Add an option to skip non-existent files in diff-related dl-git commands

datalens-tech · Oct 24, 2023 · d3995f8 · d3995f8
1 parent 0501d44
commit d3995f8
Show file tree

Hide file tree

Showing 3 changed files with 51 additions and 19 deletions.
diff --git a/terrarium/dl_gitmanager/README.md b/terrarium/dl_gitmanager/README.md
@@ -47,12 +47,13 @@ List files that have changed between two given revisions including the changes i
 dl-git range-diff-paths --base <base> --head <head>
 dl-git range-diff-paths --base <base> --head <head> --absolute
 dl-git range-diff-paths --base <base> --head <head> --only-added-commits
+dl-git range-diff-paths --base <base> --head <head> --without-deleted-files
 ```
 
 Here `base` and `head` can be a commit ID, branch name, `HEAD~3` or any similar notation
 that is usually accepted by git.
 These arguments are optional. By default `head` is `HEAD` and `base` is `HEAD~1`.
-Thgis means you can use the following command to see the diff of the last commit in the current branch:
+This means you can use the following command to see the diff of the last commit in the current branch:
 ```
 dl-git diff-paths
 ```
@@ -63,13 +64,16 @@ By default they are printed relative to the repository root.
 The `--only-added-commits` option makes the tool inspect only commits
 that have been added in the head version.
 
+The `--without-deleted-files` option makes the tool omit deleted (non-existent) files in the resulting list.
+
 ### list-diff-paths
 
 List files that have changed in commits passed on as input
 
 ```
 echo <commit-id> | dl-git list-diff-paths
-echo <commit-id>> | dl-git list-diff-paths --absolute
+echo <commit-id> | dl-git list-diff-paths --absolute
+echo <commit-id> | dl-git list-diff-paths --without-deleted-files
 ```
 
-Option `--absolute` has the same meaning as in `range-diff-paths`.
+Options `--absolute` and `--without-deleted-files` have the same meaning as in `range-diff-paths`.
diff --git a/terrarium/dl_gitmanager/dl_gitmanager/git_manager.py b/terrarium/dl_gitmanager/dl_gitmanager/git_manager.py
@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import os
 from pathlib import Path
 from typing import (
     Collection,
@@ -114,23 +115,35 @@ def _iter_list_diffs(
             for _, diff_item in submodule_manager._iter_list_diffs(commits=sm_commits_for_all_commits):
                 yield submodule_base_path, diff_item
 
-    def _collect_paths_from_diffs(self, diff_iterable: Iterable[tuple[Path, Diff]]) -> list[str]:
+    def _collect_paths_from_diffs(
+        self, diff_iterable: Iterable[tuple[Path, Diff]], without_deleted: bool = False
+    ) -> list[str]:
         result: set[str] = set()
         for base_path, diff_item in diff_iterable:
-            if diff_item.a_path:
-                result.add(str(base_path / diff_item.a_path))
-            if diff_item.b_path:
-                result.add(str(base_path / diff_item.b_path))
+            for diff_path in (diff_item.a_path, diff_item.b_path):
+                if not diff_path:
+                    continue
+                absolute_path = base_path / diff_path
+                if not without_deleted or absolute_path.exists():
+                    result.add(str(absolute_path))
 
         return sorted(result)
 
-    def get_range_diff_paths(self, base: str, head: str, absolute: bool = False) -> list[str]:
+    def get_range_diff_paths(
+        self, base: str, head: str, absolute: bool = False, without_deleted: bool = False
+    ) -> list[str]:
         return self._collect_paths_from_diffs(
-            diff_iterable=self._iter_range_diffs(base=base, head=head, absolute=absolute)
+            diff_iterable=self._iter_range_diffs(base=base, head=head, absolute=absolute),
+            without_deleted=without_deleted,
         )
 
-    def get_list_diff_paths(self, commits: Collection[str], absolute: bool = False) -> list[str]:
-        return self._collect_paths_from_diffs(diff_iterable=self._iter_list_diffs(commits=commits, absolute=absolute))
+    def get_list_diff_paths(
+        self, commits: Collection[str], absolute: bool = False, without_deleted: bool = False
+    ) -> list[str]:
+        return self._collect_paths_from_diffs(
+            diff_iterable=self._iter_list_diffs(commits=commits, absolute=absolute),
+            without_deleted=without_deleted,
+        )
 
     def get_all_ancestor_commits(self, commit: str) -> set[str]:
         result: set[str] = {commit}

diff --git a/terrarium/dl_gitmanager/dl_gitmanager/scripts/gitmanager_cli.py b/terrarium/dl_gitmanager/dl_gitmanager/scripts/gitmanager_cli.py
@@ -49,6 +49,9 @@ def get_parser(cls) -> argparse.ArgumentParser:
         range_diff_paths_parser.add_argument(
             "--only-added-commits", action="store_true", help="Inspect only commits that are added in head"
         )
+        range_diff_paths_parser.add_argument(
+            "--without-deleted-files", action="store_true", help="Omit deleted files in the resulting list"
+        )
 
         subparsers.add_parser(
             "list-diff-paths",
@@ -58,18 +61,26 @@ def get_parser(cls) -> argparse.ArgumentParser:
 
         return parser
 
-    def range_diff_paths(self, base: str, head: Optional[str], absolute: bool, only_added_commits: bool) -> None:
+    def range_diff_paths(
+        self, base: str, head: Optional[str], absolute: bool, only_added_commits: bool, without_deleted: bool = False
+    ) -> None:
         diff_name_list: list[str]
         if only_added_commits:
             commits = self.git_manager.get_missing_commits(base=base, head=head)
-            diff_name_list = self.git_manager.get_list_diff_paths(commits=commits, absolute=absolute)
+            diff_name_list = self.git_manager.get_list_diff_paths(
+                commits=commits, absolute=absolute, without_deleted=without_deleted
+            )
         else:
-            diff_name_list = self.git_manager.get_range_diff_paths(base=base, head=head, absolute=absolute)
+            diff_name_list = self.git_manager.get_range_diff_paths(
+                base=base, head=head, absolute=absolute, without_deleted=without_deleted
+            )
         print("\n".join(diff_name_list))
 
-    def list_diff_paths(self, absolute: bool) -> None:
+    def list_diff_paths(self, absolute: bool, without_deleted: bool = False) -> None:
         commits = [line.strip() for line in self.input_text_io if line.strip()]
-        diff_name_list = self.git_manager.get_list_diff_paths(commits=commits, absolute=absolute)
+        diff_name_list = self.git_manager.get_list_diff_paths(
+            commits=commits, absolute=absolute, without_deleted=without_deleted
+        )
         print("\n".join(diff_name_list))
 
     @classmethod
@@ -86,10 +97,14 @@ def run_parsed_args(cls, args: argparse.Namespace) -> None:
         match args.command:
             case "range-diff-paths":
                 tool.range_diff_paths(
-                    base=args.base, head=args.head, absolute=args.absolute, only_added_commits=args.only_added_commits
+                    base=args.base,
+                    head=args.head,
+                    absolute=args.absolute,
+                    only_added_commits=args.only_added_commits,
+                    without_deleted=args.without_deleted_files,
                 )
             case "list-diff-paths":
-                tool.list_diff_paths(absolute=args.absolute)
+                tool.list_diff_paths(absolute=args.absolute, without_deleted=args.without_deleted_files)
             case _:
                 raise RuntimeError(f"Got unknown command: {args.command}")