[Qemacs-commit] qemacs charset.c

[Charlie Gordon]

CVSROOT:        /sources/qemacs
Module name:    qemacs
Changes by:     Charlie Gordon <chqrlie>        14/03/24 09:33:50

Modified files:
        .              : charset.c 

Log message:
        improve utf8 and ucsx probing functions

CVSWeb URLs:
http://cvs.savannah.gnu.org/viewcvs/qemacs/charset.c?cvsroot=qemacs&r1=1.34&r2=1.35

Patches:
Index: charset.c
===================================================================
RCS file: /sources/qemacs/qemacs/charset.c,v
retrieving revision 1.34
retrieving revision 1.35
diff -u -b -r1.34 -r1.35
--- charset.c   15 Mar 2014 14:55:05 -0000      1.34
+++ charset.c   24 Mar 2014 09:33:49 -0000      1.35
@@ -596,7 +596,7 @@
             return 0;
         }
     }
-    if (count_spaces | count_lines)
+    if (count_spaces + count_lines > size / 16)
         return 1;
     else
         return 0;
@@ -732,7 +732,7 @@
             return 0;
         }
     }
-    if (count_spaces | count_lines)
+    if (count_spaces + count_lines > size / 16)
         return 1;
     else
         return 0;
@@ -876,7 +876,7 @@
             return 0;
         }
     }
-    if (count_spaces | count_lines)
+    if (count_spaces + count_lines > size / 16)
         return 1;
     else
         return 0;
@@ -1006,7 +1006,7 @@
             return 0;
         }
     }
-    if (count_spaces | count_lines)
+    if (count_spaces + count_lines > size / 16)
         return 1;
     else
         return 0;
@@ -1425,8 +1425,10 @@
     has_utf8 = 0;
     for (i = 0; i < size;) {
         c = buf[i++];
-        if ((c >= 0x80 && c < 0xc0) || c >= 0xfe)
-            goto no_utf8;
+        if ((c >= 0x80 && c < 0xc0) || c >= 0xfe) {
+            has_utf8 = -1;
+            goto done_utf8;
+        }
         l = utf8_length[c];
         while (l > 1) {
             has_utf8 = 1;
@@ -1434,14 +1436,14 @@
                 break;
             c = buf[i++];
             if (!(c >= 0x80 && c < 0xc0)) {
-            no_utf8:
-                has_utf8 = 0;
-                break;
+                has_utf8 = -1;
+                goto done_utf8;
             }
             l--;
         }
     }
-    if (has_utf8) {
+done_utf8:
+    if (has_utf8 > 0) {
         return detect_eol_type(buf, size, &charset_utf8, eol_typep);
     }
 
@@ -1523,7 +1525,7 @@
 
     detect_eol_type(buf, size, &charset_raw, eol_typep);
 
-    if (*eol_typep == EOL_DOS) {
+    if (*eol_typep == EOL_DOS || has_utf8 < 0) {
         /* XXX: default DOS files to Latin1, should be selectable */
         return &charset_8859_1;
     }