Issue 2415: Tolerate unicode strings when using intern()

4 files changed, 39 insertions, 18 deletions

diff --git a/src/engine/SCons/Node/FS.py b/src/engine/SCons/Node/FS.py
index bd8314b..c34cfed 100644
--- a/src/engine/SCons/Node/FS.py
+++ b/src/engine/SCons/Node/FS.py
@@ -588,22 +588,22 @@ class Base(SCons.Node.Node):
 
         # Filenames and paths are probably reused and are intern'ed to
         # save some memory.
-        self.name = intern(name)
-        self.suffix = intern(SCons.Util.splitext(name)[1])
+        self.name = SCons.Util.silent_intern(name)
+        self.suffix = SCons.Util.silent_intern(SCons.Util.splitext(name)[1])
         self.fs = fs
 
         assert directory, "A directory must be provided"
 
-        self.abspath = intern(directory.entry_abspath(name))
-        self.labspath = intern(directory.entry_labspath(name))
+        self.abspath = SCons.Util.silent_intern(directory.entry_abspath(name))
+        self.labspath = SCons.Util.silent_intern(directory.entry_labspath(name))
         if directory.path == '.':
-            self.path = intern(name)
+            self.path = SCons.Util.silent_intern(name)
         else:
-            self.path = intern(directory.entry_path(name))
+            self.path = SCons.Util.silent_intern(directory.entry_path(name))
         if directory.tpath == '.':
-            self.tpath = intern(name)
+            self.tpath = SCons.Util.silent_intern(name)
         else:
-            self.tpath = intern(directory.entry_tpath(name))
+            self.tpath = SCons.Util.silent_intern(directory.entry_tpath(name))
         self.path_elements = directory.path_elements + [self]
 
         self.dir = directory
diff --git a/src/engine/SCons/Scanner/__init__.py b/src/engine/SCons/Scanner/__init__.py
index d0c7d3c..cbf829c 100644
--- a/src/engine/SCons/Scanner/__init__.py
+++ b/src/engine/SCons/Scanner/__init__.py
@@ -358,10 +358,7 @@ class Classic(Current):
             includes = self.find_include_names (node)
             # Intern the names of the include files. Saves some memory
             # if the same header is included many times.
-            try:
-                node.includes = map(intern, includes)
-            except TypeError:
-                node.includes = includes
+            node.includes = map(SCons.Util.silent_intern, includes)
 
         # This is a hand-coded DSU (decorate-sort-undecorate, or
         # Schwartzian transform) pattern.  The sort key is the raw name
@@ -405,12 +402,7 @@ class ClassicCPP(Classic):
 
         n = SCons.Node.FS.find_file(include[1], paths)
 
-        i = include[1]
-        try:
-            i = intern(i)
-        except TypeError:
-            # Probably a unicode object; just don't worry about intern().
-            pass
+        i = SCons.Util.silent_intern(include[1])
         return n, i
 
     def sort_key(self, include):
diff --git a/src/engine/SCons/Util.py b/src/engine/SCons/Util.py
index ac949a0..ebdd1cd 100644
--- a/src/engine/SCons/Util.py
+++ b/src/engine/SCons/Util.py
@@ -1569,6 +1569,27 @@ def MD5collect(signatures):
 
 
 
+# Wrap the intern() function so it doesn't throw exceptions if ineligible
+# arguments are passed. The intern() function was moved into the sys module in
+# Python 3.
+try:
+    intern
+except NameError:
+    from sys import intern
+
+def silent_intern(x):
+    """
+    Perform intern() on the passed argument and return the result.
+    If the input is ineligible (e.g. a unicode string) the original argument is
+    returned and no exception is thrown.
+    """
+    try:
+        return intern(x)
+    except TypeError:
+        return x
+
+
+
 # From Dinu C. Gherman,
 # Python Cookbook, second edition, recipe 6.17, p. 277.
 # Also:
diff --git a/src/engine/SCons/UtilTests.py b/src/engine/SCons/UtilTests.py
index ee2c898..7d417d3 100644
--- a/src/engine/SCons/UtilTests.py
+++ b/src/engine/SCons/UtilTests.py
@@ -695,6 +695,14 @@ bling
             'bling\n',
         ], lines
 
+    def test_intern(self):
+        s1 = silent_intern("spam")
+        s2 = silent_intern(u"unicode spam")
+        s3 = silent_intern(42)
+        s4 = silent_intern("spam")
+        assert id(s1) == id(s4)
+
+
 class MD5TestCase(unittest.TestCase):
 
     def test_collect(self):