Raw codepoint test backend. Tweaks to constructor arguments. Try to handle the case...

[python-collate.git] / collate / codepoint.py

diff --git a/collate/codepoint.py b/collate/codepoint.py
new file mode 100644 (file)
index 0000000..e0bbcbd
--- /dev/null
+++ b/collate/codepoint.py
@@ -0,0 +1,43 @@
+"""Codepoint-based collation.
+
+This collation backend sorts using only the basic codepoint order. It
+is primarily intended to be used as a baseline and example for other
+collation backends.
+
+Use this collation backend if...
+ - You are writing tests for pycollate.
+ - You are writing specialized Unicode software.
+ - You are on a system with no locale module.
+
+Avoid this backend if...
+ - You are writing a normal program for a normal runtime environment.
+ - You are sorting strings to show normal humans.
+
+"""
+
+import collate._abcollator
+import collate._locale
+
+__all__ = ["Collate"]
+
+class Collator(collate._abcollator.Collator):
+    """Codepoint-based collation.
+
+    Arguments
+    locale - all parts but encoding ignored, always 'C'
+    encoding - try to use this string encoding
+    """
+
+    def __init__(self, locale=None, encoding=None):
+        dummy, self.encoding = collate._locale.getpair(locale, encoding)
+        self.locale = "C"
+
+    def key(self, string):
+        """Sort key for a string.
+
+        If string is a str instance, it is first decoded according to
+        the 'encoding' attribute of the Collator.
+        """
+        if isinstance(string, str):
+            string = string.decode(self.encoding, 'replace')
+        return string