starting python core

src/.hashes

@@ -0,0 +1 @@
+{"caff\u00e8": 1, "briosche": 2}

src/hash/gen.py

@@ -1,8 +0,0 @@
-import hashlib 
-
-def generate_file_hash(file_path, hexdigest=True):
-  with open(file_path, "rb") as f:
-    buf = f.read()
-
-  if hexdigest: return hashlib.md5(buf).hexdigest()
-  return hashlib.md5(buf).digest()

src/snapshot/dump.py

@@ -0,0 +1,32 @@
+import gzip
+import json
+from snapshot import gen as _gen
+
+DUMP_FILE_NAME = '.snapshot.json.gz'
+SNAPSHOT_PATH = '../'
+
+def dump_snapshot(snapshot, path=None, dump_file_name=None):
+  path = path or SNAPSHOT_PATH
+  path = _gen.check_isdir(path)
+
+  dump = json.dumps(snapshot)
+  dump = gzip.compress(dump.encode())
+
+  dump_file_name = dump_file_name or DUMP_FILE_NAME
+
+  file = '%s%s' % (path, dump_file_name)
+  with open(file, "wb") as f:
+    f.write(dump)
+  
+def decode_snapshot(path=None, dump_file_name=None):
+  path = path or SNAPSHOT_PATH
+  path = _gen.check_isdir(path)
+
+  dump_file_name = dump_file_name or DUMP_FILE_NAME
+
+  file = '%s%s' % (path, dump_file_name)
+  with open(file, "rb") as f:
+    buf = f.read()
+
+  buf = gzip.decompress(buf)
+  return json.loads(buf)

src/snapshot/gen.py

@@ -0,0 +1,72 @@
+import os
+import hashlib 
+import json
+import gzip
+
+
+def generate_file_hash(file_path, hexdigest=True):
+  with open(file_path, "rb") as f:
+    buf = f.read()
+
+  if hexdigest: return hashlib.md5(buf).hexdigest()
+  return hashlib.md5(buf).digest()
+
+def check_isdir(path: str):
+  if not os.path.isdir(path):
+    raise Exception('Provide a valid folder to start the hashing')
+
+  if not path.endswith(os.path.sep):
+    path = path + os.path.sep
+  return path
+
+def generate_tree_hash(root_path :str):
+  '''
+    @param             root_path        string
+    generate a map of hashes starting from `root_path` recursively
+  '''
+  # if not os.path.isdir(root_path):
+  #   raise Exception('Provide a valid folder to start the hashing')
+  #
+  # if not root_path.endswith(os.path.sep):
+  #   root_path = root_path + os.path.sep
+  root_path = check_isdir(root_path)
+
+  treemap = {}
+
+  items = os.listdir(root_path)
+  for item in items:
+    absolute_item_path = root_path + item
+    print('absolute_item_path: %s, item %s, isdir: %s' % (absolute_item_path, item, os.path.isdir(absolute_item_path)))
+    if os.path.isdir(absolute_item_path):
+      treemap[item] = generate_tree_hash(absolute_item_path)
+    else:
+      treemap[item] = generate_file_hash(absolute_item_path)
+
+  return treemap
+
+
+'''
+a
+ - b
+   - c
+     - k.txt
+   i.txt
+   g.txt
+j.txt
+k.txt
+
+tree['a'] = {
+    'b' : {
+
+      },
+    'j.txt' : '012349jasdfh9934',
+
+
+    }
+
+'''
+
+root_path = '/home/luca/rsyn_test_fap'
+
+
+

src/snapshot/snap.py

@@ -0,0 +1,34 @@
+from snapshot import gen as _gen
+from snapshot import dump as _dump
+
+def dmpsnap(root_tree: str):
+  snapshot = _gen.generate_recursive_filemap_hash(root_tree)
+  _dump.dump_snapshot(snapshot)
+
+def dcsnap(path: str, filename=None):
+  return _dump.decode_snapshot(path, dump_file_name = filename)
+
+
+def diffsnap(snapshot_a, snapshot_b, path='./'):
+  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 {hsh_b} removed in snapshot_b')
+      continue
+    if type(hsh_a) != type(hsh_a):
+      print(f'{path}{name_a} present in snapshot_a and in snapshot_b, but different 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')
+