From 9fd284ce1a9c310571049aa1d7ad0a6fa89b8a26 Mon Sep 17 00:00:00 2001
From: Thomas Waldmann <tw@waldmann-edv.de>
Date: Fri, 8 Jan 2021 18:38:35 +0100
Subject: [PATCH] refactor new zero chunk handling to be reusable

---
 src/borg/archive.py  | 20 ++------------------
 src/borg/chunker.pyx | 30 +++++++++++++++++++++++++++++-
 2 files changed, 31 insertions(+), 19 deletions(-)

diff --git a/src/borg/archive.py b/src/borg/archive.py
index cc5c33fb9..eff10b4bf 100644
--- a/src/borg/archive.py
+++ b/src/borg/archive.py
@@ -19,7 +19,7 @@ from .logger import create_logger
 logger = create_logger()
 
 from . import xattr
-from .chunker import get_chunker, max_chunk_size, Chunk
+from .chunker import get_chunker, max_chunk_size, Chunk, chunk_to_id_data
 from .cache import ChunkListEntry
 from .crypto.key import key_factory
 from .compress import Compressor, CompressionSpec
@@ -43,7 +43,6 @@ from .helpers import msgpack
 from .helpers import sig_int
 from .patterns import PathPrefixPattern, FnmatchPattern, IECommand
 from .item import Item, ArchiveItem, ItemDiff
-from .lrucache import LRUCache
 from .platform import acl_get, acl_set, set_flags, get_flags, swidth, hostname
 from .remote import cache_if_remote
 from .repository import Repository, LIST_SCAN_LIMIT
@@ -1105,8 +1104,6 @@ class ChunksProcessor:
         self.checkpoint_interval = checkpoint_interval
         self.last_checkpoint = time.monotonic()
         self.rechunkify = rechunkify
-        self.zero_chunk_ids = LRUCache(10, dispose=lambda _: None)  # length of all-zero chunk -> chunk_id
-        self.zeros = memoryview(bytes(MAX_DATA_SIZE))
 
     def write_part_file(self, item, from_chunk, number):
         item = Item(internal_dict=item.as_dict())
@@ -1139,20 +1136,7 @@ class ChunksProcessor:
     def process_file_chunks(self, item, cache, stats, show_progress, chunk_iter, chunk_processor=None):
         if not chunk_processor:
             def chunk_processor(chunk):
-                allocation = chunk.meta['allocation']
-                if allocation == CH_DATA:
-                    data = chunk.data
-                    chunk_id = self.key.id_hash(data)
-                elif allocation in (CH_HOLE, CH_ALLOC):
-                    size = chunk.meta['size']
-                    data = self.zeros[:size]
-                    try:
-                        chunk_id = self.zero_chunk_ids[size]
-                    except KeyError:
-                        chunk_id = self.key.id_hash(data)
-                        self.zero_chunk_ids[size] = chunk_id
-                else:
-                    raise ValueError('unexpected allocation type')
+                chunk_id, data = chunk_to_id_data(chunk, self.key.id_hash)
                 chunk_entry = cache.add_chunk(chunk_id, data, stats, wait=False)
                 self.cache.repository.async_response(wait=False)
                 return chunk_entry
diff --git a/src/borg/chunker.pyx b/src/borg/chunker.pyx
index 6bf64784c..1fd316a03 100644
--- a/src/borg/chunker.pyx
+++ b/src/borg/chunker.pyx
@@ -6,7 +6,8 @@ import errno
 import os
 from collections import namedtuple
 
-from .constants import CH_DATA, CH_ALLOC, CH_HOLE
+from .constants import CH_DATA, CH_ALLOC, CH_HOLE, MAX_DATA_SIZE
+from .lrucache import LRUCache
 
 from libc.stdlib cimport free
 
@@ -52,6 +53,33 @@ def Chunk(data, **meta):
     return _Chunk(meta, data)
 
 
+zeros = bytes(MAX_DATA_SIZE)
+
+# remember a few recently used all-zero chunk hashes in this mapping.
+# (hash_func, chunk_length) -> chunk_hash
+# we play safe and have the hash_func in the mapping key, in case we
+# have different hash_funcs within the same borg run.
+zero_chunk_ids = LRUCache(10, dispose=lambda _: None)
+
+def chunk_to_id_data(chunk, id_hash):
+    allocation = chunk.meta['allocation']
+    if allocation == CH_DATA:
+        data = chunk.data
+        chunk_id = id_hash(data)
+    elif allocation in (CH_HOLE, CH_ALLOC):
+        size = chunk.meta['size']
+        assert size <= len(zeros)
+        data = memoryview(zeros)[:size]
+        try:
+            chunk_id = zero_chunk_ids[(id_hash, size)]
+        except KeyError:
+            chunk_id = id_hash(data)
+            zero_chunk_ids[(id_hash, size)] = chunk_id
+    else:
+        raise ValueError('unexpected allocation type')
+    return chunk_id, data
+
+
 def dread(offset, size, fd=None, fh=-1):
     use_fh = fh >= 0
     if use_fh: