Add merge strategy (#2922)

* add merge strategy * add merge strategy * generate files * Adding merge strategies support to merge test * refactoring merge test running engine to external function * merge strategies - completing conflict handling * merge strategy tests * Adding explanations of conflicts and their resolution to non trivial conflict tests * Updating description for merge-strategy flag * Additional test - merge conflict resolved by strategies * Adding documentation to handleAll function, to explain the new strategyToInclude parameter * changing 'theirs' and 'ours' to 'source-wins' and 'dest-wins' * renaming lakectl test functions for easier isolated run * lakectl tests - parameterizing file name, checksum and commit message * lakectl test for merge strategies * fixing merge issues in lakectl test Co-authored-by: Itai David <itai.david@treeverse.io> Co-authored-by: itai-david <90712874+itai-david@users.noreply.github.com>
treeverse · Mar 2, 2022 · d2db61c · d2db61c
1 parent 4726a32
commit d2db61c
Show file tree

Hide file tree

Showing 37 changed files with 1,297 additions and 512 deletions.
diff --git a/api/swagger.yml b/api/swagger.yml
@@ -439,6 +439,9 @@ components:
           type: object
           additionalProperties:
             type: string
+        strategy:
+          description: In case of a merge conflict, this option will force the merge process to automatically favor changes from the dest branch ('dest-wins') or from the source branch('source-wins'). In case no selection is made, the merge process will fail in case of a conflict
+          type: string
 
     BranchCreation:
       type: object

diff --git a/clients/java/api/openapi.yaml b/clients/java/api/openapi.yaml
diff --git a/clients/java/docs/Merge.md b/clients/java/docs/Merge.md
diff --git a/clients/java/src/main/java/io/lakefs/clients/api/model/Merge.java b/clients/java/src/main/java/io/lakefs/clients/api/model/Merge.java
diff --git a/clients/java/src/test/java/io/lakefs/clients/api/model/MergeTest.java b/clients/java/src/test/java/io/lakefs/clients/api/model/MergeTest.java
diff --git a/clients/python/docs/Merge.md b/clients/python/docs/Merge.md
diff --git a/clients/python/docs/RefsApi.md b/clients/python/docs/RefsApi.md
diff --git a/clients/python/lakefs_client/model/merge.py b/clients/python/lakefs_client/model/merge.py
diff --git a/cmd/lakectl/cmd/merge.go b/cmd/lakectl/cmd/merge.go
@@ -28,12 +28,17 @@ var mergeCmd = &cobra.Command{
 		client := getClient()
 		sourceRef := MustParseRefURI("source ref", args[0])
 		destinationRef := MustParseRefURI("destination ref", args[1])
+		strategy := MustString(cmd.Flags().GetString("strategy"))
 		Fmt("Source: %s\nDestination: %s\n", sourceRef.String(), destinationRef)
 		if destinationRef.Repository != sourceRef.Repository {
 			Die("both references must belong to the same repository", 1)
 		}
 
-		resp, err := client.MergeIntoBranchWithResponse(cmd.Context(), destinationRef.Repository, sourceRef.Ref, destinationRef.Ref, api.MergeIntoBranchJSONRequestBody{})
+		if strategy != "dest-wins" && strategy != "source-wins" && strategy != "" {
+			Die("Invalid strategy value. Expected \"dest-wins\" or \"source-wins\"", 1)
+		}
+
+		resp, err := client.MergeIntoBranchWithResponse(cmd.Context(), destinationRef.Repository, sourceRef.Ref, destinationRef.Ref, api.MergeIntoBranchJSONRequestBody{Strategy: &strategy})
 		if resp != nil && resp.JSON409 != nil {
 			Die("Conflict found.", 1)
 		}
@@ -52,4 +57,5 @@ var mergeCmd = &cobra.Command{
 //nolint:gochecknoinits
 func init() {
 	rootCmd.AddCommand(mergeCmd)
+	mergeCmd.Flags().String("strategy", "", "In case of a merge conflict, this option will force the merge process to automatically favor changes from the dest branch (\"dest-wins\") or from the source branch(\"source-wins\"). In case no selection is made, the merge process will fail in case of a conflict")
 }
diff --git a/cmd/lakefs/cmd/import.go b/cmd/lakefs/cmd/import.go
@@ -216,7 +216,7 @@ func runImport(cmd *cobra.Command, args []string) (statusCode int) {
 	if withMerge {
 		fmt.Printf("Merging import changes into lakefs://%s/%s/\n", repoName, repo.DefaultBranch)
 		msg := fmt.Sprintf(onboard.CommitMsgTemplate, stats.CommitRef)
-		commitLog, err := c.Merge(ctx, repoName, onboard.DefaultImportBranchName, repo.DefaultBranch, CommitterName, msg, nil)
+		commitLog, err := c.Merge(ctx, repoName, onboard.DefaultImportBranchName, repo.DefaultBranch, CommitterName, msg, nil, "")
 		if err != nil {
 			fmt.Printf("Merge failed: %s\n", err)
 			return 1

diff --git a/docs/assets/js/swagger.yml b/docs/assets/js/swagger.yml
@@ -439,6 +439,9 @@ components:
           type: object
           additionalProperties:
             type: string
+        strategy:
+          description: In case of a merge conflict, this option will force the merge process to automatically favor changes from the dest branch ('dest-wins') or from the source branch('source-wins'). In case no selection is made, the merge process will fail in case of a conflict
+          type: string
 
     BranchCreation:
       type: object

diff --git a/docs/reference/commands.md b/docs/reference/commands.md
@@ -2102,7 +2102,8 @@ lakectl merge <source ref> <destination ref> [flags]
 {:.no_toc}
 
 ```
-  -h, --help   help for merge
+  -h, --help              help for merge
+      --strategy string   In case of a merge conflict, this option will force the merge process to automatically favor changes from the dest branch ("dest-wins") or from the source branch("source-wins"). In case no selection is made, the merge process will fail in case of a conflict
 ```
 
 

diff --git a/nessie/files/ro_1k b/nessie/files/ro_1k
diff --git a/nessie/files/ro_1k_other b/nessie/files/ro_1k_other
diff --git a/nessie/golden/lakectl_branch_create.golden b/nessie/golden/lakectl_branch_create.golden
@@ -0,0 +1,2 @@
+Source ref: lakefs://${REPO}/${SOURCE_BRANCH}
+created branch '${DEST_BRANCH}' <COMMIT_ID>
diff --git a/nessie/golden/lakectl_commit.golden b/nessie/golden/lakectl_commit.golden
@@ -2,7 +2,7 @@ Branch: lakefs://${REPO}/${BRANCH}
 Commit for branch "${BRANCH}" completed.
 
 ID: <COMMIT_ID>
-Message: nessie_lakectl:TestCommit
+Message: ${MESSAGE}
 Timestamp: <DATE> <TIME> <TZ>
 Parents: <COMMIT_ID>
 
diff --git a/nessie/golden/lakectl_fs_ls_1_file.golden b/nessie/golden/lakectl_fs_ls_1_file.golden
@@ -0,0 +1 @@
+object          <DATE> <TIME> <TZ>    1.0 kB          ${FILE_PATH}
diff --git a/nessie/golden/lakectl_fs_ls_2_file.golden b/nessie/golden/lakectl_fs_ls_2_file.golden
@@ -0,0 +1,2 @@
+object          <DATE> <TIME> <TZ>    1.0 kB          ${FILE_PATH_1}
+object          <DATE> <TIME> <TZ>    1.0 kB          ${FILE_PATH_2}
diff --git a/nessie/golden/lakectl_fs_second_upload.golden b/nessie/golden/lakectl_fs_second_upload.golden
diff --git a/nessie/golden/lakectl_fs_upload.golden b/nessie/golden/lakectl_fs_upload.golden
@@ -1,7 +1,7 @@
-Path: upload_file1.txt
+Path: ${FILE_PATH}
 Modified Time: <DATE> <TIME> <TZ>
-Size: 20 bytes
-Human Size: 20 B
+Size: 1024 bytes
+Human Size: 1.0 kB
 Physical Address: ${STORAGE}/<OBJECT_KEY>
-Checksum: bcd52aa37b0c6151513579d6f21c4414
+Checksum: <CHECKSUM>
 Content-Type: application/octet-stream
diff --git a/nessie/golden/lakectl_log_with_commit.golden b/nessie/golden/lakectl_log_with_commit.golden
@@ -3,7 +3,7 @@ ID:            <COMMIT_ID>
 Author:        nessie
 Date:          <DATE> <TIME> <TZ>
 
-	nessie_lakectl:TestCommit
+	${MESSAGE}
 
 
 ID:            <COMMIT_ID>

diff --git a/nessie/golden/lakectl_merge_conflict.golden b/nessie/golden/lakectl_merge_conflict.golden
@@ -0,0 +1,4 @@
+Source: lakefs://${REPO}/${SOURCE_BRANCH}
+Destination: lakefs://${REPO}/${DEST_BRANCH}
+Conflict found.
+Error executing command.
diff --git a/nessie/golden/lakectl_merge_source_wins.golden b/nessie/golden/lakectl_merge_source_wins.golden
@@ -0,0 +1,3 @@
+Source: lakefs://${REPO}/${SOURCE_BRANCH}
+Destination: lakefs://${REPO}/${DEST_BRANCH}
+Merged "${SOURCE_BRANCH}" into "${DEST_BRANCH}" to get "<COMMIT_ID>".
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		Source ref: lakefs://${REPO}/${SOURCE_BRANCH}
		created branch '${DEST_BRANCH}' <COMMIT_ID>
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		object <DATE> <TIME> <TZ> 1.0 kB ${FILE_PATH_1}
		object <DATE> <TIME> <TZ> 1.0 kB ${FILE_PATH_2}