Better chars

stdlib/Matcher.expr

@@ -1,27 +1,59 @@
--- Matcher library for Ostap
+-- (C) Dmitry Boulytchev, St. Petersburg State University, JetBrains Research, 2020
+-- Matcher: simple string matching library.
 
-fun matcherCreate (pos, buf, line, col) {
+
+-- Create a regular expression representation.
+-- Arguments:
+--    r    --- a string representation for regular expression (as per GNU regexp)
+--    name --- a string describing the meaning of the expression in free form
+--             (e.g. "identifier", "string constant", etc.), used for error
+--             reporting
+fun createRegexp (r, name) {
+  return [regexp (r), name]
+}
+
+-- Create an immutable matcher.
+-- Arguments:
+--    buf --- a string to match in
+--    pos --- an integer beginning position to match from
+--    line, col --- line and column numbers
+-- This function is internal, do not use it directly.
+-- To initially create a matcher use initMatcher function (see below).
+fun matcherCreate (buf, pos, line, col) {
+  -- Shows a matcher in a readable form
   fun show () {
     return sprintf ("buf : %-40s\npos : %d\nline: %d\ncol : %d\n", buf, pos, line, col)
   }
-  
+
+  -- Calculates the number of remaining unmatched characters in the buffer
   fun rest () {
     return buf.length - pos
   }
-  
+
+  -- Moves the position pointer on given number of characters within one line (i.e. 
   fun shift (n) {
-    return matcherCreate (pos + n, buf, line, col + n)
+    return matcherCreate (buf, pos + n, line, col + n)
   }
-  
+
   fun matchString (s) {
     return
       if s.length > rest ()
-      then None
-      elif matchSubString (buf, s, pos) then Some (shift (s.length)) 
-      else None
+      then Fail (sprintf ("""%s"" expected at %d:%d", s, line, col))
+      elif matchSubString (buf, s, pos) then Succ (shift (s.length), s) 
+      else Fail (sprintf ("""%s"" expected at %d:%d", s, line, col))
       fi
   }
 
+  fun matchRegexp (r) {
+    local n;
+    
+    return
+      if (n := regexpMatch (r[0], buf, pos)) > 0
+      then Succ (shift (n), substring (buf, pos, n))
+      else Fail (sprintf ("%s expected at %d:%d", r[1], line, col))
+      fi
+  }
+  
   fun eof () {
     return rest () == 0
   }
@@ -29,32 +61,39 @@ fun matcherCreate (pos, buf, line, col) {
   return [
      show,
      eof,
-     matchString
+     matchString,
+     matchRegexp
   ]
 }
 
 fun show (m) {
-  return m [0]
+  return m [0] ()
 }
 
 fun eof (m) {
-  return m [1]
+  return m [1] ()
 }
 
 fun matchString (m, s) {
   return m [2] (s)
 }
 
-fun matcherInit (buf) {
-  return matcherCreate (0, buf, 1, 1)
+fun matchRegexp (m, r) {
+  return m [3] (r)
 }
 
+fun matcherInit (buf) {
+  return matcherCreate (buf, 0, 1, 1)
+}
+
+--fun parse (a) {
+ 
+--}
+
 local m = matcherInit ("abc");
 
-printf ("%s", m.show ());
-printf ("eof: %s\n", m.eof ().string);
-printf ("matchString(""u""): %s\n", case m.matchString ("u") of Some (m) -> m.show () | _ -> "None" esac);
-printf ("matchString(""a""): %s\n", case m.matchString ("a") of Some (m) -> m.show () | _ -> "None" esac);
-printf ("matchString(""ab""): %s\n", case m.matchString ("ab") of Some (m) -> m.show () | _ -> "None" esac);
-printf ("matchString(""abc""): %s\n", case m.matchString ("abc") of Some (m) -> m.show () | _ -> "None" esac);
-printf ("matchString(""abcd""): %s\n", case m.matchString ("abcd") of Some (m) -> m.show () | _ -> "None" esac)
+local ident = createRegexp ("[a-z][a-zA-Z_]*", "identifier");
+local ws    = createRegexp ("[ \n\t]+", "whitespace");
+local str   = createRegexp ("""\([^""]\|""""\)*""", "string literal");
+
+printf ("ident: %s\n", case m.matchRegexp (ident) of Succ (m, s) -> "(" ++ m.show  ++ ", " ++ s ++ ")" | Fail (err) -> err.string esac)