From: Martin Mares <mj@ucw.cz>
Date: Sun, 21 Jan 2001 17:59:58 +0000 (+0000)
Subject: Added functions for reading/writing UTF-8 characters on fastbuf streams.
X-Git-Tag: holmes-import~1577
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=dd79339e61140f94f3602e8bbdee78e838f9090c;p=libucw.git

Added functions for reading/writing UTF-8 characters on fastbuf streams.
---

diff --git a/charset/unistream.h b/charset/unistream.h
new file mode 100644
index 00000000..5483d809
--- /dev/null
+++ b/charset/unistream.h
@@ -0,0 +1,72 @@
+/*
+ *	The UniCode Library: Reading and writing of UTF-8 on Fastbuf Streams
+ *
+ *	(c) 2001 Martin Mares <mj@ucw.cz>
+ */
+
+#ifndef _UNISTREAM_H
+#define _UNISTREAM_H
+
+/* FIXME: Do these need to be inline? */
+
+static inline int
+bget_utf8(struct fastbuf *b)
+{
+  int c = bgetc(b);
+  int code;
+
+  if (c < 0x80)				/* Includes EOF */
+    return c;
+  if (c < 0xc0)				/* Incorrect combination */
+    return UNI_REPLACEMENT;
+  if (c >= 0xf0)			/* Too large, skip it */
+    {
+      while ((c = bgetc(b)) >= 0x80 && c < 0xc0)
+	;
+      goto wrong;
+    }
+  if (c >= 0xe0)			/* 3 bytes */
+    {
+      code = c & 0x0f;
+      if ((c = bgetc(b)) < 0x80 || c >= 0xc0)
+	goto wrong;
+      code = (code << 6) | (c & 0x3f);
+      if ((c = bgetc(b)) < 0x80 || c >= 0xc0)
+	goto wrong;
+      code = (code << 6) | (c & 0x3f);
+    }
+  else					/* 2 bytes */
+    {
+      code = c & 0x1f;
+      if ((c = bgetc(b)) < 0x80 || c >= 0xc0)
+	goto wrong;
+      code = (code << 6) | (c & 0x3f);
+    }
+  return code;
+
+ wrong:
+  if (c >= 0)
+    bungetc(b, c);
+  return UNI_REPLACEMENT;
+}
+
+static inline void
+bput_utf8(struct fastbuf *b, uns u)
+{
+  ASSERT(u < 65536);
+  if (u < 0x80)
+    bputc(b, u);
+  else
+    {
+      if (u < 0x800)
+	bputc(b, 0xc0 | (u >> 6));
+      else
+	{
+	  bputc(b, 0xe0 | (u >> 12));
+	  bputc(b, 0x80 | ((u >> 6) & 0x3f));
+	}
+      bputc(b, 0x80 | (u & 0x3f));
+    }
+}
+
+#endif