Add two new tools to compare codecs and show differences and to

list all installed codecs.

2 files changed, 94 insertions, 0 deletions

diff --git a/Tools/unicode/comparecodecs.py b/Tools/unicode/comparecodecs.py
new file mode 100644
index 0000000..cd417a4
--- /dev/null
+++ b/Tools/unicode/comparecodecs.py
@@ -0,0 +1,53 @@
+#!/usr/bin/env python
+
+""" Compare the output of two codecs.
+
+(c) Copyright 2005, Marc-Andre Lemburg (mal@lemburg.com).
+
+    Licensed to PSF under a Contributor Agreement.
+
+"""
+import sys
+
+def compare_codecs(encoding1, encoding2):
+
+    print 'Comparing encoding/decoding of   %r and   %r' % (encoding1, encoding2)
+    mismatch = 0
+    # Check encoding
+    for i in range(sys.maxunicode):
+        u = unichr(i)
+        try:
+            c1 = u.encode(encoding1)
+        except UnicodeError, reason:
+            c1 = '<undefined>'
+        try:
+            c2 = u.encode(encoding2)
+        except UnicodeError, reason:
+            c2 = '<undefined>'
+        if c1 != c2:
+            print ' * encoding mismatch for 0x%04X: %-14r != %r' % \
+                  (i, c1, c2)
+            mismatch += 1
+    # Check decoding
+    for i in range(256):
+        c = chr(i)
+        try:
+            u1 = c.decode(encoding1)
+        except UnicodeError:
+            u1 = u'<undefined>'
+        try:
+            u2 = c.decode(encoding2)
+        except UnicodeError:
+            u2 = u'<undefined>'
+        if u1 != u2:
+            print ' * decoding mismatch for 0x%04X: %-14r != %r' % \
+                  (i, u1, u2)
+            mismatch += 1
+    if mismatch:
+        print
+        print 'Found %i mismatches' % mismatch
+    else:
+        print '-> Codecs are identical.'
+
+if __name__ == '__main__':
+    compare_codecs(sys.argv[1], sys.argv[2])
diff --git a/Tools/unicode/listcodecs.py b/Tools/unicode/listcodecs.py
new file mode 100644
index 0000000..19d21e1
--- /dev/null
+++ b/Tools/unicode/listcodecs.py
@@ -0,0 +1,41 @@
+""" List all available codec modules.
+
+(c) Copyright 2005, Marc-Andre Lemburg (mal@lemburg.com).
+
+    Licensed to PSF under a Contributor Agreement.
+
+"""
+
+import os, codecs, encodings
+
+_debug = 0
+
+def listcodecs(dir):
+    names = []
+    for filename in os.listdir(dir):
+        if filename[-3:] != '.py':
+            continue
+        name = filename[:-3]
+        # Check whether we've found a true codec
+        try:
+            codecs.lookup(name)
+        except LookupError:
+            # Codec not found
+            continue
+        except Exception, reason:
+            # Probably an error from importing the codec; still it's
+            # a valid code name
+            if _debug:
+                print '* problem importing codec %r: %s' % \
+                      (name, reason)
+        names.append(name)
+    return names
+
+
+if __name__ == '__main__':
+    names = listcodecs(encodings.__path__[0])
+    names.sort()
+    print 'all_codecs = ['
+    for name in names:
+        print '    %r,' % name
+    print ']'