sub: libguess support for -subcp

Actually this is rather disappointing.

DOCS/man/en/options.rst

@@ -2044,6 +2044,15 @@
     - ``--subcp=enca:pl`` guess the encoding for Polish, fall back on UTF-8.
     - ``--subcp=enca`` try universal detection, fall back on UTF-8.
 
+    If the player was compiled with libguess support you can use it with:
+
+    ``--subcp=guess:<language>:<fallback codepage>``
+
+    Note that libguess always needs a language. There is no universal detection
+    mode. Use ``--subcp=guess:help`` to get a list of languages (like with ENCA,
+    it will be printed only if the conversion code is somehow called, for
+    example when loading an external subtitle).
+
 --sub-delay=<sec>
     Delays subtitles by <sec> seconds. Can be negative.
 

configure

@@ -292,6 +292,7 @@ Installation directories:
 
 Optional features:
   --disable-encoding     disable encoding functionality [enable]
+  --disable-libguess     disable libguess [autodetect]
   --enable-termcap       use termcap database for key codes [autodetect]
   --enable-termios       use termios database for key codes [autodetect]
   --disable-iconv        disable iconv for encoding conversion [autodetect]
@@ -463,6 +464,7 @@ networking=yes
 _winsock2_h=auto
 _smb=auto
 _libquvi=auto
+_libguess=auto
 _joystick=no
 _lirc=auto
 _lircc=auto
@@ -663,6 +665,8 @@ for ac_option do
   --disable-smb)        _smb=no ;;
   --enable-libquvi)     _libquvi=yes    ;;
   --disable-libquvi)    _libquvi=no     ;;
+  --enable-libguess)    _libguess=yes   ;;
+  --disable-libguess)   _libguess=no    ;;
   --enable-joystick)    _joystick=yes   ;;
   --disable-joystick)   _joystick=no    ;;
   --enable-libav)       ffmpeg=yes      ;;
@@ -1685,6 +1689,21 @@ else
 fi
 
 
+echocheck "libguess support"
+if test "$_libguess" = auto ; then
+    _libguess=no
+    if pkg_config_add 'libguess >= 1.0' ; then
+        _libguess=yes
+    fi
+fi
+if test "$_libguess" = yes; then
+    def_libguess="#define CONFIG_LIBGUESS 1"
+else
+    def_libguess="#undef CONFIG_LIBGUESS"
+fi
+echores "$_libguess"
+
+
 echocheck "Samba support (libsmbclient)"
 if test "$_smb" = yes; then
   libs_mplayer="$libs_mplayer -lsmbclient"
@@ -3128,6 +3147,7 @@ VF_LAVFI = $vf_lavfi
 AF_LAVFI = $af_lavfi
 LIBSMBCLIENT = $_smb
 LIBQUVI = $_libquvi
+LIBGUESS = $_libguess
 LIBTHEORA = $_theora
 LIRC = $_lirc
 MACOSX_BUNDLE = $_macosx_bundle
@@ -3326,6 +3346,7 @@ $def_inet_pton
 $def_networking
 $def_smb
 $def_libquvi
+$def_libguess
 $def_socklen_t
 $def_vstream
 

core/charset_conv.c

@@ -31,6 +31,10 @@
 #include <enca.h>
 #endif
 
+#ifdef CONFIG_LIBGUESS
+#include <libguess.h>
+#endif
+
 #ifdef CONFIG_ICONV
 #include <iconv.h>
 #endif
@@ -67,7 +71,8 @@ bool mp_charset_requires_guess(const char *user_cp)
 {
     bstr res[2] = {{0}};
     split_colon(user_cp, 2, res);
-    return bstrcasecmp0(res[0], "enca") == 0;
+    return bstrcasecmp0(res[0], "enca") == 0 ||
+           bstrcasecmp0(res[0], "guess") == 0;
 }
 
 #ifdef CONFIG_ENCA
@@ -102,6 +107,23 @@ static const char *enca_guess(bstr buf, const char *language)
 }
 #endif
 
+#ifdef CONFIG_LIBGUESS
+static const char *libguess_guess(bstr buf, const char *language)
+{
+    if (libguess_validate_utf8(buf.start, buf.len))
+        return "UTF-8";
+
+    if (!language || !language[0] || strcmp(language, "help") == 0) {
+        mp_msg(MSGT_SUBREADER, MSGL_ERR, "libguess needs a language: "
+               "japanese taiwanese chinese korean russian arabic turkish "
+               "greek hebrew polish baltic\n");
+        return NULL;
+    }
+
+    return libguess_determine_encoding(buf.start, buf.len, language);
+}
+#endif
+
 // Runs charset auto-detection on the input buffer, and returns the result.
 // If auto-detection fails, NULL is returned.
 // If user_cp doesn't refer to any known auto-detection (for example because
@@ -126,6 +148,10 @@ const char *mp_charset_guess(bstr buf, const char *user_cp)
     if (bstrcasecmp0(type, "enca") == 0)
         res = enca_guess(buf, lang);
 #endif
+#ifdef CONFIG_LIBGUESS
+    if (bstrcasecmp0(type, "guess") == 0)
+        res = libguess_guess(buf, lang);
+#endif
 
     if (res) {
         mp_msg(MSGT_SUBREADER, MSGL_DBG2, "%.*s detected charset: '%s'\n",