Add option to show MD5 checksums in dvc ls (#10573)

dvc/commands/ls/__init__.py

@@ -9,18 +9,25 @@ from dvc.ui import ui
 logger = logger.getChild(__name__)
 
 
-def _format_entry(entry, fmt):
+def _format_entry(entry, fmt, with_size=True, with_md5=False):
     from dvc.utils.humanize import naturalsize
 
+    ret = []
+    if with_size:
         size = entry.get("size")
         if size is None:
             size = ""
         else:
             size = naturalsize(size)
-    return size, fmt(entry)
+        ret.append(size)
+    if with_md5:
+        md5 = entry.get("md5", "")
+        ret.append(md5)
+    ret.append(fmt(entry))
+    return ret
 
 
-def show_entries(entries, with_color=False, with_size=False):
+def show_entries(entries, with_color=False, with_size=False, with_md5=False):
     if with_color:
         ls_colors = LsColors()
         fmt = ls_colors.format
@@ -29,8 +36,13 @@ def show_entries(entries, with_color=False, with_size=False):
         def fmt(entry):
             return entry["path"]
 
-    if with_size:
-        ui.table([_format_entry(entry, fmt) for entry in entries])
+    if with_size or with_md5:
+        ui.table(
+            [
+                _format_entry(entry, fmt, with_size=with_size, with_md5=with_md5)
+                for entry in entries
+            ]
+        )
         return
 
     # NOTE: this is faster than ui.table for very large number of entries
@@ -55,7 +67,12 @@ class CmdList(CmdBaseNoRepo):
             if self.args.json:
                 ui.write_json(entries)
             elif entries:
-                show_entries(entries, with_color=True, with_size=self.args.size)
+                show_entries(
+                    entries,
+                    with_color=True,
+                    with_size=self.args.size,
+                    with_md5=self.args.show_hash,
+                )
             return 0
         except FileNotFoundError:
             logger.exception("")
@@ -123,6 +140,12 @@ def add_parser(subparsers, parent_parser):
         ),
     )
     list_parser.add_argument("--size", action="store_true", help="Show sizes.")
+    list_parser.add_argument(
+        "--show-hash",
+        help="Display hash value for each item.",
+        action="store_true",
+        default=False,
+    )
     list_parser.add_argument(
         "path",
         nargs="?",

dvc/repo/ls.py

@@ -110,6 +110,7 @@ def _ls(
             "isdir": info["type"] == "directory",
             "isexec": info.get("isexec", False),
             "size": info.get("size"),
+            "md5": dvc_info.get("md5") or dvc_info.get("md5-dos2unix"),
         }
 
     return ret

tests/func/test_ls.py

@@ -222,7 +222,12 @@ def test_ls_repo_with_missed_path_dvc_only(tmp_dir, dvc, scm):
     tmp_dir.dvc_gen(DVC_STRUCTURE, commit="dvc")
 
     with pytest.raises(FileNotFoundError):
-        Repo.ls(os.fspath(tmp_dir), path="missed_path", recursive=True, dvc_only=True)
+        Repo.ls(
+            os.fspath(tmp_dir),
+            path="missed_path",
+            recursive=True,
+            dvc_only=True,
+        )
 
 
 def test_ls_repo_with_removed_dvc_dir(tmp_dir, dvc, scm):
@@ -467,20 +472,49 @@ def test_ls_granular(erepo_dir, M):
 
     entries = Repo.ls(os.fspath(erepo_dir), os.path.join("dir", "subdir"))
     assert entries == [
-        {"isout": True, "isdir": False, "isexec": False, "path": "bar", "size": 3},
-        {"isout": True, "isdir": False, "isexec": False, "path": "foo", "size": 3},
+        {
+            "isout": True,
+            "isdir": False,
+            "isexec": False,
+            "path": "bar",
+            "size": 3,
+            "md5": "37b51d194a7513e45b56f6524f2d51f2",
+        },
+        {
+            "isout": True,
+            "isdir": False,
+            "isexec": False,
+            "path": "foo",
+            "size": 3,
+            "md5": "acbd18db4cc2f85cedef654fccc4a4d8",
+        },
     ]
 
     entries = Repo.ls(os.fspath(erepo_dir), "dir")
     assert entries == [
-        {"isout": True, "isdir": False, "isexec": False, "path": "1", "size": 1},
-        {"isout": True, "isdir": False, "isexec": False, "path": "2", "size": 1},
+        {
+            "isout": True,
+            "isdir": False,
+            "isexec": False,
+            "path": "1",
+            "size": 1,
+            "md5": "c4ca4238a0b923820dcc509a6f75849b",
+        },
+        {
+            "isout": True,
+            "isdir": False,
+            "isexec": False,
+            "path": "2",
+            "size": 1,
+            "md5": "c81e728d9d4c2f636f067f89cc14862c",
+        },
         {
             "isout": True,
             "isdir": True,
             "isexec": False,
             "path": "subdir",
             "size": M.instance_of(int),
+            "md5": None,
         },
     ]
 
@@ -506,14 +540,42 @@ def test_ls_target(erepo_dir, use_scm):
         return Repo.ls(os.fspath(erepo_dir), path)
 
     assert _ls(os.path.join("dir", "1")) == [
-        {"isout": isout, "isdir": False, "isexec": False, "path": "1", "size": 1}
+        {
+            "isout": isout,
+            "isdir": False,
+            "isexec": False,
+            "path": "1",
+            "size": 1,
+            "md5": "c4ca4238a0b923820dcc509a6f75849b" if not use_scm else None,
+        }
     ]
     assert _ls(os.path.join("dir", "subdir", "foo")) == [
-        {"isout": isout, "isdir": False, "isexec": False, "path": "foo", "size": 3}
+        {
+            "isout": isout,
+            "isdir": False,
+            "isexec": False,
+            "path": "foo",
+            "size": 3,
+            "md5": "acbd18db4cc2f85cedef654fccc4a4d8" if not use_scm else None,
+        }
     ]
     assert _ls(os.path.join("dir", "subdir")) == [
-        {"isdir": False, "isexec": 0, "isout": isout, "path": "bar", "size": 3},
-        {"isdir": False, "isexec": 0, "isout": isout, "path": "foo", "size": 3},
+        {
+            "isdir": False,
+            "isexec": 0,
+            "isout": isout,
+            "path": "bar",
+            "size": 3,
+            "md5": "37b51d194a7513e45b56f6524f2d51f2" if not use_scm else None,
+        },
+        {
+            "isdir": False,
+            "isexec": 0,
+            "isout": isout,
+            "path": "foo",
+            "size": 3,
+            "md5": "acbd18db4cc2f85cedef654fccc4a4d8" if not use_scm else None,
+        },
     ]
 
 
@@ -576,6 +638,7 @@ def test_broken_symlink(tmp_dir, dvc, M):
             "isexec": False,
             "path": ".dvcignore",
             "size": M.instance_of(int),
+            "md5": None,
         },
         {
             "isout": False,
@@ -583,6 +646,7 @@ def test_broken_symlink(tmp_dir, dvc, M):
             "isexec": False,
             "path": "link",
             "size": 0,
+            "md5": None,
         },
     ]
 
@@ -614,14 +678,23 @@ def test_ls_broken_dir(tmp_dir, dvc, M):
             "isout": False,
             "path": ".dvcignore",
             "size": M.instance_of(int),
+            "md5": None,
+        },
+        {
+            "isdir": True,
+            "isexec": False,
+            "isout": True,
+            "path": "broken",
+            "size": 3,
+            "md5": "630bd47b538d2a513c7d267d07e0bc44.dir",
         },
-        {"isdir": True, "isexec": False, "isout": True, "path": "broken", "size": 3},
         {
             "isdir": False,
             "isexec": False,
             "isout": False,
             "path": "broken.dvc",
             "size": M.instance_of(int),
+            "md5": None,
         },
         {
             "isdir": True,
@@ -629,6 +702,7 @@ def test_ls_broken_dir(tmp_dir, dvc, M):
             "isout": True,
             "path": "dir",
             "size": M.instance_of(int),
+            "md5": "91aaa9bb58b657d623ef143b195a67e4.dir",
         },
         {
             "isdir": False,
@@ -636,14 +710,23 @@ def test_ls_broken_dir(tmp_dir, dvc, M):
             "isout": False,
             "path": "dir.dvc",
             "size": M.instance_of(int),
+            "md5": None,
+        },
+        {
+            "isdir": False,
+            "isexec": False,
+            "isout": True,
+            "path": "foo",
+            "size": 3,
+            "md5": "acbd18db4cc2f85cedef654fccc4a4d8",
         },
-        {"isdir": False, "isexec": False, "isout": True, "path": "foo", "size": 3},
         {
             "isdir": False,
             "isexec": False,
             "isout": False,
             "path": "foo.dvc",
             "size": M.instance_of(int),
+            "md5": None,
         },
     ]