OK - fixes: some bugs in the text parser

     ( ]} meaning end of code and somehting else instead of end of precode)
OK - add: in Odoc_info: text_of_string, text_string_of_text, info_of_string


git-svn-id: http://caml.inria.fr/svn/ocaml/trunk@5974 f963ae5c-01c2-4b8c-9fe0-0dff7051ff02

7 files changed, 195 insertions, 21 deletions

diff --git a/ocamldoc/Changes.txt b/ocamldoc/Changes.txt
index 7b92295039..3f432819a3 100644
--- a/ocamldoc/Changes.txt
+++ b/ocamldoc/Changes.txt
@@ -1,4 +1,7 @@
 Current :
+OK - fixes: some bugs in the text parser 
+     ( ]} meaning end of code and somehting else instead of end of precode)
+OK - add: in Odoc_info: text_of_string, text_string_of_text, info_of_string
 OK - fix: better output of titles in html (use more the style)
 OK - add: -intro option to use a file content as ocamldoc comment to use as
 OK   introduction for LaTeX document and HTML index page
diff --git a/ocamldoc/Makefile b/ocamldoc/Makefile
index 9e223cfb1b..772305bc4a 100644
--- a/ocamldoc/Makefile
+++ b/ocamldoc/Makefile
@@ -328,6 +328,14 @@ stdlib_man/Pervasives.o: $(STDLIB_MLIS)
 	-t "OCaml library" -man-mini -man-suffix 3o \
 	$(STDLIB_MLIS)
 
+autotest_stdlib: dummy
+	$(MKDIR) $@ 
+	$(OCAMLDOC_RUN) -g autotest/odoc_test.cmo\
+	$(INCLUDES) -keep-code \
+	../stdlib/pervasives.ml ../stdlib/*.mli \
+	../otherlibs/unix/unix.mli \
+	../otherlibs/str/str.mli
+
 # backup, clean and depend :
 ############################
 
diff --git a/ocamldoc/odoc_info.ml b/ocamldoc/odoc_info.ml
index ffb016a148..4e11dfe2ed 100644
--- a/ocamldoc/odoc_info.ml
+++ b/ocamldoc/odoc_info.ml
@@ -50,6 +50,8 @@ and text_element = Odoc_types.text_element =
 
 and text = text_element list
 
+exception Text_syntax = Odoc_text.Text_syntax
+
 type see_ref = Odoc_types.see_ref =
     See_url of string
   | See_file of string
@@ -176,7 +178,11 @@ let apply_if_equal f v1 v2 =
   else
     v2
 
-let info_of_comment_file f =
+let text_of_string = Odoc_text.Texter.text_of_string
+
+let text_string_of_text = Odoc_text.Texter.string_of_text
+
+let info_of_string s =
   let dummy =
     {
       i_desc = None ;
@@ -191,19 +197,19 @@ let info_of_comment_file f =
       i_custom = [] ;
     } 
   in
+  let s2 = Printf.sprintf "(** %s *)" s in
+  let (_, i_opt) = Odoc_comments.Basic_info_retriever.first_special "-" s2 in
+  match i_opt with
+    None -> dummy
+  | Some i -> i
+      
+let info_of_comment_file f =
   try
-    let s = Printf.sprintf "(** %s *)" (Odoc_misc.input_file_as_string f) in
-    let (_, i_opt) = Odoc_comments.Basic_info_retriever.first_special f s in
-    (
-     match i_opt with
-       None -> dummy
-     | Some i -> i
-    )
+    let s = Odoc_misc.input_file_as_string f in
+    info_of_string s
   with
     Sys_error s -> 
-      prerr_endline s;
-      incr errors;
-      dummy
+      failwith s
 
 module Search = 
   struct 
diff --git a/ocamldoc/odoc_info.mli b/ocamldoc/odoc_info.mli
index 02ed187336..a2b8529fee 100644
--- a/ocamldoc/odoc_info.mli
+++ b/ocamldoc/odoc_info.mli
@@ -59,6 +59,10 @@ type see_ref = Odoc_types.see_ref =
   | See_file of string
   | See_doc of string
 
+(** Raised when parsing string to build a {!Odoc_info.text}
+   structure. [(line, char, string) *)
+exception Text_syntax of int * int * string 
+
 (** The information in a \@see tag. *)
 type see = see_ref * text
 
@@ -725,9 +729,26 @@ val apply_opt : ('a -> 'b) -> 'a option -> 'b option
    are different, return the second one.*)
 val apply_if_equal : ('a -> 'a) -> 'a -> 'a -> 'a
 
+(** [text_of_string s] returns the text structure from the
+   given string. 
+   @raise Text_syntax if a syntax error is encountered. *)
+val text_of_string : string -> text
+
+(** [string_text_of_text text] returns the string representing
+   the given [text]. This string can then be parsed again
+   by {!Odoc_info.text_of_string}.*)   
+val text_string_of_text : text -> string
+
+(** [info_of_string s] parses the given string
+   like a regular ocamldoc comment and return an 
+   {!Odoc_info.info} structure.
+   @return an empty structure if there was a syntax error. TODO: change this
+*)
+val info_of_string : string -> info
+
 (** [info_of_comment_file file] parses the given file
-   and return an info structure. The content of the file
-   must have the same syntax as the content of a special comment.
+   and return an {!Odoc_info.info} structure. The content of the 
+   file must have the same syntax as the content of a special comment.
    @raise Failure is the file could not be opened or there is a
    syntax error.
 *)
diff --git a/ocamldoc/odoc_text.ml b/ocamldoc/odoc_text.ml
index aa675692f7..b83c88a19f 100644
--- a/ocamldoc/odoc_text.ml
+++ b/ocamldoc/odoc_text.ml
@@ -13,6 +13,8 @@
 
 exception Text_syntax of int * int * string (* line, char, string *)
 
+open Odoc_types
+
 module Texter =
   struct
     (* builds a text structure from a string. *)
@@ -27,5 +29,116 @@ module Texter =
                               !Odoc_text_lexer.char_number, 
                               s)
                 )
+
+    let count s c =
+      let count = ref 0 in
+      for i = 0 to String.length s - 1 do
+	if s.[i] = c then incr count
+      done;
+      !count
+
+    let escape_n s c n =
+      let remain = ref n in
+      let len = String.length s in
+      let b = Buffer.create (len + n) in
+      for i = 0 to len - 1 do
+	if s.[i] = c && !remain > 0 then
+	  (
+	   Printf.bprintf b "\\%c" c;
+	   decr remain
+	  )
+	else
+	  Buffer.add_char b s.[i]
+      done;
+      Buffer.contents b
+
+    let escape_code s =
+      let open_brackets = count s '[' in
+      let close_brackets = count s ']' in
+      if open_brackets > close_brackets then
+	escape_n s '[' (open_brackets - close_brackets)
+      else
+	if close_brackets > open_brackets then
+	  escape_n s ']' (close_brackets - open_brackets) 
+	else
+	  s
+
+    let escape_raw s =
+      let len = String.length s in
+      let b = Buffer.create len in
+      for i = 0 to len - 1 do
+	match s.[i] with
+	  '[' | ']' | '{' | '}' ->
+	    Printf.bprintf b "\\%c" s.[i]
+	| c ->
+	    Buffer.add_char b c
+      done;
+      Buffer.contents b
+
+    let p = Printf.bprintf
+
+    let rec p_text b t =
+      List.iter (p_text_element b) t
+
+    and p_list b l =
+      List.iter
+	(fun t -> p b "{- " ; p_text b t ; p b "}\n")
+	l
+
+    and p_text_element b = function
+      | Raw s -> p b "%s" (escape_raw s)
+      | Code s -> p b "[%s]" (escape_code s)
+      | CodePre s -> p b "{[%s]}" s
+      | Verbatim s -> p b "{v %s v}" s
+      | Bold t -> p b "{b " ; p_text b t ; p b "}"
+      | Italic t -> p b "{i " ; p_text b t ; p b "}"
+      | Emphasize t -> p b "{e " ; p_text b t ; p b "}"
+      | Center t -> p b "{C " ; p_text b t ; p b "}"
+      | Left t -> p b "{L " ; p_text b t ; p b "}"
+      | Right t -> p b "{R " ; p_text b t ; p b "}"
+      | List l -> p b "{ul\n"; p_list b l; p b "}"
+      | Enum l -> p b "{ol\n"; p_list b l; p b "}"
+      | Newline -> p b "\n" 
+      | Block  t -> p_text b t
+      | Title (n, l_opt, t) ->
+	  p b "{%d%s " 
+	    n
+	    (match l_opt with
+	      None -> ""
+	    | Some s -> ":"^s
+	    );
+	  p_text b t ; 
+	  p b "}"
+      | Latex s -> p b "{%% %s%%}" s
+      | Link (s,t) ->
+	  p b "{{:%s}" s;
+	  p_text b t ;
+	  p b "}"
+      | Ref (s,None) ->
+	  p b "{!%s}" s
+      |	Ref (s, Some k) ->
+	  (
+	   let sk = match k with
+	     RK_module -> "module"
+	   | RK_module_type -> "modtype"
+	   | RK_class -> "class"
+	   | RK_class_type -> "classtype"
+	   | RK_value -> "val"
+	   | RK_type -> "type"
+	   | RK_exception -> "exception"
+	   | RK_attribute -> "attribute"
+	   | RK_method -> "method"
+	   | RK_section _ -> "section"
+	   in
+	   p b "{!%s:%s}" sk s	   
+	  )
+      | Superscript t -> p b "{^" ; p_text b t ; p b "}"
+      | Subscript t -> p b "{_" ; p_text b t ; p b "}"
+	    
+    let string_of_text s =
+      let b = Buffer.create 256 in
+      p_text b s;
+      Buffer.contents b
+	
   end
  
diff --git a/ocamldoc/odoc_text.mli b/ocamldoc/odoc_text.mli
index 6ed5bd0e44..bc121791dd 100644
--- a/ocamldoc/odoc_text.mli
+++ b/ocamldoc/odoc_text.mli
@@ -18,4 +18,7 @@ exception Text_syntax of int * int * string (* line, char, string *)
 
 (** Transformation of strings to text structures. *)
 module Texter : 
-    sig val text_of_string : string -> Odoc_types.text end
+    sig 
+      val text_of_string : string -> Odoc_types.text 
+      val string_of_text : Odoc_types.text -> string
+    end
diff --git a/ocamldoc/odoc_text_lexer.mll b/ocamldoc/odoc_text_lexer.mll
index 44f0e8c88b..f21ec9ae39 100644
--- a/ocamldoc/odoc_text_lexer.mll
+++ b/ocamldoc/odoc_text_lexer.mll
@@ -418,13 +418,33 @@ rule main = parse
       if !verb_mode or !latex_mode or !ele_ref_mode then
         Char (Lexing.lexeme lexbuf)
       else
-        if !code_pre_mode then 
-          (
-           code_pre_mode := false;
-           END_CODE_PRE 
-          )
-        else
-          Char (Lexing.lexeme lexbuf)
+	if !open_brackets >= 1 then
+	  (
+	   lexbuf.Lexing.lex_curr_pos <- lexbuf.Lexing.lex_curr_pos - 1;
+           lexbuf.Lexing.lex_curr_p <- 
+	     { lexbuf.Lexing.lex_curr_p with
+	       pos_cnum = lexbuf.Lexing.lex_curr_p.pos_cnum - 1
+	     } ;
+	   decr char_number ;
+	   if !open_brackets > 1 then
+	     (
+	      decr open_brackets;
+              Char "]"
+	     )
+	   else
+	     (
+              open_brackets := 0;
+              END_CODE 
+             )
+	  )
+	else
+          if !code_pre_mode then 
+            (
+             code_pre_mode := false;
+             END_CODE_PRE 
+            )
+          else
+            Char (Lexing.lexeme lexbuf)
     }
 
 | begin_ele_ref end