# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1481648253 -3600
# Node ID b8782209d44595cf33317ea0e3ac60dc31d1fe16
# Parent  a39afb95fb2610a1085979c6671f3ec4a8be234c
Introduced initial support for Unicode constants and literals.

diff -r a39afb95fb26 -r b8782209d445 common.py
--- a/common.py	Tue Dec 13 00:59:23 2016 +0100
+++ b/common.py	Tue Dec 13 17:57:33 2016 +0100
@@ -218,6 +218,14 @@
 
     # Constant and literal recording.
 
+    def get_constant_value(self, value):
+
+        "Encode the 'value' if appropriate."
+
+        if isinstance(value, unicode):
+            value = value.encode("utf-8")
+        return value
+
     def get_constant_reference(self, ref, value):
 
         "Return a constant reference for the given 'ref' type and 'value'."
@@ -923,13 +931,25 @@
 
     # NOTE: This makes assumptions about the __builtins__ structure.
 
-    return name == "string" and "str" or name == "NoneType" and "none" or name
+    if name == "string":
+        return "str"
+    elif name == "utf8string":
+        return "unicode"
+    elif name == "NoneType":
+        return "none"
+    else:
+        return name
 
 def get_builtin_type(name):
 
     "Return the type name provided by the given Python value 'name'."
 
-    return name == "str" and "string" or name
+    if name == "str":
+        return "string"
+    elif name == "unicode":
+        return "utf8string"
+    else:
+        return name
 
 # Useful data.
 
diff -r a39afb95fb26 -r b8782209d445 inspector.py
--- a/inspector.py	Tue Dec 13 00:59:23 2016 +0100
+++ b/inspector.py	Tue Dec 13 17:57:33 2016 +0100
@@ -1406,7 +1406,8 @@
         # Constant values are independently recorded.
 
         else:
-            return self.get_constant_reference(ref, n.value)
+            value = self.get_constant_value(n.value)
+            return self.get_constant_reference(ref, value)
 
     # Special names.
 
diff -r a39afb95fb26 -r b8782209d445 tests/unicode.py
--- a/tests/unicode.py	Tue Dec 13 00:59:23 2016 +0100
+++ b/tests/unicode.py	Tue Dec 13 17:57:33 2016 +0100
@@ -4,15 +4,29 @@
 
 # Print bytes.
 
-s = "æøå"
+s = b"æøå"
 print s                             # æøå
 
 # Obtain text and print it.
 
+# Explicitly from bytes.
+
 u = unicode(s, "ISO-8859-1")
 print u                             # æøå
 print u.encode("ISO-8859-1")        # æøå
 
+# Explicitly from Unicode literals.
+
+u2 = u"æøå"
+print u2                            # æøå
+print u2.encode("ISO-8859-1")       # æøå
+
+# Implicitly from string literals.
+
+#u3 = "æøå"
+#print u3                            # æøå
+#print u3.encode("ISO-8859-1")       # æøå
+
 # Inspect and update the encoding of stdout.
 
 print sys.stdout.encoding           # None
diff -r a39afb95fb26 -r b8782209d445 translator.py
--- a/translator.py	Tue Dec 13 00:59:23 2016 +0100
+++ b/translator.py	Tue Dec 13 17:57:33 2016 +0100
@@ -421,12 +421,13 @@
         if name in ("dict", "list", "tuple"):
             return self.process_literal_sequence_node(n, name, ref, TrLiteralSequenceRef)
         else:
+            value = self.get_constant_value(n.value)
             path = self.get_namespace_path()
-            local_number = self.importer.all_constants[path][n.value]
+            local_number = self.importer.all_constants[path][value]
             constant_name = "$c%d" % local_number
             objpath = self.get_object_path(constant_name)
             number = self.optimiser.constant_numbers[objpath]
-            return TrConstantValueRef(constant_name, ref.instance_of(), n.value, number)
+            return TrConstantValueRef(constant_name, ref.instance_of(), value, number)
 
     # Namespace translation.