From 1379af22de6693b21252fb7682e9fe7eb13dcaf8 Mon Sep 17 00:00:00 2001
From: Marian Beermann <public@enkore.de>
Date: Sun, 4 Jun 2017 19:50:25 +0200
Subject: [PATCH] HashIndexCompactTestCase

---
 src/borg/testsuite/hashindex.py | 131 ++++++++++++++++++++++++++++++++
 1 file changed, 131 insertions(+)

diff --git a/src/borg/testsuite/hashindex.py b/src/borg/testsuite/hashindex.py
index 5550e1adc..179f13a64 100644
--- a/src/borg/testsuite/hashindex.py
+++ b/src/borg/testsuite/hashindex.py
@@ -1,5 +1,6 @@
 import base64
 import hashlib
+import io
 import os
 import tempfile
 import zlib
@@ -18,6 +19,11 @@ def H(x):
     return bytes('%-0.32d' % x, 'ascii')
 
 
+def H2(x):
+    # like H(x), but with pseudo-random distribution of the output value
+    return hashlib.sha256(H(x)).digest()
+
+
 class HashIndexTestCase(BaseTestCase):
 
     def _generic_test(self, cls, make_value, sha):
@@ -357,6 +363,131 @@ def test_integrity_checked_file(self):
                     ChunkIndex.read(fd)
 
 
+class HashIndexCompactTestCase(HashIndexDataTestCase):
+    def index(self, num_entries, num_buckets):
+        index_data = io.BytesIO()
+        index_data.write(b'BORG_IDX')
+        # num_entries
+        index_data.write(num_entries.to_bytes(4, 'little'))
+        # num_buckets
+        index_data.write(num_buckets.to_bytes(4, 'little'))
+        # key_size
+        index_data.write((32).to_bytes(1, 'little'))
+        # value_size
+        index_data.write((3 * 4).to_bytes(1, 'little'))
+
+        self.index_data = index_data
+
+    def index_from_data(self):
+        self.index_data.seek(0)
+        index = ChunkIndex.read(self.index_data)
+        return index
+
+    def index_to_data(self, index):
+        data = io.BytesIO()
+        index.write(data)
+        return data.getvalue()
+
+    def index_from_data_compact_to_data(self):
+        index = self.index_from_data()
+        index.compact()
+        compact_index = self.index_to_data(index)
+        return compact_index
+
+    def write_entry(self, key, *values):
+        self.index_data.write(key)
+        for value in values:
+            self.index_data.write(value.to_bytes(4, 'little'))
+
+    def write_empty(self, key):
+        self.write_entry(key, 0xffffffff, 0, 0)
+
+    def write_deleted(self, key):
+        self.write_entry(key, 0xfffffffe, 0, 0)
+
+    def test_simple(self):
+        self.index(num_entries=3, num_buckets=6)
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_deleted(H2(1))
+        self.write_empty(H2(2))
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_entry(H2(4), 8, 9, 10)
+        self.write_empty(H2(5))
+
+        compact_index = self.index_from_data_compact_to_data()
+
+        self.index(num_entries=3, num_buckets=3)
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_entry(H2(4), 8, 9, 10)
+        assert compact_index == self.index_data.getvalue()
+
+    def test_first_empty(self):
+        self.index(num_entries=3, num_buckets=6)
+        self.write_deleted(H2(1))
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_empty(H2(2))
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_entry(H2(4), 8, 9, 10)
+        self.write_empty(H2(5))
+
+        compact_index = self.index_from_data_compact_to_data()
+
+        self.index(num_entries=3, num_buckets=3)
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_entry(H2(4), 8, 9, 10)
+        assert compact_index == self.index_data.getvalue()
+
+    def test_last_used(self):
+        self.index(num_entries=3, num_buckets=6)
+        self.write_deleted(H2(1))
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_empty(H2(2))
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_empty(H2(5))
+        self.write_entry(H2(4), 8, 9, 10)
+
+        compact_index = self.index_from_data_compact_to_data()
+
+        self.index(num_entries=3, num_buckets=3)
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_entry(H2(4), 8, 9, 10)
+        assert compact_index == self.index_data.getvalue()
+
+    def test_too_few_empty_slots(self):
+        self.index(num_entries=3, num_buckets=6)
+        self.write_deleted(H2(1))
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_empty(H2(2))
+        self.write_empty(H2(5))
+        self.write_entry(H2(4), 8, 9, 10)
+
+        compact_index = self.index_from_data_compact_to_data()
+
+        self.index(num_entries=3, num_buckets=3)
+        self.write_entry(H2(0), 1, 2, 3)
+        self.write_entry(H2(3), 5, 6, 7)
+        self.write_entry(H2(4), 8, 9, 10)
+        assert compact_index == self.index_data.getvalue()
+
+    def test_empty(self):
+        self.index(num_entries=0, num_buckets=6)
+        self.write_deleted(H2(1))
+        self.write_empty(H2(0))
+        self.write_deleted(H2(3))
+        self.write_empty(H2(2))
+        self.write_empty(H2(5))
+        self.write_deleted(H2(4))
+
+        compact_index = self.index_from_data_compact_to_data()
+
+        self.index(num_entries=0, num_buckets=0)
+        assert compact_index == self.index_data.getvalue()
+
+
 class NSIndexTestCase(BaseTestCase):
     def test_nsindex_segment_limit(self):
         idx = NSIndex()