Fixed unicode handling

3 files changed, 100 insertions, 14 deletions

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index af9fafe..4e7f0b2 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -5,6 +5,7 @@ link_directories(${GEE_LIBRARY_DIRS} ${GIO_LIBRARY_DIRS} ${ERL_LIBRARY_DIRS})
 vala_precompile(EVA_C
   PacketHandler.vala
   Term.vala
+  Util.vala
 PACKAGES
   gee-1.0
   gio-2.0
diff --git a/src/Term.vala b/src/Term.vala
index 7037e48..46c03b3 100644
--- a/src/Term.vala
+++ b/src/Term.vala
@@ -37,7 +37,7 @@ namespace Eva {
       case Erl.TermType.ATOM:
         char[] value = new char[size+1];
         assert(Erl.decode_atom(buffer, ref index, value) == 0);
-        return new Atom(((string)value).ndup(size));
+        return new Atom(array_to_string(value, size));
         
       case Erl.TermType.SMALL_TUPLE:
       case Erl.TermType.LARGE_TUPLE:
@@ -68,7 +68,7 @@ namespace Eva {
       case Erl.TermType.STRING:
         char[] value = new char[size+1];
         assert(Erl.decode_string(buffer, ref index, value) == 0);
-        return new String(((string)value).ndup(size));
+        return new String(array_to_string(value, size));
         
       case Erl.TermType.BINARY:
         char[] value = new char[size];
@@ -184,11 +184,33 @@ namespace Eva {
     public string value {get; construct;}
     
     public Atom(string v) {
+      assert(v.validate());
       Object(value: v);
     }
     
     public string to_string() {
-      return value.to_string();
+      if(value.length == 0)
+        return "''";
+      
+      bool special = false;
+      unichar c = value.get_char();
+      
+      if(!c.islower())
+        special = true;
+      
+      for(unowned string rest = value.next_char(); rest.length != 0 && !special; rest = rest.next_char()) {
+        c = rest.get_char();
+        
+        if(!(c == '_' || c.isalnum()))
+          special = true;
+      }
+      
+      if(!special) {
+        return value;
+      }
+      else {
+        return "'" + value.replace("\\", "\\\\").replace("'", "\\'") + "'";
+      }
     }
     
     public bool equals(Term o) {
@@ -201,7 +223,9 @@ namespace Eva {
     }
     
     public void encode(Erl.Buffer buffer) {
-      buffer.encode_atom(value.to_utf8());
+      char[]? array = string_to_array(value);
+      assert(array != null);
+      buffer.encode_atom(array);
     }
   }
   
@@ -312,23 +336,30 @@ namespace Eva {
         string s = "";
         
         foreach(Term t in list) {
+          unichar c;
+          
           if(t is Int) {
             Int i = t as Int;
             if(i.value < 0 || i.value > 255)
               return false;
             
-            s += i.value.to_string("%c");
+            c = (unichar)i.value;
           }
           else if(t is UInt) {
             UInt i = t as UInt;
             if(i.value > 255)
               return false;
             
-            s += i.value.to_string("%c");
+            c = (unichar)i.value;
           }
           else {
             return false;
           }
+          
+          string buf = string.nfill(6, 0);
+          c.to_utf8(buf);
+          
+          s += buf;
         }
         
         return (s == (o as String).value);
@@ -382,7 +413,7 @@ namespace Eva {
     }
     
     public string to_string() {
-      return value.to_string();
+      return "\"" + value.replace("\\", "\\\\").replace("\"", "\\\"") + "\"";
     }
     
     public bool equals(Term o) {
@@ -398,7 +429,13 @@ namespace Eva {
     }
     
     public void encode(Erl.Buffer buffer) {
-      buffer.encode_string(value.to_utf8());
+      char[]? array = string_to_array(value);
+      if(array != null) {
+        buffer.encode_string(array);
+      }
+      else {
+        string_to_list(value).encode(buffer);
+      }
     }
   }
   
@@ -442,7 +479,7 @@ namespace Eva {
     public uint creation {get; construct;}
     
     public Pid(Erl.Pid pid) {
-      this.create(((string)pid.node).ndup(Erl.MAXATOMLEN), pid.num, pid.serial, pid.creation);
+      this.create(array_to_string(pid.node, Erl.MAXATOMLEN), pid.num, pid.serial, pid.creation);
     }
     
     private Pid.create(string node0, uint num0, uint serial0, uint creation0) {
@@ -464,7 +501,8 @@ namespace Eva {
     
     public void encode(Erl.Buffer buffer) {
       Erl.Pid pid = Erl.Pid();
-      char[] nodedata = node.to_utf8();
+      char[]? nodedata = string_to_array(node);
+      assert(nodedata != null);
       
       Memory.copy(pid.node, nodedata, int.min(Erl.MAXATOMLEN, nodedata.length));
       pid.node[int.min(Erl.MAXATOMLEN, nodedata.length)] = 0;
@@ -483,7 +521,7 @@ namespace Eva {
     public uint creation {get; construct;}
     
     public Port(Erl.Port port) {
-      this.create(((string)port.node).ndup(Erl.MAXATOMLEN), port.id, port.creation);
+      this.create(array_to_string(port.node, Erl.MAXATOMLEN), port.id, port.creation);
     }
     
     private Port.create(string node0, uint id0, uint creation0) {
@@ -505,7 +543,8 @@ namespace Eva {
     
     public void encode(Erl.Buffer buffer) {
       Erl.Port port = Erl.Port();
-      char[] nodedata = node.to_utf8();
+      char[]? nodedata = string_to_array(node);
+      assert(nodedata != null);
       
       Memory.copy(port.node, nodedata, int.min(Erl.MAXATOMLEN, nodedata.length));
       port.node[int.min(Erl.MAXATOMLEN, nodedata.length)] = 0;
@@ -524,7 +563,7 @@ namespace Eva {
     public uint creation {get; construct;}
     
     public Ref(Erl.Ref reference) {
-      this.create(((string)reference.node).ndup(Erl.MAXATOMLEN), reference.len, reference.n, reference.creation);
+      this.create(array_to_string(reference.node, Erl.MAXATOMLEN), reference.len, reference.n, reference.creation);
     }
     
     private Ref.create(string node0, int len0, uint* n0, uint creation0) {
@@ -560,7 +599,8 @@ namespace Eva {
     
     public void encode(Erl.Buffer buffer) {
       Erl.Ref reference = Erl.Ref();
-      char[] nodedata = node.to_utf8();
+      char[]? nodedata = string_to_array(node);
+      assert(nodedata != null);
       
       Memory.copy(reference.node, nodedata, int.min(Erl.MAXATOMLEN, nodedata.length));
       reference.node[int.min(Erl.MAXATOMLEN, nodedata.length)] = 0;
diff --git a/src/Util.vala b/src/Util.vala
new file mode 100644
index 0000000..3666f96
--- /dev/null
+++ b/src/Util.vala
@@ -0,0 +1,45 @@
+namespace Eva {
+  private string array_to_string(uchar* array, ulong len) {
+    string ret = "";
+    
+    for(ulong i = 0; i < len; ++i) {
+      if(array[i] == 0)
+        break;
+      
+      string buf = string.nfill(6, 0);
+      ((unichar)array[i]).to_utf8(buf);
+      
+      ret += buf;
+    }
+    
+    return ret;
+  }
+  
+  private char[]? string_to_array(string str) {
+    char[] ret = new char[str.len()];
+    int index = 0;
+    
+    for(unowned string rest = str; rest.length > 0; rest = rest.next_char()) {
+      unichar c = rest.get_char();
+      
+      if(c > 255)
+        return null;
+      
+      ret[index++] = (char)c;
+    }
+    
+    return ret;
+  }
+  
+  private List string_to_list(string str) {
+    List ret = new List();
+    
+    for(unowned string rest = str; rest.length > 0; rest = rest.next_char()) {
+      unichar c = rest.get_char();
+      
+      ret.list.add(new UInt(c));
+    }
+    
+    return ret;
+  }
+}