- changes defaults of getxml (format gzipped, don't patch dtd)

[helm.git] / helm / ocaml / cic_disambiguation / cicTextualLexer2.ml

diff --git a/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml b/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
index b62230a335aa84b1da78e8d3eb0687b6c00b534a..650b6972375f76c5c49885cfd51efdb05a3b8a5b 100644 (file)
--- a/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
+++ b/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
@@ -37,6 +37,8 @@ let regexp symbol_char =
      ' ' '\t' '\n'
      '\\' '(' '[' '{' ')' ']' '}' '?'
   ]
+let regexp comment_char = [^ '\n' ]
+let regexp comment = "%%" comment_char*
 let regexp blanks = blank+
 let regexp num = digit+
 let regexp tex_token = '\\' alpha+
@@ -104,8 +106,11 @@ let rec token' = lexer
 
 and token = lexer
 *)
-let rec token = lexer
-  | blanks -> token lexbuf
+
+let remove_quotes s = String.sub s 1 (String.length s - 2)
+
+let rec token comments = lexer
+  | blanks -> token comments lexbuf
   | uri -> return lexbuf ("URI", Ulexing.utf8_lexeme lexbuf)
   | ident ->
       let lexeme = Ulexing.utf8_lexeme lexbuf in
@@ -117,9 +122,7 @@ let rec token = lexer
   | meta -> return lexbuf ("META", Ulexing.utf8_lexeme lexbuf)
   | implicit -> return lexbuf ("IMPLICIT", Ulexing.utf8_lexeme lexbuf)
   | qstring ->
-      let lexeme = Ulexing.utf8_lexeme lexbuf in
-      let s = String.sub lexeme 1 (String.length lexeme - 2) in
-      return lexbuf ("QSTRING", s)
+      return lexbuf ("QSTRING", remove_quotes (Ulexing.utf8_lexeme lexbuf))
   | symbol -> return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf)
   | tex_token ->
       let macro =
@@ -129,22 +132,30 @@ let rec token = lexer
         return lexbuf ("SYMBOL", Utf8Macro.expand macro)
       with Utf8Macro.Macro_not_found _ ->
         return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf))
+  | comment ->
+      if comments then
+        let comment =
+          Ulexing.utf8_sub_lexeme lexbuf 2 (Ulexing.lexeme_length lexbuf - 2)
+        in
+        return lexbuf ("COMMENT", comment)
+      else
+        token comments lexbuf
   | eof -> return lexbuf ("EOI", "")
   | _ -> error lexbuf "Invalid character"
 
-let tok_func stream =
+let tok_func comments stream =
   let lexbuf = Ulexing.from_utf8_stream stream in
   Token.make_stream_and_flocation
     (fun () ->
       try
-       token lexbuf
+       token comments lexbuf
       with
       | Ulexing.Error -> error_at_end lexbuf "Unexpected character"
       | Ulexing.InvalidCodepoint i -> error_at_end lexbuf "Invalid code point")
 
-let cic_lexer =
-  { 
-    Token.tok_func = tok_func;
+let cic_lexer ?(comments = false) () =
+  {
+    Token.tok_func = tok_func comments;
     Token.tok_using = (fun _ -> ());
     Token.tok_removing = (fun _ -> ()); 
     Token.tok_match = Token.default_match;