snapdiff

src/client/agent.py

@@ -27,6 +27,7 @@ class SyncAgent():
     self.pkey = paramiko.RSAKey.from_private_key_file(PKEY_PATH)
     self.client = paramiko.SSHClient()
     self.client.load_host_keys(filename=HOSTS_KEYS_PATH)
+    self.sftpc = None
 
   def connect(self):
     '''
@@ -49,11 +50,14 @@ class SyncAgent():
     return ostream.strip()
 
   def get_sftp_client(self):
+    if self.sftpc:
+      return self.sftpc
     if not self.client.get_transport():
       print('Transport not present proceed to create it... ')
       self.connect()
       print('--> OK!')
-    return self.client.open_sftp()
+    self.sftpc = self.client.open_sftp()
+    return self.sftpc
 
   def test_is_dir(self, path='.'):
     sftpc = self.get_sftp_client()

src/snapshot/generate/remote.py

@@ -120,13 +120,13 @@ async def generate_tree_hash_oversftp_async(root_path :str):
   await asyncio.gather(*tasks)
   return rtreemap
 
-
 def test_sync_rtree():
   start = time.time()
   print('Start task')
   rtree = generate_tree_hash_oversftp('/home/notanamber/notes_dev/')
   end = time.time()
   print('task done in %.2f' % (end - start))
+  return rtree
 
 def test_async_rtree():
   start = time.time()
@@ -134,6 +134,7 @@ def test_async_rtree():
   rtree = asyncio.run(generate_tree_hash_oversftp_async('/home/notanamber/notes_dev/')) 
   end = time.time()
   print('task done in %.2f' % (end - start))
+  return rtree
 
 
 

src/snapshot/snap.py

@@ -8,36 +8,71 @@ def dmpsnap(root_tree: str):
 def dcsnap(path: str, filename=None):
   return _dump.decode_snapshot(path, dump_file_name = filename)
 
-
-def diffsnap(snapshot_a, snapshot_b, path='./'):
+def diffsnap(last_tree, current_tree, path='./', removed=[], changed=[], added=[]):
   '''
     snapshot is a tree represented by dictionary {k, val} when k can be folder or file
     and val can be a dictionary or an hash
 
-    @param                 snapshot_a                 snapshot computed at time t0
-    @param                 snapshot_b                 snapshot computed at time t1
+    @param                 last_tree                 snapshot computed at time t0
+    @param                 current_tree              snapshot computed at time t1
   '''
-  for name_a, hsh_a in snapshot_a.items():
-    hsh_b = snapshot_b.get(name_a)
-    if hsh_b is None:
-      print(f'{path}{name_a} present in snapshot_a but removed in snapshot_b, subtree {name_a} removed in snapshot_b')
+  removed = []
+  changed = []
+  added = []
+
+  for name_last, hsh_last in last_tree.items():
+    hsh_current = current_tree.get(name_last)
+
+    # is_dir_last = True if isinstance(hsh_last, dict) else False
+    last_type = 'dir' if isinstance(hsh_last, dict) else 'file'
+
+    item = {
+        'name' : name_last, 
+        'path' : path,
+        }
+
+    if hsh_current is None:
+      ## the name in not founded in current snapshot - Subtree removed (or moved)
+      print(f'{path}{name_last} present in last_tree but removed in current_tree, subtree {name_last} removed in current_tree')
+      removed.append(item.update({
+        'last_type' : 'dir',
+        }))
       continue
-    if type(hsh_a) != type(hsh_a):
-      print(f'{path}{name_a} present in snapshot_a and in snapshot_b, but different type')
+
+    current_type = 'dir' if isinstance(hsh_current, dict) else 'file'
+
+    if type(hsh_last) != type(hsh_current):
+      print(f'{path}{name_last} changed his type in {current_type}')
+      changed.append(item.update({
+        'last_type' : last_type,
+        'current_type' : current_type,
+        }))
       continue
-    if isinstance(hsh_a, dict):
-      # name is dir
-      diffsnap(hsh_a, hsh_b, path='%s%s/' % (path,name_a))
-    else:
-      # name is a file
-      if hsh_b is None:
-        print(f'{path}{name_a} present in snapshot_a but not in snapshot_b, removed in snapshot_b')
-        continue
-      if hsh_a != hsh_b: 
-        print(f'{path}{name_a} present in snapshot_a but changed  in snapshot_b, snapshot_a: {hsh_a}, snapshot_b: {hsh_b}')
-        continue
-      else:
-        print(f'{path}{name_a} present in snapshot_a and in snapshot_b, snapshot_a: {hsh_a}, snapshot_b: {hsh_b}, unchanged')
+
+    if isinstance(hsh_last, str):
+      if hsh_last != hsh_current:
+        print(f'file {path}{name_last} changed his hash')
+        changed.append(item.update({
+          'last_type' : last_type,
+          'current_type' : current_type,
+          'last_hash' : hsh_last,
+          'cur_hash' : hsh_current,
+          }))
+      continue
+
+    # name is dir
+    if (hsh_last == hsh_current):
+      print(f'file {path}{name_last} subtree unchanged ')
+      continue
+
+    print(f'file {path}{name_last} subtree changed in current snapshot')
+    keys_added = set(hsh_current.keys()) - set(hsh_last.keys())
+    ## name last is a dir
+    if keys_added:
+      for key_added in keys_added:
+        print(f'file {path}{name_last}/{key_added} subtree added in current snapshot')
+
+    diffsnap(hsh_last, hsh_current, path='%s%s/' % (path,name_last))
 
 def findsubtree(snapshot, subtree, path='./'):
   '''