Con oracle

stanza/models/constituency/dynamic_oracle.py

@@ -16,6 +16,48 @@ def advance_past_constituents(gold_sequence, cur_index):
         cur_index = cur_index + 1
     return None
 
+def find_previous_open(gold_sequence, cur_index):
+    """
+    Go backwards from cur_index to find the open which opens the previous block of stuff.
+
+    Return None if it can't be found.
+    """
+    count = 0
+    cur_index = cur_index - 1
+    while cur_index >= 0:
+        if isinstance(gold_sequence[cur_index], OpenConstituent):
+            count = count + 1
+            if count > 0:
+                return cur_index
+        elif isinstance(gold_sequence[cur_index], CloseConstituent):
+            count = count - 1
+        cur_index = cur_index - 1
+    return None
+
+def find_in_order_constituent_end(gold_sequence, cur_index):
+    """
+    Advance cur_index through gold_sequence until the next block has ended
+
+    This is different from advance_past_constituents in that it will
+    also return when there is a Shift when count == 0.  That way, we
+    return the first block of things we know attach to the left
+    """
+    count = 0
+    saw_shift = False
+    while cur_index < len(gold_sequence):
+        if isinstance(gold_sequence[cur_index], OpenConstituent):
+            count = count + 1
+        elif isinstance(gold_sequence[cur_index], CloseConstituent):
+            count = count - 1
+            if count == -1: return cur_index
+        elif isinstance(gold_sequence[cur_index], Shift):
+            if saw_shift and count == 0:
+                return cur_index
+            else:
+                saw_shift = True
+        cur_index = cur_index + 1
+    return None
+
 class DynamicOracle():
     def __init__(self, root_labels, oracle_level, repair_types, additional_levels, deactivated_levels):
         self.root_labels = root_labels
@@ -46,11 +88,18 @@ def fix_error(self, gold_transition, pred_transition, gold_sequence, gold_index)
         for repair_type in self.repair_types:
             if repair_type.fn is None:
                 continue
-            if self.oracle_level is not None and repair_type.value > self.oracle_level and repair_type not in self.additional_levels:
+            if self.oracle_level is not None and repair_type.value > self.oracle_level and repair_type not in self.additional_levels and not repair_type.debug:
                 continue
             if repair_type in self.deactivated_levels:
                 continue
             repair = repair_type.fn(gold_transition, pred_transition, gold_sequence, gold_index, self.root_labels)
+            if repair is None:
+                continue
+
+            if isinstance(repair, tuple) and len(repair) == 2:
+                return repair
+
+            # TODO: could update all of the returns to be tuples of length 2
             if repair is not None:
                 return repair_type, repair
 

stanza/models/constituency/ensemble.py

@@ -76,7 +76,7 @@ def __init__(self, args, filenames=None, models=None, foundation_cache=None):
             if self.models[0].transition_scheme() != model.transition_scheme():
                 raise ValueError("Models {} and {} are incompatible.  {} vs {}".format(filenames[0], filenames[model_idx], self.models[0].transition_scheme(), model.transition_scheme()))
             if self.models[0].transitions != model.transitions:
-                raise ValueError("Models %s and %s are incompatible: different transitions" % (filenames[0], filenames[model_idx]))
+                raise ValueError(f"Models {filenames[0]} and {filenames[model_idx]} are incompatible: different transitions\n{filenames[0]}:\n{self.models[0].transitions}\n{filenames[model_idx]}:\n{model.transitions}")
             if self.models[0].constituents != model.constituents:
                 raise ValueError("Models %s and %s are incompatible: different constituents" % (filenames[0], filenames[model_idx]))
             if self.models[0].root_labels != model.root_labels: