[bug-idutils] bug in lisp scanner

[matthew green]

hello.

i had to apply this patch to stop mkid from cpu-spinning on some
.el inputs (eg, gettext's gettext-tools/misc/po-mode.el).  i didn't
track exactly what was happening, just noticed that it was possible
for the EOF handling to be done poorly.

it would be great if you included this or a similar patch in future
releases.  thanks!


.mrg.


--- libidu/scanners.c.orig      2010-12-11 15:27:10.000000000 -0800
+++ libidu/scanners.c   2010-12-11 15:26:59.000000000 -0800
@@ -1618,6 +1618,7 @@
   switch (c)
     {
     case EOF:
+    out:
       obstack_free (&tokens_obstack, obstack_finish (&tokens_obstack));
       return 0;
       
@@ -1634,19 +1635,25 @@
     case ';':                          /* comment */
       do {
        c = getc (in_FILE);
-      } while ( (c != EOF) && (c != '\n'));
+       if (c == EOF)
+         goto out;
+      } while (c != '\n');
       goto top;
       
     case '"':                          /* string with or without ansi-C 
escapes */
     string:
       do {
        c = getc (in_FILE);
+        if (c == EOF)
+         goto out;
        if (c == '\\')
          {
            c = getc (in_FILE);
+            if (c == EOF)
+             goto out;
            continue;
          }
-      } while ( (c != EOF) && (c != '"'));
+      } while (c != '"');
       goto top;
       
     case '.':
@@ -1654,11 +1661,12 @@
       id = scanner_buffer;
       *id++ = c;
       c = getc (in_FILE);
+      if (c == EOF)
+       goto out;
       if (is_DIGIT (c) ||
          (scanner_buffer[0] != '.' && (c == '.' || c == 'i' || c == 'I')))
        goto number;
-      if (c != EOF)
-       ungetc (c, in_FILE);
+      ungetc (c, in_FILE);
       goto ident;
       
     case '#':
@@ -1667,20 +1675,23 @@
       
       c = getc (in_FILE);
       if (c == EOF)
-       goto top;
+       goto out;
       else if (is_RADIX (c))
        goto number;
       else if (c == '\\')      /* #\... literal Character */
        {
          *id++ = c;
          c = getc (in_FILE);
+          if (c == EOF)
+           goto out;
          *id++ = c;
          if (is_LETTER (c))
            {
              while (is_LETTER (c = getc (in_FILE)))
                *id++ = c;
-             if (c != EOF)
-               ungetc (c, in_FILE);
+              if (c == EOF)
+               goto out;
+             ungetc (c, in_FILE);
            }
          *flags = TOK_LITERAL;
          obstack_grow0 (&tokens_obstack, scanner_buffer, id - scanner_buffer);
@@ -1694,8 +1705,9 @@
        {
          while (is_LETTER (c = getc (in_FILE)))
            *id++ = c;
-         if (c != EOF)
-           ungetc (c, in_FILE);
+          if (c == EOF)
+           goto out;
+         ungetc (c, in_FILE);
          *flags = TOK_LITERAL;
          obstack_grow0 (&tokens_obstack, scanner_buffer, id - scanner_buffer);
          return (struct token *) obstack_finish (&tokens_obstack);
@@ -1712,13 +1724,17 @@
                  break;
              }
          } while (c != EOF);
+          if (c == EOF)
+           goto out;
          goto top;
        }
       else if (c == '@')       /* address@hidden ...^_ EMACS byte-code comment 
*/
        {
          do {
            c = getc (in_FILE);
-         } while ( (c != EOF) && (c != '\037'));
+            if (c == EOF)
+             goto out;
+         } while (c != '\037');
          goto top;
        }
       else if (c == '[')       /* #[ ... ] EMACS byte-code object */
@@ -1742,23 +1758,24 @@
            {
              while (is_IDENT (c = getc (in_FILE)))
                *id++ = c;
+              if (c == EOF)
+               goto out;
              if (c == '[')
                {
                  c = getc (in_FILE);
+                  if (c == EOF)
+                   goto out;
                  if (c == ']')
                    {
                      *id++ = '[';
                      *id++ = ']';
                      continue;
                    }
-                 if (c != EOF)
-                   ungetc (c, in_FILE);
-                 ungetc ('[', in_FILE);
+                 ungetc (c, in_FILE);
                }
              break;
            }
-         if (c != EOF)
-           ungetc (c, in_FILE);
+         ungetc (c, in_FILE);
          *flags = TOK_NAME | TOK_LITERAL;
          obstack_grow0 (&tokens_obstack, scanner_buffer, id - scanner_buffer);
          return (struct token *) obstack_finish (&tokens_obstack);
@@ -1770,8 +1787,9 @@
          *id++ = c;
          while (is_NUMBER (c = getc (in_FILE)))
            *id++ = c;
-         if (c != EOF)
-           ungetc (c, in_FILE);
+          if (c == EOF)
+           goto out;
+         ungetc (c, in_FILE);
          *flags = TOK_NUMBER | TOK_LITERAL;
          obstack_grow0 (&tokens_obstack, scanner_buffer, id - scanner_buffer);
          return (struct token *) obstack_finish (&tokens_obstack);