relationalai 0.11.3__py3-none-any.whl → 0.11.4__py3-none-any.whl

relationalai/semantics/rel/rel_utils.py

@@ -125,7 +125,7 @@ OPERATORS = {
     "ends_with": "::std::common::ends_with",
     "contains": "::std::common::contains",
     "substring": "::std::common::substring",
-    "num_chars": "pyrel_num_chars",
+    "num_chars": "::std::common::num_chars",
     "like_match": "::std::common::like_match",
     "lower": "::std::common::lowercase",
     "upper": "::std::common::uppercase",
@@ -134,7 +134,14 @@ OPERATORS = {
     "replace": "rel_primitive_replace",
     "split": "::std::common::string_split",
     "split_part": "::std::common::string_split",
+
+    # regex
     "regex_match": "rel_primitive_regex_match",
+    "regex_match_all": "::std::common::regex_match_all",
+    "regex_search": "pyrel_regex_search",
+    "capture_group_by_index": "::std::common::capture_group_by_index",
+    "capture_group_by_name": "::std::common::capture_group_by_name",
+    "escape_regex_metachars": "::std::common::escape_regex_metachars",
 
     # ints
     "parse_int64": "::std::common::parse_int",

relationalai/semantics/rel/rewrite/extract_common.py

@@ -7,6 +7,7 @@ from relationalai.semantics.metamodel import ir, factory as f, helpers
 from relationalai.semantics.metamodel.compiler import Pass, group_tasks
 from relationalai.semantics.metamodel.util import OrderedSet, ordered_set
 from relationalai.semantics.metamodel import dependency
+from relationalai.semantics.metamodel import builtins
 
 class ExtractCommon(Pass):
     """
@@ -35,6 +36,17 @@ class ExtractCommon(Pass):
                 Logical2 ...
     """
 
+    # The extraction plan heuristic is as follows:
+    #
+    # Given a set of binder tasks B and a set of extractable tasks E, we find:
+    #   - A subset of common tasks C in B, and
+    #   - A subset of exposed variables V output from tasks in C
+    # where:
+    #   - The intersection of common dependencies of all tasks in E are contained in C
+    #     (including transitive dependencies)
+    #   - The union of input variables for all tasks in E intersected with the output
+    #     variables of tasks in C are contained in V
+
     #--------------------------------------------------
     # Public API
     #--------------------------------------------------
@@ -77,112 +89,124 @@ class ExtractCommon(Pass):
             return task
 
     def handle_logical(self, task: ir.Logical, ctx: Context):
-        # create frame to process the children
-        # with ctx_frame(ctx, task) as frame:
-            # process the original body
-            groups = group_tasks(task.body, {
-                "binders": helpers.BINDERS,
-                "composites": helpers.COMPOSITES
-            })
-            binders = groups["binders"]
-            composites = groups["composites"]
-
-            # the new body of the rewritten task
-            body:OrderedSet[ir.Task] = ordered_set()
-
-            # quick check to see if it's worth doing more analysis; we only want to extract
-            # common binders if there are multiple, and there are also multiple composites
-            # that will be extracted by the flatten pass later (so that they can share the
-            # extracted logic).
-            plan = None
-            if len(binders) > 1 and composites:
-                extractables = flatten.extractables(composites)
-                # only makes sense to extract common if at least one nested composite will be extracted
-                if extractables:
-                    # make a plan to extract common tasks from the logical
-                    plan = self._create_extraction_plan(binders, composites, extractables, ctx)
-                    if plan and len(composites) > 1:
-                        # plan is worthwhile and there are multiple composites, extract the common body and add the connection to the body
-                        exposed_vars = plan.exposed_vars.get_list()
-                        plan.common_reference = f.lookup(helpers.extract(task, plan.common_body, exposed_vars, ctx.rewrite_ctx, "common"), exposed_vars)
-                        # if we are not distributing the reference, add to the main body
-                        if not plan.distribute_common_reference:
-                            body.add(plan.common_reference)
-
-            # if we have a plan and will distribute the common reference, keep track of
-            # variables still needed by the remaining tasks, as they need to be hoisted by
-            # the remaining composites that get the common reference
-            remaining_vars = None
-            if plan and plan.distribute_common_reference:
-                # add variables hoisted by this logical that are in the exposed vars, to
-                # make sure they are hoisted all the way through
-                remaining_vars = OrderedSet.from_iterable(helpers.hoisted_vars(task.hoisted)) & plan.exposed_vars
-                for child in task.body:
-                    if child in groups["other"] or child not in plan.remaining_body or child in composites:
-                        continue
-                    remaining_vars.update(ctx.info.task_inputs(child))
-                    remaining_vars.update(ctx.info.task_outputs(child))
-                remaining_vars = remaining_vars & plan.exposed_vars
-
-            # if the plan was not used in one of the cases above, ignore it completely, we
-            # are not extracting common nor distributing it around
-            if plan and not plan.distribute_common_reference and not len(composites) > 1:
-                plan = None
-
-            # recursively handle children
+        # Process the original body to find binders and extractables. The Flatten pass later
+        # will extract out both composites and effects, so we group them together here
+        groups = group_tasks(task.body, {
+            "binders": helpers.BINDERS,
+            "composites_and_effects": helpers.COMPOSITES + helpers.EFFECTS,
+        })
+        binders = groups["binders"]
+        composites_and_effects = groups["composites_and_effects"]
+
+        # the new body of the rewritten task
+        body:OrderedSet[ir.Task] = ordered_set()
+
+        # quick check to see if it's worth doing more analysis; we only want to extract
+        # common binders if there are multiple, and there are also multiple composites
+        # that will be extracted by the flatten pass later (so that they can share the
+        # extracted logic).
+        plan = None
+        if len(binders) > 1 and composites_and_effects:
+            extractables = flatten.extractables(composites_and_effects)
+            # only makes sense to extract common if at least one nested composite will be
+            # extracted during Flatten
+            if extractables:
+                # make a plan to extract common tasks from the logical
+                plan = self._create_extraction_plan(binders, composites_and_effects, extractables, ctx)
+                if plan and len(composites_and_effects) > 1:
+                    # plan is worthwhile and there are multiple composites, extract the common body and add the connection to the body
+                    exposed_vars = plan.exposed_vars.get_list()
+                    plan.common_reference = f.lookup(helpers.extract(task, plan.common_body, exposed_vars, ctx.rewrite_ctx, "common"), exposed_vars)
+                    # if we are not distributing the reference, add to the main body
+                    if not plan.distribute_common_reference:
+                        body.add(plan.common_reference)
+
+        # if we have a plan and will distribute the common reference, keep track of
+        # variables still needed by the remaining tasks, as they need to be hoisted by
+        # the remaining composites that get the common reference
+        remaining_vars = None
+        if plan and plan.distribute_common_reference:
+            # add variables hoisted by this logical that are in the exposed vars, to
+            # make sure they are hoisted all the way through
+            remaining_vars = OrderedSet.from_iterable(helpers.hoisted_vars(task.hoisted)) & plan.exposed_vars
             for child in task.body:
-                # skip children that were extracted
-                if plan and child not in groups["other"] and child not in plan.remaining_body and child not in composites:
+                if child in groups["other"] or child not in plan.remaining_body or child in composites_and_effects:
                     continue
+                remaining_vars.update(ctx.info.task_inputs(child))
+                remaining_vars.update(ctx.info.task_outputs(child))
+            remaining_vars = remaining_vars & plan.exposed_vars
 
-                # no plan or child is not a composite, so just add the handled to the body
-                if not plan or child not in composites:
-                    body.add(self.handle(child, ctx))
-                    continue
+        # if the plan was not used in one of the cases above, ignore it completely, we
+        # are not extracting common nor distributing it around
+        if plan and not plan.distribute_common_reference and not len(composites_and_effects) > 1:
+            plan = None
+
+        # recursively handle children
+        for child in task.body:
+            # skip children that were extracted
+            if plan and child not in groups["other"] and child not in plan.remaining_body and child not in composites_and_effects:
+                continue
+
+            # no plan or child is not a composite, so just add the handled to the body
+            if not plan or child not in composites_and_effects:
+                body.add(self.handle(child, ctx))
+                continue
 
-                # there is a plan and the child is in composites, so...
-                replacement = self.handle(child, ctx)
-
-                # this child needs either extra local dependencies or the common reference
-                if child in plan.local_dependencies or plan.distribute_common_reference:
-                    # the new body will have maybe the common reference and the local deps
-                    replacement_body = ordered_set()
-                    hoisted = OrderedSet.from_iterable(replacement.hoisted if isinstance(replacement, helpers.COMPOSITES) else [])
-                    if plan.distribute_common_reference:
-                        if len(composites) == 1:
-                            # if there's a single composite, just insert the whole common body into it
-                            replacement_body.update(plan.common_body)
-                        else:
-                            # otherwise insert a clone of the reference on the extracted rule
-                            assert(plan.common_reference)
-                            replacement_body.add(plan.common_reference.clone())
-                        # add remaining vars to hoisted, making sure there's no duplicates (due to VarOrDefault)
-                        hoisted_vars = helpers.hoisted_vars(hoisted)
-                        if remaining_vars:
-                            hoisted = OrderedSet.from_iterable(filter(lambda v: v not in hoisted_vars, remaining_vars)) | hoisted
-
-                    if child in plan.local_dependencies:
-                        for local_dep in plan.local_dependencies[child]:
-                            replacement_body.add(local_dep.clone())
-                        #replacement_body.update(plan.local_dependencies[child])
-
-                    if isinstance(replacement, ir.Logical):
-                        # if the replacements is a logical, we can just add to the body
-                        body.add(replacement.reconstruct(
-                            replacement.engine,
-                            tuple(hoisted.get_list()),
-                            tuple(replacement_body.update(replacement.body).get_list()),
-                            replacement.annotations
-                        ))
+            # there is a plan and the child is in composites, so...
+            replacement = self.handle(child, ctx)
+
+            # this child needs either extra local dependencies or the common reference
+            if child in plan.local_dependencies or plan.distribute_common_reference:
+                # the new body will have maybe the common reference and the local deps
+                replacement_body = ordered_set()
+
+                hoisted = OrderedSet()
+                if isinstance(replacement, ir.Logical):
+                    # if replacement is a logical, just keep the same hoisted vars
+                    hoisted.update(replacement.hoisted)
+                else:
+                    # otherwise, we need to hoist the vars that are output from local deps
+                    # and input to the replacement task
+                    dep_outputs = OrderedSet()
+                    for d in plan.local_dependencies.get(child, ordered_set()):
+                        dep_outputs.update(ctx.info.task_outputs(d))
+                    hoisted.update(dep_outputs & ctx.info.task_inputs(replacement))
+
+                if plan.distribute_common_reference:
+                    if len(composites_and_effects) == 1:
+                        # if there's a single composite, just insert the whole common body into it
+                        replacement_body.update(plan.common_body)
                     else:
-                        # anything else is now wrapped in a logical so that we can add the
-                        # local dependencies
-                        body.add(f.logical(replacement_body.add(replacement).get_list(), hoisted.get_list(), replacement.engine))
+                        # otherwise insert a clone of the reference on the extracted rule
+                        assert(plan.common_reference)
+                        replacement_body.add(plan.common_reference.clone())
+                    # add remaining vars to hoisted, making sure there's no duplicates (due to VarOrDefault)
+                    hoisted_vars = helpers.hoisted_vars(hoisted)
+                    if remaining_vars:
+                        hoisted = OrderedSet.from_iterable(filter(lambda v: v not in hoisted_vars, remaining_vars)) | hoisted
+
+                if child in plan.local_dependencies:
+                    for local_dep in plan.local_dependencies[child]:
+                        replacement_body.add(local_dep.clone())
+
+                if isinstance(replacement, ir.Logical):
+                    # if the replacements is a logical, we can just add to the body
+                    body.add(replacement.reconstruct(
+                        replacement.engine,
+                        tuple(hoisted.get_list()),
+                        tuple(replacement_body.update(replacement.body).get_list()),
+                        replacement.annotations
+                    ))
                 else:
-                    # child does not need extras in the body, just add it to the main body
+                    # Otherwise, wrap the local dependencies in a Lookup where the output
+                    # variables are hoisted, and keep the computed replacement.
+                    body.add(f.logical(replacement_body.get_list(), hoisted.get_list(), replacement.engine))
                     body.add(replacement)
+            else:
+                # child does not need extras in the body, just add it to the main body
+                body.add(replacement)
 
-            return ir.Logical(task.engine, task.hoisted, tuple(body))
+        return ir.Logical(task.engine, task.hoisted, tuple(body))
 
     @dataclass
     class ExtractionPlan():
@@ -206,16 +230,23 @@ class ExtractCommon(Pass):
         Compute a plan to extract tasks in this frame that are common dependencies
         across these composite tasks.
         """
-        # compute intersection of task dependencies and inputs
+        # If there are any pragma lookups, then don't extract anything. Pragma lookups are
+        # designed to control execution order, and extracting them may affect their
+        # semantics.
+        for b in binders:
+            if isinstance(b, ir.Lookup) and builtins.is_pragma(b.relation):
+                return None
+
+        # Compute intersection of task dependencies
         sample = composites.some()
         deps = ctx.info.task_dependencies(sample)
         if deps is None:
             return None
         # only get sibling dependencies
         common_body = binders & deps
-        exposed_vars = OrderedSet.from_iterable(ctx.info.task_inputs(sample))
 
-        # now extract from the original sets
+        # For other composites, remove their sibling dependencies so that we end up with
+        # the intersection of dependencies
         for composite in composites:
             if composite is sample:
                 continue
@@ -226,11 +257,23 @@ class ExtractCommon(Pass):
                 for task in common_body:
                     if task not in deps:
                         common_body.remove(task)
+
+        # Compute union of input vars
+        # Start with the output vars of the common body. We only want to expose vars that
+        # are output from the common body
+        body_output_vars = OrderedSet()
+        for child in common_body:
+            body_output_vars.update(ctx.info.task_outputs(child))
+
+        # Compute the union of input vars across all composites, intersected with output
+        # vars of the common body
+        exposed_vars = OrderedSet.from_iterable(ctx.info.task_inputs(sample)) & body_output_vars
+        for composite in composites:
+            if composite is sample:
+                continue
             # compute common input vars
-            t_inputs = ctx.info.task_inputs(composite)
-            for v in exposed_vars:
-                if v not in t_inputs:
-                    exposed_vars.remove(v)
+            t_inputs = OrderedSet.from_iterable(ctx.info.task_inputs(composite))
+            exposed_vars.update(t_inputs & body_output_vars)
 
         # no vars in common, not worth to extract
         if not exposed_vars:
@@ -263,7 +306,7 @@ class ExtractCommon(Pass):
         for binder in binders:
             if binder not in common_body:
                 remaining.add(binder)
-                deps = self._compute_local_dependencies(ctx, binders, binder, common_body, exposed_vars)
+                deps = self._compute_local_dependencies(ctx, binders, binder, exposed_vars)
                 if deps:
                     remaining.update(deps)
 
@@ -271,16 +314,17 @@ class ExtractCommon(Pass):
         # depends on it but it is not exposed by the vars
         local_dependencies: dict[ir.Task, OrderedSet[ir.Task]] = dict()
         for composite in composites:
-            local = self._compute_local_dependencies(ctx, binders, composite, common_body, exposed_vars)
+            local = self._compute_local_dependencies(ctx, binders, composite, exposed_vars)
             if local:
                 local_dependencies[composite] = local
 
         # distribute the common reference only if all of the composites are extractable and there's nothing else remaining
         distribute_common_reference = len(extractables) == len(composites) and not remaining
+
         return ExtractCommon.ExtractionPlan(common_body, remaining, exposed_vars, local_dependencies, distribute_common_reference)
 
 
-    def _compute_local_dependencies(self, ctx: Context, binders: OrderedSet[ir.Task], composite: ir.Task, common_body: OrderedSet[ir.Task], exposed_vars: OrderedSet[ir.Var]):
+    def _compute_local_dependencies(self, ctx: Context, binders: OrderedSet[ir.Task], composite: ir.Task, exposed_vars: OrderedSet[ir.Var]):
         """
         The tasks in common_body will be extracted into a logical that will expose the exposed_vars.
         Compute which additional dependencies are needed specifically for this composite, because
@@ -298,145 +342,12 @@ class ExtractCommon(Pass):
         if not vars_needed:
             return None
 
-        # needs = ctx.info.local_dependencies(composite) & binders | common_body
-        # if needs:
-        #     return needs
-        # return None
-
-        # this is a greedy algorithm that uses the first task in the common body that provides
-        # a variable needed; it may result in sub-optimal extraction, but should be correct
-        local_body = ordered_set()
-        while(vars_needed):
-            v = vars_needed.pop()
-            for x in common_body:
-                if x not in local_body:
-                    # an x that is not yet in local_body can fulfill v
-                    x_outputs = ctx.info.task_outputs(x)
-                    if x_outputs and v in x_outputs:
-                        # add it to local_body and add its outputs to vars exposed
-                        local_body.add(x)
-                        vars_exposed.add(x_outputs)
-                        # but add its inputs the vars now needed
-                        inputs = ctx.info.task_inputs(x)
-                        if inputs:
-                            vars_needed.update(inputs - vars_exposed)
-        return local_body
-
-
-
-
-
-    def _create_extraction_plan_old(self, binders: OrderedSet[ir.Task], composites: OrderedSet[ir.Task], extractables: list[ir.Task], ctx: Context):
-        """
-        Compute a plan to extract tasks in this frame that are common dependencies
-        across these composite tasks.
-        """
-        # compute intersection of task dependencies and inputs
-        sample = composites.some()
-        deps = ctx.info.task_dependencies(sample)
-        if deps is None:
-            return None
-        # only get sibling dependencies
-        common_body = binders & deps
-        exposed_vars = OrderedSet.from_iterable(ctx.info.task_inputs(sample))
-
-        # now extract from the original sets
-        for composite in composites:
-            if composite is sample:
-                continue
-
-            # compute sibling dependencies
-            deps = ctx.info.task_dependencies(composite)
-            if deps:
-                for task in common_body:
-                    if task not in deps:
-                        common_body.remove(task)
-            # compute common input vars
-            t_inputs = ctx.info.task_inputs(composite)
-            for v in exposed_vars:
-                if v not in t_inputs:
-                    exposed_vars.remove(v)
-
-        # no vars in common, not worth to extract
-        if not exposed_vars:
-            return None
-
-        # pull the transitive closure of the intersected tasks in the common body
-        chasing = True
-        while(chasing):
-            chasing = False
-            for task in common_body:
-                deps = ctx.info.task_dependencies(task)
-                if deps:
-                    for hop in binders & deps:
-                        if hop not in common_body:
-                            common_body.add(hop)
-                            chasing = True
-
-        # not useful to extract common tasks if there's a single one
-        if len(common_body) < 2:
-            return None
-
-        # check if some variable used in the common body is needed by some binder that is
-        # not going to be extracted. In that case, we need to expose this variable from the
-        # common body
-        common_vars = ordered_set()
-        for task in common_body:
-            common_vars.update(ctx.info.task_outputs(task))
-        common_vars = common_vars - exposed_vars
-        for v in common_vars:
-            for binder in binders:
-                if binder not in common_body and ctx.info.task_inputs(binder) and v in ctx.info.task_inputs(binder):
-                    exposed_vars.add(v)
-                    break
-
-        # check with of the original binders remain, and make sure their dependencies also stay
-        remaining = ordered_set()
-        for binder in binders:
-            if binder not in common_body:
-                remaining.add(binder)
-                deps = self._compute_local_dependencies_old(ctx, binder, common_body, exposed_vars)
-                if deps:
-                    remaining.update(deps)
-
-        # for each composite, check if there are additional tasks needed, because the task
-        # depends on it but it is not exposed by the vars
-        local_dependencies: dict[ir.Task, OrderedSet[ir.Task]] = dict()
-        for composite in composites:
-            local = self._compute_local_dependencies_old(ctx, composite, common_body, exposed_vars)
-            if local:
-                local_dependencies[composite] = local
-
-        # distribute the common reference only if all of the composites are extractable and there's nothing else remaining
-        distribute_common_reference = len(extractables) == len(composites) and not remaining
-        # TODO: we are forcing this for now, but we should remove the whole analysis
-        distribute_common_reference = False
-        return ExtractCommon.ExtractionPlan(common_body, remaining, exposed_vars, local_dependencies, distribute_common_reference)
-
-
-    def _compute_local_dependencies_old(self, ctx: Context, composite: ir.Task, common_body: OrderedSet[ir.Task], exposed_vars: OrderedSet[ir.Var]):
-        """
-        The tasks in common_body will be extracted into a logical that will expose the exposed_vars.
-        Compute which additional dependencies are needed specifically to this composite, because
-        it depends on some tasks that are extracted to common_body but not exposed by exposed_vars.
-        """
-
-        # vars exposed by exposed vars + tasks added to the local body
-        vars_exposed = OrderedSet.from_iterable(exposed_vars)
-        # working list of vars we still need to fulfill
-        inputs = ctx.info.task_inputs(composite)
-        if not inputs:
-            return None
-        vars_needed = (inputs - vars_exposed)
-        if not vars_needed:
-            return None
-
         # this is a greedy algorithm that uses the first task in the common body that provides
         # a variable needed; it may result in sub-optimal extraction, but should be correct
         local_body = ordered_set()
         while(vars_needed):
             v = vars_needed.pop()
-            for x in common_body:
+            for x in binders:
                 if x not in local_body:
                     # an x that is not yet in local_body can fulfill v
                     x_outputs = ctx.info.task_outputs(x)