[PATCH] id3v2: add support for UTF-16 encoding.

[Anton Khirnov]

---
 libavformat/id3v2.c |   34 ++++++++++++++++++++++++++++++++++
 libavutil/common.h  |   26 ++++++++++++++++++++++++++
 2 files changed, 60 insertions(+), 0 deletions(-)

diff --git a/libavformat/id3v2.c b/libavformat/id3v2.c
index 0cf2cb1..fe79c01 100644
--- a/libavformat/id3v2.c
+++ b/libavformat/id3v2.c
@@ -81,6 +81,7 @@ static void read_ttag(AVFormatContext *s, int taglen, const char *key)
     char *q, dst[512];
     int len, dstlen = sizeof(dst) - 1;
     unsigned genre;
+    unsigned int (*get)(ByteIOContext*) = NULL;
 
     dst[0] = 0;
     if (taglen < 1)
@@ -99,11 +100,44 @@ static void read_ttag(AVFormatContext *s, int taglen, const char *key)
         *q = '\0';
         break;
 
+    case 1:  /* UTF-16 with BOM */
+        taglen -= 2;
+        switch (get_be16(s->pb)) {
+        case 0xfeff:
+            get = get_be16;
+            break;
+        case 0xfffe:
+            get = get_le16;
+            break;
+        default:
+            av_log(s, AV_LOG_ERROR, "Incorrect BOM value.\n");
+            return;
+        }
+        // fall-through
+
+    case 2:  /* UTF-16BE without BOM */
+        if (!get)
+            get = get_be16;
+
+        q = dst;
+        while (taglen > 1) {
+            uint32_t ch;
+            uint8_t tmp;
+
+            GET_UTF16(ch, get(s->pb), break;, len)
+            PUT_UTF8(ch, tmp, if (q - dst < dstlen -1) *q++ = tmp;)
+            taglen -= len;
+        }
+        *q = '\0';
+        break;
+
     case 3:  /* UTF-8 */
         len = FFMIN(taglen, dstlen - 1);
         get_buffer(s->pb, dst, len);
         dst[len] = 0;
         break;
+    default:
+        av_log(s, AV_LOG_WARNING, "Unknown encoding in tag %s\n.", key);
     }
 
     if (!strcmp(key, "genre")
diff --git a/libavutil/common.h b/libavutil/common.h
index 0797a79..458e32f 100644
--- a/libavutil/common.h
+++ b/libavutil/common.h
@@ -266,6 +266,32 @@ static inline av_const int av_ceil_log2(int x)
     }
 
 /*!
+ * \def GET_UTF16(val, GET_BYTE, ERROR)
+ * Converts a UTF-16 character (2 or 4 bytes) to its 32-bit UCS-4 encoded form
+ * \param val is the output and should be of type uint32_t. It holds the converted
+ * UCS-4 character and should be a left value.
+ * \param GET_BYTE gets UTF-16 encoded bytes from any proper source. It can be
+ * a function or a statement whose return value or evaluated value is of type
+ * uint16_t. It will be executed up to 2 times.
+ * \param ERROR action that should be taken when an invalid UTF-16 surrogate is
+ * returned from GET_BYTE. It should be a statement that jumps out of the macro,
+ * like exit(), goto, return, break, or continue.
+ * \param read should be an int and is set to the number of bytes read (2 or 4).
+ */
+#define GET_UTF16(val, GET_BYTE, ERROR, read)\
+    {\
+        val = GET_BYTE;\
+        read = 2;\
+        if (val >= 0xD800 && val <= 0xDBFF) {\
+            val = ((val - 0xD800)<<10) + (get(s->pb) - 0xDC00) + 0x0010000UL;\
+            read += 2;\
+        }\
+        else if (val > 0xDBFF && val <= 0xDFFF) {\
+            ERROR\
+        }\
+    }
+
+/*!
  * \def PUT_UTF8(val, tmp, PUT_BYTE)
  * Converts a 32-bit Unicode character to its UTF-8 encoded form (up to 4 bytes long).
  * \param val is an input-only argument and should be of type uint32_t. It holds
-- 
1.6.3.3


--cNdxnHkX5QqsyA0e--