4 files changed, 701 insertions, 0 deletions
diff --git a/bsfs/query/__init__.py b/bsfs/query/__init__.py
new file mode 100644
index 0000000..21c7389
--- /dev/null
+++ b/bsfs/query/__init__.py
@@ -0,0 +1,20 @@
+"""
+
+Part of the BlackStar filesystem (bsfs) module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# inner-module imports
+from . import ast
+from . import validator as validate
+
+# exports
+__all__: typing.Sequence[str] = (
+    'ast',
+    'validate',
+    )
+
+## EOF ##
diff --git a/bsfs/query/ast/__init__.py b/bsfs/query/ast/__init__.py
new file mode 100644
index 0000000..704d051
--- /dev/null
+++ b/bsfs/query/ast/__init__.py
@@ -0,0 +1,24 @@
+"""Query AST components.
+
+The query AST consists of a Filter syntax tree.
+
+Classes beginning with an underscore (_) represent internal type hierarchies
+and should not be used for parsing. Note that the AST structures do not
+(and cannot) check semantic validity or consistency with a given schema.
+
+Part of the BlackStar filesystem (bsfs) module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# inner-module imports
+from . import filter_ as filter # pylint: disable=redefined-builtin
+
+# exports
+__all__: typing.Sequence[str] = (
+    'filter',
+    )
+
+## EOF ##
diff --git a/bsfs/query/ast/filter_.py b/bsfs/query/ast/filter_.py
new file mode 100644
index 0000000..b129ded
--- /dev/null
+++ b/bsfs/query/ast/filter_.py
@@ -0,0 +1,433 @@
+"""Filter AST.
+
+Note that it is easily possible to construct an AST that is inconsistent with
+a given schema. Furthermore, it is possible to construct a semantically invalid
+AST which that cannot be parsed correctly or includes contradicting statements.
+The AST nodes do not (and cannot) check such issues.
+
+For example, consider the following AST:
+
+>>> Any(ns.bse.collection,
+...     And(
+...         Equals('hello'),
+...         Any(ns.bsm.guid, Any(ns.bsm.guid, Equals('hello'))),
+...         Any(ns.bst.label, Equals('world')),
+...         All(ns.bst.label, Not(Equals('world'))),
+...     )
+... )
+
+This AST has multiple issues that are not verified upon its creation:
+* A condition on a non-literal.
+* A Filter on a literal.
+* Conditions exclude each other
+* The predicate along the branch have incompatible domains and ranges.
+
+Part of the BlackStar filesystem (bsfs) module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+from collections import abc
+import typing
+
+# bsfs imports
+from bsfs.utils import URI, typename, normalize_args
+
+# inner-module imports
+#from . import utils
+
+# exports
+__all__ : typing.Sequence[str] = (
+    # base classes
+    'FilterExpression',
+    'PredicateExpression',
+    # predicate expressions
+    'OneOf',
+    'Predicate',
+    # branching
+    'All',
+    'Any',
+    # aggregators
+    'And',
+    'Or',
+    # value matchers
+    'Equals',
+    'Substring',
+    'EndsWith',
+    'StartsWith',
+    # range matchers
+    'GreaterThan',
+    'LessThan',
+    # misc
+    'Has',
+    'Is',
+    'Not',
+    )
+
+
+## code ##
+
+# pylint: disable=too-few-public-methods # Many expressions use mostly magic methods
+
+class _Expression(abc.Hashable):
+    def __repr__(self) -> str:
+        """Return the expressions's string representation."""
+        return f'{typename(self)}()'
+
+    def __hash__(self) -> int:
+        """Return the expression's integer representation."""
+        return hash(type(self))
+
+    def __eq__(self, other: typing.Any) -> bool:
+        """Return True if *self* and *other* are equivalent."""
+        return isinstance(other, type(self))
+
+
+class FilterExpression(_Expression):
+    """Generic Filter expression."""
+
+
+class PredicateExpression(_Expression):
+    """Generic Predicate expression."""
+
+
+class _Branch(FilterExpression):
+    """Branch the filter along a predicate."""
+
+    # predicate to follow.
+    predicate: PredicateExpression
+
+    # child expression to evaluate.
+    expr: FilterExpression
+
+    def __init__(
+            self,
+            predicate: typing.Union[PredicateExpression, URI],
+            expr: FilterExpression,
+            ):
+        # process predicate argument
+        if isinstance(predicate, URI):
+            predicate = Predicate(predicate)
+        elif not isinstance(predicate, PredicateExpression):
+            raise TypeError(predicate)
+        # process expression argument
+        if not isinstance(expr, FilterExpression):
+            raise TypeError(expr)
+        # assign members
+        self.predicate = predicate
+        self.expr = expr
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.predicate}, {self.expr})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), self.predicate, self.expr))
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) \
+           and self.predicate == other.predicate \
+           and self.expr == other.expr
+
+class Any(_Branch):
+    """Any (and at least one) triple matches."""
+
+
+class All(_Branch):
+    """All (and at least one) triples match."""
+
+
+class _Agg(FilterExpression, abc.Collection):
+    """Combine multiple expressions."""
+
+    # child expressions
+    expr: typing.Set[FilterExpression]
+
+    def __init__(
+            self,
+            *expr: typing.Union[FilterExpression,
+                                typing.Iterable[FilterExpression],
+                                typing.Iterator[FilterExpression]]
+            ):
+        # unfold arguments
+        unfolded = set(normalize_args(*expr))
+        # check type
+        if not all(isinstance(e, FilterExpression) for e in unfolded):
+            raise TypeError(expr)
+        # assign member
+        self.expr = unfolded
+
+    def __contains__(self, expr: typing.Any) -> bool:
+        """Return True if *expr* is among the child expressions."""
+        return expr in self.expr
+
+    def __iter__(self) -> typing.Iterator[FilterExpression]:
+        """Iterator over child expressions."""
+        return iter(self.expr)
+
+    def __len__(self) -> int:
+        """Number of child expressions."""
+        return len(self.expr)
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.expr})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), tuple(self.expr))) # FIXME: Unique hash of different orders over self.expr
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) and self.expr == other.expr
+
+
+class And(_Agg):
+    """All conditions match."""
+
+
+class Or(_Agg):
+    """At least one condition matches."""
+
+
+class Not(FilterExpression):
+    """Invert a statement."""
+
+    # child expression
+    expr: FilterExpression
+
+    def __init__(self, expr: FilterExpression):
+        # check argument
+        if not isinstance(expr, FilterExpression):
+            raise TypeError(expr)
+        # assign member
+        self.expr = expr
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.expr})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), self.expr))
+
+    def __eq__(self, other: typing.Any) -> bool:
+        return super().__eq__(other) and self.expr == other.expr
+
+
+class Has(FilterExpression):
+    """Has predicate N times"""
+
+    # predicate to follow.
+    predicate: PredicateExpression
+
+    # target count
+    count: FilterExpression
+
+    def __init__(
+            self,
+            predicate: typing.Union[PredicateExpression, URI],
+            count: typing.Optional[typing.Union[FilterExpression, int]] = None,
+            ):
+        # check predicate
+        if isinstance(predicate, URI):
+            predicate = Predicate(predicate)
+        elif not isinstance(predicate, PredicateExpression):
+            raise TypeError(predicate)
+        # check count
+        if count is None:
+            count = GreaterThan(1, strict=False)
+        elif isinstance(count, int):
+            count = Equals(count)
+        elif not isinstance(count, FilterExpression):
+            raise TypeError(count)
+        # assign members
+        self.predicate = predicate
+        self.count = count
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.predicate}, {self.count})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), self.predicate, self.count))
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) \
+           and self.predicate == other.predicate \
+           and self.count == other.count
+
+
+class _Value(FilterExpression):
+    """
+    """
+
+    # target value.
+    value: typing.Any
+
+    def __init__(self, value: typing.Any):
+        self.value = value
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.value})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), self.value))
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) and self.value == other.value
+
+
+class Is(_Value):
+    """Match the URI of a node."""
+
+
+class Equals(_Value):
+    """Value matches exactly.
+    NOTE: Value format must correspond to literal type; can be a string, a number, or a Node
+    """
+
+
+class Substring(_Value):
+    """Value matches a substring
+    NOTE: value format must be a string
+    """
+
+
+class StartsWith(_Value):
+    """Value begins with a given string."""
+
+
+class EndsWith(_Value):
+    """Value ends with a given string."""
+
+
+class _Bounded(FilterExpression):
+    """
+    """
+
+    # bound.
+    threshold: float
+
+    # closed (True) or open (False) bound.
+    strict: bool
+
+    def __init__(
+            self,
+            threshold: float,
+            strict: bool = True,
+            ):
+        self.threshold = float(threshold)
+        self.strict = bool(strict)
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.threshold}, {self.strict})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), self.threshold, self.strict))
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) \
+           and self.threshold == other.threshold \
+           and self.strict == other.strict
+
+
+
+class LessThan(_Bounded):
+    """Value is (strictly) smaller than threshold.
+    NOTE: only on numerical literals
+    """
+
+
+class GreaterThan(_Bounded):
+    """Value is (strictly) larger than threshold
+    NOTE: only on numerical literals
+    """
+
+
+class Predicate(PredicateExpression):
+    """A single predicate."""
+
+    # predicate URI
+    predicate: URI
+
+    # reverse the predicate's direction
+    reverse: bool
+
+    def __init__(
+            self,
+            predicate: URI,
+            reverse: typing.Optional[bool] = False,
+            ):
+        # check arguments
+        if not isinstance(predicate, URI):
+            raise TypeError(predicate)
+        # assign members
+        self.predicate = predicate
+        self.reverse = bool(reverse)
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.predicate}, {self.reverse})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), self.predicate, self.reverse))
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) \
+           and self.predicate == other.predicate \
+           and self.reverse == other.reverse
+
+
+class OneOf(PredicateExpression, abc.Collection):
+    """A set of predicate alternatives.
+
+    The predicates' domains must be ascendants or descendants of each other.
+    The overall domain is the most specific one.
+
+    The predicate's domains must be ascendants or descendants of each other.
+    The overall range is the most generic one.
+    """
+
+    # predicate alternatives
+    expr: typing.Set[PredicateExpression]
+
+    def __init__(self, *expr: typing.Union[PredicateExpression, URI]):
+        # unfold arguments
+        unfolded = set(normalize_args(*expr)) # type: ignore [arg-type] # this is getting too complex...
+        # check arguments
+        if len(unfolded) == 0:
+            raise AttributeError('expected at least one expression, found none')
+        # ensure PredicateExpression
+        unfolded = {Predicate(e) if isinstance(e, URI) else e for e in unfolded}
+        # check type
+        if not all(isinstance(e, PredicateExpression) for e in unfolded):
+            raise TypeError(expr)
+        # assign member
+        self.expr = unfolded
+
+    def __contains__(self, expr: typing.Any) -> bool:
+        """Return True if *expr* is among the child expressions."""
+        return expr in self.expr
+
+    def __iter__(self) -> typing.Iterator[PredicateExpression]:
+        """Iterator over child expressions."""
+        return iter(self.expr)
+
+    def __len__(self) -> int:
+        """Number of child expressions."""
+        return len(self.expr)
+
+    def __repr__(self) -> str:
+        return f'{typename(self)}({self.expr})'
+
+    def __hash__(self) -> int:
+        return hash((super().__hash__(), tuple(self.expr))) # FIXME: Unique hash of different orders over self.expr
+
+    def __eq__(self, other) -> bool:
+        return super().__eq__(other) and self.expr == other.expr
+
+
+# Helpers
+
+def IsIn(*values): # pylint: disable=invalid-name # explicitly mimics an expression
+    """Match any of the given URIs."""
+    return Or(Is(value) for value in normalize_args(*values))
+
+def IsNotIn(*values): # pylint: disable=invalid-name # explicitly mimics an expression
+    """Match none of the given URIs."""
+    return Not(IsIn(*values))
+
+## EOF ##
diff --git a/bsfs/query/validator.py b/bsfs/query/validator.py
new file mode 100644
index 0000000..352203a
--- /dev/null
+++ b/bsfs/query/validator.py
@@ -0,0 +1,224 @@
+"""
+
+Part of the BlackStar filesystem (bsfs) module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# bsfs imports
+from bsfs import schema as bsc
+from bsfs.namespace import ns
+from bsfs.utils import errors, typename
+
+# inner-module imports
+from . import ast
+
+# exports
+__all__ : typing.Sequence[str] = (
+    'Filter',
+    )
+
+
+## code ##
+
+class Filter():
+    """Validate a `bsfs.query.ast.filter` query's structure and schema compliance.
+
+    * Conditions (Bounded, Value) can only be applied on literals
+    * Branches, Id, and Has can only be applied on nodes
+    * Predicates' domain and range must match
+    * Predicate paths must follow the schema
+    * Referenced types are present in the schema
+
+    """
+
+    # vertex types
+    T_VERTEX = typing.Union[bsc.Node, bsc.Literal] # FIXME: Shouldn't this be in the schema?
+
+    # schema to validate against.
+    schema: bsc.Schema
+
+    def __init__(self, schema: bsc.Schema):
+        self.schema = schema
+
+    def __call__(self, root_type: bsc.Node, query: ast.filter.FilterExpression):
+        """Validate a filter *query*, assuming the subject having *root_type*.
+
+        Raises a `bsfs.utils.errors.ConsistencyError` if the query violates the schema.
+        Raises a `bsfs.utils.errors.BackendError` if the query structure is invalid.
+
+        """
+        # root_type must be a schema.Node
+        if not isinstance(root_type, bsc.Node):
+            raise TypeError(f'Expected a node, found {typename(root_type)}')
+        # root_type must exist in the schema
+        if root_type not in self.schema.nodes():
+            raise errors.ConsistencyError(f'{root_type} is not defined in the schema')
+        # check root expression
+        self._parse_filter_expression(root_type, query)
+        # all tests passed
+        return True
+
+
+    ## routing methods
+
+    def _parse_filter_expression(self, type_: T_VERTEX, node: ast.filter.FilterExpression):
+        """Route *node* to the handler of the respective FilterExpression subclass."""
+        if isinstance(node, ast.filter.Is):
+            return self._is(type_, node)
+        if isinstance(node, ast.filter.Not):
+            return self._not(type_, node)
+        if isinstance(node, ast.filter.Has):
+            return self._has(type_, node)
+        if isinstance(node, (ast.filter.Any, ast.filter.All)):
+            return self._branch(type_, node)
+        if isinstance(node, (ast.filter.And, ast.filter.Or)):
+            return self._agg(type_, node)
+        if isinstance(node, (ast.filter.Equals, ast.filter.Substring, ast.filter.StartsWith, ast.filter.EndsWith)):
+            return self._value(type_, node)
+        if isinstance(node, (ast.filter.LessThan, ast.filter.GreaterThan)):
+            return self._bounded(type_, node)
+        # invalid node
+        raise errors.BackendError(f'expected filter expression, found {node}')
+
+    def _parse_predicate_expression(self, node: ast.filter.PredicateExpression) -> typing.Tuple[T_VERTEX, T_VERTEX]:
+        """Route *node* to the handler of the respective PredicateExpression subclass."""
+        if isinstance(node, ast.filter.Predicate):
+            return self._predicate(node)
+        if isinstance(node, ast.filter.OneOf):
+            return self._one_of(node)
+        # invalid node
+        raise errors.BackendError(f'expected predicate expression, found {node}')
+
+
+    ## predicate expressions
+
+    def _predicate(self, node: ast.filter.Predicate) -> typing.Tuple[T_VERTEX, T_VERTEX]:
+        # predicate exists in the schema
+        if not self.schema.has_predicate(node.predicate):
+            raise errors.ConsistencyError(f'predicate {node.predicate} is not in the schema')
+        # determine domain and range
+        pred = self.schema.predicate(node.predicate)
+        dom, rng = pred.domain, pred.range
+        if rng is None:
+            # FIXME: It is a design error that Predicates can have a None range...
+            raise errors.BackendError(f'predicate {pred} has no range')
+        if node.reverse:
+            dom, rng = rng, dom # type: ignore [assignment] # variable re-use confuses mypy
+        # return domain and range
+        return dom, rng
+
+    def _one_of(self, node: ast.filter.OneOf) -> typing.Tuple[T_VERTEX, T_VERTEX]:
+        # determine domain and range types
+        # NOTE: select the most specific domain and the most generic range
+        dom, rng = None, None
+        for pred in node:
+            # parse child expression
+            subdom, subrng = self._parse_predicate_expression(pred)
+            try:
+                # determine overall domain
+                if dom is None or subdom < dom: # pick most specific domain
+                    dom = subdom
+                # domains must be related across all child expressions
+                if not subdom <= dom and not subdom >= dom:
+                    raise errors.ConsistencyError(f'domains {subdom} and {dom} are not related')
+            except TypeError as err: # compared literal vs. node
+                raise errors.ConsistencyError(f'domains {subdom} and {dom} are not of the same type') from err
+
+            try:
+                # determine overall range
+                if rng is None or subrng > rng: # pick most generic range
+                    rng = subrng
+                # ranges must be related across all child expressions
+                if not subrng <= rng and not subrng >= rng:
+                    raise errors.ConsistencyError(f'ranges {subrng} and {rng} are not related')
+            except TypeError as err: # compared literal vs. node
+                raise errors.ConsistencyError(f'ranges {subrng} and {rng} are not of the same type') from err
+        # check domain and range
+        if dom is None or rng is None:
+            # OneOf guarantees at least one expression, these two cases cannot happen
+            raise errors.UnreachableError()
+        # return domain and range
+        return dom, rng
+
+
+    ## intermediates
+
+    def _branch(self, type_: T_VERTEX, node: ast.filter._Branch):
+        # type is a Node
+        if not isinstance(type_, bsc.Node):
+            raise errors.ConsistencyError(f'expected a Node, found {type_}')
+        # type exists in the schema
+        # FIXME: Isn't it actually guaranteed that the type (except the root type) is part of the schema?
+        # all types can be traced back to (a) root_type, (b) predicate, or (c) manually set (e.g. in _is).
+        # For (a), we do (and have to) perform a check. For (c), the code base should be consistent throughout
+        # the module, so this is an assumption that has to be ensured in schema.Schema. For (b), we know (and
+        # check) that the predicate is in the schema, hence all node/literals derived from it are also in the
+        # schema by construction of the schema.Schema class. So, why do we check this every time?
+        if type_ not in self.schema.nodes():
+            raise errors.ConsistencyError(f'node {type_} is not in the schema')
+        # predicate is valid
+        dom, rng = self._parse_predicate_expression(node.predicate)
+        # type_ is a subtype of the predicate's domain
+        if not type_ <= dom:
+            raise errors.ConsistencyError(f'expected type {dom} or subtype thereof, found {type_}')
+        # child expression is valid
+        self._parse_filter_expression(rng, node.expr)
+
+    def _agg(self, type_: T_VERTEX, node: ast.filter._Agg):
+        for expr in node:
+            # child expression is valid
+            self._parse_filter_expression(type_, expr)
+
+    def _not(self, type_: T_VERTEX, node: ast.filter.Not):
+        # child expression is valid
+        self._parse_filter_expression(type_, node.expr)
+
+    def _has(self, type_: T_VERTEX, node: ast.filter.Has):
+        # type is a Node
+        if not isinstance(type_, bsc.Node):
+            raise errors.ConsistencyError(f'expected a Node, found {type_}')
+        # type exists in the schema
+        if type_ not in self.schema.nodes():
+            raise errors.ConsistencyError(f'node {type_} is not in the schema')
+        # predicate is valid
+        dom, _= self._parse_predicate_expression(node.predicate)
+        # type_ is a subtype of the predicate's domain
+        if not type_ <= dom:
+            raise errors.ConsistencyError(f'expected type {dom}, found {type_}')
+        # node.count is a numerical expression
+        # FIXME: We have to ensure that ns.xsd.integer is always known in the schema!
+        self._parse_filter_expression(self.schema.literal(ns.xsd.integer), node.count)
+
+
+    ## conditions
+
+    def _is(self, type_: T_VERTEX, node: ast.filter.Is): # pylint: disable=unused-argument # (node)
+        if not isinstance(type_, bsc.Node):
+            raise errors.ConsistencyError(f'expected a Node, found {type_}')
+        if type_ not in self.schema.nodes():
+            raise errors.ConsistencyError(f'node {type_} is not in the schema')
+
+    def _value(self, type_: T_VERTEX, node: ast.filter._Value): # pylint: disable=unused-argument # (node)
+        # type is a literal
+        if not isinstance(type_, bsc.Literal):
+            raise errors.ConsistencyError(f'expected a Literal, found {type_}')
+        # type exists in the schema
+        if type_ not in self.schema.literals():
+            raise errors.ConsistencyError(f'literal {type_} is not in the schema')
+        # FIXME: Check if node.value corresponds to type_
+        # FIXME: A specific literal might be requested (i.e., a numeric type when used in Has)
+
+    def _bounded(self, type_: T_VERTEX, node: ast.filter._Bounded): # pylint: disable=unused-argument # (node)
+        # type is a literal
+        if not isinstance(type_, bsc.Literal):
+            raise errors.ConsistencyError(f'expected a Literal, found {type_}')
+        # type exists in the schema
+        if type_ not in self.schema.literals():
+            raise errors.ConsistencyError(f'literal {type_} is not in the schema')
+        # FIXME: Check if node.value corresponds to type_
+
+
+## EOF ##