complete small fixes to allow all cards to pass through _source_to_dataset

Signed-off-by: dafnapension <dafnashein@yahoo.com>

Author: dafnapension

prepare/cards/bfcl.py

@@ -5,9 +5,10 @@
 from unitxt.operators import (
     Copy,
     ExecuteExpression,
+    FilterByExpression,
     Set,
 )
-from unitxt.stream_operators import JoinStreams
+from unitxt.stream_operators import DeleteSplits, JoinStreams
 from unitxt.test_utils.card import test_card
 
 base_path = "https://raw.githubusercontent.com/ShishirPatil/gorilla/70b6a4a2144597b1f99d1f4d3185d35d7ee532a4/berkeley-function-call-leaderboard/data/"
@@ -31,6 +32,7 @@
                     on="id",
                     new_stream_name="test",
                 ),
+                DeleteSplits(splits=["questions", "answers"]),
                 Copy(field="question/0/0/content", to_field="query"),
                 Copy(field="function", to_field="tools"),
                 "operators.fix_json_schema",
@@ -100,9 +102,13 @@
                     on="id",
                     new_stream_name="test",
                 ),
+                DeleteSplits(splits=["questions", "answers"]),
                 Copy(field="question/*/0", to_field="dialog"),
                 Copy(field="function", to_field="tools"),
                 "operators.fix_json_schema",
+                FilterByExpression(
+                    expression="all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+                ),
                 ExecuteExpression(
                     expression='[{"name": k, "arguments": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != ""] for vval in v.values()])]',
                     to_field="reference_calls",

prepare/operators/fix_json_schema.py

@@ -4,6 +4,7 @@
 operator = RecursiveReplace(
     key="type",
     map_values={
+        "": "object",
         "dict": "object",
         "float": "number",
         "tuple": "array",

src/unitxt/catalog/cards/bfcl/multi_turn/java_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/javascript_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/live_multiple_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/live_parallel_multiple_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/live_parallel_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/live_simple_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/multiple_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",

src/unitxt/catalog/cards/bfcl/multi_turn/parallel_multiple_v3.json

@@ -20,6 +20,13 @@
             "on": "id",
             "new_stream_name": "test"
         },
+        {
+            "__type__": "delete_splits",
+            "splits": [
+                "questions",
+                "answers"
+            ]
+        },
         {
             "__type__": "copy",
             "field": "question/*/0",
@@ -31,6 +38,10 @@
             "to_field": "tools"
         },
         "operators.fix_json_schema",
+        {
+            "__type__": "filter_by_expression",
+            "expression": "all(isinstance(v, dict) for d in ground_truth for k, v in d.items())"
+        },
         {
             "__type__": "execute_expression",
             "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() for vals in itertools.product(*[[w for w in vval if w != \"\"] for vval in v.values()])]",