From e27438bcec04e5e9a0325e92fbf75b00b7c0da32 Mon Sep 17 00:00:00 2001
From: Stein Magnus Jodal <stein.magnus@jodal.no>
Date: Mon, 22 Mar 2010 21:36:24 +0100
Subject: [PATCH] Add spotify_uri_to_int(uri) util function

---
 mopidy/utils.py | 46 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

diff --git a/mopidy/utils.py b/mopidy/utils.py
index 99a050be..6b47e02b 100644
--- a/mopidy/utils.py
+++ b/mopidy/utils.py
@@ -37,3 +37,49 @@ def unpickle_connection(pickled_connection):
     # From http://stackoverflow.com/questions/1446004
     (func, args) = pickle.loads(pickled_connection)
     return func(*args)
+
+def spotify_uri_to_int(uri, output_bits=31):
+    """
+    Stable one-way translation from Spotify URI to 31-bit integer.
+
+    Spotify track URIs has 62^22 possible values, which requires 131 bits of
+    storage. The original MPD server uses 32-bit unsigned integers for track
+    IDs. GMPC seems to think the track ID is a signed integer, thus we use 31
+    output bits.
+
+    In other words, this function throws away 100 bits of information. Since we
+    only use the track IDs to identify a track within a single Mopidy instance,
+    this information loss is acceptable. The chances of getting two different
+    tracks with the same track ID loaded in the same Mopidy instance is still
+    rather slim. 1 to 2,147,483,648 to be exact.
+
+    Normal usage, with data loss::
+
+        >>> spotify_uri_to_int('spotify:track:5KRRcT67VNIZUygEbMoIC1')
+        624351954
+
+    No data loss, may be converted back into a Spotify URI::
+
+        >>> spotify_uri_to_int('spotify:track:5KRRcT67VNIZUygEbMoIC1',
+        ...     output_bits=131)
+        101411513484007705241035418492696638725L
+
+    :param uri: Spotify URI on the format ``spotify:track:*``
+    :type uri: string
+    :param output_bits: number of bits of information kept in the return value
+    :type output_bits: int
+    :rtype: int
+    """
+
+    CHARS = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz'
+    BITS_PER_CHAR = 6 # int(math.ceil(math.log(len(CHARS), 2)))
+
+    key = uri.split(':')[-1]
+    full_id = 0
+    for i, char in enumerate(key):
+        full_id ^= CHARS.index(char) << BITS_PER_CHAR * i
+    compressed_id = 0
+    while full_id != 0:
+        compressed_id ^= (full_id & (2 ** output_bits - 1))
+        full_id >>= output_bits
+    return int(compressed_id)