Add biom validator

app/grandchallenge/components/models.py

@@ -52,6 +52,7 @@
     validate_docker_image,
 )
 from grandchallenge.components.validators import (
+    validate_biom_format,
     validate_newick_tree_format,
     validate_no_slash_at_ends,
     validate_safe_path,
@@ -1490,6 +1491,8 @@ def validate_user_upload(self, user_upload):
             self.interface.validate_against_schema(value=value)
         elif self.interface.kind == InterfaceKindChoices.NEWICK:
             validate_newick_tree_format(tree=user_upload.read_object())
+        elif self.interface.kind == InterfaceKindChoices.BIOM:
+            validate_biom_format(user_upload=user_upload)
 
         self._user_upload_validated = True
 

app/grandchallenge/components/validators.py

@@ -1,3 +1,8 @@
+from functools import wraps
+from io import BytesIO
+
+import biom
+import h5py
 from billiard.exceptions import SoftTimeLimitExceeded, TimeLimitExceeded
 from Bio.Phylo import NewickIO
 from django.core.exceptions import SuspiciousFileOperation, ValidationError
@@ -24,10 +29,22 @@ def validate_no_slash_at_ends(value):
         raise ValidationError("Path must not begin or end with '/'")
 
 
+def _handle_validation_resource_errors(func):
+    @wraps(func)
+    def wrapper(*args, **kwargs):
+        try:
+            func(*args, **kwargs)
+        except (MemoryError, SoftTimeLimitExceeded, TimeLimitExceeded):
+            raise ValidationError("The file is too large")
+
+    return wrapper
+
+
 def _newick_parser(tree):
     return NewickIO.Parser.from_string(tree)
 
 
+@_handle_validation_resource_errors
 def validate_newick_tree_format(tree):
     """Validates a Newick tree by passing it through a parser"""
     parser = _newick_parser(tree)
@@ -37,10 +54,34 @@ def validate_newick_tree_format(tree):
     try:
         for _ in parser.parse():
             has_tree = True
-    except (MemoryError, SoftTimeLimitExceeded, TimeLimitExceeded):
-        raise ValidationError("The file is too large")
     except NewickIO.NewickError as e:
         raise ValidationError(f"Invalid Newick tree format: {e}")
 
     if not has_tree:
         raise ValidationError("No Newick tree found")
+
+
+@_handle_validation_resource_errors
+def validate_biom_format(*, user_upload):
+    """Validates an uploaded BIOM file by passing it's content through a parser"""
+
+    with BytesIO() as fileobj:
+        # Get the object into memory
+        user_upload.download_fileobj(fileobj)
+        fileobj.seek(0)
+
+        # Attempt to wrap it in a hdf5 handler
+        try:
+            hdf5_file = h5py.File(fileobj, "r")
+        except OSError:
+            raise ValidationError(
+                "Only BIOM in valid HDF5 binary file format are supported"
+            )
+
+        # Attempt to parse it as a BIOM table
+        try:
+            _handle_validation_resource_errors(biom.Table.from_hdf5)(hdf5_file)
+        except ValidationError as e:
+            raise e
+        except Exception:
+            raise ValidationError("Does not appear to be a BIOM-format file")