dayhoff-tools 1.1.12__tar.gz → 1.1.14__tar.gz

{dayhoff_tools-1.1.12 → dayhoff_tools-1.1.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dayhoff-tools
-Version: 1.1.12
+Version: 1.1.14
 Summary: Common tools for all the repos at Dayhoff Labs
 Author: Daniel Martin-Alarcon
 Author-email: dma@dayhofflabs.com

{dayhoff_tools-1.1.12 → dayhoff_tools-1.1.14}/dayhoff_tools/cli/utility_commands.py

@@ -6,11 +6,14 @@ import shutil
 import subprocess
 import sys
 from pathlib import Path
+from typing import Optional
 
 import toml
 import typer
 import yaml
 
+# Import cloud helper lazily inside functions to avoid heavy deps at module load
+
 
 def test_github_actions_locally():
     """Run the script test_pytest_in_github_actions_container.sh.sh."""
@@ -52,9 +55,16 @@ 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 +101,16 @@ 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 +566,68 @@ def update_dependencies(
     except Exception as e:
         print(f"An unexpected error occurred: {e}")
         sys.exit(1)
+
+
+# ----------------------
+# Cloud credential guard
+# ----------------------
+
+
+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.
+
+    This situation prevents gsutil/DVC from accessing private buckets and is
+    almost never what we want when interacting with the warehouse.  The
+    function prints a coloured warning and, when *force_prompt* is *True* and
+    the session is interactive, asks the user whether to continue or abort the
+    command.
+
+    Parameters
+    ----------
+    force_prompt : bool, default False
+        When *True* and running in an interactive shell (stdin & stdout are
+        TTYs), display a yes/no prompt (via *questionary.confirm*) asking the
+        user whether to proceed.  If the user declines, ``SystemExit(1)`` is
+        raised to stop the CLI command immediately.  In non-interactive
+        contexts the function merely prints the warning without prompting.
+    """
+
+    # Delay heavy import until the function is actually needed
+    from dayhoff_tools.cli import cloud_commands as _cc  # local import to ease testing
+
+    try:
+        impersonation = _cc._get_current_gcp_impersonation()
+        user = _cc._get_current_gcp_user()
+        active = impersonation if impersonation != "None" else user
+        short_name = _cc._get_short_name(active)
+    except Exception:
+        # If gcloud is missing or some other unexpected error occurs, fail
+        # gracefully by doing nothing.
+        return
+
+    if short_name != "default VM service account":
+        # All good – nothing to warn about
+        return
+
+    # Colour constants – fall back to empty strings if not available
+    RED = getattr(_cc, "RED", "\033[0;31m")
+    YELLOW = getattr(_cc, "YELLOW", "\033[0;33m")
+    BLUE = getattr(_cc, "BLUE", "\033[0;36m")
+    NC = getattr(_cc, "NC", "\033[0m")
+
+    warning_msg = (
+        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}"
+    )
+    print(warning_msg, file=sys.stderr)
+
+    interactive = sys.stdin.isatty() and sys.stdout.isatty()
+    if force_prompt and interactive:
+        import questionary
+
+        proceed = questionary.confirm("Proceed anyway?", default=False).ask()
+        if not proceed:
+            print(f"{RED}Aborted due to unsafe GCP credentials.{NC}", file=sys.stderr)
+            raise SystemExit(1)

{dayhoff_tools-1.1.12 → dayhoff_tools-1.1.14}/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
@@ -406,32 +405,19 @@ class MMSeqsProfileProcessor(Processor):
                 run_base_dir,
             )
 
-            # 9. Create subdatabase of hits from original target_db
+            # 9. Extract hit sequences directly to FASTA using result2flat
             self._run_mmseqs_command(
                 [
                     "mmseqs",
-                    "createsubdb",
-                    str(
-                        target_db
-                    ),  # inputDB: database to filter (all target sequences)
-                    str(result_db),  # filterDB: IDs appearing in result_db (2nd column)
-                    str(hits_db),
-                    "--subdb-mode",
-                    "1",  # Use mode 1 to extract target sequences (2nd column)
-                ],
-                "Create hits subDB from target_db",
-                run_base_dir,
-            )
-
-            # 10. Convert hit sequences to FASTA -> to intermediate file
-            self._run_mmseqs_command(
-                [
-                    "mmseqs",
-                    "convert2fasta",
-                    str(hits_db),
-                    str(intermediate_hits_fasta_file),
+                    "result2flat",
+                    str(profile_db),  # queryDB used in search
+                    str(target_db),  # targetDB
+                    str(result_db),  # resultDB
+                    str(intermediate_hits_fasta_file),  # output FASTA
+                    "--use-fasta-header",
+                    "1",
                 ],
-                "Convert hits to FASTA",
+                "Extract hit sequences to FASTA via result2flat",
                 run_base_dir,
             )
 

{dayhoff_tools-1.1.12 → dayhoff_tools-1.1.14}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "poetry.core.masonry.api"
 
 [project]
 name = "dayhoff-tools"
-version = "1.1.12"
+version = "1.1.14"
 description = "Common tools for all the repos at Dayhoff Labs"
 authors = [
     {name = "Daniel Martin-Alarcon", email = "dma@dayhofflabs.com"}