_after
 119    +> List.iter (fun ((t1, before), (t2, after)) ->
 120
 121      assert(t1 =*= t2);
 122
 123      let before' = before +> List.map convert_relevant_tokens in
 124      let after' = after  +> List.map convert_relevant_tokens in
 125
 126      let info = Token_helpers.info_of_tok t1 in
 127      info.Ast_c.comments_tag :=
 128        { Ast_c.mbefore = before';
 129          Ast_c.mafter = after';
 130          mbefore2 = [];
 131          mafter2 = [];
 132        };
 133
 134    );
 135    (* modified via side effect. I return it just to have a
 136     * clean signature.
 137     *)
 138    asts
 139
 140
 141
 142