UCW: Added UTF-8 routines with custom replacement char (lib/ff-unicode.h).

diff --git a/lib/ff-unicode.c b/lib/ff-unicode.c
index 7f46aaab845cd6d58fcd369e8fd06c54120bcb0c..69a5247b0500c109c89c5dc1c36f5b06186e29e9 100644 (file)
--- a/lib/ff-unicode.c
+++ b/lib/ff-unicode.c
@@ -14,7 +14,7 @@
 #include "lib/ff-unicode.h"
 
 int
-bget_utf8_slow(struct fastbuf *b)
+bget_utf8_slow(struct fastbuf *b, uns repl)
 {
   int c = bgetc(b);
   int code;
@@ -22,7 +22,7 @@ bget_utf8_slow(struct fastbuf *b)
   if (c < 0x80)                                /* Includes EOF */
     return c;
   if (c < 0xc0)                                /* Incorrect combination */
-    return UNI_REPLACEMENT;
+    return repl;
   if (c >= 0xf0)                       /* Too large, skip it */
     {
       while ((c = bgetc(b)) >= 0x80 && c < 0xc0)
@@ -51,11 +51,11 @@ bget_utf8_slow(struct fastbuf *b)
  wrong:
   if (c >= 0)
     bungetc(b);
-  return UNI_REPLACEMENT;
+  return repl;
 }
 
 int
-bget_utf8_32_slow(struct fastbuf *b)
+bget_utf8_32_slow(struct fastbuf *b, uns repl)
 {
   int c = bgetc(b);
   int code;
@@ -64,7 +64,7 @@ bget_utf8_32_slow(struct fastbuf *b)
   if (c < 0x80)                                /* Includes EOF */
     return c;
   if (c < 0xc0)                                /* Incorrect combination */
-    return UNI_REPLACEMENT;
+    return repl;
   if (c < 0xe0)
     {
       code = c & 0x1f;
@@ -107,7 +107,7 @@ bget_utf8_32_slow(struct fastbuf *b)
  wrong:
   if (c >= 0)
     bungetc(b);
-  return UNI_REPLACEMENT;
+  return repl;
 }
 
 void

diff --git a/lib/ff-unicode.h b/lib/ff-unicode.h
index 31510ff4411b5aceb85c9c921f20ff9d8f086474..e3838b1c1a691d365b9b6c485d7ee13326e6ee81 100644 (file)
--- a/lib/ff-unicode.h
+++ b/lib/ff-unicode.h
@@ -14,47 +14,57 @@
 #include "lib/fastbuf.h"
 #include "lib/unicode.h"
 
-int bget_utf8_slow(struct fastbuf *b);
-int bget_utf8_32_slow(struct fastbuf *b);
+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 void
-bput_utf8(struct fastbuf *b, uns u)
+static inline int
+bget_utf8_32_repl(struct fastbuf *b, uns repl)
 {
-  ASSERT(u < 65536);
-  if (bavailw(b) >= 3)
-    PUT_UTF8(b->bptr, u);
+  uns u;
+  if (bavailr(b) >= 6)
+    {
+      b->bptr = utf8_32_get_repl(b->bptr, &u, repl);
+      return u;
+    }
   else
-    bput_utf8_slow(b, u);
+    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)
 {
-  uns u;
+  return bget_utf8_32_repl(b, UNI_REPLACEMENT);
+}
 
-  if (bavailr(b) >= 6)
-    {
-      GET_UTF8_32(b->bptr, u);
-      return u;
-    }
+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
-    return bget_utf8_32_slow(b);
+    bput_utf8_slow(b, u);
 }
 
 static inline void
@@ -62,7 +72,7 @@ bput_utf8_32(struct fastbuf *b, uns u)
 {
   ASSERT(u < (1U<<31));
   if (bavailw(b) >= 6)
-    PUT_UTF8_32(b->bptr, u);
+    b->bptr = utf8_32_put(b->bptr, u);
   else
     bput_utf8_32_slow(b, u);
 }