From bcc7983492dde927bf184938aba60f995e1df8b7 Mon Sep 17 00:00:00 2001
From: Edwin Eefting <edwin@datux.nl>
Date: Mon, 21 Feb 2022 17:51:23 +0100
Subject: [PATCH] tree compare

---
 zfs_autobackup/BlockHasher.py | 10 +++++++--
 zfs_autobackup/TreeHasher.py  | 38 +++++++++++++++++++++++++++++++++++
 zfs_autobackup/ZfsCheck.py    | 20 ++++++++++++++++++
 3 files changed, 66 insertions(+), 2 deletions(-)

diff --git a/zfs_autobackup/BlockHasher.py b/zfs_autobackup/BlockHasher.py
index 487cd86..22e2b61 100644
--- a/zfs_autobackup/BlockHasher.py
+++ b/zfs_autobackup/BlockHasher.py
@@ -44,11 +44,17 @@ class BlockHasher():
         checked=0
         with open(fname, "rb") as f:
             for ( chunk_nr, hexdigest ) in generator:
+                print ("comparing {} {} {}".format(fname, chunk_nr, hexdigest))
+
                 checked=checked+1
                 hash = self.hash_class()
                 f.seek(chunk_nr * self.bs * self.count)
-                for block_nr in range(0,self.count):
-                    hash.update(f.read(self.bs))
+                block_nr=0
+                for block in iter(lambda: f.read(self.bs), b""):
+                    hash.update(block)
+                    block_nr=block_nr+1
+                    if block_nr == self.count:
+                        break
 
                 if (hash.hexdigest()!=hexdigest):
                     raise Exception("Block {} mismatched! Hash is {}, but should be {}".format(chunk_nr, hash.hexdigest(), hexdigest))
diff --git a/zfs_autobackup/TreeHasher.py b/zfs_autobackup/TreeHasher.py
index 1241255..e976bdf 100644
--- a/zfs_autobackup/TreeHasher.py
+++ b/zfs_autobackup/TreeHasher.py
@@ -5,6 +5,10 @@ class TreeHasher():
     """uses BlockHasher recursively on a directory tree"""
 
     def __init__(self, block_hasher):
+        """
+
+        :type block_hasher: BlockHasher
+        """
         self.block_hasher=block_hasher
 
     def generate(self, start_path):
@@ -31,3 +35,37 @@ class TreeHasher():
         finally:
             os.chdir(cwd)
 
+
+    def compare(self, start_path, generator):
+        """reads from generator and compares blocks, raises exception on error
+        """
+
+        cwd=os.getcwd()
+        os.chdir(start_path)
+
+        try:
+            current= [None]
+            def per_file_generator():
+
+                (current_file_path, chunk_nr, hash)=current[0]
+                yield ( chunk_nr, hash)
+
+                for (file_path, chunk_nr, hash) in generator:
+                    if file_path==current_file_path:
+                        yield ( chunk_nr, hash)
+                    else:
+                        current[0] = (file_path, chunk_nr, hash)
+                        return
+
+                current[0]=None
+
+
+            current[0] = generator.next()
+            while current[0] is not None:
+                self.block_hasher.compare(current[0][0], per_file_generator())
+
+        finally:
+            os.chdir(cwd)
+
+
+
diff --git a/zfs_autobackup/ZfsCheck.py b/zfs_autobackup/ZfsCheck.py
index 283762c..cf54d66 100644
--- a/zfs_autobackup/ZfsCheck.py
+++ b/zfs_autobackup/ZfsCheck.py
@@ -68,6 +68,11 @@ class ZfsCheck(CliBase):
 
             self.debug("Hashing tree: {}".format(mnt))
             if not self.args.test:
+
+                # generator=tree_hasher.generate(mnt)
+                # tree_hasher.compare(mnt, generator)
+
+
                 for (file, block, hash) in tree_hasher.generate(mnt):
                     print("{}\t{}\t{}".format(file, block, hash))
                     sys.stdout.flush() #important, to generate SIGPIPES on ssh disconnect
@@ -158,6 +163,21 @@ def cli():
 
     sys.exit(ZfsCheck(sys.argv[1:], False).run())
 
+    # block_hasher=BlockHasher()
+
+    # if sys.argv[1]=="s":
+    #     for ( fname, nr, hash ) in TreeHasher(block_hasher).generate("/usr/src/linux-headers-5.14.14-051414"):
+    #         print("{}\t{}\t{}".format(fname, nr, hash))
+    #
+    # if sys.argv[1]=="r":
+    #
+    #     def gen():
+    #         for line in sys.stdin:
+    #             ( fname, nr, hash)=line.rstrip().split('\t')
+    #             yield (fname, int(nr), hash)
+    #
+    #     TreeHasher(block_hasher).compare("/usr/src/linux-headers-5.14.14-051414", gen())
+
 
 if __name__ == "__main__":