X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;ds=sidebyside;f=lib%2Fff-utf8.c;h=a7e40d3c0cf35bd177a0e33d68a5bf29845d37f8;hb=c5fbc7b75705d1f7a322ad73e6055284a3b94e73;hp=f55719b1ada99c02811bee8d2da1141912f9d89b;hpb=414288b47086de5aafa715dd9c794f62f94f8ab2;p=libucw.git

diff --git a/lib/ff-utf8.c b/lib/ff-utf8.c
index f55719b1..a7e40d3c 100644
--- a/lib/ff-utf8.c
+++ b/lib/ff-utf8.c
@@ -1,7 +1,8 @@
 /*
- *	Sherlock Library: Reading and writing of UTF-8 on Fastbuf Streams
+ *	UCW Library: Reading and writing of UTF-8 on Fastbuf Streams
  *
  *	(c) 2001--2004 Martin Mares <mj@ucw.cz>
+ *	(c) 2004 Robert Spalek <robert@ucw.cz>
  *
  *	This software may be freely distributed and used according to the terms
  *	of the GNU Lesser General Public License.
@@ -53,6 +54,62 @@ bget_utf8_slow(struct fastbuf *b)
   return UNI_REPLACEMENT;
 }
 
+int
+bget_utf8_32_slow(struct fastbuf *b)
+{
+  int c = bgetc(b);
+  int code;
+  int nr;
+
+  if (c < 0x80)				/* Includes EOF */
+    return c;
+  if (c < 0xc0)				/* Incorrect combination */
+    return UNI_REPLACEMENT;
+  if (c < 0xe0)
+    {
+      code = c & 0x1f;
+      nr = 1;
+    }
+  else if (c < 0xf0)
+    {
+      code = c & 0x0f;
+      nr = 2;
+    }
+  else if (c < 0xf8)
+    {
+      code = c & 0x07;
+      nr = 3;
+    }
+  else if (c < 0xfc)
+    {
+      code = c & 0x03;
+      nr = 4;
+    }
+  else if (c < 0xfe)
+    {
+      code = c & 0x01;
+      nr = 5;
+    }
+  else					/* Too large, skip it */
+    {
+      while ((c = bgetc(b)) >= 0x80 && c < 0xc0)
+	;
+      goto wrong;
+    }
+  while (nr-- > 0)
+    {
+      if ((c = bgetc(b)) < 0x80 || c >= 0xc0)
+	goto wrong;
+      code = (code << 6) | (c & 0x3f);
+    }
+  return code;
+
+ wrong:
+  if (c >= 0)
+    bungetc(b);
+  return UNI_REPLACEMENT;
+}
+
 void
 bput_utf8_slow(struct fastbuf *b, uns u)
 {
@@ -71,3 +128,40 @@ bput_utf8_slow(struct fastbuf *b, uns u)
       bputc(b, 0x80 | (u & 0x3f));
     }
 }
+
+void
+bput_utf8_32_slow(struct fastbuf *b, uns u)
+{
+  ASSERT(u < (1U<<31));
+  if (u < 0x80)
+    bputc(b, u);
+  else
+    {
+      if (u < 0x800)
+	bputc(b, 0xc0 | (u >> 6));
+      else
+	{
+	  if (u < (1<<16))
+	    bputc(b, 0xe0 | (u >> 12));
+	  else
+	    {
+	      if (u < (1<<21))
+		bputc(b, 0xf0 | (u >> 18));
+	      else
+		{
+		  if (u < (1<<26))
+		    bputc(b, 0xf8 | (u >> 24));
+		  else
+		    {
+		      bputc(b, 0xfc | (u >> 30));
+		      bputc(b, 0x80 | ((u >> 24) & 0x3f));
+		    }
+		  bputc(b, 0x80 | ((u >> 18) & 0x3f));
+		}
+	      bputc(b, 0x80 | ((u >> 12) & 0x3f));
+	    }
+	  bputc(b, 0x80 | ((u >> 6) & 0x3f));
+	}
+      bputc(b, 0x80 | (u & 0x3f));
+    }
+}