- factorized DisambiguateChoices module

[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 21f0205203aea6e410b54d283a8ed60f1a92bf1f..b339b8dc47101cfc4ef9b85deeaf28480e1a94ee 100644 (file)
--- a/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
+++ b/helm/ocaml/cic_disambiguation/cicTextualLexer2.ml
@@ -41,6 +41,10 @@ let regexp ident = (alpha ident_cont*) | ('_' ident_cont+)
 let regexp ident' = ((alpha | tex_token) ident_cont'*) | ('_' ident_cont'+)
 let regexp paren = [ '(' '[' '{' ')' ']' '}' ]
 let regexp meta = '?' num
+let regexp qstring = '"' [^ '"']* '"'
+let regexp uri =
+  (*      schema      *) (*     path     *) (*  ext   *) (*    xpointer     *)
+  ("cic:/" | "theory:/") ident ('/' ident)* ('.' ident)+ ('#' num ('/' num)*)?
 (* let regexp catchall = .* *)
 
 let keywords = Hashtbl.create 17
@@ -86,6 +90,7 @@ and token = lexer
 *)
 let rec token = lexer
   | blanks -> token lexbuf
+  | uri -> return lexbuf ("URI", Ulexing.utf8_lexeme lexbuf)
   | ident ->
       let lexeme = Ulexing.utf8_lexeme lexbuf in
       (try
@@ -94,6 +99,10 @@ let rec token = lexer
   | num -> return lexbuf ("NUM", Ulexing.utf8_lexeme lexbuf)
   | paren -> return lexbuf ("PAREN", Ulexing.utf8_lexeme lexbuf)
   | meta -> return lexbuf ("META", 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)
   | symbol -> return lexbuf ("SYMBOL", Ulexing.utf8_lexeme lexbuf)
   | tex_token ->
       let macro =
@@ -116,7 +125,7 @@ let tok_func stream =
       | Ulexing.Error -> error_at_end lexbuf "Unexpected character"
       | Ulexing.InvalidCodepoint i -> error_at_end lexbuf "Invalid code point")
 
-let lex =
+let cic_lexer =
   { 
     Token.tok_func = tok_func;
     Token.tok_using = (fun _ -> ());
@@ -125,3 +134,4 @@ let lex =
     Token.tok_text = Token.lexer_text;
     Token.tok_comm = None;
   }
+