UCW: Added bget_utf16_{le,be} routines.

diff --git a/lib/ff-unicode.c b/lib/ff-unicode.c
index 69a5247b0500c109c89c5dc1c36f5b06186e29e9..047fff84cf16bb57d05748d38b2f9d8c6e6dfe6a 100644 (file)
--- a/lib/ff-unicode.c
+++ b/lib/ff-unicode.c
@@ -12,6 +12,7 @@
 #include "lib/fastbuf.h"
 #include "lib/unicode.h"
 #include "lib/ff-unicode.h"
+#include "lib/ff-binary.h"
 
 int
 bget_utf8_slow(struct fastbuf *b, uns repl)
@@ -165,3 +166,33 @@ bput_utf8_32_slow(struct fastbuf *b, uns u)
       bputc(b, 0x80 | (u & 0x3f));
     }
 }
+
+int
+bget_utf16_be_slow(struct fastbuf *b, uns repl)
+{
+  if (bpeekc(b) < 0)
+    return -1;
+  uns u = bgetw_be(b), x, y;
+  if ((int)u < 0)
+    return repl;
+  if ((x = u - 0xd800) >= 0x800)
+    return u;
+  if (x >= 0x400 || bpeekc(b) < 0 || (y = bgetw_be(b) - 0xdc00) >= 0x400)
+    return repl;
+  return 0x10000 + (x << 10) + y;
+}
+
+int
+bget_utf16_le_slow(struct fastbuf *b, uns repl)
+{
+  if (bpeekc(b) < 0)
+    return -1;
+  uns u = bgetw_le(b), x, y;
+  if ((int)u < 0)
+    return repl;
+  if ((x = u - 0xd800) >= 0x800)
+    return u;
+  if (x >= 0x400 || bpeekc(b) < 0 || (y = bgetw_le(b) - 0xdc00) >= 0x400)
+    return repl;
+  return 0x10000 + (x << 10) + y;
+}

diff --git a/lib/ff-unicode.h b/lib/ff-unicode.h
index e3838b1c1a691d365b9b6c485d7ee13326e6ee81..f577c6359685c42a685715f089a24f4d5f6e1fd6 100644 (file)
--- a/lib/ff-unicode.h
+++ b/lib/ff-unicode.h
@@ -18,6 +18,8 @@ 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)
@@ -77,4 +79,42 @@ bput_utf8_32(struct fastbuf *b, uns u)
     bput_utf8_32_slow(b, u);
 }
 
+static inline int
+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);
+}
+
+static inline int
+bget_utf16_le_repl(struct fastbuf *b, uns repl)
+{
+  uns u;
+  if (bavailr(b) >= 4)
+    {
+      b->bptr = utf16_le_get_repl(b->bptr, &u, repl);
+      return u;
+    }
+  else
+    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);
+}
+
 #endif

diff --git a/sherlock/xml/xml.c b/sherlock/xml/xml.c
index b365c683d12cd9ebfaf47fda88c72d866bfc5513..73b3a92f32e0b3b693a716138e84c44e0feb3fbf 100644 (file)
--- a/sherlock/xml/xml.c
+++ b/sherlock/xml/xml.c
@@ -110,64 +110,6 @@ xml_char_cat(uns c)
     return 1;
 }
 
-/*** Generic UTF decoding ***/
-
-static uns
-bget_utf16_le_slow(struct fastbuf *fb, uns repl)
-{
-  if ((int)bpeekc(fb) < 0)
-    return ~0U;
-  uns u = bgetw_le(fb), x, y;
-  if ((int)u < 0)
-    return repl;
-  if ((x = u - 0xd800) >= 0x800)
-    return u;
-  if (x >= 0x400 || (int)bpeekc(fb) < 0 || (y = bgetw_le(fb) - 0xdc00) >= 0x400)
-    return repl;
-  return 0x10000 + (x << 10) + y;
-}
-
-static uns
-bget_utf16_be_slow(struct fastbuf *fb, uns repl)
-{
-  if ((int)bpeekc(fb) < 0)
-    return ~0U;
-  uns u = bgetw_be(fb), x, y;
-  if ((int)u < 0)
-    return repl;
-  if ((x = u - 0xd800) >= 0x800)
-    return u;
-  if (x >= 0x400 || (int)bpeekc(fb) < 0 || (y = bgetw_be(fb) - 0xdc00) >= 0x400)
-    return repl;
-  return 0x10000 + (x << 10) + y;
-}
-
-static inline uns
-bget_utf16_le_repl(struct fastbuf *fb, uns repl)
-{
-  uns u;
-  if (bavailr(fb) >= 4)
-    {
-      fb->bptr = utf16_le_get_repl(fb->bptr, &u, repl);
-      return u;
-    }
-  else
-    return bget_utf16_le_slow(fb, repl);
-}
-
-static inline uns
-bget_utf16_be_repl(struct fastbuf *fb, uns repl)
-{
-  uns u;
-  if (bavailr(fb) >= 4)
-    {
-      fb->bptr = utf16_be_get_repl(fb->bptr, &u, repl);
-      return u;
-    }
-  else
-    return bget_utf16_be_slow(fb, repl);
-}
-
 /*** Memory management ***/
 
 static void NONRET