dayhoff-tools 1.1.13__tar.gz → 1.1.15__tar.gz

{dayhoff_tools-1.1.13 → dayhoff_tools-1.1.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dayhoff-tools
-Version: 1.1.13
+Version: 1.1.15
 Summary: Common tools for all the repos at Dayhoff Labs
 Author: Daniel Martin-Alarcon
 Author-email: dma@dayhofflabs.com

{dayhoff_tools-1.1.13 → dayhoff_tools-1.1.15}/dayhoff_tools/cli/utility_commands.py

@@ -11,6 +11,46 @@ import toml
 import typer
 import yaml
 
+# Import cloud helper lazily inside functions to avoid heavy deps at module load
+
+
+def _warn_if_gcp_default_sa(force_prompt: bool = False) -> None:
+    """Warn the user when the active gcloud principal is the default VM service
+    account.  See detailed docstring later in file (duplicate for early
+    availability)."""
+
+    from dayhoff_tools.cli import cloud_commands as _cc
+
+    try:
+        impersonation = _cc._get_current_gcp_impersonation()
+        user = _cc._get_current_gcp_user()
+        active = impersonation if impersonation != "None" else user
+        short = _cc._get_short_name(active)
+    except Exception:
+        return
+
+    if short != "default VM service account":
+        return
+
+    YELLOW = getattr(_cc, "YELLOW", "\033[0;33m")
+    BLUE = getattr(_cc, "BLUE", "\033[0;36m")
+    RED = getattr(_cc, "RED", "\033[0;31m")
+    NC = getattr(_cc, "NC", "\033[0m")
+
+    print(
+        f"{YELLOW}⚠  You are currently authenticated as the *default VM service account*.{NC}\n"
+        f"{YELLOW}   This will block gsutil/DVC access to private buckets (e.g. warehouse).{NC}\n"
+        f"{YELLOW}   Run {BLUE}dh gcp login{YELLOW} or {BLUE}dh gcp use-devcon{YELLOW} before retrying.{NC}",
+        file=sys.stderr,
+    )
+
+    if force_prompt and sys.stdin.isatty() and sys.stdout.isatty():
+        import questionary
+
+        if not questionary.confirm("Proceed anyway?", default=False).ask():
+            print(f"{RED}Aborted due to unsafe GCP credentials.{NC}", file=sys.stderr)
+            raise SystemExit(1)
+
 
 def test_github_actions_locally():
     """Run the script test_pytest_in_github_actions_container.sh.sh."""
@@ -52,9 +92,16 @@ def get_ancestry(filepath: str) -> None:
 
 def import_from_warehouse_typer() -> None:
     """Import a file from warehouse.
-    This is a thin wrapper around `cli.utils.import_from_warehouse`,
-    with interactive prompts using questionary.
+
+    Emits an early warning if the active GCP credentials are the *default VM
+    service account* because this will prevent DVC/gsutil from accessing the
+    warehouse bucket.  The user can abort the command when running
+    interactively.
     """
+
+    # Early-exit guard for wrong GCP credentials
+    _warn_if_gcp_default_sa(force_prompt=True)
+
     # Import only when the function is called
     import questionary
     from dayhoff_tools.warehouse import import_from_warehouse
@@ -91,8 +138,16 @@ def import_from_warehouse_typer() -> None:
 
 
 def add_to_warehouse_typer() -> None:
-    """Add a new data file to warehouse, and expand its .dvc file with
-    metadata, including ancestor files."""
+    """Add a new data file to warehouse and enrich its generated .dvc file.
+
+    As with *dh wimport*, this command fails when the user is logged in with
+    the default VM service account.  A guard therefore warns the user first
+    and allows them to abort interactively.
+    """
+
+    # Early-exit guard for wrong GCP credentials
+    _warn_if_gcp_default_sa(force_prompt=True)
+
     # Import only when the function is called
     import questionary
     from dayhoff_tools.warehouse import add_to_warehouse
@@ -548,3 +603,8 @@ def update_dependencies(
     except Exception as e:
         print(f"An unexpected error occurred: {e}")
         sys.exit(1)
+
+
+# ----------------------
+# Cloud credential guard
+# ----------------------

{dayhoff_tools-1.1.13 → dayhoff_tools-1.1.15}/dayhoff_tools/deployment/processors.py

@@ -283,7 +283,6 @@ class MMSeqsProfileProcessor(Processor):
         aln_db = mmseqs_temp_dir / "alnDB"
         profile_db = mmseqs_temp_dir / "profileDB"
         result_db = mmseqs_temp_dir / "resultDB"
-        hits_db = mmseqs_temp_dir / "hitsDB"
 
         try:
             # 1. Create query database
@@ -417,8 +416,6 @@ class MMSeqsProfileProcessor(Processor):
                     str(intermediate_hits_fasta_file),  # output FASTA
                     "--use-fasta-header",
                     "1",
-                    "--threads",
-                    self.num_threads,
                 ],
                 "Extract hit sequences to FASTA via result2flat",
                 run_base_dir,

{dayhoff_tools-1.1.13 → dayhoff_tools-1.1.15}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "poetry.core.masonry.api"
 
 [project]
 name = "dayhoff-tools"
-version = "1.1.13"
+version = "1.1.15"
 description = "Common tools for all the repos at Dayhoff Labs"
 authors = [
     {name = "Daniel Martin-Alarcon", email = "dma@dayhofflabs.com"}