fixed SHERLOCK_VERSION

[libucw.git] / lib / ff-unicode.h

diff --git a/lib/ff-unicode.h b/lib/ff-unicode.h
index 31510ff4411b5aceb85c9c921f20ff9d8f086474..af39bbda7ab50fd706072c61a432fcdc1013aaf4 100644 (file)
--- a/lib/ff-unicode.h
+++ b/lib/ff-unicode.h
@@ -1,70 +1,144 @@
 /*
- *     UCW Library: Reading and writing of UTF-8 on Fastbuf Streams
+ *     UCW Library: Reading and writing of UTF-8 and UTF-16 on Fastbuf Streams
  *
  *     (c) 2001--2004 Martin Mares <mj@ucw.cz>
  *     (c) 2004 Robert Spalek <robert@ucw.cz>
+ *     (c) 2007--2008 Pavel Charvat <pchar@ucw.cz>
  *
  *     This software may be freely distributed and used according to the terms
  *     of the GNU Lesser General Public License.
  */
 
-#ifndef _UCW_FF_UTF8_H
-#define _UCW_FF_UTF8_H
+#ifndef _UCW_FF_UNICODE_H
+#define _UCW_FF_UNICODE_H
 
 #include "lib/fastbuf.h"
 #include "lib/unicode.h"
 
-int bget_utf8_slow(struct fastbuf *b);
-int bget_utf8_32_slow(struct fastbuf *b);
+/*** 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);
 
 static inline int
-bget_utf8(struct fastbuf *b)
+bget_utf8_repl(struct fastbuf *b, uns repl)
 {
   uns u;
-
   if (bavailr(b) >= 3)
     {
-      GET_UTF8(b->bptr, u);
+      b->bptr = utf8_get_repl(b->bptr, &u, repl);
       return u;
     }
   else
-    return bget_utf8_slow(b);
+    return bget_utf8_slow(b, repl);
+}
+
+static inline int
+bget_utf8_32_repl(struct fastbuf *b, uns repl)
+{
+  uns u;
+  if (bavailr(b) >= 6)
+    {
+      b->bptr = utf8_32_get_repl(b->bptr, &u, repl);
+      return u;
+    }
+  else
+    return bget_utf8_32_slow(b, repl);
+}
+
+static inline int
+bget_utf8(struct fastbuf *b)
+{
+  return bget_utf8_repl(b, UNI_REPLACEMENT);
+}
+
+static inline int
+bget_utf8_32(struct fastbuf *b)
+{
+  return bget_utf8_32_repl(b, UNI_REPLACEMENT);
 }
 
 static inline void
 bput_utf8(struct fastbuf *b, uns u)
 {
-  ASSERT(u < 65536);
   if (bavailw(b) >= 3)
-    PUT_UTF8(b->bptr, u);
+    b->bptr = utf8_put(b->bptr, u);
   else
     bput_utf8_slow(b, u);
 }
 
+static inline void
+bput_utf8_32(struct fastbuf *b, uns u)
+{
+  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_utf8_32(struct fastbuf *b)
+bget_utf16_be_repl(struct fastbuf *b, uns repl)
 {
   uns u;
+  if (bavailr(b) >= 4)
+    {
+      b->bptr = utf16_be_get_repl(b->bptr, &u, repl);
+      return u;
+    }
+  else
+    return bget_utf16_be_slow(b, repl);
+}
 
-  if (bavailr(b) >= 6)
+static inline int
+bget_utf16_le_repl(struct fastbuf *b, uns repl)
+{
+  uns u;
+  if (bavailr(b) >= 4)
     {
-      GET_UTF8_32(b->bptr, u);
+      b->bptr = utf16_le_get_repl(b->bptr, &u, repl);
       return u;
     }
   else
-    return bget_utf8_32_slow(b);
+    return bget_utf16_le_slow(b, repl);
+}
+
+static inline int
+bget_utf16_be(struct fastbuf *b)
+{
+  return bget_utf16_be_repl(b, UNI_REPLACEMENT);
+}
+
+static inline int
+bget_utf16_le(struct fastbuf *b)
+{
+  return bget_utf16_le_repl(b, UNI_REPLACEMENT);
 }
 
 static inline void
-bput_utf8_32(struct fastbuf *b, uns u)
+bput_utf16_be(struct fastbuf *b, uns u)
 {
-  ASSERT(u < (1U<<31));
-  if (bavailw(b) >= 6)
-    PUT_UTF8_32(b->bptr, u);
+  if (bavailw(b) >= 4)
+    b->bptr = utf16_be_put(b->bptr, u);
   else
-    bput_utf8_32_slow(b, u);
+    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