analyze: sum up plaintext sizes rather than count

src/borg/archiver/analyze_cmd.py

@@ -56,22 +56,26 @@ def analyze_archives(self) -> None:
     def analyze_archive(self, id):
         """compute the set of chunks for each directory in this archive"""
         archive = Archive(self.manifest, id)
-        chunks_by_path = defaultdict(set)  # collect all chunk IDs generated from files in this directory path
+        chunks_by_path = defaultdict(dict)  # collect all chunk IDs generated from files in this directory path
         for item in archive.iter_items():
             if "chunks" in item:
-                item_chunks = set(id for id, size in item.chunks)
+                item_chunks = dict(item.chunks)  # chunk id -> plaintext size
                 directory_path = os.path.dirname(item.path)
                 chunks_by_path[directory_path].update(item_chunks)
         return chunks_by_path
 
     def analyze_change(self, base, new):
-        """for each directory path, count the chunks changed (removed or added chunks) between base and new."""
+        """for each directory path, sum up the changed (removed or added) chunks' sizes between base and new."""
 
         def analyze_path_change(path):
             base_chunks = base[path]
             new_chunks = new[path]
-            different_chunks = base_chunks.symmetric_difference(new_chunks)  # removed or added chunks
-            self.difference_by_path[directory_path] += len(different_chunks)
+            # add up added chunks' sizes
+            for id in new_chunks.keys() - base_chunks.keys():
+                self.difference_by_path[directory_path] += new_chunks[id]
+            # add up removed chunks' sizes
+            for id in base_chunks.keys() - new_chunks.keys():
+                self.difference_by_path[directory_path] += base_chunks[id]
 
         for directory_path in base:
             analyze_path_change(directory_path)
@@ -85,8 +89,7 @@ def report(self):
         print("=========================================")
         for directory_path in sorted(self.difference_by_path, key=lambda p: self.difference_by_path[p], reverse=True):
             difference = self.difference_by_path[directory_path]
-            if difference > 0:
-                print(f"{directory_path}: {difference}")
+            print(f"{directory_path}: {difference}")
 
 
 class AnalyzeMixIn:

src/borg/testsuite/archiver/analyze_cmd.py

@@ -19,23 +19,23 @@ def analyze_archives():
     input_path = pathlib.Path(archiver.input_path)
 
     # 1st archive
-    (input_path / "file1").write_text("foo")
+    (input_path / "file1").write_text("1")
     create_archive()
 
     # 2nd archive
-    (input_path / "file2").write_text("bar")
+    (input_path / "file2").write_text("22")
     create_archive()
 
-    assert "/input: 1" in analyze_archives()  # 2nd archive added 1 chunk for input path
+    assert "/input: 2" in analyze_archives()  # 2nd archive added 1 chunk for input path
 
     # 3rd archive
-    (input_path / "file3").write_text("baz")
+    (input_path / "file3").write_text("333")
     create_archive()
 
-    assert "/input: 2" in analyze_archives()  # 2nd/3rd archives added 2 chunks for input path
+    assert "/input: 5" in analyze_archives()  # 2nd/3rd archives added 2 chunks for input path
 
     # 4th archive
     (input_path / "file2").unlink()
     create_archive()
 
-    assert "/input: 3" in analyze_archives()  # 2nd/3rd archives added 2, 4th archive removed 1
+    assert "/input: 7" in analyze_archives()  # 2nd/3rd archives added 2, 4th archive removed 1