From: Pavel Charvat <pavel.charvat@netcentrum.cz>
Date: Mon, 10 Dec 2007 11:36:50 +0000 (+0100)
Subject: UCW: Implemented bput_utf16_*.
X-Git-Tag: holmes-import~488
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=abd00be4bf89ed0ad7bdecfa1b0130f0b1994c42;p=libucw.git

UCW: Implemented bput_utf16_*.
---

diff --git a/lib/ff-unicode.c b/lib/ff-unicode.c
index 047fff84..280b3b75 100644
--- a/lib/ff-unicode.c
+++ b/lib/ff-unicode.c
@@ -14,6 +14,8 @@
 #include "lib/ff-unicode.h"
 #include "lib/ff-binary.h"
 
+/*** UTF-8 ***/
+
 int
 bget_utf8_slow(struct fastbuf *b, uns repl)
 {
@@ -167,6 +169,8 @@ bput_utf8_32_slow(struct fastbuf *b, uns u)
     }
 }
 
+/*** UTF-16 ***/
+
 int
 bget_utf16_be_slow(struct fastbuf *b, uns repl)
 {
@@ -196,3 +200,41 @@ bget_utf16_le_slow(struct fastbuf *b, uns repl)
     return repl;
   return 0x10000 + (x << 10) + y;
 }
+
+void
+bput_utf16_be_slow(struct fastbuf *b, uns u)
+{
+  if (u < 0xd800 || (u < 0x10000 && u >= 0xe000))
+    {
+      bputc(b, u >> 8);
+      bputc(b, u & 0xff);
+    }
+  else if ((u -= 0x10000) < 0x100000)
+    {
+      bputc(b, 0xd8 | (u >> 18));
+      bputc(b, (u >> 10) & 0xff);
+      bputc(b, 0xdc | ((u >> 8) & 0x3));
+      bputc(b, u & 0xff);
+    }
+  else
+    ASSERT(0);
+}
+
+void
+bput_utf16_le_slow(struct fastbuf *b, uns u)
+{
+  if (u < 0xd800 || (u < 0x10000 && u >= 0xe000))
+    {
+      bputc(b, u & 0xff);
+      bputc(b, u >> 8);
+    }
+  else if ((u -= 0x10000) < 0x100000)
+    {
+      bputc(b, (u >> 10) & 0xff);
+      bputc(b, 0xd8 | (u >> 18));
+      bputc(b, u & 0xff);
+      bputc(b, 0xdc | ((u >> 8) & 0x3));
+    }
+  else
+    ASSERT(0);
+}
diff --git a/lib/ff-unicode.h b/lib/ff-unicode.h
index f577c635..93f8a06e 100644
--- a/lib/ff-unicode.h
+++ b/lib/ff-unicode.h
@@ -14,12 +14,12 @@
 #include "lib/fastbuf.h"
 #include "lib/unicode.h"
 
+/*** UTF-8 ***/
+
 int bget_utf8_slow(struct fastbuf *b, uns repl);
 int bget_utf8_32_slow(struct fastbuf *b, uns repl);
 void bput_utf8_slow(struct fastbuf *b, uns u);
 void bput_utf8_32_slow(struct fastbuf *b, uns u);
-int bget_utf16_be_slow(struct fastbuf *b, uns repl);
-int bget_utf16_le_slow(struct fastbuf *b, uns repl);
 
 static inline int
 bget_utf8_repl(struct fastbuf *b, uns repl)
@@ -62,7 +62,6 @@ bget_utf8_32(struct fastbuf *b)
 static inline void
 bput_utf8(struct fastbuf *b, uns u)
 {
-  ASSERT(u < 65536);
   if (bavailw(b) >= 3)
     b->bptr = utf8_put(b->bptr, u);
   else
@@ -72,13 +71,19 @@ bput_utf8(struct fastbuf *b, uns u)
 static inline void
 bput_utf8_32(struct fastbuf *b, uns u)
 {
-  ASSERT(u < (1U<<31));
   if (bavailw(b) >= 6)
     b->bptr = utf8_32_put(b->bptr, u);
   else
     bput_utf8_32_slow(b, u);
 }
 
+/*** UTF-16 ***/
+
+int bget_utf16_be_slow(struct fastbuf *b, uns repl);
+int bget_utf16_le_slow(struct fastbuf *b, uns repl);
+void bput_utf16_be_slow(struct fastbuf *b, uns u);
+void bput_utf16_le_slow(struct fastbuf *b, uns u);
+
 static inline int
 bget_utf16_be_repl(struct fastbuf *b, uns repl)
 {
@@ -117,4 +122,22 @@ bget_utf16_le(struct fastbuf *b)
   return bget_utf16_le_repl(b, UNI_REPLACEMENT);
 }
 
+static inline void
+bput_utf16_be(struct fastbuf *b, uns u)
+{
+  if (bavailw(b) >= 4)
+    b->bptr = utf16_be_put(b->bptr, u);
+  else
+    bput_utf16_be_slow(b, u);
+}
+
+static inline void
+bput_utf16_lbe(struct fastbuf *b, uns u)
+{
+  if (bavailw(b) >= 4)
+    b->bptr = utf16_le_put(b->bptr, u);
+  else
+    bput_utf16_le_slow(b, u);
+}
+
 #endif
diff --git a/lib/unicode.h b/lib/unicode.h
index fb8722d0..9a3fe07a 100644
--- a/lib/unicode.h
+++ b/lib/unicode.h
@@ -239,8 +239,8 @@ utf16_le_put(void *p, uns u)
     }
   else if ((u -= 0x10000) < 0x100000)
     {
-      put_u16_le(p, u >> 10);
-      put_u16_le(p + 2, u & 0x7ff);
+      put_u16_le(p, 0xd800 | (u >> 10));
+      put_u16_le(p + 2, 0xdc00 | (u & 0x3ff));
       return p + 4;
     }
   else
@@ -257,8 +257,8 @@ utf16_be_put(void *p, uns u)
     }
   else if ((u -= 0x10000) < 0x100000)
     {
-      put_u16_be(p, u >> 10);
-      put_u16_be(p + 2, u & 0x7ff);
+      put_u16_be(p, 0xd800 | (u >> 10));
+      put_u16_be(p + 2, 0xdc00 | (u & 0x3ff));
       return p + 4;
     }
   else