Convert labels from utf-8 to EBU

5 files changed, 89 insertions, 19 deletions

diff --git a/doc/example.mux b/doc/example.mux
index 556a4c9..632772d 100644
--- a/doc/example.mux
+++ b/doc/example.mux
@@ -81,8 +81,13 @@ ensemble {
     ; automatic calculation of the local time offset, set the environment variable TZ
     ; to your timezone (e.g. TZ=Europe/Rome) before you launch ODR-DabMux
 
-    ; all labels are maximum 16 characters in length
+    ; All labels are maximum 16 characters in length.
+    ; Labels that are valid utf-8 will be converted to EBU Latin Character set
+    ; as defined in ETSI TS 101 756, in Annex C. If it's not valid utf-8, the
+    ; label is taken as-is, byte per byte. Characters that cannot be
+    ; represented in EBU Latin will be replaced by a space character.
     label "OpenDigitalRadio"
+
     ; The short label is built from the label by erasing letters, and cannot
     ; be longer than 8 characters. If omitted, it will be truncated from the
     ; label
@@ -95,8 +100,8 @@ services {
     ; used throughout the configuration file and for the RC.
     srv-fu {
         id 0x8daa
-        label "Funk"
-        ; you can define a shortlabel too.
+        label "Fünk"
+        ; You can define a shortlabel too.
     }
     srv-ri {
         id 0x8dab
diff --git a/lib/charset/charset.cpp b/lib/charset/charset.cpp
index 2ee14f8..5cc773d 100644
--- a/lib/charset/charset.cpp
+++ b/lib/charset/charset.cpp
@@ -52,3 +52,23 @@ const char* utf8_encoded_EBU_Latin[CHARSET_TABLE_ENTRIES] = {
 "Â", "Ä", "Ê", "Ë", "Î", "Ï", "Ô", "Ö", "Û", "Ü", "ř", "č", "š", "ž", "đ", "ŀ",
 "Ã", "Å", "Æ", "Œ", "ŷ", "Ý", "Õ", "Ø", "Þ", "Ŋ", "Ŕ", "Ć", "Ś", "Ź", "Ť", "ð",
 "ã", "å", "æ", "œ", "ŵ", "ý", "õ", "ø", "þ", "ŋ", "ŕ", "ć", "ś", "ź", "ť", "ħ"};
+
+std::string CharsetConverter::convert_ebu_to_utf8(const std::string& str)
+{
+    using namespace std;
+
+    string utf8_str;
+    for (const uint8_t c : str) {
+        // Table offset because NUL is not represented
+        if (c >= CHARSET_TABLE_OFFSET) {
+            string utf8_char(utf8_encoded_EBU_Latin[c - CHARSET_TABLE_OFFSET]);
+            utf8_str += utf8_char;
+        }
+        else {
+            utf8_str += "⁇";
+        }
+    }
+
+    return utf8_str;
+}
+
diff --git a/lib/charset/charset.h b/lib/charset/charset.h
index 6ff19bc..78dc94e 100644
--- a/lib/charset/charset.h
+++ b/lib/charset/charset.h
@@ -81,9 +81,14 @@ class CharsetConverter
             return encoded_line;
         }
 
-    private:
+        /* Convert a EBU Latin byte stream to a UTF-8 encoded string.
+         * Invalid input characters are converted to ⁇ (unicode U+2047).
+         */
+        std::string convert_ebu_to_utf8(const std::string& str);
 
+    private:
         std::vector<uint32_t> m_conversion_table;
 };
 
+
 #endif
diff --git a/src/MuxElements.cpp b/src/MuxElements.cpp
index cdfcc42..a31791a 100644
--- a/src/MuxElements.cpp
+++ b/src/MuxElements.cpp
@@ -30,6 +30,7 @@
 #include <algorithm>
 
 #include "MuxElements.h"
+#include "lib/charset/charset.h"
 #include <boost/algorithm/string.hpp>
 #include <boost/format.hpp>
 
@@ -44,7 +45,7 @@ const unsigned short Sub_Channel_SizeTable[64] = {
     232, 280, 160, 208, 280, 192, 280, 416
 };
 
-
+static CharsetConverter charset_converter;
 
 using namespace std;
 
@@ -177,7 +178,14 @@ int DabLabel::setLabel(const std::string& label)
 
     m_flag = 0xFF00; // truncate the label to the eight first characters
 
-    m_label = label;
+    try {
+        m_label = charset_converter.convert(label);
+    }
+    catch (const utf8::exception& e) {
+        etiLog.level(warn) << "Failed to convert label '" << label <<
+            "' to EBU Charset";
+        m_label = label;
+    }
 
     return 0;
 }
@@ -185,20 +193,43 @@ int DabLabel::setLabel(const std::string& label)
 int DabLabel::setLabel(const std::string& label, const std::string& short_label)
 {
     DabLabel newlabel;
+    newlabel.m_flag = 0xFF00;
+
+    try {
+        newlabel.m_label = charset_converter.convert(label);
+
+        int flag = newlabel.setShortLabel(charset_converter.convert(short_label));
+        if (flag < 0) {
+            return flag;
+        }
+
+        m_flag = flag & 0xFFFF;
+    }
+    catch (const utf8::exception& e) {
+        etiLog.level(warn) << "Failed to convert label '" << label <<
+            " or short label '" << short_label << "' to EBU Charset";
+
+        // Use label as-is
 
-    int result = newlabel.setLabel(label);
-    if (result < 0)
-        return result;
+        newlabel.m_label = label;
+        newlabel.m_flag = 0xFF00;
 
-    /* First check if we can actually create the short label */
-    int flag = newlabel.setShortLabel(short_label);
-    if (flag < 0)
-        return flag;
+        int result = newlabel.setLabel(label);
+        if (result < 0) {
+            return result;
+        }
+
+        /* First check if we can actually create the short label */
+        int flag = newlabel.setShortLabel(short_label);
+        if (flag < 0) {
+            return flag;
+        }
+
+        m_flag = flag & 0xFFFF;
+    }
 
     // short label is valid.
-    m_flag = flag & 0xFFFF;
     m_label = newlabel.m_label;
-
     return 0;
 }
 
@@ -258,6 +289,11 @@ int DabLabel::setShortLabel(const std::string& slabel)
     return flag;
 }
 
+const string DabLabel::long_label() const
+{
+    return charset_converter.convert_ebu_to_utf8(m_label);
+}
+
 const string DabLabel::short_label() const
 {
     stringstream shortlabel;
@@ -267,7 +303,7 @@ const string DabLabel::short_label() const
         }
     }
 
-    return shortlabel.str();
+    return charset_converter.convert_ebu_to_utf8(shortlabel.str());
 }
 
 void DabLabel::writeLabel(uint8_t* buf) const
diff --git a/src/MuxElements.h b/src/MuxElements.h
index ba9941f..1f3f548 100644
--- a/src/MuxElements.h
+++ b/src/MuxElements.h
@@ -147,7 +147,11 @@ struct dabOutput {
 class DabLabel
 {
     public:
-        /* Set a new label and short label.
+        /* Set a new label and short label. If the label parses as valid UTF-8, it
+         * will be converted to EBU Latin. If utf-8 decoding fails, the label
+         * will be used as is. Characters that cannot be converted are replaced
+         * by a space.
+         *
          * returns:  0 on success
          *          -1 if the short_label is not a representable
          *          -2 if the short_label is too long
@@ -170,7 +174,7 @@ class DabLabel
         void writeLabel(uint8_t* buf) const;
 
         uint16_t flag() const { return m_flag; }
-        const std::string long_label() const { return m_label; }
+        const std::string long_label() const;
         const std::string short_label() const;
 
     private:
@@ -182,7 +186,7 @@ class DabLabel
         /* The m_label is not padded in any way */
         std::string m_label;
 
-        /* Checks and calculates the flag */
+        /* Checks and calculates the flag. slabel must be EBU Latin Charset */
         int setShortLabel(const std::string& slabel);
 };