clarifai 11.2.3rc4__py3-none-any.whl → 11.2.3rc6__py3-none-any.whl

clarifai/__init__.py

@@ -1 +1 @@
-__version__ = "11.2.3rc4"
+__version__ = "11.2.3rc6"

clarifai/runners/models/model_class.py

@@ -1,6 +1,5 @@
 import inspect
 import itertools
-import json
 import logging
 import os
 import traceback
@@ -12,6 +11,7 @@ from clarifai_grpc.grpc.api.status import status_code_pb2, status_pb2
 from google.protobuf import json_format
 
 from clarifai.runners.utils import data_types
+from clarifai.runners.utils.data_utils import DataConverter
 from clarifai.runners.utils.method_signatures import (build_function_signature, deserialize,
                                                       get_stream_from_signature, serialize,
                                                       signatures_to_json)
@@ -71,141 +71,6 @@ class ModelClass(ABC):
     output.data.text.raw = signatures_to_json(signatures)
     return resp
 
-  def _convert_input_data_to_new_format(
-      self, data: resources_pb2.Data,
-      input_fields: List[resources_pb2.ModelTypeField]) -> resources_pb2.Data:
-    """Convert input data to new format."""
-    new_data = resources_pb2.Data()
-    for field in input_fields:
-      part_data = self._convert_field(data, field)
-      part = new_data.parts.add()
-      part.id = field.name
-      part.data.CopyFrom(part_data)
-    return new_data
-
-  def _convert_field(self, old_data: resources_pb2.Data,
-                     field: resources_pb2.ModelTypeField) -> resources_pb2.Data:
-    data_type = field.type
-    if data_type == resources_pb2.ModelTypeField.DataType.STR:
-      new_data = resources_pb2.Data()
-      new_data.string_value = old_data.text.raw
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.IMAGE:
-      new_data = resources_pb2.Data()
-      new_data.image.CopyFrom(old_data.image)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.VIDEO:
-      new_data = resources_pb2.Data()
-      new_data.video.CopyFrom(old_data.video)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.BOOL:
-      new_data = resources_pb2.Data()
-      new_data.bool_value = old_data.bool_value
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.INT:
-      new_data = resources_pb2.Data()
-      new_data.int_value = old_data.int_value
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.FLOAT:
-      new_data = resources_pb2.Data()
-      new_data.float_value = old_data.float_value
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.BYTES:
-      new_data = resources_pb2.Data()
-      new_data.bytes_value = old_data.bytes_value
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.NDARRAY:
-      new_data = resources_pb2.Data()
-      new_data.ndarray.CopyFrom(old_data.ndarray)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.JSON_DATA:
-      new_data = resources_pb2.Data()
-      struct_dict = old_data.text.raw
-      new_data.string_value = json.dumps(struct_dict)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.TEXT:
-      new_data = resources_pb2.Data()
-      new_data.text.CopyFrom(old_data.text)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.CONCEPT:
-      new_data = resources_pb2.Data()
-      new_data.concepts.extend(old_data.concepts)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.REGION:
-      new_data = resources_pb2.Data()
-      new_data.regions.extend(old_data.regions)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.FRAME:
-      new_data = resources_pb2.Data()
-      new_data.frames.extend(old_data.frames)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.AUDIO:
-      new_data = resources_pb2.Data()
-      new_data.audio.CopyFrom(old_data.audio)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.LIST:
-      new_data = resources_pb2.Data()
-      if not field.type_args:
-        raise ValueError("LIST type requires type_args")
-      element_field = field.type_args[0]
-      element_data = self._convert_field(old_data, element_field)
-      part = new_data.parts.add()
-      part.data.CopyFrom(element_data)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.TUPLE:
-      new_data = resources_pb2.Data()
-      for element_field in field.type_args:
-        element_data = self._convert_field(old_data, element_field)
-        part = new_data.parts.add()
-        part.data.CopyFrom(element_data)
-      return new_data
-    elif data_type == resources_pb2.ModelTypeField.DataType.NAMED_FIELDS:
-      new_data = resources_pb2.Data()
-      for named_field in field.type_args:
-        part_data = self._convert_field(old_data, named_field)
-        part = new_data.parts.add()
-        part.id = named_field.name
-        part.data.CopyFrom(part_data)
-      return new_data
-    else:
-      raise ValueError(f"Unsupported data type: {data_type}")
-
-  def is_old_format(self, data: resources_pb2.Data) -> bool:
-    """Check if the Data proto is in the old format (without parts)."""
-    if len(data.parts) > 0:
-      return False  # New format uses parts
-
-    # Check if any singular field is set
-    singular_fields = [
-        'image', 'video', 'metadata', 'geo', 'text', 'audio', 'ndarray', 'int_value',
-        'float_value', 'bytes_value', 'bool_value', 'string_value'
-    ]
-    for field in singular_fields:
-      if data.HasField(field):
-        return True
-
-    # Check if any repeated field has elements
-    repeated_fields = [
-        'concepts', 'colors', 'clusters', 'embeddings', 'regions', 'frames', 'tracks',
-        'time_segments', 'hits', 'heatmaps'
-    ]
-    for field in repeated_fields:
-      if getattr(data, field):
-        return True
-
-    return False
-
-  def _convert_output_data_to_old_format(self, data: resources_pb2.Data) -> resources_pb2.Data:
-    """Convert output data to old format."""
-    old_data = resources_pb2.Data()
-    part_data = data.parts[0].data
-    # Handle text.raw specially (common case for text outputs)
-    old_data = part_data
-    if old_data.string_value:
-      old_data.text.raw = old_data.string_value
-
-    return old_data
-
   def _batch_predict(self, method, inputs: List[Dict[str, Any]]) -> List[Any]:
     """Batch predict method for multiple inputs."""
     outputs = []
@@ -239,10 +104,10 @@ class ModelClass(ABC):
       python_param_types = method_info.python_param_types
       for input in request.inputs:
         # check if input is in old format
-        is_convert = self.is_old_format(input.data)
+        is_convert = DataConverter.is_old_format(input.data)
         if is_convert:
           # convert to new format
-          new_data = self._convert_input_data_to_new_format(input.data, signature.input_fields)
+          new_data = DataConverter.convert_input_data_to_new_format(input.data, signature.input_fields)
           input.data.CopyFrom(new_data)
       # convert inputs to python types
       inputs = self._convert_input_protos_to_python(request.inputs, inference_params,
@@ -285,10 +150,10 @@ class ModelClass(ABC):
       python_param_types = method_info.python_param_types
       for input in request.inputs:
         # check if input is in old format
-        is_convert = self.is_old_format(input.data)
+        is_convert = DataConverter.is_old_format(input.data)
         if is_convert:
           # convert to new format
-          new_data = self._convert_input_data_to_new_format(input.data, signature.input_fields)
+          new_data = DataConverter.convert_input_data_to_new_format(input.data, signature.input_fields)
           input.data.CopyFrom(new_data)
       inputs = self._convert_input_protos_to_python(request.inputs, inference_params,
                                                     signature.input_fields, python_param_types)
@@ -346,10 +211,10 @@ class ModelClass(ABC):
 
       for input in request.inputs:
         # check if input is in old format
-        is_convert = self.is_old_format(input.data)
+        is_convert = DataConverter.is_old_format(input.data)
         if is_convert:
           # convert to new format
-          new_data = self._convert_input_data_to_new_format(input.data, signature.input_fields)
+          new_data = DataConverter.convert_input_data_to_new_format(input.data, signature.input_fields)
           input.data.CopyFrom(new_data)
       # convert all inputs for the first request, including the first stream value
       inputs = self._convert_input_protos_to_python(request.inputs, inference_params,
@@ -423,7 +288,7 @@ class ModelClass(ABC):
     serialize({'return': output}, variables_signature, proto.data, is_output=True)
     if convert_old_format:
       # convert to old format
-      data = self._convert_output_data_to_old_format(proto.data)
+      data = DataConverter.convert_output_data_to_old_format(proto.data)
       proto.data.CopyFrom(data)
     proto.status.code = status_code_pb2.SUCCESS
     if hasattr(self, "_prompt_tokens") and self._prompt_tokens is not None:

clarifai/runners/utils/data_utils.py

@@ -6,6 +6,8 @@ from clarifai_grpc.grpc.api.resources_pb2 import ModelTypeRangeInfo
 from PIL import Image
 
 from clarifai.runners.utils.data_types import MessageData
+from clarifai_grpc.grpc.api import resources_pb2
+from typing import List
 
 
 def image_to_bytes(img: Image.Image, format="JPEG") -> bytes:
@@ -178,3 +180,158 @@ class InputField(MessageData):
         import json
         proto.default = json.dumps(default)
     return proto
+
+
+class DataConverter:
+  """A class that can be used to convert data to and from a specific format."""
+
+  @classmethod
+  def convert_input_data_to_new_format(
+      cls, data: resources_pb2.Data,
+      input_fields: List[resources_pb2.ModelTypeField]) -> resources_pb2.Data:
+    """Convert input data to new format."""
+    new_data = resources_pb2.Data()
+    for field in input_fields:
+      part_data = cls._convert_field(data, field)
+      if cls._is_data_set(part_data):
+        # if the field is set, add it to the new data part
+        part = new_data.parts.add()
+        part.id = field.name
+        part.data.CopyFrom(part_data)
+      else:
+        if field.required:
+          raise ValueError(f"Field {field.name} is required but not set")
+    return new_data
+  
+  @classmethod
+  def _convert_field(cls, old_data: resources_pb2.Data,
+                     field: resources_pb2.ModelTypeField) -> resources_pb2.Data:
+    data_type = field.type
+    new_data = resources_pb2.Data()
+    if data_type == resources_pb2.ModelTypeField.DataType.STR:
+      if old_data.HasField('text'):
+        new_data.string_value = old_data.text.raw
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.IMAGE:
+      if old_data.HasField('image'):
+        new_data.image.CopyFrom(old_data.image)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.VIDEO:
+      if old_data.HasField('video'):
+        new_data.video.CopyFrom(old_data.video)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.BOOL:
+      if old_data.HasField('bool_value'):
+        new_data.bool_value = old_data.bool_value
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.INT:
+      if old_data.HasField('int_value'):
+        new_data.int_value = old_data.int_value
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.FLOAT:
+      if old_data.HasField('float_value'):
+        new_data.float_value = old_data.float_value
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.BYTES:
+      if old_data.HasField('bytes_value'):
+        new_data.bytes_value = old_data.bytes_value
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.NDARRAY:
+      if old_data.HasField('ndarray'):
+        new_data.ndarray.CopyFrom(old_data.ndarray)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.TEXT:
+      if old_data.HasField('text'):
+        new_data.text.CopyFrom(old_data.text)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.AUDIO:
+      if old_data.HasField('audio'):
+        new_data.audio.CopyFrom(old_data.audio)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.CONCEPT:
+      if old_data.concepts:
+        new_data.concepts.extend(old_data.concepts)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.REGION:
+      if old_data.regions:
+        new_data.regions.extend(old_data.regions)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.FRAME:
+      if old_data.frames:
+        new_data.frames.extend(old_data.frames)
+      return new_data
+    elif data_type == resources_pb2.ModelTypeField.DataType.LIST:
+      if not field.type_args:
+        raise ValueError("LIST type requires type_args")
+      element_field = field.type_args[0]
+      if element_field in (resources_pb2.ModelTypeField.DataType.CONCEPT,
+                           resources_pb2.ModelTypeField.DataType.REGION,
+                           resources_pb2.ModelTypeField.DataType.FRAME):
+        # convert to new format
+        new_data = cls._convert_field(old_data, element_field)
+      return new_data
+    else:
+      return new_data
+      # raise ValueError(f"Unsupported data type: {data_type}")
+
+  @classmethod
+  def is_old_format(cls, data: resources_pb2.Data) -> bool:
+    """Check if the Data proto is in the old format (without parts)."""
+    if len(data.parts) > 0:
+      return False  # New format uses parts
+
+    # Check if any singular field is set
+    singular_fields = [
+        'image', 'video', 'metadata', 'geo', 'text', 'audio', 'ndarray', 'int_value',
+        'float_value', 'bytes_value', 'bool_value', 'string_value'
+    ]
+    for field in singular_fields:
+      if data.HasField(field):
+        return True
+
+    # Check if any repeated field has elements
+    repeated_fields = [
+        'concepts', 'colors', 'clusters', 'embeddings', 'regions', 'frames', 'tracks',
+        'time_segments', 'hits', 'heatmaps'
+    ]
+    for field in repeated_fields:
+      if getattr(data, field):
+        return True
+
+    return False
+  
+  @classmethod
+  def convert_output_data_to_old_format(cls, data: resources_pb2.Data) -> resources_pb2.Data:
+    """Convert output data to old format."""
+    old_data = resources_pb2.Data()
+    part_data = data.parts[0].data
+    # Handle text.raw specially (common case for text outputs)
+    old_data = part_data
+    if old_data.string_value:
+      old_data.text.raw = old_data.string_value
+
+    return old_data
+  
+  @classmethod
+  def _is_data_set(cls, data_msg):
+    # Singular message fields
+    singular_fields = ["image", "video", "metadata", "geo", "text", "audio", "ndarray"]
+    for field in singular_fields:
+      if data_msg.HasField(field):
+        return True
+
+    # Repeated fields
+    repeated_fields = [
+        "concepts", "colors", "clusters", "embeddings", "regions", "frames", "tracks",
+        "time_segments", "hits", "heatmaps", "parts"
+    ]
+    for field in repeated_fields:
+      if getattr(data_msg, field):  # checks if the list is not empty
+        return True
+
+    # Scalar fields (proto3 default: 0 for numbers, empty for strings/bytes, False for bool)
+    if (data_msg.int_value != 0 or data_msg.float_value != 0.0 or data_msg.bytes_value != b"" or
+        data_msg.bool_value is True or data_msg.string_value != ""):
+      return True
+
+    return False

{clarifai-11.2.3rc4.dist-info → clarifai-11.2.3rc6.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: clarifai
-Version: 11.2.3rc4
+Version: 11.2.3rc6
 Summary: Clarifai Python SDK
 Home-page: https://github.com/Clarifai/clarifai-python
 Author: Clarifai

{clarifai-11.2.3rc4.dist-info → clarifai-11.2.3rc6.dist-info}/RECORD

@@ -1,4 +1,4 @@
-clarifai/__init__.py,sha256=M11rm1B4Hwgzbp8nZy-JnEJfer9JDpDqQzMqNG3d39I,26
+clarifai/__init__.py,sha256=Dm7rGVjbmVqBK20TBYmaDHvsteQJbwgl47LQuzh9zek,26
 clarifai/cli.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 clarifai/errors.py,sha256=RwzTajwds51wLD0MVlMC5kcpBnzRpreDLlazPSBZxrg,2605
 clarifai/versions.py,sha256=jctnczzfGk_S3EnVqb2FjRKfSREkNmvNEwAAa_VoKiQ,222
@@ -139,7 +139,7 @@ clarifai/runners/dockerfile_template/Dockerfile.template,sha256=5cjv7U8PmWa3DB_5
 clarifai/runners/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 clarifai/runners/models/base_typed_model.py,sha256=0QCWxch8CcyJSKvE1D4PILd2RSnQZHTmx4DXlQQ6dpo,7856
 clarifai/runners/models/model_builder.py,sha256=vJXYrxBwuayanMquAPwFOv9z7U5EQ2aIdppl8DfTGgQ,36900
-clarifai/runners/models/model_class.py,sha256=clowWGQ9Kh3Q25K0StnL4RUPsOW59T0q7HvcLgNAT5w,19801
+clarifai/runners/models/model_class.py,sha256=JVgUAeVeQTAoE09k2uQZXPpMCEcmAyAwaQvKXrxNFpI,14458
 clarifai/runners/models/model_class_refract.py,sha256=HxuozxSW7ag5yWCPxjNwgLArQ6dORhyGXlnpPaZz2-c,3211
 clarifai/runners/models/model_run_locally.py,sha256=m5uLlXpWVGuQIJv-XBt-sHGx1gD8w6d77f9b85DOZqs,18170
 clarifai/runners/models/model_runner.py,sha256=T4Qn_x0vky7XdeS54bvipzEmKZMra1tQdAu_u01yyjc,6503
@@ -161,7 +161,7 @@ clarifai/runners/utils/const.py,sha256=bwj-Pcw558-pasdIFbNhnkn-9oiCdojYH1fNTTUG2
 clarifai/runners/utils/data_handler.py,sha256=b7k6MWYPXSgjrfw6wsDf82xFYa0D7UjYmjE4mw5HzHM,8499
 clarifai/runners/utils/data_handler_refract.py,sha256=3M-V4hkOoF-9Ix4hE6ocXWiTJPc9dewtu6FMtddd-jQ,6343
 clarifai/runners/utils/data_types.py,sha256=bFRKVl2s6Mmx0bQn_lKwbqxBET9wtIEf0Tk6PAIc6ko,12465
-clarifai/runners/utils/data_utils.py,sha256=RoPsi5szPvYLxU5DSO0ZmF-zA33QH7mJpQhZ7It4Y4Q,5289
+clarifai/runners/utils/data_utils.py,sha256=OE0yyx9TF0tTIdva-sQDmTxoARLjOdgE6SFXQ-rnODM,11375
 clarifai/runners/utils/loader.py,sha256=SgNHMwRmCCymFQm8aDp73NmIUHhM-N60CBlTKbPzmVc,7470
 clarifai/runners/utils/logger.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 clarifai/runners/utils/method_signatures.py,sha256=mr0d0hs-og2icwZn7DmiZqzOMnEqY-ZBW0mheselsZE,18134
@@ -230,9 +230,9 @@ clarifai/workflows/__pycache__/__init__.cpython-39.pyc,sha256=9nA--jULSW7OFrYOcs
 clarifai/workflows/__pycache__/export.cpython-310.pyc,sha256=phEGwi2gAojCUhRTqjZVeTDn7Gk6LCVBeSTjAj4m9iY,2418
 clarifai/workflows/__pycache__/utils.cpython-310.pyc,sha256=M9_KTM7GOOS5SPrWwAzqHDqyGvgKi3xuSGvyw6MNf-I,1925
 clarifai/workflows/__pycache__/validate.cpython-310.pyc,sha256=c18Jgp_-CAm8RD_tmUpDCPoqZeexaoWELG0yBzb9rjw,2149
-clarifai-11.2.3rc4.dist-info/LICENSE,sha256=mUqF_d12-qE2n41g7C5_sq-BMLOcj6CNN-jevr15YHU,555
-clarifai-11.2.3rc4.dist-info/METADATA,sha256=dcuOr9_2JIzWPcooBh_qUCMUGByy7TqQuBaPi4DINw0,22453
-clarifai-11.2.3rc4.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-clarifai-11.2.3rc4.dist-info/entry_points.txt,sha256=X9FZ4Z-i_r2Ud1RpZ9sNIFYuu_-9fogzCMCRUD9hyX0,51
-clarifai-11.2.3rc4.dist-info/top_level.txt,sha256=wUMdCQGjkxaynZ6nZ9FAnvBUCgp5RJUVFSy2j-KYo0s,9
-clarifai-11.2.3rc4.dist-info/RECORD,,
+clarifai-11.2.3rc6.dist-info/LICENSE,sha256=mUqF_d12-qE2n41g7C5_sq-BMLOcj6CNN-jevr15YHU,555
+clarifai-11.2.3rc6.dist-info/METADATA,sha256=992jW3H0kdqAE8KffQoOFM7DalzjHEnf7MLixhJubns,22453
+clarifai-11.2.3rc6.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+clarifai-11.2.3rc6.dist-info/entry_points.txt,sha256=X9FZ4Z-i_r2Ud1RpZ9sNIFYuu_-9fogzCMCRUD9hyX0,51
+clarifai-11.2.3rc6.dist-info/top_level.txt,sha256=wUMdCQGjkxaynZ6nZ9FAnvBUCgp5RJUVFSy2j-KYo0s,9
+clarifai-11.2.3rc6.dist-info/RECORD,,