fix: improve validation of bulk imports

src/obsidian_metadata/_utils/__init__.py

@@ -6,10 +6,12 @@ from obsidian_metadata._utils.utilities import (
     clean_dictionary,
     clear_screen,
     dict_contains,
+    dict_keys_to_lower,
     dict_values_to_lists_strings,
     docstring_parameter,
     merge_dictionaries,
     remove_markdown_sections,
+    validate_csv_bulk_imports,
     version_callback,
 )
 
@@ -18,11 +20,12 @@ __all__ = [
     "clean_dictionary",
     "clear_screen",
     "dict_contains",
+    "dict_keys_to_lower",
     "dict_values_to_lists_strings",
     "docstring_parameter",
     "LoggerManager",
     "merge_dictionaries",
     "remove_markdown_sections",
-    "vault_validation",
+    "validate_csv_bulk_imports",
     "version_callback",
 ]

src/obsidian_metadata/_utils/utilities.py

@@ -1,11 +1,15 @@
 """Utility functions."""
+import csv
 import re
 from os import name, system
+from pathlib import Path
 from typing import Any
 
 import typer
 
 from obsidian_metadata.__version__ import __version__
+from obsidian_metadata._utils import alerts
+from obsidian_metadata._utils.alerts import logger as log
 from obsidian_metadata._utils.console import console
 
 
@@ -63,6 +67,18 @@ def dict_contains(
     return key in dictionary and value in dictionary[key]
 
 
+def dict_keys_to_lower(dictionary: dict) -> dict:
+    """Convert all keys in a dictionary to lowercase.
+
+    Args:
+        dictionary (dict): Dictionary to convert
+
+    Returns:
+        dict: Dictionary with all keys converted to lowercase
+    """
+    return {key.lower(): value for key, value in dictionary.items()}
+
+
 def dict_values_to_lists_strings(
     dictionary: dict,
     strip_null_values: bool = False,
@@ -182,6 +198,55 @@ def remove_markdown_sections(
     return text
 
 
+def validate_csv_bulk_imports(csv_path: Path, note_paths: list) -> dict[str, list[dict[str, str]]]:
+    """Validate the bulk import CSV file.
+
+    Args:
+        csv_path (dict): Dictionary to validate
+        note_paths (list): List of paths to all notes in vault
+
+    Returns:
+        dict: Validated dictionary
+    """
+    csv_dict: dict[str, Any] = {}
+    with csv_path.expanduser().open("r") as csv_file:
+        csv_reader = csv.DictReader(csv_file, delimiter=",")
+        row_num = 0
+        for row in csv_reader:
+            if row_num == 0:
+                if "path" not in row:
+                    raise typer.BadParameter("Missing 'path' column in CSV file")
+                if "type" not in row:
+                    raise typer.BadParameter("Missing 'type' column in CSV file")
+                if "key" not in row:
+                    raise typer.BadParameter("Missing 'key' column in CSV file")
+                if "value" not in row:
+                    raise typer.BadParameter("Missing 'value' column in CSV file")
+            row_num += 1
+
+            if row["path"] not in csv_dict:
+                csv_dict[row["path"]] = []
+
+            csv_dict[row["path"]].append(
+                {"type": row["type"], "key": row["key"], "value": row["value"]}
+            )
+
+        if row_num == 0 or row_num == 1:
+            raise typer.BadParameter("Empty CSV file")
+
+        paths_to_remove = [x for x in csv_dict if x not in note_paths]
+
+        for _path in paths_to_remove:
+            alerts.warning(f"'{_path}' does not exist in vault. Skipping...")
+            del csv_dict[_path]
+
+        if len(csv_dict) == 0:
+            log.error("No paths in the CSV file matched paths in the vault")
+            raise typer.Exit(1)
+
+    return csv_dict
+
+
 def version_callback(value: bool) -> None:
     """Print version and exit."""
     if value:

src/obsidian_metadata/models/application.py

@@ -1,7 +1,6 @@
 """Questions for the cli."""
 
 
-import csv
 from pathlib import Path
 from typing import Any
 
@@ -11,7 +10,7 @@ from rich import box
 from rich.table import Table
 
 from obsidian_metadata._config import VaultConfig
-from obsidian_metadata._utils import alerts
+from obsidian_metadata._utils import alerts, validate_csv_bulk_imports
 from obsidian_metadata._utils.console import console
 from obsidian_metadata.models import InsertLocation, Vault, VaultFilter
 from obsidian_metadata.models.enums import MetadataType
@@ -301,18 +300,12 @@ class Application:
             alerts.error("File must be a CSV file")
             return
 
-        csv_dict: dict[str, Any] = {}
-        with csv_path.open("r") as csv_file:
-            csv_reader = csv.DictReader(csv_file, delimiter=",")
-            for row in csv_reader:
-                if row["path"] not in csv_dict:
-                    csv_dict[row["path"]] = []
+        note_paths = [
+            str(n.note_path.relative_to(self.vault.vault_path)) for n in self.vault.all_notes
+        ]
 
-                csv_dict[row["path"]].append(
-                    {"type": row["type"], "key": row["key"], "value": row["value"]}
-                )
-
-        num_changed = self.vault.update_from_dict(csv_dict)
+        dict_from_csv = validate_csv_bulk_imports(csv_path, note_paths)
+        num_changed = self.vault.update_from_dict(dict_from_csv)
 
         if num_changed == 0:
             alerts.warning("No notes were changed")

src/obsidian_metadata/models/vault.py

@@ -572,7 +572,7 @@ class Vault:
         for _note in self.all_notes:
             path = _note.note_path.relative_to(self.vault_path)
             if str(path) in dictionary:
-                log.debug(f"Updating metadata for {path}")
+                log.info(f"Updating metadata for '{path}'")
                 num_changed += 1
                 _note.delete_all_metadata()
                 for row in dictionary[str(path)]:
@@ -590,7 +590,6 @@ class Vault:
                         )
 
                     if row["type"].lower() == "tag" or row["type"].lower() == "tags":
-                        console.print(f"Adding tag {row['value']}")
                         _note.add_metadata(
                             area=MetadataType.TAGS,
                             value=row["value"],