last touches in bfcl, reviewing whole datasets

dafnapension · dafnapension · commit ca55da99286c · 2025-07-22T20:30:18.000+03:00
Signed-off-by: dafnapension &lt;dafnashein@yahoo.com&gt;
diff --git a/docs/docs/adding_operator.rst b/docs/docs/adding_operator.rst
@@ -40,11 +40,11 @@ There are several types of operators.
 1. Field Operators - Operators that modify individual fields of the instances in the input streams.  Example of such operators are operators that
 cast field values, uppercase string fields, or translate text between languages.
 
-1. Instance Operators - Operators that modify individual instances in the input streams. For example, operators that add or remove fields.
+2. Instance Operators - Operators that modify individual instances in the input streams. For example, operators that add or remove fields.
 
-2. Stream Operators - Operators that perform operations on full streams. For example, operators that remove instances based on some condition.
+3. Stream Operators - Operators that perform operations on full streams. For example, operators that remove instances based on some condition.
 
-3. MultiStream Operators - Operator that perform operations on multiple streams.  For example, operators that repartition the instances between train and test splits.
+4. MultiStream Operators - Operator that perform operations on multiple streams.  For example, operators that repartition the instances between train and test splits.
 
 Unitxt comes with a large collection of built in operators - that were design to cover most common requirements of dataset processing.
 
diff --git a/prepare/cards/bfcl.py b/prepare/cards/bfcl.py
@@ -75,12 +75,12 @@
     for subset in [
         "simple",
         "multiple",
-        "live_multiple",
+        "live_multiple",  # instances above 900 reach size of hundreds of MBs
         "live_simple",
         "java",
         "javascript",
         "parallel",
-        "parallel_multiple",
+        "parallel_multiple",  # error caused by instance 179, hence expression now constrains: if isinstance(v, dict)
         "live_parallel",
         "live_parallel_multiple",
     ]:
@@ -105,7 +105,7 @@
                 Copy(field="function", to_field="tools"),
                 FixJsonSchemaOfToolParameterTypes(),
                 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(*v.values())]',
+                    expression='[{"name": k, "arguments": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]',
                     to_field="reference_calls",
                     imports_list=["itertools"],
                 ),
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/java_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/java_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/javascript_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/javascript_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/live_multiple_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/live_multiple_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/live_parallel_multiple_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/live_parallel_multiple_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/live_parallel_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/live_parallel_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/live_simple_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/live_simple_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/multiple_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/multiple_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/parallel_multiple_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/parallel_multiple_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/parallel_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/parallel_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/catalog/cards/bfcl/multi_turn/simple_v3.json b/src/unitxt/catalog/cards/bfcl/multi_turn/simple_v3.json
@@ -35,7 +35,7 @@
         },
         {
             "__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(*v.values())]",
+            "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",
             "to_field": "reference_calls",
             "imports_list": [
                 "itertools"
diff --git a/src/unitxt/operators.py b/src/unitxt/operators.py
@@ -2629,7 +2629,6 @@ def prepare(self):
             "any": "object",
             "Any": "object",
             "Array": "array",
-            "array": "array",
             "ArrayList": "array",
             "Bigint": "integer",
             "bool": "boolean",
@@ -2643,11 +2642,9 @@ def prepare(self):
             "HashMap": "object",
             "Hashtable": "object",
             "int": "integer",
-            "integer": "integer",
             "list": "array",
             "List": "array",
             "long": "integer",
-            "number": "number",
             "Queue": "array",
             "short": "integer",
             "Stack": "array",
@@ -2656,7 +2653,6 @@ def prepare(self):
             "set": "array",
             "str": "string",
             "String": "string",
-            "string": "string",
         }
 
     def dict_type_of(self, type_str: str) -> dict:
@@ -2695,6 +2691,8 @@ def type_str_to_jsonschema_dict(self, type_str: str) -> dict:
             for i in range(len(args)):
                 args[i] = args[i].strip()
             return {"anyOf": [self.type_str_to_jsonschema_dict(arg) for arg in args]}
+        if re.match(r"^(Callable)\[(.*?)\]$", type_str):
+            return self.dict_type_of("object")
         if "," in type_str:
             sub_types = type_str.split(",")
             for i in range(len(sub_types)):
@@ -2712,7 +2710,7 @@ def type_str_to_jsonschema_dict(self, type_str: str) -> dict:
                     return {"anyOf": [basic_type, self.dict_type_of("null")]}
             return basic_type
 
-        return self.dict_type_of("object")  # otherwise - fall back to a safe zone
+        return self.dict_type_of(type_str)  # otherwise - return what arrived
 
     def process(
         self, instance: Dict[str, Any], stream_name: Optional[str] = None

Original file line number	Diff line number	Diff line change
`@@ -35,7 +35,7 @@`
`35`	`35`	`},`
`36`	`36`	`{`
`37`	`37`	`"__type__": "execute_expression",`
`38`		`- "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(*v.values())]",`
	`38`	`+ "expression": "[{\"name\": k, \"arguments\": dict(zip(v.keys(), vals))} for d in ground_truth for k, v in d.items() if isinstance(v, dict) for vals in itertools.product(*v.values())]",`
`39`	`39`	`"to_field": "reference_calls",`
`40`	`40`	`"imports_list": [`
`41`	`41`	`"itertools"`