GraphQL commit deletion

src/doc_builder/commands/push.py

@@ -60,6 +60,57 @@ def create_additions(library_name: str) -> List[Dict]:
     return additions
 
 
+def create_deletions(repo_id: str, library_name: str, token: str) -> List[Dict]:
+    """
+    Given `repo_id/library_name` path, returns [FileDeletion!]!: [{path: "some_path"}, ...]
+    see more here: https://docs.github.com/en/graphql/reference/input-objects#filechanges
+    """
+    # 1. find url for `doc-build-dev/{library_name}` ex: doc-build-dev/accelerate
+    res = requests.get(
+        f"https://api.github.com/repos/{repo_id}/git/trees/heads/main", headers={"Authorization": f"bearer {token}"}
+    )
+    if res.status_code != 200:
+        raise Exception(f"create_deletions failed (GET tree root): {res.message}")
+    json = res.json()
+    node = next(filter(lambda node: node["path"] == library_name, json["tree"]), None)
+    url = node["url"]
+
+    # 2. find url for `doc-build-dev/{library_name}/{doc_version}` ex: doc-build-dev/accelerate/pr_365
+    root_folder = Path(library_name)
+    doc_version_folder = next(filter(lambda x: not x.is_file(), root_folder.glob("*")), None).relative_to(root_folder)
+    doc_version_folder = str(doc_version_folder)
+    res = requests.get(url, headers={"Authorization": f"bearer {token}"})
+    if res.status_code != 200:
+        raise Exception(f"create_deletions failed (GET tree root/{repo_id}): {res.message}")
+    json = res.json()
+    node = next(filter(lambda node: node["path"] == doc_version_folder, json["tree"]), None)
+    if node is None:
+        # there is no need to delete since the path does not exist
+        return []
+    url = node["url"]
+
+    # 3. list paths in `doc-build-dev/{library_name}/{doc_version}/**/*` ex: doc-build-dev/accelerate/pr_365/**/*
+    res = requests.get(f"{url}?recursive=true", headers={"Authorization": f"bearer {token}"})
+    if res.status_code != 200:
+        raise Exception(f"create_deletions failed (GET tree root/{repo_id}/{doc_version_folder}): {res.message}")
+    json = res.json()
+    tree = json["tree"]
+
+    # 4. list paths in currently built doc folder
+    built_docs_path = Path(f"{library_name}/{doc_version_folder}").absolute()
+    built_docs_files = [x for x in built_docs_path.glob("**/*") if x.is_file()]
+    built_docs_files_relative = set([str(f.relative_to(built_docs_path)) for f in built_docs_files])
+
+    # 5. deletions = set difference between step 3 & 4
+    deletions = [
+        {"path": f"{library_name}/{doc_version_folder}/{node['path']}"}
+        for node in tree
+        if node["type"] == "blob" and node["path"] not in built_docs_files_relative
+    ]
+
+    return deletions
+
+
 MAX_CHUNK_LEN = 3e7  # 30 Megabytes
 
 
@@ -95,14 +146,15 @@ def create_additions_chunks(additions: List[Dict]) -> List[List[Dict]]:
 mutation (
   $repo_id: String!
   $additions: [FileAddition!]!
+  $deletions: [FileDeletion!]!
   $head_oid: GitObjectID!
   $commit_msg: String!
 ) {
   createCommitOnBranch(
     input: {
       branch: { repositoryNameWithOwner: $repo_id, branchName: "main" }
       message: { headline: $commit_msg }
-      fileChanges: { additions: $additions }
+      fileChanges: { additions: $additions, deletions: $deletions }
       expectedHeadOid: $head_oid
     }
   ) {
@@ -112,7 +164,9 @@ def create_additions_chunks(additions: List[Dict]) -> List[List[Dict]]:
 """
 
 
-def create_commit(client: Client, repo_id: str, additions: List[Dict], token: str, commit_msg: str):
+def create_commit(
+    client: Client, repo_id: str, additions: List[Dict], deletions: List[Dict], token: str, commit_msg: str
+):
     """
     Commits additions and/or deletions to a repository using Github GraphQL mutation `createCommitOnBranch`
     see more here: https://docs.github.com/en/graphql/reference/mutations#createcommitonbranch
@@ -122,6 +176,7 @@ def create_commit(client: Client, repo_id: str, additions: List[Dict], token: st
     head_oid = get_head_oid(repo_id, token)
     params = {
         "additions": additions,
+        "deletions": deletions,
         "repo_id": repo_id,
         "head_oid": head_oid,
         "commit_msg": commit_msg,
@@ -143,7 +198,9 @@ def push_command(args):
     number_of_retries = args.n_retries
     n_seconds_sleep = 5
 
-    # commit file additions
+    # file deletions
+    deletions = create_deletions(args.doc_build_repo_id, args.library_name, args.token)
+    # file additions
     time_start = time()
     additions = create_additions(args.library_name)
     time_end = time()
@@ -158,9 +215,13 @@ def push_command(args):
                 url="https://api.github.com/graphql", headers={"Authorization": f"bearer {args.token}"}, verify=True
             )
             with Client(transport=transport, fetch_schema_from_transport=True, execute_timeout=None) as gql_client:
+                # commit file deletions
+                create_commit(
+                    gql_client, args.doc_build_repo_id, [], deletions, args.token, f"Deleteions: {args.commit_msg}"
+                )
                 # commit push chunks additions
                 for i, additions in enumerate(additions_chunks):
-                    create_commit(gql_client, args.doc_build_repo_id, additions, args.token, args.commit_msg)
+                    create_commit(gql_client, args.doc_build_repo_id, additions, [], args.token, args.commit_msg)
                     print(f"Committed additions chunk: {i+1}/{len(additions_chunks)}")
             break
         except Exception as e: