From d14ff302dbd206da3f4163bc38148efef64f9f19 Mon Sep 17 00:00:00 2001
From: Juho Inkinen <34240031+juhoinkinen@users.noreply.github.com>
Date: Fri, 12 Apr 2024 15:15:11 +0300
Subject: [PATCH] Refactor to address quality complains

---
 annif/cli.py      | 19 ++-----------------
 annif/cli_util.py | 28 ++++++++++++++++++++++++++--
 2 files changed, 28 insertions(+), 19 deletions(-)

diff --git a/annif/cli.py b/annif/cli.py
index 95553f858..a0178bfe3 100644
--- a/annif/cli.py
+++ b/annif/cli.py
@@ -609,7 +609,7 @@ def run_upload(project_ids_pattern, repo_id, token, commit_message):
     Hub repository. An authentication token and commit message can be given with
     options.
     """
-    from huggingface_hub import HfApi, preupload_lfs_files
+    from huggingface_hub import HfApi
     from huggingface_hub.utils import HfHubHTTPError, HFValidationError
 
     projects = cli_util.get_matching_projects(project_ids_pattern)
@@ -621,24 +621,9 @@ def run_upload(project_ids_pattern, repo_id, token, commit_message):
         else f"Upload project(s) {project_ids_pattern} with Annif"
     )
 
-    project_dirs = {p.datadir for p in projects}
-    vocab_dirs = {p.vocab.datadir for p in projects}
-    data_dirs = project_dirs.union(vocab_dirs)
-
     fobjs, operations = [], []
     try:
-        for data_dir in data_dirs:
-            logger.debug(f"Archiving directory {data_dir}")
-            fobj, operation = cli_util.prepare_datadir_commit(data_dir)
-            logger.debug(f"Preuploading to {operation.path_in_repo}")
-            preupload_lfs_files(repo_id, additions=[operation])
-            fobjs.append(fobj)
-            operations.append(operation)
-        for project in projects:
-            fobj, operation = cli_util.prepare_config_commit(project)
-            fobjs.append(fobj)
-            operations.append(operation)
-
+        fobjs, operations = cli_util.prepare_commits(projects, repo_id)
         api = HfApi()
         api.create_commit(
             repo_id=repo_id,
diff --git a/annif/cli_util.py b/annif/cli_util.py
index 978214483..7a67d1f8a 100644
--- a/annif/cli_util.py
+++ b/annif/cli_util.py
@@ -252,7 +252,31 @@ def get_matching_projects(pattern: str) -> list[AnnifProject]:
     ]
 
 
-def prepare_datadir_commit(data_dir: str) -> tuple[io.BufferedRandom, Any]:
+def prepare_commits(projects: list[AnnifProject], repo_id: str) -> tuple[list, list]:
+    """Prepare and pre-upload data and config commit operations for projects to a
+    Hugging Face Hub repository."""
+    from huggingface_hub import preupload_lfs_files
+
+    fobjs, operations = [], []
+    data_dirs = {p.datadir for p in projects}
+    vocab_dirs = {p.vocab.datadir for p in projects}
+    all_dirs = data_dirs.union(vocab_dirs)
+
+    for data_dir in all_dirs:
+        fobj, operation = _prepare_datadir_commit(data_dir)
+        preupload_lfs_files(repo_id, additions=[operation])
+        fobjs.append(fobj)
+        operations.append(operation)
+
+    for project in projects:
+        fobj, operation = _prepare_config_commit(project)
+        fobjs.append(fobj)
+        operations.append(operation)
+
+    return fobjs, operations
+
+
+def _prepare_datadir_commit(data_dir: str) -> tuple[io.BufferedRandom, Any]:
     from huggingface_hub import CommitOperationAdd
 
     zip_repo_path = data_dir.split(os.path.sep, 1)[1] + ".zip"
@@ -261,7 +285,7 @@ def prepare_datadir_commit(data_dir: str) -> tuple[io.BufferedRandom, Any]:
     return fobj, operation
 
 
-def prepare_config_commit(project: AnnifProject) -> tuple[io.BytesIO, Any]:
+def _prepare_config_commit(project: AnnifProject) -> tuple[io.BytesIO, Any]:
     from huggingface_hub import CommitOperationAdd
 
     config_repo_path = project.project_id + ".cfg"