RF: Initial steps to start RFing validate-bids to use ValidationResul…

…t records

dandi/bids_utils.py

@@ -1,46 +1,14 @@
 from .utils import pluralize
+from .validate import ValidationResult
 
 
-def is_valid(
-    validation_result: dict,
-    allow_invalid_filenames: bool = False,
-    allow_missing_files: bool = False,
-) -> bool:
-    """Determine whether a dataset validation result marks it as valid.
-
-    Parameters
-    ----------
-    validation_result: dict
-        Dictionary as returned by `dandi.support.bids.validator.validate_bids()`.
-    allow_missing_files: bool, optional
-        Whether to consider the dataset invalid if any mandatory files are not present.
-    allow_invalid_filenames: bool, optional
-        Whether to consider the dataset invalid if any filenames inside are invalid.
-
-    Returns
-    -------
-    bool: whether the dataset validation result marks it as valid.
-
-    """
-
-    if allow_invalid_filenames and allow_missing_files:
-        return True
-    missing_files = [
-        i["regex"] for i in validation_result["schema_tracking"] if i["mandatory"]
-    ]
-    invalid_filenames = validation_result["path_tracking"]
-
-    if missing_files and not allow_missing_files:
-        return False
-    if invalid_filenames and not allow_invalid_filenames:
-        return False
-    else:
-        return True
-
-
-def report_errors(
-    validation_result: dict,
+def print_validation_results(
+    validation_result: list[ValidationResult],
+    # TODO: options for control
+    #  - either report warnings, hints, ...
+    #  - either report groupped by severity, record.id
 ) -> None:
+    raise NotImplementedError("TODO: RF to use ValidationResult records")
     import click
 
     missing_files = [

dandi/cli/cmd_validate.py

@@ -36,19 +36,23 @@ def validate_bids(
         dandi validate-bids /my/path
     """
 
-    from ..bids_utils import is_valid, report_errors
+    from ..bids_utils import print_validation_results
+    from ..validate import Severity
     from ..validate import validate_bids as validate_bids_
 
-    validator_result = validate_bids_(
+    validator_result = validate_bids_(  # Controller
         *paths,
         report=report,
         report_path=report_path,
         schema_version=schema,
     )
-    valid = is_valid(validator_result)
-    report_errors(validator_result)
 
-    if not valid:
+    if validator_result:
+        print_validation_results(validator_result)  # View
+
+    validation_errors = [e for e in validator_result if e.severity == Severity.ERROR]
+
+    if validation_errors:
         raise SystemExit(1)
 
 

dandi/files/bids.py

@@ -66,6 +66,7 @@ def _validate(self) -> None:
                 bids_paths = [str(self.filepath)] + [
                     str(asset.filepath) for asset in self.dataset_files
                 ]
+                # TODO: use RFed data structures, avoid duplicating logic
                 results = validate_bids(*bids_paths)
                 self._dataset_errors: list[str] = []
                 if len(results["path_listing"]) == len(results["path_tracking"]):

dandi/validate.py

@@ -44,7 +44,7 @@ def validate_bids(
     schema_version: Optional[str] = None,
     report: bool = False,
     report_path: str = "",
-) -> dict:
+) -> list[ValidationResult]:
     """Validate BIDS paths.
 
     Parameters
@@ -67,6 +67,7 @@ def validate_bids(
         patterns.
     """
 
+    import bidsschematools
     from bidsschematools.validator import validate_bids as validate_bids_
 
     if report and not report_path:
@@ -80,7 +81,45 @@ def validate_bids(
         schema_version=schema_version,
         report_path=report_path,
     )
-    return dict(validation_result)
+    our_validation_result = []
+    origin = ValidationOrigin(
+        name="bidsschematools",
+        version=bidsschematools.__version__,
+    )
+    for path in validation_result["path_tracking"]:
+        our_validation_result.append(
+            ValidationResult(
+                origin=origin,
+                severity=Severity.ERROR,
+                id="BIDS.WRONG_PATH_TODO",
+                scope=Scope.FILE,
+                path=path,
+                message="TODO",
+                # TODO - discover dandiset or actually BIDS dataset
+                # might want separate the two
+                # dandiset_path="TODO",  # might contain multiple datasets
+                # dataset_path="TODO",  # BIDS dataset in this case
+                # asset_paths: Optional[list[str]] = None
+            )
+        )
+
+    for pattern in validation_result["schema_tracking"]:
+        if pattern["mandatory"]:  # TODO: future proof if gets renamed to required
+            our_validation_result.append(
+                ValidationResult(
+                    origin=origin,
+                    severity=Severity.ERROR,
+                    id="BIDS.MANDATORY_FILE_MISSING",  # we decided to generalize, and not have
+                    scope=Scope.FILE,
+                    message="TODO",
+                    # TODO - discover dandiset or actually BIDS dataset
+                    # might want separate the two
+                    # dandiset_path="TODO",  # might contain multiple datasets
+                    # dataset_path="TODO",  # BIDS dataset in this case
+                    # asset_paths: Optional[list[str]] = None
+                )
+            )
+    return our_validation_result
 
 
 def validate(

setup.cfg

@@ -30,7 +30,7 @@ project_urls =
 python_requires = >=3.7
 install_requires =
     appdirs
-    bidsschematools
+    bidsschematools ~= 0.4.0
     click
     click-didyoumean
     dandischema ~= 0.7.0